L-WRN cells were obtained from the ATCC (CRL-3276). HEK293-derived Adherent-293 (AD-293) cells were obtained from Stratagene (240085). CBRLuc-mCherry reporter murine duodenal organoids were a gift from Dr. Helen Piwnica-Worms (28). Murine duodenal organoids were cultured at 37°C in 5% CO₂ and 5% O₂, while all other cell lines were cultured at 37°C in standard 5% CO₂ incubators. All cell lines were authenticated by short tandem repeat profiling and were confirmed to be Mycoplasma free. Recombinant human/mouse Wnt5a was purchased from R&D Biosystems (645-WN-010-CF).

All experimental mouse work adhered to the standards articulated in the Animal Research: Reporting of In Vivo Experiments guidelines. Additionally, all mouse work was approved by the Institutional Animal Care and Use Committee of The University of Texas MD Anderson Cancer Center. Mice were maintained on a 12-hour light/dark cycle and were provided with sterilized water and standard rodent chow (Prolab Isopro RMH 3000 irradiated feed) ad libitum. C57BL/6 mice (RRID : IMSR_JAX:000664), Wnt5a^fl/fl mice (RRID : IMSR_JAX:026626) (29), and R26-LSL-hHIF2a^dPA mice (RRID : IMSR_JAX:009674) (25) were obtained from Jackson Laboratories.

L-WRN conditioned media was prepared as previously described (24, 28). Briefly, L-WRN cells were maintained in DMEM high glucose media (Sigma, D6429) supplemented with 10% (v/v) FBS (Sigma, F4135, 1% (v/v) Penicillin/Streptomycin (Sigma, P4333), 500 μg/ml Hygromycin B Gold (In vivoGen, ant-hg-1), and 500 μg/ml G418 (Sigma, G8168). Once cells were confluent, the media was replaced with Advanced DMEM/F12 media (Gibco, 12634010) supplemented with 10% (v/v) FBS, 1% (v/v) Penicillin/Streptomycin, and 2 mM L-glutamine (Sigma, G7513). Conditioned media was collected for six days, centrifuged at 3,000 rpm for 5 minutes, vacuum filtered through a 20 μM PES membrane (Thermo Scientific, 567-0020), and stored at -80°C.

Duodenal crypts were isolated from C57BL/6 mice, R26-LSL-hHIF2a^dPA mice, and Wnt5a^fl/fl mice, and 3D organoid cultures were established as previously described (24, 28). For all steps in this protocol, EDTA (Sigma-Aldrich, E7889) was added fresh to both PBS (Cytiva, SH30256.LS) and HBSS without calcium and magnesium (Gibco, 14025092) to a final concentration of 2 mM, and kept on ice. Mice were humanely euthanized by CO₂ inhalation followed by cervical dislocation. The duodenum was measured 1 cm below the pylorus, and 4 cm were resected and flushed with PBS/EDTA, then incubated on fresh PBS/EDTA for 10 min on ice, and finally transferred to ice-cold HBSS/EDTA. Duodenal samples were then serially vortexed at 1,600 rpm at 4°C, in fresh HBSS each time, for 5 min, 3 min, and 8 min. Supernatants from the second and third vortexes were combined and passed through 70-μm strainers (Corning, 431751) to isolate crypts and remove any villi that might remain after the washes. Duodenal crypts were pelleted at 1,000 rpm at 4°C, then washed in Advanced DMEM/F12 supplemented with 10% (v/v) FBS, 1% (v/v) Penicillin/Streptomycin, and 2 mM L-glutamine, and re-centrifuged as before. The pelleted crypts were resuspended in 50% (v/v) Matrigel (Corning, 354234) diluted with the crypt washing media, and then were seeded as domes into 24-well plates (Corning, 3524). After Matrigel solidification at 37°C, the duodenal organoids were cultured in 50% (v/v) L-WRN conditioned media supplemented with 10 mM Y27632 (ROCK inhibitor; Sigma-Aldrich, Y0503) and 10 μM SB431542 (TGF-β RI Kinase Inhibitor VI; Sigma-Aldrich, 616461). The culture media was refreshed every other day and the organoids were passaged every third day.

