Abstract

   Upon chemotherapy, excessive reactive oxygen species (ROS) often lead
   to the production of massive lipid peroxides in cancer cells and induce
   cell death, namely ferroptosis. The elimination of ROS is pivotal for
   tumor cells to escape from ferroptosis and acquire drug resistance.
   Nevertheless, the precise functions of long non-coding RNAs (lncRNAs)
   in ROS metabolism and tumor drug-resistance remain elusive. In this
   study, we identify LncRNA-HMG as a chemoresistance-related lncRNA in
   colorectal cancer (CRC) by high-throughput screening. Abnormally high
   expression of LncRNA-HMG predicts poorer prognosis in CRC patients.
   Concurrently, we found that LncRNA-HMG protects CRC cells from
   ferroptosis upon chemotherapy, thus enhancing drug resistance of CRC
   cells. LncRNA-HMG binds to p53 and facilitates MDM2-mediated
   degradation of p53. Decreased p53 induces upregulation of SLC7A11 and
   VKORC1L1, which contribute to increase the supply of reducing agents
   and eliminate excessive ROS. Consequently, CRC cells escape from
   ferroptosis and acquire chemoresistance. Importantly, inhibition of
   LncRNA-HMG by anti-sense oligo (ASO) dramatically sensitizes CRC cells
   to chemotherapy in patient-derived xenograft (PDX) model. LncRNA-HMG is
   also a transcriptional target of β-catenin/TCF and activated Wnt
   signals trigger the marked upregulation of LncRNA-HMG. Collectively,
   these findings demonstrate that LncRNA-HMG promotes CRC chemoresistance
   and might be a prognostic or therapeutic target for CRC.

   Keywords: LncRNA-HMG, Colorectal cancer, Chemoresistance, Ferroptosis,
   p53

Graphical abstract

   [39]Image 1
   [40]Open in a new tab

1. Introduction

   Colorectal cancer (CRC) is one of the most popular malignancies all
   over the world. In the past decades, the application of anticancer
   drugs (5-fluorouracil or oxaliplatin as the first-line drugs)
   monotherapy or combined regimens has markedly prolonged the survival
   time of advanced CRC patients [[41]1,[42]2]. However, frequent
   occurrence of chemoresistance causes important limitations to
   pharmacological therapy [[43]3]. This urges us to uncover the key
   molecules and elucidate potential mechanisms responsible for
   chemoresistance in order to develop new therapeutic strategies.

   Usually, tumor cells are shown to carry more reactive oxygen species
   (ROS) than their normal counterparts [[44]4]. However, excessive ROS
   lead to the production of massive lipid peroxides, which give rise to
   lipid membrane dysregulation as well as cause cell death, defined as
   ferroptosis [[45]5]. Upon chemotherapy or radiotherapy, a large number
   of ROS are produced to help kill cancer cells [[46]6,[47]7]. In
   chemoresistant cancer cells, it's important to eliminate ROS or lipid
   peroxides for protecting cells from ferroptosis and acquiring drug
   resistance [[48]8]. The study of the cooperative mechanisms underlying
   ROS balance and cancer chemoresistance will bring us a better cognition
   about tumor treatment.

   Metabolic alteration is critical for tumor cells to balance the
   production and elimination of ROS [[49]9]. The reducing agents are
   regularly employed to detoxify ROS or lipid peroxides in order to keep
   cell survival, such as glutathione (GSH), and reduced nicotinamide
   adenine dinucleotide phosphate oxidase (NADPH) [[50]10,[51]11]. GSH
   assists GPX4 to transform lipid peroxides into non-toxic lipid alcohol,
   thereby suppressing ferroptosis [[52]12]. The synthesis of cellular GSH
   depends on cystine transporter system xc- (mainly consisting of
   SLC7A11), which is essential for GPX4-mediated ferroptosis prevention
   [[53]13]. Tumor suppressor, p53, transcriptionally inhibits expression
   of SLC7A11 to initiate ferroptosis [[54]14]. Recently, it's reported
   that p53 also regulates vitamin K metabolism and induces ferroptosis
   [[55]15]. VKORC1L1 is the critical enzyme for producing the reduced
   form of vitamin K to decrease lipid peroxides and protect cells from
   ferroptosis [[56]16]. Activation of p53 reduces the expression of
   VKORC1L1, thus inducing cancer cell ferroptosis for tumor suppression.
   These studies indicated that p53-mediated metabolic alteration is
   connected with tumor suppression. For chemoresistant cells, evasion
   from p53-related ferroptosis is essential for gaining resistance to
   chemotherapy, which generates redundant ROS and lipid peroxides.
   However, the underlying mechanism remains elusive.

   Long non-coding RNAs (lncRNAs) are defined as the non-coding RNA
   transcripts with a length of more than 200 nucleotides. Recently, a
   large number of studies have reported that lncRNAs are tightly
   associated with tumorigenesis and cancer progression such as cancer
   cell growth, metastasis, and drug resistance [[57][17], [58][18],
   [59][19]]. Targeting lncRNAs displays the potential to develop novel
   strategies for cancer treatment such as inhibiting lncRNAs by small
   interfering RNAs (siRNAs), antisense oligonucleotides (ASOs), or the
   CRISPR/Cas9 system [[60]20,[61]21]. It's reported that some lncRNAs are
   involved in regulating tumor metabolism. However, the roles of lncRNAs
   in modulating p53-mediated metabolic adaption to balance ROS level upon
   CRC chemotherapy are still unclear.

   In our study, we found that LncRNA-HMG aggravates CRC chemoresistance.
   Mechanistically, LncRNA-HMG, induced by Wnt/β-catenin signaling,
   disrupts p53-mediated transcriptional repression to eliminate ROS and
   lipid peroxides, thereby suppressing ferroptosis. These findings
   uncover the pivotal roles of LncRNA-HMG in CRC chemoresistance and
   imply its potential value as a treatment target in CRC treatment.

2. Results

2.1. LncRNA-HMG is identified as a novel ferroptosis-related lncRNA

   To dissect the ferroptosis-related lncRNA in CRC, we download TCGA data
   and performed the analysis as shown in [62]Supplementary Fig. S1A.
   Finally, eleven lncRNAs were screened as the ferroptosis-related
   lncRNAs ([63]Fig. 1A). Next, the expression of these 11 lncRNAs was
   evaluated in patients with CRC in another cohort (Peking University
   Cancer Hospital, PKU Cancer). As shown in [64]Fig. 1B, three lncRNAs
   were increased in CRC tissues compared with the adjacent non-cancerous
   tissues, including [65]AC005841.1, [66]AC104819.3 and [67]AC107308.1
   (we referred it as LncRNA-HMG since it located near to HMG gene on
   chromosome). We also established the drug resistant CRC cells
   ([68]Supplementary Fig. S1B) and analyzed the expression of these
   lncRNAs. Two lncRNAs were upregulated in 5-Fu (fluorouracil) resistant
   cells ([69]AP003555.1 and LncRNA-HMG) and four lncRNAs were
   overexpressed in oxaliplatin resistant cells (AC245100.5,
   [70]AC156455.1, [71]AP003555.1 and LncRNA-HMG) ([72]Fig. 1C). Since
   LncRNA-HMG was upregulated in all these analyses, we explored the
   functions of LncRNA-HMG in CRC chemoresistance and ferroptosis.

