miR-34a and IRE1A/XBP-1(S) Form a Double-Negative Feedback Loop to Regulate Hypoxia-Induced EMT, Metastasis, Chemo-Resistance and Autophagy

Simple Summary The hypoxic tumor microenvironment is a key factor in the formation of metastasis and treatment resistance. We recently identified a new regulatory network consisting of the hypoxia-inducible transcription factor HIF-1A and the p53-inducible miR-34a that determines whether a tumor cell undergoes EMT (epithelial-mesenchymal transition) or MET (mesenchymal-epithelial transition) under hypoxic conditions. Here, we characterized XBP-1 and IRE1A as new miR-34a targets, which are relevant in this context. In addition, we found that the activation of the IRE1A/XBP-1 arm of the unfolded protein response (UPR) by hypoxia results in repression of miR-34a and thereby mediates hypoxia-induced EMT, migration, invasion, and chemo-resistance in p53 mutated/deficient CRC lines and ultimately contributes to lung metastasis formation. In this context, the restoration of miR-34a may be of therapeutic value in the future. Abstract Tumor-associated hypoxia, i.e., decreased availability of oxygen, results in a poor clinical outcome since it promotes EMT, metastasis, and chemotherapy-resistance. We have previously identified p53 and its target miR-34a, as critical determinants of the effect of hypoxia on colorectal cancer (CRC). Here, we aimed to characterize mechanisms that contribute to the selective advantage of cells with loss of p53/miR-34a function in a hypoxic environment. Using in silico prediction, we identified XBP-1 and IRE1A as potential miR-34a targets. IRE1A and XBP-1 are central components of the unfolded protein response that is activated by ER stress, which is also induced in tumor cells as a response to harsh conditions surrounding tumors such as hypoxia and a limited supply of nutrients. Here we characterized the XBP-1(S) transcription factor and its regulator IRE1A as direct, conserved miR-34a targets in CRC cells. After hypoxia and DNA damage, IRE1A and XBP-1 were repressed by p53 in a miR-34a-dependent manner, whereas p53-deficient cells showed induction of IRE1A and XBP-1(S). Furthermore, miR-34a expression was directly suppressed by XBP-1(S). In p53-deficient CRC cells, hypoxia-induced EMT, migration, invasion, metastases formation, and resistance to 5-FU were dependent on IRE1A/XBP-1(S) activation. Hypoxia-induced autophagy was identified as an XBP-1(S)-dependent mediator of 5-FU resistance and was reversed by ectopic miR-34a expression. The HIF1A/IRE1A/XBP-1(S)/p53/miR-34a feedback loop described here represents a central regulator of the response to hypoxia and ER stress that maintains cellular homeostasis. In tumors, the inactivation of p53 and miR-34a may result in IRE1A/XPB-1(S)-mediated EMT and autophagy, which ultimately promotes metastasis and chemoresistance.


RNA Isolation and qPCR
Total RNA was isolated from human cell lines with the Total RNA Isolation Kit (Roche, Basel, Switzerland) and from mouse tissues with the RNeasy Total RNA Isolation Kit (Qiagen) according to manufacturer's instructions. 1µg of total RNA per sample was used to generate cDNA using the Verso cDNA synthesis kit (Thermo Scientific, Waltham, MA, USA). Quantitative real-time PCR (qPCR) was performed with the Fast SYBR Green Master Mix (Applied Biosystems, Waltham, MA, USA) on a LightCycler 480 (Roche). The sequences of oligonucleotides used as qPCR primers are listed in Supplementary Table S1. Expression of mRNA and primary miRNA was normalized to GAPDH. The qPCR results were calculated using the ∆∆Ct method. Experiments were performed in triplicate.