Ad-GFP (VVC-U of Iowa-4, Ad5CMVeGFP), Ad-Cre (VVC-U of Iowa-5, Ad5CMVCre), and Ad-Cre-GFP (VVC-U of Iowa-1174, Ad5CMVCre-eGFP) viral vectors were provided by the University of Iowa Viral Vector Core (http://www.medicine.uiowa.edu/vectorcore). Ad-human HIF1 and Ad-human HIF2 viral vectors were previously produced (30) using hHIF1 (Addgene #18955) and hHIF2 plasmids (Addgene #18956) that contain double proline-to-alanine substitutions which render them nondegradable by VHL (31). Duodenal organoids were transduced after at least three passages. First, organoids were harvested by incubation with Cell Recovery Solution (Corning, 354253) for 30 min on ice, then centrifuged at 1,000 rpm for 5 min at 4°C, and then washed with cold PBS. Organoids were then digested into single intestinal stem cells (ISCs) via mechanical digestion while incubating in TrypLE (Gibco, 12605010) supplemented with 10 μM Y27632 and 500 μM N-acetylcysteine (Sigma-Aldrich, A0737) for 5 min at 37°C. To neutralize TrypLE, cold Advanced DMEM/F12 media supplemented with 10% (v/v) FBS, 1% (v/v)  Penicillin/Streptomycin, 2 mM L-glutamine, 10 μM Y27632, and 10 μM SB431542 was added and the ISC suspension was centrifuged at 1,500 rpm for 5 min at 4°C. Single ISCs were resuspended in Advanced DMEM/F12 supplemented as described above and passed through a 35-μM strainer (Corning, 352235). Single ISCs were stained with Trypan Blue (Bio-Rad, 1450021) and counted on a TC20 Automated Cell Counter (Bio-Rad). Duodenal ISCs were transduced with adenoviral particles at MOI rates of 50 particles per cell for 1.5 h at 37°C. Afterwards, transduced ISCs and leftover adenoviral particles were resuspended in 50% (v/v) Matrigel, seeded into 24-well plates, and cultured in 50% (v/v) L-WRN conditioned media supplemented with 10 μM Y27632, 10 μM SB431542, and 10 mM Nicotinamide (Sigma-Aldrich, N3376). RNA purification of duodenal organoids was done approximately 72 hours post-transduction.

Duodenal organoids were released from Matrigel by incubation with Cell Recovery Solution for 30 min on ice, then centrifugation at 1,000 rpm for 5 min at 4°C, followed by washing with cold PBS. Organoids were then lysed in M-PER mammalian protein extraction reagent (Thermo Fisher Scientific, 78501) supplemented with protease inhibitor (Roche Life Science, 11836170001) and phosphatase inhibitor (Roche Life Science, 4906837001), and then denatured with 4x Laemmli sample buffer (Bio-Rad, 1610747). Equal cell protein lysate amounts were resolved on SDS-PAGE gel (Bio-Rad, 4568034) and transferred onto a nitrocellulose membrane (Bio-Rad, 1620215) using a Trans-Blot Turbo transfer system (Bio-Rad). After blocking with 5% (w/v) skim milk powder (Bio-Rad, 1706404XTU) in TBS-T for 1h at room temperature, the membrane was probed with primary antibodies diluted in Superblock T20 (ThermoFisher Scientific, 37536) at 4°C overnight. Anti-HIF2α rabbit (Novus Biologicals, NB100-122) was used at 1:500 and anti-β-actin rabbit (Cell Signaling Technology, 4970S) was used at 1:1,000. The membrane was then washed and incubated with HRP-conjugated secondary antibody for 1h. The membrane was developed with Clarity Western ECL Substrate kit (Bio-Rad, 1705061) and visualized using a ChemiDoc imaging system (Bio-Rad). Relative protein expression was quantified based on band intensity using ImageJ software (RRID : SCR_003070) and normalized to control group.

Duodenal organoids were released from Matrigel as indicated above, then homogenized by vortexing and vigorously pipetting, and then RNA was purified using an RNeasy mini kit following the manufacturer’s handbook (Qiagen, 74106). Reverse transcription was performed with a mix of random primers and oligos using an iScript cDNA Synthesis kit (Bio-Rad, 1708891). Next, qRT-PCR was performed using SYBR Green Master mix (Bio-Rad, 1725124) and the primer assays listed on Supplementary Table 1 , on a CFX384 Real-time system (Bio-Rad). Relative gene expression and fold change was calculated using Hprt and Tbp as reference genes.