Fig. 1.

   [73]Fig. 1
   [74]Open in a new tab

   LncRNA-HMG is associated with the prognosis of CRC patients. A The
   expression level of 11 lncRNAs in TCGA colon cancer patients
   (normal = 39, tumor = 39) was shown. B The expression levels of 11
   lncRNAs in CRC tissues and adjacent tissues from Peking University
   Cancer hospital (PKU Cancer, normal = 12, tumor = 12) were detected. C
   The expression levels of 11 lncRNAs were detected by RT-qPCR in 5-Fu
   resistant and oxaliplatin-resistant cells. D The Cox regression
   analysis was employed for multivariate analyses of various prognostic
   parameters in CRC patients (PKU Cancer cohort). E Overall survival
   analysis based on LncRNA-HMG level was analyzed using data from PKU
   Cancer cohort. F Overall survival analysis based on LncRNA-HMG level
   was analyzed using data from TCGA colon cancer cohort. G The 5′ and 3′
   rapid amplification of cDNA ends (RACE) assays in LoVo cells were
   performed for the comprehensive detection of the entire sequence of
   LncRNA-HMG. On the left, a gel electrophoresis image illustrated the
   PCR products obtained from the 5′-RACE and 3′-RACE assays. On the
   right, sequencing of these PCR products revealed the precise boundary
   between the universal anchor primer and the sequences specific to
   LncRNA-HMG. H The expression levels of LncRNA-HMG were detected by
   RT-qPCR in CRC cell lines (n = 3). I Subcellular localization of
   LncRNA-HMG was analyzed by FISH. J RT-qPCR detection of LncRNA-HMG
   expression in the cytoplasmic and nuclear fractions (n = 3). K The
   coding probability of GSCAR was predicted by CPAT
   ([75]http://lilab.research.bcm.edu/). All the data are presented as
   mean ± S.D. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

2.2. LncRNA-HMG expression level predicts the poor prognosis of CRC

   Next, we evaluated the relationship between LncRNA-HMG level and
   prognosis of CRC patients. As shown in [76]Fig. 1D, univariate analysis
   and multivariate Cox regression analyses revealed that besides
   chemotherapeutic effect, the level of LncRNA-HMG was also screened as
   an independent prognostic indicator for CRC patients ([77]Fig. 1D and
   [78]Tables S1–3). High LncRNA-HMG level was associated with a shorter
   overall survival in CRC patients in PKU cancer cohort ([79]Fig. 1E).
   Moreover, the survival analysis using the TCGA dataset confirmed that
   high LncRNA-HMG expression indicated a worse overall survival in
   patients with CRC ([80]Fig. 1F). Together, our data suggested that
   LncRNA-HMG is increased in CRC tissues and predicts a poor prognosis
   for CRC patients.

   Through rapid amplification of the 5′ and 3’ cDNA ends (RACE) assays,
   we successfully obtained the full length of LncRNA-HMG ([81]Fig. 1G and
   [82]Supplementary Fig. S1C). LncRNA-HMG was markedly upregulated in CRC
   cell lines compared to colon epithelial cell, NCM460 ([83]Fig. 1H). RNA
   fluorescence in situ hybridization (FISH) assays were employed to
   determine the subcellular localization of LncRNA-HMG in CRC cells. The
   data revealed that LncRNA-HMG was localized in the cytoplasm, with some
   localization in the nucleus ([84]Fig. 1I). The results were further
   confirmed by cytoplasmic and nuclear fraction analysis ([85]Fig. 1J).
   In addition, we confirmed that LncRNA-HMG could not be translated into
   coding-proteins ([86]Fig. 1K).

2.3. LncRNA-HMG protects CRC cells from ferroptosis

   To investigate the biological role of LncRNA-HMG in ferroptosis, we
   inhibited the expression of LncRNA-HMG by specific shRNAs in HCT116 and
   LoVo cells, and cells expressing scrambled shRNA were used as controls
   ([87]Fig. 2A). Erastin was frequently employed to induce ferroptosis
   [[88]22]. Inhibition of LncRNA-HMG sensitized cells to erastin-induced
   death ([89]Fig. 2B). Colony formation analysis further confirmed that
   downregulation of LncRNA-HMG sensitized cells to erastin ([90]Fig. 2C).
   To verify the results, we generated LncRNA-HMG-overexpressing cells
   ([91]Fig. 2D). Consistently, LncRNA-HMG overexpression attenuates
   erastin-induced cell death ([92]Fig. 2E and F).

Fig. 2.

   [93]Fig. 2
   [94]Open in a new tab

   The LncRNA-HMG protects CRC cells from ferroptosis. A The stable
   LncRNA-HMG knockdown cell line was established and LncRNA-HMG level was
   assessed by RT-qPCR (n = 3). B The proliferation capacity of cells
   stably expressing shHMG treated with different concentrations of
   Erastin (5 μM) for 8 h was examined using the cell counting kit (CCK-8)
   kit (n = 4). C The indicated cells were treated or untreated with
   Erastin (5 μM) and cell colony formation experiments were performed.
   The results were summarized using data derived from three independent
   experiments (n = 3). D The cells stably expressing LncRNA-HMG were
   constructed and LncRNA-HMG level was evaluated by RT-qPCR (n = 3). E
   The proliferation capacity of the indicated cells treated with
   different concentrations of Erastin was assessed (n = 4). F The colony
   formation experiment was performed using the indicated cells and
   results derived from three independent experiments were presented in a
   violin plot (n = 3). G The relative level of ROS was detected in the
   indicated cells (n = 3). H The lipid peroxidation level was analyzed in
   the indicated cells. I GSH/GSSG ratio was detected in the indicated
   cells after Erastin (5 μM) treatment (n = 3). J C11-BODIPY level was
   analyzed in the indicated cells (n = 3). K-M The ROS level (K), lipid
   peroxidation level (L), and GSH/GSSG ratio (M) were analyzed in
   LncRNA-HMG-overexpressed CRC cells as indicated (n = 3). N The
   indicated cells were subjected to TEM. Representative TEM images of
   mitochondrial morphology were presented. Red arrows indicate the
   mitochondria. All the data are presented as mean ± SD. ∗P < 0.05,
   ∗∗P < 0.01, ∗∗∗P < 0.001.

   Ferroptosis is a lipid peroxidation-driven cell death cascade in an
   iron-dependent way [[95]23]. Thus, we evaluated the lipid peroxidation
   and ROS level after LncRNA-HMG depletion. Knockdown of LncRNA-HMG
   upregulated ROS and lipid peroxidation of CRC cells ([96]Fig. 2G and
   H). Notably, the intracellular GSH/GSSG ratio was decreased in
   LncRNA-HMG-depleted cells ([97]Fig. 2I). To verify the results, the
   lipid ROS level was also analyzed by C11-BODIPY and flow cytometry
   analysis. As shown in [98]Fig. 2J, LncRNA-HMG depletion upregulated ROS
   level of CRC cells. Furthermore, in LncRNA-HMG-overexpressing cells,
   ROS and lipid peroxidation was significantly downregulated while GSH
   was upregulated ([99]Fig. 2K–M). Transmission electron microscopy (TEM)
   analysis suggested that mitochondria in LncRNA-HMG-depleted cells
   showed obvious characteristic of ferroptosis, displaying mitochondrial
   atrophy, reduction of mitochondrial size, and smaller cristae
   ([100]Fig. 2N). Together, these data suggested that LncRNA-HMG
   suppresses ferroptosis of CRC cells.