Western Blot Analysis
Western blot analyses were performed as described previously. In brief, cells were lysed in RIPA lysis buffer with Complete Mini Protease and Phosphatase Inhibitors (Roche) added, sonicated, and centrifuged at 16.060× g for 15 min at 4 • C. For mouse tissues, samples were lysed in ice-cold lysis buffer by mechanical homogenization as described previously [24]. To complete lysis, samples were incubated 10 min on ice and centrifuged at 13,000 rpm at 4 • C to separate cell debris and the lysate for 15 min. Protein concentration was measured with BCA Protein Assay Kit (Thermo Fisher Scientific) according to manufacturer's instructions. 30-60 µg of whole cell lysate per lane was separated on SDS-acrylamide gels and transferred into Immobilon PVDF membranes (Millipore, Burlington, MA, USA). For immunodetection, membranes were incubated with antibodies listed in Supplementary  Table S2. Enhanced chemiluminescence (ECL, Millipore) signals were recorded with a 440CF imaging system (Eastman Kodak Co., Rochester, NY, USA). For quantification of Western blot signals, intensities of protein expression signals were quantified by densitometric analysis. The resulting values of the protein of interest were normalized to the corresponding loading controls.

Modified Boyden-Chamber Assay for Analysis of Migration and Invasion
Migration and invasion analyses were conducted as described previously. In short, cells were serum-starved by cultivation in 0.1% serum for 24 h. To analyze migration, 5 × 10 4 cells were seeded in the upper chamber (8.0 µm pore size membrane; Corning, Corning, NY, USA) in serum-free medium. To analyze invasion, chamber membranes were first coated with Matrigel (BD Bioscience, Franklin Lakes, NJ, USA) at a dilution of 3.3 ng/mL in medium without serum. Then 8 × 10 4 cells were seeded on the Matrigel in the upper chamber in serum-free medium. 10% FBS was placed as a chemo-attractant in the lower chamber. After cells were cultured for 48 h, non-motile cells at the top of the filter were removed and the cells in the bottom chamber were fixed with methanol, stained with DAPI, and counted using immunofluorescence microscopy. Relative invasion/migration was normalized to the corresponding control.

Chromatin Immunoprecipitation (ChIP) Assay
Chromatin immunoprecipitation in DLD-1 cells was performed according to instructions provided in iDeal ChIP-qPCR kit (Diagenode, Ougrée, Belgium). The sequences of oligonucleotides used as qChIP primers are listed in Supplementary Table S3.

Dual 3 -UTR Luciferase Reporter Assays
The full-length 3 -UTRs of the human and mouse XBP-1 and IRE1A mRNA were PCR-amplified from cDNA of human diploid fibroblasts (HDFs). The PCR product was cloned into the shuttle vector pGEM-T-Easy (Promega, Madison, WI, USA), and then transferred into the pGL3-control-MCS vector and verified by sequencing. Mutagenesis of the miR-34a seed-matching sequences in human and mouse was achieved with the Quik-Change II XL Site-Directed Mutagenesis Kit (Stratagene, San Diego, CA, USA) according to manufacturer's instructions and verified by sequencing. H1299 cells were seeded in 12-well format dishes at 3 × 10 4 cells/well for 24 h, transfected 100 ng of the respective firefly luciferase reporter plasmid, 20 ng of Renilla reporter plasmid as a normalization control, and 25 nM of pre-miR-34a (Ambion, PM11030,) or a negative control oligonucleotide (Ambion, neg. control #1) with HiPerFect Transfection Reagent (Qiagen) for 48 h. The analysis was performed with Dual Luciferase Reporter assay (Promega) according to manufacturer's instructions. Luminescence intensities were measured with an Orion II luminometer (Berthold, Berthold, ND, USA) in 96-well format and analyzed with the SIMPLICITY software package (DLR) version 3.0. The sequences of oligonucleotides used as primers are listed in Supplementary Table S4.

Wound Healing Assay
The wound healing assay was performed as described previously. In brief, Mitomycin C [10 ng/mL] was added two hours before generating a scratch using a Culture-Insert (IBIDI, 80241). After washing twice with HBSS to remove Mitomycin C and detached cells, medium was added. Cells were allowed to close the wound for the indicated periods. Images were captured on an Axiovert Observer Z.1 microscope connected to an AxioCam MRm camera using the Axiovision software (Zeiss) at the respective time-points.