Duodenal organoids from C57BL/6 mice transduced with either Ad-GFP, Ad-hHIF1, or Ad-hHIF2 were harvested and RNA was purified using an RNeasy mini kit. RNA purity and concentration were measured using an Epoch Microplate Spectrophotometer with a Take3 Micro-Volume Plate, and Gen5 software (v2), all from BioTek Instruments, Inc. Library preparation and sequencing were performed in the Sequencing and Microarray Facility at MD Anderson Cancer Center. The raw sequencing data was downloaded from the core server and low-quality reads were removed and Q20 and GC contents were calculated. Transcript abundance was quantified using the RSEM software package (RRID : SCR_013027) (32). Differential expression analysis was performed using EBSeq software package (RRID : SCR_003526) (33). Supplementary Tables 2 and 3 show the output differential expression gene matrix for Ad-HIF2 and Ad-HIF1, respectively, compared to Ad-GFP.

The human WNT5a promoter from −2000 to +200 nucleotides from the +1 transcriptional start site (34) was synthesized directly into the KpnI- and XhoI-flanked region of the pUC57 vector (Genscript, SD1176). All mutant promoters were synthesized in a similar fashion. The specific sequences of the mutations can be found in Supplementary Figure 2 . The KpnI- and XhoI-flanked fragments of each of the promoter constructs were excised and then ligated into the compatible BglII- and HindIII-flanked sites in the pGL4.10[luc2] vector (Promega, E6651). AD-293 cells were cultured in DMEM supplemented with 10% FBS and 2 mM L-Glutamine. Cells were transfected using Xfect Transfection Reagent (TakaraBio, 631318) using the manufacturer’s standard protocol. For instance, a 96-well opaque plate was seeded with 9.5 x 10³ cells/well and transfected with 0.08 μg of hHIF1 plasmid, hHIF2 plasmid, or control GFP plasmid (Addgene #26822), 0.09 μg of WNT5A-luciferase construct, and 9 ng of Renilla luciferase, along with Xfect polymer. The transfected cells were incubated for 48 h in Opti-MEM media (Gibco, 31985070). Dual luciferase assays were performed using the Dual-Glo^® Reagent (Promega, E2940) kit according to manufacturer’s instructions. The luminescence data was measured on a Cytation 3 luminometer (Biotek Instruments, Inc.). Luciferase signal was normalized to Renilla signal, and then WNT5A promotor transactivation was calculated as a fold change over control GFP plasmid transfection.

Spheroid Formation Assays were performed as previously described (28), using a murine duodenal Click Beetle Red Luciferase-mCherry (CBRLuc-mCherry) reporter organoid line and Wnt5a^fl/fl duodenal organoids. Briefly, duodenal organoid cultures were exposed to the indicated pre-radiation treatments and then irradiated using an X-Rad 320 cell irradiator (Precision X-Ray). Immediately after irradiation, the duodenal organoids were harvested using Cell Recovery Solution and then digested into single ISCs using TrypLE supplemented with Y27632 and N-Acetylcysteine, followed by filtering through a cell strainer as detailed above for Adenoviral Transduction. Live cells were quantified using ViaStain™ AO/PI Staining Solution (Nexcelom, CS2-0106) in a Cellometer^® Vision CBA Image Cytometer (Nexcelom). Live duodenal ISCs were seeded in Matrigel in 24-well culture plates at a density of 5,000 cells/well. Organoids were maintained in 50% L-WRN conditioned media supplemented with 10 μM Y27632 and 10 μM SB431542 for six days, then surviving cells were assessed via bioluminescence or z-stack imaging. For quantification of bioluminescence, organoids were incubated with 300 μg/ml of D-Luciferin Firefly (Gold Biotechnology, L-123) in the media for 20 min at 37°C, and measurements were taken using a CLARIOstar plate reader (BMG Labtech). Sample wells were first normalized to blank Matrigel wells, and then to unirradiated controls. Z-stack images covering the entire Matrigel dome area were taken using a Cytation 3 Cell Imaging Multi-Mode Reader (Biotek Instruments, Inc.). Z-stack images were stitched using Photoshop version 19.1.7 (RRID : SCR_014199) and viable organoids measuring at least 150 μm were manually quantified using ImageJ version 1.52q (RRID : SCR_003070). For organoids analyzed via z-stack imaging, the surviving fraction was calculated as described previously (35).

All statistical analyses were performed using GraphPad Prism^® V.8 (San Diego, CA; RRID : SCR_002798), with a significance level of α = 0.05. Dunnett’s multiple comparisons test was used to analyze Dual Luciferase Assays.