2.4. LncRNA-HMG promotes chemoresistance of CRC cells

   Next, we explored the roles of LncRNA-HMG in CRC progression. As shown
   in [101]Fig. 3A and B, LncRNA-HMG depletion decreased cell
   proliferation and colony formation. Consistently, LncRNA-HMG
   overexpression enhanced cell proliferation and colony formation
   ([102]Supplementary Figs. S2A and S2B). The response of
   LncRNA-HMG-depleted cells to 5-Fu or oxaliplatin was further evaluated.
   Inhibition of LncRNA-HMG sensitized the CRC cells to 5-Fu or
   oxaliplatin treatment ([103]Fig. 3C and [104]Supplementary Fig. S2C).
   In drug-resistant cells, knockdown of LncRNA-HMG also improved
   chemosensitivity of these cells ([105]Supplementary Fig. S2D).
   Overexpression of LncRNA-HMG enhanced chemoresistance of CRC cells
   ([106]Fig. 3D and [107]Supplementary Fig. S2E). Colony formation
   analysis confirmed these results ([108]Fig. 3E and F). Together, these
   data indicated that LncRNA-HMG enhances chemoresistance of CRC cells.

Fig. 3.

   [109]Fig. 3
   [110]Open in a new tab

   LncRNA-HMG promotes the drug resistance of CRC Cells. A The
   proliferation capacity of cells stably expressing LncRNA-HMG was
   assessed by CCK-8 experiment (n = 4). B The cell colony formation
   experiment was conducted with the indicated cells (n = 3). C The IC[50]
   values of the indicated cells were detected after treatment of the
   indicated drugs (n = 3). D The IC[50] values of the cells
   overexpressing LncRNA-HMG were detected after treatment of the
   indicated drugs (n = 3). E-F The colony formation experiment was
   conducted with the indicated cells treated or untreated with 5-Fu
   (5 μM) and oxaliplatin (40 μM) for 24 h. The results were summarized
   with the data derived from three independent experiments (n = 3). G ROS
   level was analyzed by C11-BODIPY assay in LncRNA-HMG-depleted cells
   treated or untreated with 5-Fu (5 μM) and oxaliplatin (40 μM) for 24 h
   (n = 3). H GSH/GSSG ratio was detected in LncRNA-HMG overexpressed CRC
   cells (n = 3). I The indicated cells were harvested and GSH/GSSG ratio
   was analyzed (n = 3). J The indicated cells were treated with or
   without 5-Fu (5 μM) and oxaliplatin (40 μM) for 24 h. Then, TEM was
   employed to analyze the mitochondrial morphology. Red arrows indicate
   the mitochondria. All the data are presented as mean ± SD. ∗P < 0.05,
   ∗∗P < 0.01, ∗∗∗P < 0.001.

   Upon chemotherapy, cancer cells are often exposed to high level of ROS,
   which causes lipid peroxidation and ferroptosis. Since LncRNA-HMG plays
   important role in eliminating ROS in cancer cells, we hypothesize that
   LncRNA-HMG enhances chemoresistance by decreasing ROS level after drug
   treatment. However, the lipid ROS level was grammatically elevated in
   LncRNA-HMG-depleted cells upon drug treatment ([111]Fig. 3G and
   [112]Supplementary Fig. S2F). Accordingly, LncRNA-HMG changed the
   intracellular GSH/GSSG ratio upon chemotherapy ([113]Fig. 3H and I).
   Meanwhile, TEM results confirmed that LncRNA-HMG alleviated
   drug-induced dysfunction of mitochondria in CRC cells ([114]Fig. 3J).
   Ferrostatin-1, a ferroptosis inhibitor, reversed cell death caused by
   LncRNA-HMG shRNA upon chemotherapy ([115]Supplementary Fig. S2G). Thus,
   these data revealed that LncRNA-HMG increases the production of
   reducing substance and impairs drug-induced ferroptosis, thereby
   decreasing chemosensitivity of CRC cells.

2.5. LncRNA-HMG weakens p53-mediated ferroptosis

   To explore the related molecular mechanisms underlying
   LncRNA-HMG-mediated chemoresistance of CRC cells, we analyzed the
   potential LncRNA-HMG-involved pathways. According to the LncRNA-HMG
   level, CRC samples were divided into two groups, including the
   LncRNA-HMG-low group and the LncRNA-HMG-high group. Next, we identified
   the differentially expressed genes (DEGs) between the two groups
   ([116]Fig. 4A). The gene enrichment analysis demonstrated that these
   genes were significantly enriched in gene sets related to p53 signaling
   ([117]Fig. 4B and [118]Supplementary Fig. S3A). Recently, p53 signaling
   is reported to be involved in regulating ferroptosis [[119]24]. Next,
   we purified LncRNA-HMG-binding proteins and mass spectrometry analysis
   was conducted to identify these proteins. Totally, 104 proteins were
   identified as LncRNA-HMG-binding proteins ([120]Table S4).
   Protein-protein interaction analysis by STRING screened the top 25 hub
   genes among the 104 proteins ([121]Fig. 4C), indicating that p53 is one
   of the hub factors. Together, it's possible that LncRNA-HMG binds to
   p53 and regulates p53 signaling pathway.

Fig. 4.

   [122]Fig. 4
   [123]Open in a new tab

   LncRNA-HMG regulates the expression levels of SLC7A11 and VKORC1L1. A
   The differentially expressed genes were identified between
   LncRNA-HMG-high samples and LncRNA-HMG-low samples based on TCGA data.
   B Pathway enrichment analysis was performed with GSEA method and the
   result suggested that LncRNA-HMG was negatively correlated with p53
   signaling. C LncRNA-HMG binding proteins were identified through mass
   spectrometry analysis and protein-protein interaction (PPI) analysis
   was employed to uncover the top 25 hub genes. D The control or shHMG
   cells were lysed and the indicated proteins were detected by Western
   blot. E LoVo cells were transfected with the indicated vectors and
   western bot was performed using the indicated antibodies. F-G The total
   RNA was extracted from the indicated cells and RT-qPCR was performed to
   detect the indicated genes (n = 3). H–I CRC cells were treated with
   5-Fu (5 μM) and oxaliplatin (40 μM) for 24 h. Then, Western blot was
   performed using the indicated antibodies. J Relative mRNA expression
   levels of SLC7A11 and VKORC1L1 were detected by RT-qPCR in the
   indicated CRC cells treated with or without 5-Fu (5 μM) and oxaliplatin
   (40 μM) for 24 h (n = 3). K LncRNA-HMG was depleted in 5-Fu and
   oxaliplatin-resistant cells and RNA sequencing analysis was performed
   to analyze the indicated mRNA levels. L The indicated cells were
   treated with CHX (50 μg/mL) and the p53 level was analyzed by Western
   blot (n = 3). All the data are presented as mean ± SD. ∗P < 0.05,
   ∗∗P < 0.01, ∗∗∗P < 0.001.