Animal Experiments
The generation of miR-34a −/− mice with a C57BL6/SV129 background has been described previously [25]. To delete Mir34a and p53 in IECs, Mir34a Fl/Fl [25] and p53 Fl/Fl [24] mice crossed to Villin-Cre mice [26]. p53 Fl/Fl mice were obtained from Anton Berns (NKI, Amsterdam, The Netherlands) and Villin-Cre mice from Jackson Laboratories (Bar Harbor, ME, USA). All mice were crossed to at least 5 generations to FVB background. In all experiments, littermate controls were used. Mice were injected i.p. with Azoxymethane (AOM, 10 mg/kg, Sigma-Aldrich) at the age of 6 to 10 weeks, six times in weekly intervals, and on week 16 all mice were sacrificed. Mice were housed in individually ventilated cages (IVC) using "Lingocel Select" bedding. All animal protocols were approved by the local authorities (Regierung von Oberbayern, AZ: 55.2-1-54-2532-201-2014). All experiments involving mice were conducted with approval by the local Animal Experimentation Committee (Regierung of Oberbayern).

Analysis of Metastases Formation in NOD/SCID Mice
NOD/SCID mice were purchased from the Jackson Laboratory. DLD-1-luc2 cells transfected with XBP-1 or control siRNAs, or treated with DMSO or STF-083010 (50 µM) for 24 h and cultured at 20% O 2 or 0.5% O 2 for 30 h. 4 × 10 6 DLD-1-luc2 cells were dissolved in 0.2 mL HBSS and were injected into the lateral tail vein of 6-8 week old male immunocompromised NOD/SCID mice using 25-gauge needles. Anesthetized mice were injected intraperitoneally with D-luciferin (150 mg/kg) and imaged 10 min after injection using the IVIS Illumina System (Caliper Life Sciences, Hopkinton, MA, USA) in weekly intervals during light cycles. The acquisition time was 2 min. After 9 weeks mice were sacrificed and resected lungs examined for metastases using H&E staining. All experiments involving mice were conducted with approval by the local Animal Experimentation Committee (ROB-55.2-2532.Vet_02-18-57, Regierung of Oberbayern). All experiments were performed in accordance with the ARRIVE guidelines and regulations.

Tissue Preparation and Immunohistochemistry
The colon was opened longitudinally and rolled to form a "swiss roll". Tissues were fixed overnight in formalin, dehydrated, and embedded in paraffin and 2 µm sections were obtained. For immunohistochemistry (IHC) staining, tissue sections were deparaffinized/rehydrated, then boiled in citrate-based antigen retrieval solution (S2369, Dako, Hamburg, Germany) for 20 min, incubated in 3% H 2 O 2 /PBS (Carl Roth, 9681.1) for 10 min, and then blocked with 2.5% Normal Horse Serum (Vector Labs, San Francisco, CA, USA). The serial sections were incubated overnight at 4 • C with the primary antibody (anti-XBP-1(S), Cell Signaling #12782), followed by detection reagent (ImmPRESS ® HRP Horse Anti-Rabbit IgG Polymer Detection Kit, Peroxidase, MP-7401) for 1 h at room temperature. Bound antibodies were visualized by using DAB staining (Dako Liquid + Dab Substrate Chromogen System, K3468, Dako) according to the manufacturer's instructions. Sections were then counter-stained with hematoxylin (H-3401 Vector Labs). Slides were scanned with Vectra ® Polaris™ Automated Quantitative Pathology Imaging System (PerkinElmer, Waltham, MA, USA) and quantified by Image J software (U. S. National Institutes of Health, Bethesda, MD, USA).