To gain insight into how the HIFs regulate GI biology and to understand the mechanisms by which HIF2 confers radioprotection to the small intestine, we generated a 3D murine small intestinal organoid model system. We harvested duodenum crypts of wild-type C57BL/6 mice, seeded them in basement membrane matrix (Matrigel), and cultured them in conditioned media containing Wnt3a, R-spondin 3, and noggin ( Figure 1A ) (24). These conditions enrich for intestinal stem cells (ISCs), which self-assemble into crypt-like 3D structures that are able to recapitulate the small intestinal crypt ex vivo (36). We then transiently overexpressed a non-degradable HIF1 or HIF2 allele (25), which contain mutations in the two key prolyl residues that are hydroxylated by the EGLN proteins, rendering them stable even under normoxic conditions, using adenoviral particles ( Figure 1A ). We confirmed adenoviral-mediated HIF overexpression by both western blot and qRT-PCR, and then performed RNA sequencing ( Figures 1A–C ).

Among thousands of coding transcripts assessed, 1,113 genes exhibited significant differential expression between control GFP- and HIF2-overexpressing duodenal organoids ( Figure 2A and Supplementary Table 2 ). All the differentially expressed genes with statistical significance were selected with p < 0.05, False Discovery Rate (FDR) < 0.05, and at least 2-fold change. Hierarchical clustering analysis revealed a total of 461 upregulated genes and 652 downregulated genes ( Supplementary Table 2 ). The entire set of differentially expressed genes can be visualized in a volcano plot in Figure 2B . To validate our RNA sequencing results, we independently assessed the expression of 5 genes by qRT-PCR. Our results confirmed that Stat6, Aqp8, and Nos2 were upregulated and that Wnt4 and Egr1 were downregulated in HIF2-overexpressing duodenal organoids compared to wild-type duodenal organoids, indicating that the RNA sequencing dataset is reliable ( Figure 2C ).

We also examined the transcriptome of HIF1-overexpressing organoids to gain more insight into how the HIFs regulate overall ISC biology. Among the thousands of coding transcripts assessed, only 55 genes exhibited significant differential expression between control GFP- and HIF1-overexpressing duodenal organoids, in which 25 genes were upregulated and 30 genes were downregulated ( Supplementary Figure 1 and Supplementary Table 3 ). Interestingly, the only genes commonly upregulated by HIF1 and HIF2 in our dataset were Nanp, Ppp1r3c, Rasgrf1, Klhl3, Mical2, and Nos2, and the only transcripts commonly downregulated by both HIFs were Slc18a1, Gm37069, Hr, Gm12480, Gm13151, Gm21981, Klhl30, Vcp-rs, Gm4204, AY512931, and Gm28037 ( Supplementary Tables 2 and 3 ). We then focused our attention on the HIF2 dataset, as HIF2 and not HIF1, has been shown to be the main HIF isoform driving GI radioprotection (22).

Interestingly, the HIF2-induced transcriptome included known radiation modulators as well as genes involved in GI healing and homeostasis, as highlighted in Figure 3A in blue and red, respectively. We identified Wnt5a as a transcriptional target of HIF2, but not HIF1 ( Figure 3A and Supplementary Tables 2 and 3 ), and interestingly, Wnt5a has a known connection to non-canonical intestinal crypt regeneration (26). Thus, we took a candidate approach to further investigate its transcriptional regulation by HIF2. We verified that Wnt5a was upregulated by HIF2 using two approaches. First, we performed qRT-PCR to independently evaluate duodenal organoids that transiently overexpressed non-degradable HIF2 via Adeno-hHIF2 transduction (25), and found significantly increased Wnt5a expression by almost 50-fold compared to organoids transduced with Adeno-GFP ( Figure 3B ). Second, we generated duodenal organoids from mice that conditionally overexpressed non-degradable human HIF2 by knock-in into the Rosa 26 locus (R26-LSL-hHIF2) (25), transduced them with either Adeno-Cre or control Adeno-GFP vectors, and assessed Wnt5a expression using qRT-PCR. Stable HIF2 overexpression also resulted in significant upregulation of Wnt5a by 6-fold ( Figure 3C ). Together, these results suggest that HIF2 induces intestinal Wnt5a expression.