   As a critical transcriptional factor, p53 affects the intracellular ROS
   level by inhibiting the expression of ferroptosis suppressors, such as
   SLC7A11 and VKORC1L1 [[124]16]. Thus, we determined the p53 protein
   levels and its downstream targets after LncRNA-HMG ablation. As shown
   in [125]Fig. 4D and [126]Supplementary Fig. S3B, LncRNA-HMG depletion
   increased p53 level while decreased SLC7A11 and VKORC1L1 levels.
   Overexpression of LncRNA-HMG had an opposite effect on the level of
   these proteins ([127]Fig. 4E). Furthermore, the mRNA levels of SLC7A11
   and VKORC1L1 were decreased in LncRNA-HMG-depleted cells ([128]Fig. 4F)
   while were upregulated in LncRNA-HMG-overexpressed cells ([129]Fig.
   4G). These data indicated that LncRNA-HMG regulates p53 and its
   downstream targets.

   Consistent with previous reports [[130]25], p53 was upregulated in
   response to anti-tumor drug treatment ([131]Supplementary Fig. S3C).
   Therefore, it's important to elucidate whether LncRNA-HMG affects p53
   activation after drug treatment. As shown in [132]Fig. 4H, LncRNA-HMG
   depletion promoted the elevation of p53 in response to 5-Fu and
   oxaliplatin treatment. However, overexpression of LncRNA-HMG declined
   p53 activation ([133]Fig. 4I). Consistently, knockdown of LncRNA-HMG
   downregulated the expression of SLC7A11 and VKORC1L1 while
   overexpression of LncRNA-HMG enhanced the expression of these two
   targets ([134]Fig. 4H, I and 4J). Additionally, we also observed the
   effect of LncRNA-HMG on other p53 targets, such as p21. As shown in
   [135]Supplementary Fig. S3D, the mRNA level of p21 was upregulated in
   LncRNA-HMG depleted cells. Moreover, LncRNA-HMG was depleted in
   chemoresistant cells and the expression of these three genes was
   examined by RNA-seqencing (RNA-seq). As shown in [136]Fig. 4K, similar
   phenomenon was observed in chemoresistant CRC cells. To explore whether
   SLC7A11 and VKORC1L1 are required for LncRNA-HMG-repressed ferroptosis,
   we depleted the expression of SLC7A11 and VKORC1L1 in cells
   ([137]Supplementary Fig. S3E). In these cells, knockdown of LncRNA-HMG
   had little effect on lipid peroxidation, indicating the critical roles
   of SLC7A11 and VKORC1L1 in LncRNA-HMG-repressed ferroptosis
   ([138]Supplementary Fig. S3F). Intriguingly, LncRNA-HMG still regulated
   cell proliferation in SLC7A11/VKORC1L1-depleted cells
   ([139]Supplementary Fig. S3G). Since LncRNA-HMG regulated the
   expression of p21, one of the most regulators in cell proliferation
   [[140]26], it's possible that LncRNA-HMG-related cell proliferation
   mainly depends on inhibiting p21 expression. Thus, these data suggested
   that LncRNA-HMG regulates p53 level, further changing the expression of
   p53 downstream targets. The protein stability of p53 plays important
   roles in control p53 level [[141]27]. The cycloheximide (CHX, a
   ribosome-stalling drug) was employed to block protein translation and
   the stability of p53 was further analyzed. As shown in [142]Fig. 4L,
   p53 stability was compromised in LncRNA-HMG-overexpressed cells while
   inhibition of LncRNA-HMG enhanced the stability of p53. Together, our
   data revealed that LncRNA-HMG declines the stability of p53 and weakens
   p53-medicated ROS ferroptosis.

2.6. LncRNA-HMG assists MDM2 to increase ubiquitination of p53

   We next investigated how LncRNA-HMG regulates p53 stability. The
   interaction between p53 and LncRNA-HMG was confirmed by RIP assay
   ([143]Fig. 5A). The ubiquitin-proteasome system (UPS) is critical for
   modulating p53 stability [[144]28]. Thus, the proteasome inhibitor,
   MG132, was utilized to analyze if LncRNA-HMG-related p53 stabilization
   relies on UPS. p53 elevation by LncRNA-HMG depletion was reversed by
   the addition of MG132, suggesting that LncRNA-HMG modulates p53 protein
   level in a proteasome-dependent manner ([145]Fig. 5B). The
   ubiquitination level of p53 was decreased in LncRNA-HMG-depleted cells
   ([146]Fig. 5C). Overexpression of LncRNA-HMG promoted the
   ubiquitination of p53 ([147]Fig. 5D). These results suggested that
   LncRNA-HMG promotes ubiquitination of p53, leading to degradation of
   p53. It's important to analyze how LncRNA-HMG promotes ubiquitination
   of p53. MDM2 acts as the most important ubiquitin ligase for p53
   [[148]29]. It's possible that LncRNA-HMG enhances the activity of MDM2
   to p53. To test this hypothesis, we firstly evaluated the interaction
   of LncRNA-HMG and MDM2. As shown in [149]Fig. 5E, LncRNA-HMG binds to
   p53 and MDM2 simultaneously. The interaction of LncRNA-HMG and MDM2 was
   further verified by RIP assay ([150]Fig. 5F). RNA pull-down assay
   further verified that LncRNA-HMG binds to p53 or MDM2 ([151]Fig. 5G).
   Importantly, LncRNA-HMG had on effect on MDM2 level ([152]Fig. 5H).
   Furthermore, we found that depletion of LncRNA-HMG reduced the
   interaction between p53 and MDM2 ([153]Fig. 5I). LncRNA-HMG
   overexpression increased MDM2-mediated p53 ubiquitination ([154]Fig.
   5J). Moreover, Nutlin-3, the antagonist of MDM2, abrogated the effect
   of LncRNA-HMG on p53 protein level ([155]Supplementary Fig. S4A). It's
   known that the phosphorylation or acetylation of p53 affects p53-MDM2
   interaction [[156]25]. We found that LncRNA-HMG had little effect on
   the phosphorylation or acetylation of p53, indicating the direct role
   of LncRNA-HMG on p53-MDM2 interaction ([157]Supplementary Fig. S4B).
   Thus, these data suggested that LncRNA-HMG promotes the interaction of
   MDM2 with p53 and enhances ubiquitination of p53.

Fig. 5.

   [158]Fig. 5
   [159]Open in a new tab

   LncRNA-HMG promotes p53 ubiquitination. A RIP analysis was performed
   using anti-p53 antibody and the immunoprecipitates were subjected to
   RT-qPCR for analyzing LncRNA-HMG level. B The indicated cells were
   treated with MG132 (10 μM) for 4 h, and the cell lysates were subjected
   to Western blot using the indicated antibodies. C-D Ubiquitination
   levels of p53 were analyzed in the indicated cells. E The GFP-MS2 and
   LncRNA-HMG-MS2 were transfected into HCT116 cells and
   LncRNA-HMG-binding proteins were subjected to Western blot using
   anti-p53 or anti-MDM2 antibodies. F RIP assay was performed using
   anti-MDM2 antibody and RT-qPCR was conducted to detect LncRNA-HMG
   level. G RNA-pull down assay was performed to analyze the direct
   binding of LncRNA-HMG to MDM2 or p53. H The control or
   LncRNA-HMG-depleted cells were harvested and Western blot was performed
   using the indicated antibodies. I The interaction between MDM2 and p53
   was detected by immunoprecipitation in LncRNA-HMG depleted cells. J The
   cells were transfected with indicated vectors. Then, the ubiquitination
   level of p53 was evaluated using anti-p53 antibody. K LncRNA-HMG was
   depleted in HCT116 p53^+/+ or p53^−/− cells and the knockdown
   efficiency was verified by RT-qPCR (n = 3). L The p53 and SLC7A11
   levels were evaluated by Western blot in the indicated cells. M The
   mRNA level of SLC7A11 was detected by RT-qPCR in the indicated cells
   (n = 3). N GSH/GSSG ratio in the indicated cells was analyzed (n = 3).
   O The stable cells overexpressing LncRNA-HMG were established in HCT116
   p53^+/+ or p53^−/− cells. RT-qPCR was conducted to analyze the
   LncRNA-HMG level (n = 3). P The p53 and SLC7A11 levels were evaluated
   by Western blot in the indicated cells. Q The mRNA level of SLC7A11 was
   detected by RT-qPCR in the indicated cells (n = 3). R GSH/GSSG ratio in
   the indicated cells was analyzed (n = 3). All the data are presented as
   mean ± SD. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