Statistical Analysis
The GraphPad Prism 8.3.0 software was used for statistical analyses. The statistical significance of differences between group means was determined with the two-tailed unpaired Student's t-test and one-way ANOVA. Kaplan-Meier curves were used to display the overall survival time and the results were compared with a log-rank test. p values less than 0.05 were considered as statistically significant with asterisks indicated (* p < 0.05, ** p < 0.01, *** p <0.001, or **** p < 0.0001).

XBP-1 Is a Direct Target for Repression by miR-34a
When the CRC cell line DLD-1 was exposed to hypoxia for up to three days, a significant induction of XBP-1(S) at the mRNA and protein levels was observed (Figure 1a,b). In HCT-15 and HT-29 CRC cells, XBP-1(S) protein was also induced by hypoxia (Figure 1b). We have previously shown that miR-34a is directly repressed by HIF1A in CRC cell lines [10]. As a consequence, miR-34a targets are induced by hypoxia, such as INH3 [10]. Therefore, we asked whether XBP-1 contains a miR-34a seed matching site (SMS). Inspection of the human XBP-1 3 -UTRs using the TargetSCAN algorithm revealed a miR-34 seed-matching sequence (SMS), suggesting that XBP-1 may represent a miR-34a target ( Figure 1c). Indeed, a human XBP-1 3 -UTR reporter was repressed after ectopic expression of pre-miR-34a, whereas a reporter with a mutant miR-34a SMS was refractory ( Figure 1d). Furthermore, the induction of XBP-1(S) by hypoxia was prevented by ectopic pri-miR-34a expression in DLD-1 ( Figure 1e). In addition, XBP-1(S) protein levels increased after transfection of miR-34a-specific antagomir in HCT116 cells ( Figure 1f). Moreover, ectopic expression of pri-miR-34a in SW480 cells resulted in the repression of endogenous XBP-1 mRNA expression ( Figure 1g). In line with these results, ectopic pri-miR-34a expression repressed XBP-1 (S) protein in a time-dependent manner in SW620 cells ( Figure 1h). In addition, the induction of XBP-1(S) by hypoxia was prevented by ectopic pri-miR-34a expression in SW480 cells at the mRNA and protein levels ( Figure 1i,j) at the mRNA and protein levels. Taken together, these results show that XBP-1 is directly repressed by miR-34a and that this repression is alleviated during the response to hypoxia.

IRE1A Is Directly Repressed by miR-34a
Hypoxia-induced IRE1A mRNA and protein expression in DLD-1 (Figure 2a,b). XBP-1 mRNA is spliced by IRE1A in response to hypoxia-induced ER stress to produce XBP-1(S), which encodes a transcription factor [17]. Indeed, siRNA-mediated knockdown of IRE1A (Figure 2c), prevented the induction of XBP-1(S) by hypoxia ( Figure 2d). Moreover, treatment of DLD-1 cells with STF-083010, an inhibitor of the endonuclease activity of IRE1A, prevented the induction of XBP-1(S) by hypoxia ( Figure 2e). In order to determine whether repression of miR-34a contributes to the induction of IRE1A levels under hypoxic conditions, we used the TargetSCAN algorithm to identify a putative miR-34a SMS in the IRE1A 3 -UTR. (Figure 2f). A human IRE1A 3 -UTR reporter was repressed after transfection of pre-miR-34a, whereas a reporter with a mutant SMS was refractory to pre-miR-34a ( Figure 2g). Furthermore, ectopic expression of pri-miR-34a in DLD-1 and SW480 CRC cells resulted in the repression of IRE1A at the mRNA and the protein levels ( Figure 2h-j). Taken together, these results show that IRE1A is a direct target of miR-34a.

p53 Represses XBP-1(S) and IRE1A via miR-34a
Consistent with their inhibition by miR-34a, XBP-1(S) and IRE1A were repressed by ectopic p53 in SW480 CRC cells at the mRNA and protein levels (Figure 3a,b). Furthermore, hypoxia repressed XBP-1(S) and IRE1A at the protein and mRNA levels in p53-proficient HCT116 cells, whereas it resulted in their up-regulation in p53-deficient, isogenic HCT116 cells (Figure 3c,d). The repression of XBP-1(S) by p53 was alleviated by the transfection of miR-34a-specific antagomirs ( Figure 3e). Therefore, miR-34a mediates the repression of XBP-1(S) and IRE1A by p53. In addition, the repression of IRE1A after DNA damage was prevented by miR-34a-specific antagomirs ( Figure 3f). Taken together, the results demonstrate the CRC cells display a differential regulation of IRE1A and XBP-1(S) by miR-34a in response to hypoxia depending on their p53 status.

Conservation of miR-34a-Mediated Repression of XBP-1 and IRE1A in Mice
Inspection of the murine Xbp-1 and Ire1a 3 -UTRs using the TargetSCAN algorithm revealed the presence of conserved Mir-34a seed-matching sequences (SMS) in these mR-NAs (Figure 4a). Transfection with Mir-34a-specific antago-miRs, Xbp-1(S), and Ire1a protein levels increased in the CT26 murine cell line (Figure 4b). In addition, treatment of p53-proficient CT26 cells with Etoposide resulted in the repression of Xbp-1(S) and Ire1a (Figure 4c). Therefore, the negative regulation of Xbp-1 and Ire1a by p53 and Mir-34a is conserved between humans and mice.
revealed the presence of conserved Mir-34a seed-matching sequences (SMS) in thes mRNAs (Figure 4a). Transfection with Mir-34a-specific antago-miRs, Xbp-1(S), and Ire1 protein levels increased in the CT26 murine cell line (Figure 4b). In addition, treatment o p53-proficient CT26 cells with Etoposide resulted in the repression of Xbp-1(S) and Ire1 (Figure 4c). Therefore, the negative regulation of Xbp-1 and Ire1a by p53 and Mir-34a i conserved between humans and mice.  Next, we analyzed Xbp-1 expression in normal colonic epithelium and CRCs of Mir-34a-deficient mice that were previously established in our laboratory. The expression of spliced Xbp-1(S) and un-spliced Xbp-1(U) was increased in colonic epithelial cells isolated from Mir-34a knockout mice when compared with wild-type mice (Figure 4d). Moreover, we determined the expression of Xbp-1(S) in CRCs derived from mice that were treated six times with AOM (6XAOM) which represents an established mouse model for CRC [27]. Mir-34a-deficient CRCs from Mir-34a ∆IEC mice displayed robust expression of Xbp-1(S) whereas Mir-34a-proficient CRCs only showed marginal expression of Xbp-1(S) (Figure 4e). In this model, expression of Xbp-1(S) was significantly increased in CRCs from Mir-34a ∆IEC /p53 ∆IEC and Mir-34a ∆IEC mice when compared to CRCs from WT or p53 ∆IEC mice (Figure 4f). mRNA levels of IRE1A were also increased in CRCs from Mir-34a ∆IEC and p53 ∆IEC /Mir34a ∆IEC mice (Figure 4g). Moreover, the number of XBP-1(S)-positive cells was increased in CRCs of Mir-34a ∆IEC and p53 ∆IEC /Mir-34a ∆IEC mice when compared to WT and p53 ∆IEC mice (Figure 4h). Taken together, these results suggest that the negative regulation of Xbp-1 and Ire1a by miR-34a is conserved in mice and in vivo.