   It's unknown whether LncRNA-HMG regulates ROS state relying on p53. To
   address this issue, we employed HCT116 p53 null cells and generated
   LncRNA-HMG depleted cells ([160]Supplementary Fig. S4C and Fig. 5K).
   LncRNA-HMG had little effect on the expression level of SLC7A11
   ([161]Fig. 5L and M) and the intracellular GSH/GSSG ratio ([162]Fig.
   5N). We also overexpressed LncRNA-HMG in p53 null cells ([163]Fig. 5O).
   The expression level of SLC7A11 showed no changes even if LncRNA-HMG
   was overexpressed in p53 null cells ([164]Fig. 5P and Q). The similar
   phenomena were observed when we detected GSH/GSSG ratio ([165]Fig. 5R).
   These findings indicated that LncRNA-HMG modulates ROS state depending
   on p53.

2.7. Inhibition of LncRNA-HMG sensitizes CRC cells to chemotherapy in vivo

   Next, we analyzed the role of LncRNA-HMG in CRC chemoresistance in
   vivo. LncRNA-HMG-depleted or control cells were implanted into nude
   mice. LncRNA-HMG attenuated tumor growth in mice xenografts ([166]Fig.
   6A and B and [167]Supplementary S5A). The expression of Ki67 and
   SLC7A11 was also decreased in LncRNA-HMG-depleted cells ([168]Fig. 6C).
   We next determined if LncRNA-HMG modulates the chemosensitivity of CRC
   cells in vivo. Tumor with LncRNA-HMG depletion was more sensitive to
   5-Fu or oxaliplatin when compared with control group, revealing that
   LncRNA-HMG suppressed the chemosensitivity of CRC cells ([169]Fig. 6A
   and B and [170]Supplementary Fig. S5A). The expression of SLC7A11 and
   VKORC1L1 was detected in xenograft tissues. As shown in [171]Fig. 6C
   and [172]Supplementary Fig. S5B, p53 was elevated while SLC7A11 and
   VKORC1L1 were decreased when LncRNA-HMG was abrogated.

Fig. 6.

   [173]Fig. 6
   [174]Open in a new tab

   Inhibition of LncRNA-HMG promotes chemosensitivity of CRC cells in
   vivo. A Tumor formation assay in nude mice injected with
   LncRNA-HMG-depleted cells or control LoVo cells was performed. The
   right group was treated with 5-Fu (20 mg/kg) and Oxaliplatin (6 mg/kg)
   as described in the methods part. B The xenograft tumor weight was
   summarized. C The Ki67 or SLC7A11 level xenograft tissues were
   evaluated by IHC. D The experiments were designed to test the effect of
   the combination of in vivo-optimized LncRNA-HMG ASO inhibitor and
   chemotherapy in CRC PDX models. Graphic illustration was shown. E The
   image of xenograft tumor in PDX model was presented after treated as
   indicated. F The statistical analysis of tumor growth and weight was
   performed in PDX model. G The Ki67 or SLC7A11 level in PDX xenograft
   tissues were evaluated by IHC. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

   Next, we established the patient-derived xenograft (PDX) models using
   tumor tissues from CRC patients carrying wild-type p53 ([175]Fig. 6D).
   The in vivo-optimized LncRNA-HMG inhibitor (LncRNA-HMG ASO) was used to
   treat cancer combined with chemotherapy ([176]Supplementary Fig. S5C).
   LncRNA-HMG ASO downregulated the expression of LncRNA-HMG in PDX model
   ([177]Supplementary Fig. S5D). As shown in [178]Fig. 6E and F,
   LncRNA-HMG ASO significantly reduced tumor growth and increased the
   chemosensitivity in PDX model. Accordingly, as shown in the IHC and
   Western blot analyses, LncRNA-HMG knockdown affected cell proliferation
   (indicated by Ki67) and downregulated the expression of SLC7A11 and
   VKORC1L1 ([179]Fig. 6G and [180]Supplementary S5E). Together, these
   data suggested that LncRNA-HMG inhibitor promotes chemosensitity of
   CRC.

2.8. LncRNA-HMG is a transcriptional target of β-catenin

   Because LncRNA-HMG is highly expressed in CRC, we tried to uncover the
   mechanism of LncRNA-HMG regulation. The potential transcriptional
   factors bound to the promoter region of LncRNA-HMG were predicted
   ([181]Fig. 7A and [182]Supplementary Fig. S6A). Among these factors,
   TCF4/β-catenin is of interest due to the importance of Wnt/β-catenin
   signaling in CRC progression [[183]30]. β-catenin overexpression
   increased the expression level of LncRNA-HMG ([184]Fig. 7B), suggested
   that Wnt/β-catenin regulated LncRNA-HMG. To confirm the results, we
   used β-catenin activator (LiCl) to treat cells. As shown in [185]Fig.
   7C, LiCl treatment upregulated LncRNA-HMG level. Moreover, iCRT14, a
   TCF4/β-catenin inhibitor [[186]31], significantly downregulated the
   expression LncRNA-HMG ([187]Fig. 7D). Thus, these data demonstrated
   that Wnt/β-catenin upregulates the expression of LncRNA-HMG. As
   expected, the LncRNA-HMG promoter contained TCF-binding elements
   ([188]Fig. 7E), which was further verified by ChIP assay ([189]Fig.
   7F). Consistently, the transcriptional activity of luciferase reporters
   containing the region of LncRNA-HMG promoter was markedly upregulated
   by LiCl treatment or β-catenin overexpression while iCRT14 inhibited
   this effect ([190]Fig. 7G). Together, these data suggested that
   LncRNA-HMG is a transcriptional target of β-catenin.

Fig. 7.

   [191]Fig. 7
   [192]Open in a new tab

   LncRNA-HMG is a transcriptional target of Wnt/β-catenin signaling. A
   The prediction of the transcription factor binding to the LncRNA-HMG
   promoter using the PROMO database. B Cells were transfected as
   indicated and total RNA was extracted. RT-qPCR was performed using the
   indicated primers (n = 3). C Cells were treated with LiCl (10 mmol/L)
   for 24 h. Then, LncRNA-HMG was analyzed by RT-qPCR (n = 3). D Cells
   were treated with iCRT14 (50 μM) for 24 h. Then, LncRNA-HMG was
   analyzed by RT-qPCR (n = 3). E The diagram showing TCF binding elements
   (TBEs) at the promoter of LncRNA-HMG. F ChIP experiment was performed
   using anti-β-catenin or anti-TCF4 antibodies. The precipitates were
   subjected to RT-qPCR using the indicated primers. G Luciferase reporter
   assay was performed as indicated. H The working model explaining how
   LncRNA-HMG regulates CRC chemoresistance through inhibiting
   ferroptosis. All the data are presented as mean ± SD. ∗P < 0.05,
   ∗∗P < 0.01, ∗∗∗P < 0.001.