MiR-34a and IRE1A/XBP-1S Form a Double-Negative Feedback
We observed an increase in pri-miR-34a expression after the inactivation of XBP-1 by siRNAs ( Figure 5a). Moreover, down-regulation of XBP-1, IRE1A, or HIF1A by siRNAs or by inhibition of IRE1A activity by STF-083010 treatment prevented the repression of pri-miR-34a by hypoxia in DLD-1 (Figure 5b). Here, we identified a conserved XBP-1 (S) binding site within the sequence upstream of the miR-34a transcriptional start site, indicated as UPRE (UPR element) (Figure 5c), which overlaps with the HIF1A-binding site in the miR-34a promoter that we had previously characterized [10]. Thus, we hypothesized that XBP-1(S) may directly repress miR-34a. Indeed, XBP-1(S) occupancy at the UPRE was detected by chromatin-immunoprecipitation (ChIP) in hypoxic DLD-1 cells (Figure 5d). Moreover, the down-regulation of HIF1A by siRNAs did not affect the expression of IRE1A under hypoxia demonstrating that the induction of IRE1A by hypoxia is HIF1A-independent ( Figure 5e). Therefore, activation of IRE1A by hypoxia may lead to enhanced expression of XBP-1(S) and repression of miR-34a via multiple pathways as depicted in the model shown in (Figure 5f).

Roles of XBP-1(S) and IRE1A in Hypoxia-Induced EMT and Invasion
Next, we determined whether the miR-34a targets identified here mediate EMT and downstream processes, such as migration and invasion, after exposure to hypoxia. Interestingly, treatment of DLD-1 cells with STF-083010, a specific inhibitor of IRE1A and therefore XBP-1 mRNA splicing, prevented the induction of the EMT-markers Snail and Vimentin by hypoxia in DLD-1 cells (Figure 6a).
In addition, down-regulation of IRE1A or XBP-1 by treatment with a single or a pool of siRNAs prevented the induction of Snail, Slug, Zeb1, and Vimentin in DLD-1 cells under hypoxia (Figure 6b,c and Figure S2a,b). Moreover, treatment with a pool of siRNAs against XBP-1 prevented the loss of the epithelial marker E-cadherin/CDH1 from the outer membrane in DLD-1 cells by hypoxia ( Figure S1c,d). In addition, down-regulation of XBP-1 or IRE1A by treatment with a single or a pool of siRNAs also decreased cellular migration, as determined in a scratch-assay (Figure 6d,e and Figure S1e,f). Inhibition of IRE1A by STF-083010 treatment in DLD-1 cells also decreased wound closure (Figure 6f). Moreover, the down-regulation of IRE1A by STF-083010 treatment or XBP-1 by siRNAs significantly reduced invasion under hypoxia (Figure 6g). Therefore, the activation of IRE1A and XBP-1(S) is necessary for the induction of EMT, migration, and invasion by hypoxia in CRC cells.

Roles of XBP-1(S) and IRE1A in Hypoxia-Induced EMT and Invasion
Next, we determined whether the miR-34a targets identified here mediate EMT and

IRE1A/XBP-1S Activation Is Necessary for Hypoxia-Induced Metastasis
We had previously demonstrated that pretreatment of DLD-1 CRC cells with 0.5% O 2 for 48 h and their subsequent injection into tail veins of NOD/SCID mice resulted in the formation of lung metastases, whereas untreated DLD-1 cells do not form lung metastases [10]. Furthermore, ectopic miR-34a expression prevents lung metastasis formation [28]. Therefore, we determined whether inhibition of the IRE1A/XPB1 pathway is sufficient to prevent hypoxia-induced metastasis. Pre-treatment of DLD-1 cells stably expressing luciferase with STF-083010 significantly reduced the expression of the mesenchymal marker Vimentin and blocked XBP1(S) expression under hypoxia (Figure 7a). Importantly, pretreatment of these cells with STF-083010 efficiently inhibited hypoxia-induced lung metastases formation after i.v. injection into NOD/SCID mice (Figure 7b,c). Moreover, siRNA-mediated silencing of XBP-1 or IRE1A prior to exposure of DLD-1 cells to hypoxia significantly reduced the number of metastatic tumor nodules that these cells formed after injection into NOD/SCID mice (Figure 7d). Taken together, these results show that activation of the IRE1A/XBP-1(S) is required for the hypoxia-induced formation of metastases by CRC cells.

XBP-1(S) Mediates Hypoxia-Induced Chemo-Resistance and Autophagy
We have previously demonstrated that hypoxia mediates chemo-resistance towards 5-FU by suppressing miR-34a expression [10]. Importantly, p53 and miR-34a determine the response of tumor cells to 5-FU treatment under hypoxia [10]. Notably, p53-deficient cells were more resistant to 5-FU under hypoxia. Therefore, we determined whether the miR-34a targets identified here modulate the cellular response of DLD-1 cells to 5-FU under hypoxia. DLD-1 cells express mutant p53 [29] and should therefore not induce miR-34a expression after hypoxia. Interestingly, XBP-1 inhibition by siRNAs or treatment with STF-083010 significantly reduced the viability of DLD-1 cells treated with 5-FU (Figure 8a-c and Figure  S2a). Therefore, the up-regulation of XBP-1 and IRE1A is required for hypoxia-mediated resistance towards 5-FU.
It has been shown previously, that hypoxia-induced autophagy contributes to chemoresistance [30][31][32]. Consistently, the resistance of DLD-1 cells to 5-FU observed at 0.5% O 2 was reduced by treatment with SBI-0206965, a highly selective autophagy kinase ULK1 inhibitor or with chloroquine, which inhibits autophagy (Figure 8d and Figure S2b). Since XBP-1 has been linked to the induction of autophagy [33], we asked whether XBP-1 mediates the induction of autophagy by hypoxia in CRC cells. Indeed, the knockdown of XBP-1 in DLD-1 cells significantly reduced the hypoxia-induced accumulation of LC3B-II ( Figure 8e) and the number of cells positive for LC3B puncta, which indicates the formation of autophagosomes in DLD-1 cells under hypoxic conditions (Figure 8f). Notably, miR-34a is a known inhibitor of autophagy as it targets multiple factors implicated in autophagy [34,35]. Indeed, treatment of DLD-1 cells with pre-miR-34a oligonucleotides significantly reduced the number of LC3B puncta-positive cells under hypoxic conditions (Figure 8e). Moreover, ectopic expression of pri-miR-34a in SW480 cells resulted in the reduction of LC3B puncta-positive cells and decreased LC3B-II accumulation ( Figure S2c,d). Furthermore, treatment with Mir-34a-specific antagomiRs induced the transition of LC3B-I to LC3B-II in murine CT26 cells ( Figure S2e). Taken together, our findings show that during the response to hypoxia, the induction of XBP-1(S) is required for autophagy-induced chemo-resistance. In addition, repression of miR-34a by HIF1A and/or XBP-1 is presumably also important for autophagy in this context, as miR-34a would otherwise inhibit autophagy by targeting multiple components of the autophagic process.    , then exposed to 0.5% O 2 for 48 h, and subsequently treated with or without 5-FU for 72 h. (e) Indirect immunofluorescence detection of LC3B in DLD-1 cells transfected with XBP-1, control siRNAs or pre-miR-34a for 24 h then exposed to 0.5% O 2 for 48 h. (f) Western blot analysis of LC3B in DLD-1 cells transfected with XBP-1 or control siRNAs exposed to 20% or 0.5% O 2 for 48 h. (g) The HIF1A/IRS1A/XBP1/miR-34a regulatory circuit: Schematic model summarizing the findings of this study. Green arrows represent stimulatory and red lines inhibitory effects. Loss or mutation of p53/miR-34a in CRCs may enhance the indicated, protumorigenic processes. In (a,c-e) mean values ± SD (n = 3) are provided. (***) p < 0.001, (**) p < 0.01, (*) p < 0.05.