3. Discussion

   Chemoresistance is one of the major issues hindering the improvement of
   the prognosis of CRC patients. In cancer cells, various metabolic
   pathways are often altered to support tumor survival [[193]32,[194]33].
   In this study, we uncovered a novel lncRNA involved in drug resistance
   by remodeling ROS elimination ([195]Fig. 7H). LncRNA-HMG increases the
   supply of reducing agents by suppressing p53-mediated transcriptional
   regulation of SLC7A11 and VKORC1L1. This is helpful for cancer cells to
   escape from drug-induced ferroptosis and acquire chemoresistance.
   Moreover, LncRNA-HMG is transactivated by Wnt/β-catenin signaling.
   Notably, inhibition of LncRNA-HMG using specific ASO significantly
   sensitizes CRC cells to chemotherapy. Our data suggested the potential
   of LncRNA-HMG in treating CRC. However, it's unknown whether LncRNA-HMG
   is associated with cancer progression in other types of human cancers.
   It is important to further explore the expression of LncRNA-HMG in
   pan-cancer analysis and analyze the roles of LncRNA-HMG in other cancer
   treatment.

   Tumor suppressor p53 is crucial to determinate cell fate upon cellular
   stress [[196]34]. As a transcriptional factor, p53 modulates the
   transcription of diverse downstream genes in response to different
   stimuli, functioning in cell cycle arrest, DNA repair, apoptosis, and
   ferroptosis [[197]26]. Recently, it's reported that p53 regulates ROS
   metabolism to promotes lipid peroxides and ferroptosis
   [[198]35,[199]36]. Dysfunction of p53 is tightly associated with cancer
   progression, and it's important to elucidate how p53 is dysregulated.
   MDM2 is a ubiquitin ligase which acts as a critical inhibitor of p53
   [[200]37]. Small molecules, such as RG7388, have been developed to
   restore p53 signaling and treat human cancers [[201]38]. Here, we
   identified LncRNA-HMG as a linker of MDM2 and p53 in CRC cells.
   LncRNA-HMG largely promotes MDM2-mediated p53 degradation, leading to
   anti-ferroptosis of cancer cells upon chemotherapy. Our findings
   indicate the complexity of p53 regulatory mechanism. Besides small
   molecules, the inhibitors targeting lncRNAs might be alternative
   strategy to restore p53 signaling in cancer treatment.

   Besides p53, LncRNA-HMG also binds to other proteins, such as ENO1,
   HSPA5, and RUVBL1. ENO1 regulates ferroptosis in cancer cell via
   modulating AKT/STAT3 signaling [[202]39]. HSPA5 inhibits ferroptosis by
   maintaining GPX4 stability [[203]40]. RUVBL1 is also involved in CRC
   progression [[204]41]. Even if we uncovered the roles of LncRNA-HMG in
   CRC chemoresistance, the knowledge about the functions of LncRNA-HMG in
   cancer is still little. It's possible that LncRNA-HMG also binds to the
   proteins, like ENO1 and HSPA5, to regulate CRC progression. This needs
   to be investigated in the future.

   Wnt/β-catenin signaling pathway plays important roles in CRC
   progression [[205]42]. Upon Wnt signals, accumulated β-catenin
   translocates into the nucleus and helps TCF transcription factors
   activate the transcription of downstream genes such as CD44, MYC and
   LGR5, further initiating tumorigenesis [[206]30,[207]43]. We found that
   LncRNA-HMG is transactivated by Wnt/β-catenin signaling, further
   indicating the vital roles of LncRNA-HMG in CRC progression.
   Importantly, LncRNA-HMG acts as a signal between Wnt pathway and p53
   pathway. We thus provide LncRNA-HMG as a key regulator in CRC
   progression. ASO based on LncRNA-HMG shows the ability to alleviate CRC
   chemoresistance. In the future study, it will be more meaningful to try
   to develop new drug targeting LncRNA-HMG used in clinical trials.

   In summary, we show for the first time that LncRNA-HMG is associated in
   CRC chemoresistance. Wnt signals upregulate the expression LncRNA-HMG.
   Subsequently, LncRNA-HMG attenuates p53-mediated ferroptosis, thus
   facilitating the elimination of ROS or lipid peroxides. Finally,
   LncRNA-HMG protects CRC cells from ferroptosis. Therefore, LncRNA-HMG
   might be a potential target for CRC treatment.

4. Materials and methods

4.1. Patients and specimens

   The patients diagnosed with CRC and underwent surgical treatment in
   Peking University Cancer Hospital (PKU Cancer) from 2013 to 2016 were
   enrolled in this study. Exclusions were made for patients lacking
   clinical tissue sample, or with incomplete clinicopathological
   information. The follow-up evaluation for all patients was made until
   January 1, 2023. [208]Supplementary Table S1 presents the summary of
   clinical details pertaining to the samples. Written consent was
   obtained from the participating patients, and the clinical and
   histopathological data provided to researchers were anonymized. The
   investigation was approved by the Ethics Committee of Peking University
   Cancer Hospital & Institute.

4.2. Cell culture and transfection

   CRC cell lines including LoVo, HCT116 p53^+/+, HCT116 p53^−/−, and
   SW480 were cultured in DMEM medium supplemented with 10 % fetal bovine
   serum. The plasmid DNA or siRNA was transfected into cells using
   Lipofectamine 2000 (Invitrogen, USA) following the manufacturer's
   instructions. In transient transfection experiments, plasmid DNA
   remained constant for each group with the empty vector. For the
   establishment of stable LncRNA-HMG knockdown cell lines, HEK293T cells
   were transfected with pLKO.1-HMG shRNA or pLKO.1-Ctrl shRNA alongside
   packaging vectors pMD2.G and psPAX2 to generate lentiviral particles.
   After harvesting the indicated lentiviruses, LoVo or HCT116 cells were
   infected with these lentiviruses containing LncRNA-HMG shRNA or control
   shRNA. Then, puromycin was added for screening the positive stale
   clones and RT-qPCR was performed for identifying the knockdown
   efficiency.

4.3. Western blotting and antibodies

   Western blotting was conducted following previously established
   protocols [[209]25]. Primary antibodies targeting anti-SLC7A11/xCT
   (A2413) and anti-MDM2 (A13327) were purchased from ABclonal Technology
   (Wuhan, China). Additionally, antibodies including Anti-β-Catenin
   (9562), Anti-p53 (DO-1), anti-TCF4, and anti-Ub (P4D1) were acquired
   from Cell Signaling Technology (MA, USA). The antibody anti-VKORC1L1
   (abs101967) was purchased from Absin (Shanghai, China).