Discussion
The interplay between HIF-1A and p53 may serve as a critical determinant of cancer invasion and metastasis under hypoxic conditions, and a potential determinant of therapeutic outcomes for colorectal cancer patients [8]. At present, the interactions between p53 and HIF-1A are not completely understood. The scenario is made more complex by the substantial number of target genes of both factors that represent components of numerous signaling pathways and biochemical processes, which control tumor cell survival. Besides the p53-mediated regulation of genes encoding proteins, p53-induced microRNAs have emerged as important effectors of p53 functions. Importantly, miR-34a represents the miRNA, which is induced by p53 most profoundly in all tested cell types, implying outstanding importance of miR-34a among all p53-induced miRNAs [36]. We have previously shown that PPP1R11/INH3 is subject to a feed-forward regulation by HIF1A and miR-34a, which mediates its induction under hypoxic conditions. Furthermore, Inh3 expression was required for induction of EMT, invasion, and migration by hypoxia in p53-deficient colorectal cell lines [10]. Here, we characterized novel regulators and effectors of miR-34a. We found that the decision of tumor cells to undergo EMT or MET in response to hypoxia is mediated by a regulatory network involving HIF1A, p53, miR-34a, and its new targets, IRE1A and XBP-1 (see also Figure 8g for a summary model). We found that XBP-1(S)/IRE1A and miR-34a form a double-negative feedback regulatory loop, wherein miR-34a represses XBP-1 and IRE1A under normoxia, in TP53-proficient CRC cells. In TP53-defective CRC cells HIF-1A, XBP-1(S), and IRE1A may cooperate to repress and degrade miR-34a under hypoxia. The decrease in miR-34a levels may further increase the levels of IRE1A, which activates XBP-1(S). Subsequently, the resulting activation of XBP-1(S) mediates hypoxia-induced EMT, migration, invasion and ultimately contributes to metastasis.
The negative regulation of XBP-1S by the tumor suppressive p53/miR-34a axes is in line with an oncogenic/pro-tumorigenic function of the IRE1A/XBP-1(S) axes. This is also consistent with its documented role in the c-MYC network [37,38]. XBP-1(S) activation and the subsequent increase in protein folding capacities represent a central facilitator of tumor cell growth and tumor expansion/migration. Its negative regulation by the p53/miR-34a axes is therefore presumably central to the tumor suppressor functions of p53 and the miR-34 family. Here, we found that the inactivation of XBP-1(S)/IRE1A prevents the hypoxia-induced formation of lung-metastases in mice. Our findings imply that activation of the IRE1A/XBP1S pathway is necessary for tumor cells to acquire a mesenchymal/invasive phenotype under hypoxia.
Previously, we showed that miR-34a is directly repressed by HIF-1A under hypoxic conditions in the p53-deficient cells [10]. Here, we further demonstrated that XBP-1(S) enhances the repression of the tumor-suppressive miR-34a by HIF-1A. This new mechanism is in accordance with previous works showing that XBP-1(S) plays a key role in cancer cell survival under hypoxia [21,22]. Notably, XBP-1 and HIF-1A co-occupy several wellknown HIF-1A targets, and XBP-1 depletion down-regulated HIF-1A targets under hypoxic conditions [21]. In addition, XBP-1 co-localizes with hypoxia markers in tumors and the loss of XBP-1 increases the sensitivity of tumor cells to hypoxia-induced apoptosis and inhibits tumor growth [22], implicating XBP-1(S) as a critical survival factor under hypoxic conditions. Moreover, our results also revealed that the hypoxia-induced XBP-1/IRE1A activation is essential for hypoxia-mediated resistance against 5-FU through the induction of autophagy. Thereby, p53 and/or miR-34a inactivation, which is commonly found in CRC and other tumors, may promote tumor cell survival. Altogether, our results reveal a new p53/miR-34a/XBP-1/IRE1A regulatory circuitry that may play a crucial role during hypoxia-driven tumor cell survival and metastasis. In addition, up-regulation of miR-34a may prevent chemo-resistance by targeting XBP-1/IRE1A. Restauration of miR-34a function by delivery of miR-34a mimics to tumor cells using several different approaches is being clinically evaluated [39]. In the future, miR-34a restoration may be an attractive therapeutic strategy to treat cancer or an alternative approach to overcome chemo-resistance.