4.4. Quantitative real‐time PCR (RT‐qPCR) assay

   The total RNAs were isolated from the indicated cells using TRIzol
   reagent (Invitrogen, USA), followed by purification using the RNeasy
   Mini Kit (Qiagen, Germany) as per the manufacturer's instructions. cDNA
   was synthesized from 2 μg of total RNA using the Superscript
   First-Strand Synthesis System (Yeasen Biotechnology, China). The qPCR
   was conducted in accordance with the Minimal Information for
   Publication of Quantitative Real-Time PCR Experiments (MIQE)
   guidelines. The specific sequences of the qPCR primers were shown in
   [210]Supplementary Table S5.

4.5. Rapid amplification of cDNA ends (RACE)

   RACE was performed using the HiScript-TS 5/3 RACE Kit (Vazyme, China)
   according to the instruction manual. Briefly, the total RNA was
   isolated and reverse transcription was conducted using an oligo(dT)
   primer. The poly(A) tail was then added to the cDNA with Terminal
   deoxynucleotidyl transferase (TdT). The first round of PCR was
   performed with an anchor primer and an outer primer. Next, the nested
   PCR was conducted using a nested gene-specific primer (nested GSP) and
   an inner primer. The PCR products were analyzed on an agarose gel.
   Moreover, the products were cloned for sequencing to obtain the
   complete RNA sequence. The specific sequences of the primers were shown
   in [211]Supplementary Table S5.

4.6. Immunohistochemistry (IHC) assay

   The tissue sections were de-paraffinized in xylene followed by
   rehydration through a series of graded ethanol. To neutralize the
   endogenous peroxidase activity, the 0.3 % peroxidase solution was
   applied to the sections for 30 min. For antigen retrieval, the sections
   were subjected to boil in citrate buffer at 100 °C for 15 min.
   Subsequently, after treatment with the blocking buffer, the sections
   were incubated with the indicated primary antibody overnight at 4 °C.
   The visualization of the sections was accomplished using the
   UltraVision Quanto Detection System HRP DAB Kit (ZSGB-Bio, China)
   following the manufacturer's protocols. Hematoxylin was used for
   counterstaining, and evaluation was carried out by pathologists who
   were blinded to the clinical outcome.

4.7. Reduced Glutathione (GSH) assay and Oxidized Glutathione (GSSG) content
assay

   The relative concentration of GSH and GSSG in cell lysates was
   determined by utilizing Reduced Glutathione(GSH) Content Assay Kit
   (Solarbio, China) and Oxidized Glutathione(GSSG) Content Assay Kit
   (Solarbio, China) following the manufacturer's guidelines. By measuring
   the absorbance of the resulting at 412 nm using a microplate reader,
   the GSH and GSSG levels were calculated.

4.8. C11-BODIPY assay

   The cells were seeded into 6-well plates, and the culture medium was
   replaced with serum-free media containing 2 μmol/L C11-BODIPY (Thermo
   Fisher Scientific, USA) for staining. Subsequently, cells were cultured
   in the dark for 30 min to assess lipid ROS level. After incubation,
   cells were detached, washed, and suspended in 400 μl of PBS for flow
   cytometry analysis using a BD Biosciences instrument. The fluorescence
   was measured at an excitation wavelength of 488 nm and an emission
   wavelength of 525 nm. The fluorescence intensity in each group
   represented the quantity of lipid ROS.

4.9. Malondialdehyde (MDA) assay

   The levels of MDA were assessed utilizing Malondialdehyde (MDA) Content
   Assay Kit (Elabscience, China) in accordance with the provided
   instructions. About 3 × 10^6 cells were harvested and the extraction
   buffer was added to lyse the cells. The lysates were collected. Then,
   the working solution was added to the lysates and the positive control
   group. The mixtures were heated at 100 °C for 40 min. After cooling
   down to room temperature, the reaction mixtures were centrifuged and
   the supernatant was transferred to an enzyme-linked immunosorbent assay
   (ELISA) plate. The OD value was measured at 532 nm. The measured MDA
   content was then normalized to the protein concentration.

4.10. Reactive oxygen species (ROS) Fluorometric assay

   The levels of ROS were assessed by Reactive Oxygen Species (ROS)
   Fluorometric Assay Kit (Elabscience, China) in accordance with the
   provided instructions. The culture medium was removed and the cells
   were washed. The positive control wells (containing TBHP) and negative
   control wells (without DCFH-DA) were set up. The dye including DCFH-DA
   was added to the cells and incubated with the cells at 37 °C in the
   dark for 1 h. Next, the reaction was stopped and the cells were washed
   three times with wash buffer. The cells were digested and resuspended
   in wash buffer. Finally, the fluorescence values were measured using
   fluorescence spectrophotometer.

4.11. Immunoprecipitation assay

   The indicated cells were harvested and cell lysates were obtained using
   Buffer A containing specific components (25 mM Tris-Cl pH 7.5, 150 mM
   KCl, 1 mM DTT, 2 mM EDTA, 0.5 mM PMSF, and 0.2 % Nonidet P-40). These
   lysates were directly utilized for immunoprecipitation. Briefly,
   antibodies were linked with a 50 % suspension of protein A-Sepharose
   beads (GE Healthcare, USA) in IPP500 solution (500 mM NaCl, 10 mM
   Tris-Cl pH 8.0, 0.2 % Nonidet P-40). The bead-antibody complexes were
   then mixed and incubated with the cell lysates for 2 h at 4 °C. After
   washing, the isolated precipitates were subjected to analysis via
   Western blot using the specific antibodies.

4.12. Luciferase assay

   LoVo and HCT116 cells were plated at a density of 80,000 cells per
   well. Next, cells were co-transfected with either lncRNA-HMG promoter
   luciferase reporter or the pGL3-basic reporter control plasmid,
   alongside a Renilla luciferase control reporter vector (Promega, USA)
   used to standardize transfection efficiency. Then, luciferase assays
   were conducted following the manufacturer's guidelines (Dual-Luciferase
   Reporter Assay System, Promega). The data were presented as relative
   luciferase activity normalized against the expression of the Renilla
   luciferase.

4.13. RNA binding protein immunoprecipitation assay (RIP)

   The indicated cells were treated with UV light irradiation and
   subsequently lysed using a high salt lysis buffer (25 mM Tris-Cl pH
   7.5, 150 mM KCl, 2 mM EDTA, 0.5 % NP-40, 1 mM NaF, 1 mM DTT, and 100
   U/mL RNase inhibitors, along with EDTA-free protease inhibitor).
   Magnetic beads were coated with 5 mg of anti-p53 or MDM2 antibodies.
   Then, cell lysates were added and incubated overnight at 4 °C. After
   reaction, the RNA-protein complexes were washed for six times before
   digestion with proteinase K. RNA extraction was carried out using
   Trizol-chloroform RNA extraction methods. The relative RNA expression
   was assessed via RT-qPCR and normalized against the input. The specific
   sequences of the primers were shown in [212]Supplementary Table S5.

4.14. RNA pull-down assay

   Linearized DNA fragments was prepared through single enzyme digestion
   of in vitro transcription templates. The indicated DNA was purified and
   used as a template for subsequent in vitro transcription.
   Biotin-labeled RNA (Beyotime, China) is then generated and purified,
   following overnight incubation. The labeled RNA was prepared and
   incubated with magnetic beads coated with homogenous streptavidin.
   After binding and washing, the RNA-protein complexes were eluted and
   subjected to SDS-PAGE analysis.

4.15. Fluorescence in situ hybridization

   The FISH experiment was conducted using the Ribo Fluorescent In Situ
   Hybridization Kit (RiboBio, China) according to the manufacturer's
   instructions. Initially, cells were plated on slides and fixed with 4 %
   paraformaldehyde (PFA) for 15 min at room temperature. Subsequently,
   the cells underwent permeabilization using 0.5 % Triton X-100.
   Cy3-labeled probes were then introduced to detect the specific RNAs as
   indicated, followed by DAPI staining to visualize the nuclei.

4.16. Cell growth assay

   Cell growth was assessed following the manufacturer's protocol using
   Cell Counting Kit (CCK-8) (Yeasen Biotechnology Co., Ltd., China). LoVo
   and HCT116 cells, transfected with specific shRNAs, were seeded at a
   density of 1000 cells per well in 96-well plates and cultured in media
   containing 10 % serum. CCK-8 reagent was introduced to the cells, and
   the absorbance at 450 nm was measured per day.

4.17. Colony formation assay

   The cells were split and counted. Then, 500 cells were seeded into each
   well of 6-well plates and cultured in media containing 10 % serum for
   12 days. Finally, the cells were fixed and stained using crystal violet
   to count the visible colonies.

4.18. Chromatin immunoprecipitation (ChIP) assay

   The Magnetic ChIP Kit from Cell Signaling Technology (Cell Signaling
   Technology, USA) was used to perform ChIP assay. First, the cells were
   treated with formaldehyde, facilitating the cross-linking of proteins
   with genomic DNA. Glycine (0.125 M) was used to halt this process.
   Then, the cells were collected, centrifuged, and suspended in the lysis
   buffer (consisting of 1 % SDS, 10 mM EDTA, protease inhibitors, and
   50 mM Tris-HCl pH 8.0). The genomic DNA was sonicated to achieve
   fragments approximately 200–1000 bp in length. The cell lysates were
   clarified by centrifugation, followed by a 1:10 dilution in ChIP
   dilution buffer (0.01 % SDS, 1.0 % Triton X-100, 1.2 mM EDTA, 16.7 mM
   NaCl, protease inhibitors, and 16.7 mM Tris-HCl pH 8.0). Then, the
   lysates were incubated overnight at 4 °C with rotation along with
   anti-TCF4 or β-catenin antibodies. ChIP-Grade Protein A/G Magnetic
   Beads were added to each reaction and incubated for 2 h. Then, the
   Protein A/G magnetic beads were pelleted by centrifugation and washed
   sequentially with low salt wash buffer, high salt wash buffer, and LiCl
   wash buffer. Finally, the beads were washed twice with 1 ml of TE
   buffer. The reversal of protein-DNA cross-linking was achieved by
   incubating the beads with 200 mM NaCl and proteinase K for 2 h at
   65 °C. DNA was extracted and subjected to RT-qPCR. The input DNA and
   IgG were utilized for normalization purposes. The specific sequences of
   the primers were shown in [213]Supplementary Table S5.

4.19. Identification of lncRNA-HMG-binding protein and mass spectrometry
analysis

   The plasmids containing GFP-tagged-MS2 binding sites and lncRNA-HMG-MS2
   were constructed for the experiment. Cells were transfected with the
   indicated vectors, and cell lysates were prepared in RIPA buffer with
   protease inhibitors. The lysates were incubated overnight with a GFP
   antibody, and an IgG antibody was used as a control. Protein complexes
   were captured using protein A/G magnetic beads, and washed. The
   precipitates were resolved on SDS-PAGE. The specific protein bands were
   excised and subjected to mass spectrometry analysis. Mass spectrometry
   was used to identify proteins differentially associated with
   lncRNA-HMG, providing insights into potential interacting partners.

4.20. Animal study

   Five-week-old female nude mice were purchased from Beijing Vital River
   Laboratory Animal Technology Co., Ltd. and the mice were randomly
   allocated into six groups (n = 6 per group), including
   LncRNA-HMG-control shRNA, LncRNA-HMG shRNA1, LncRNA-HMG shRNA2,
   LncRNA-HMG-control shRNA+5-Fu + Oxaliplatin, LncRNA-HMG
   shRNA1+5-Fu + Oxaliplatin, and LncRNA-HMG shRNA2+5-Fu + Oxaliplatin.
   The indicated cells were subcutaneously injected into the right flank
   of each mouse. Then, the mice were treated with an i.p. injection of
   oxaliplatin (6 mg/kg) and 5-Fu (20 mg/kg) every 3 days. Tumor volume
   was detected every five days, and calculated using the formula
   (length × width^2 × 0.5). Thirty-five days later, the mice were
   sacrificed and xenografts were obtained. The protein level of SLC7A11,
   VKORC1L1 and p53 was analyzed by Western blot. Animal procedures were
   approved by the Institutional Animal Care and Use Committee of Peking
   University Health Science Center (Ethics Approval License, LA2023498).

   PDX models were established using fresh tumor samples from CRC
   patients. Tumor tissues were initially implanted subcutaneously into
   mice as the first generation (F0). Upon reaching suitable size, tumors
   were excised, divided, and subcutaneously implanted into mice as the
   second generation (F2). After the tumor was implanted subcutaneously,
   tumor-bearing mice were randomly divided into four groups, including
   LncRNA ASO (10 nmol per injection, RiboBio, China), control ASO, LncRNA
   ASO combined with oxaliplatin and 5-Fu, and control ASO combined with
   oxaliplatin and 5-Fu. ASO was administered via intratumoral injection,
   while 5-Fu (20 mg/kg) and oxaliplatin (6 mg/kg) were administered via
   intraperitoneal injection. The treatment was conducted every 3 days.
   Tumor diameter and width in PDX mice were measured every 7 days to
   calculate tumor volumes (V = 0.5 × D × W^2; V, volume; D, diameter; W,
   width). Thirty-five days later, mice were sacrificed. Then, tumor
   tissues were excised, photographed, weighted, and embedded in paraffin
   for further pathological analysis.

4.21. Statistical analyses

   To assess the significance of differences, we employed one-way ANOVA or
   a two-tailed Student's t-test. Spearman's correlation coefficient was
   used to calculate the relationships between the groups. Survival
   analysis was utilized by Kaplan-Meier analysis, and differences in
   survival probabilities were determined via the log-rank test. All
   statistical analyses were conducted using GraphPad Prism or SPSS
   version 25.0 (SPSS, Inc.). A significance level of P < 0.05 was applied
   for all analyses. Significance levels were denoted as ∗P < 0.05,
   ∗∗P < 0.01, ∗∗∗P < 0.001, and ∗∗∗∗P < 0.0001.

CRediT authorship contribution statement

   Zechang Xin: Writing – original draft, Visualization, Data curation,
   Conceptualization. Chenyu Hu: Methodology, Formal analysis. Chunfeng
   Zhang: Validation, Data curation. Ming Liu: Resources, Formal analysis.
   Juan Li: Resources, Methodology. Xiaoyan Sun: Resources, Formal
   analysis, Data curation. Yang Hu: Visualization, Software, Data
   curation. Xiaofeng Liu: Writing – review & editing, Writing – original
   draft, Supervision, Investigation, Funding acquisition,
   Conceptualization. Kun Wang: Writing – review & editing, Investigation,
   Funding acquisition, Conceptualization.

Declaration of competing interest

   The authors have declared no conflict of interest.

Acknowledgments