The HIFs are transcription factors that recognize and bind hypoxia response elements (HREs) in promoter or enhancer regions to induce gene transcription (13). Thus, we analyzed the WNT5A promoter sequence to determine whether it contained putative HRE motifs and identified multiple low- and high-stringency HRE consensus sequences ( Figure 4A and Supplemental Figure 2 ). To determine whether induction by HIF2 occurs directly or indirectly, we designed luciferase reporter constructs of the human WNT5A promoter spanning from 2,000 nucleotides upstream of the transcriptional start site (34) to 200 nucleotides downstream, containing five HRE consensus sequences that are closely associated with HIF ancillary sequences (HAS) and E-box motifs ( Figure 4A and Supplemental Figure 2 ). HAS and E-box motifs are cis-element that are required for an HRE to be functionally active. Both motifs play a role in the recruitment of transcriptional machinery that together with HIF2 induce promoter activation (37–39).We transfected this construct into human embryonic kidney-derived Adherent 293 (AD-293) cells along with an expression vector encoding GFP or constitutively active human HIF1 or HIF2 (11) and performed dual luciferase reporter assays (25). We found that HIF2 significantly increased WNT5A promoter transactivation by eight-fold over GFP controls ( Figure 4B ). HIF1 only modestly affected promoter activity ( Figure 4B ).

To understand if any specific HRE motif of the WNT5A promoter is required for its activation by HIF2, we performed mutational analyses. We engineered transversion point mutations on the guanine and cytosine nucleotides of three distal and two proximal putative HRE sites within the promoter, and transfected AD-293 cells with these constructs to perform promoter activation studies ( Figure 4C and Supplementary Figure 2 ). We found that independently mutating the two most distal HREs in the WNT5A promoter had the most impact on HIF2-induced transactivation, and that these two HREs are differentially regulated by HIF2. The point mutation at the HRE in position -1548 significantly diminished the previously observed HIF2-induced promoter activation, suggesting that the -1548HRE is necessary for HIF2’s ability to bind and positively regulate the WNT5A promoter ( Figure 4C ). Importantly, this -1548HRE resides upstream of a HAS, which is also in close proximity to an E-box ( Figure 4A and Supplementary Figure 2 ). Conversely, mutating the -1648HRE resulted in an increased ability of HIF2 to transactivate the WNT5A promoter, suggesting that it might be a repressive HRE. Although the HRE at -1648 resides nearby E-box and HAS motifs, HRE sites that are preceded by cytosine nucleotides on 5’ form E-box binding sites of other basic helix-loop-helix transcription factor families, and HIFs rarely recognize these (38). Moreover, while HIF2-induced gene upregulation is largely mediated by direct HIF2 binding to DNA motifs, HIF2-induced gene repression tends to occur indirectly through transcriptional co-repressors (37). Thus, we posit that direct binding of HIF2 to the -1548HRE promotes WNT5A transcription, whereas HIF2 interaction with the -1648HRE promotes the recruitment of repressive transcriptional machinery. The point mutations of the HREs at positions -631, -296, and +82 did not alter the activation of the WNT5A promoter by HIF2 ( Figure 4C ). Taken all together, our results indicate that HIF2 directly activates the WNT5A promoter, possibly by binding the -1548HRE.

Wnt5a is a non-canonical Wnt ligand that has been shown to promote the formation of new intestinal crypts in order to re-establish homeostasis after intestinal mucosal injury (26). To evaluate whether Wnt5a mediates HIF2-afforded GI radioprotection, we performed a spheroid formation assay, which is an ex vivo microcolony assay that allows us to evaluate potential radiation modulators in small intestinal organoid cultures (28). We treated CBRLuc-mCherry, a murine duodenal mCherry reporter organoid line (28), with recombinant Wnt5a (rWnt5a) 10 hours prior to irradiation, then exposed them to 0-8Gy of Xrays, and re-seeded single cells in Matrigel ( Figures 5A, B ). CBRLuc-mCherry organoids that were pre-treated with rWnt5a produced significantly higher relative bioluminescence levels six days after irradiation, compared to vehicle-treated organoids, indicating that rWnt5a increased the number of regenerating crypts ( Figure 5B ). Pre-treatment with rWnt5a significantly increased the cryptogenic capacity of CBRLuc-mCherry organoids exposed to 2 Gy and 4 Gy by 2-fold and 6-fold, respectively, and also increased the cryptogenic capacity of organoids exposed to 6 Gy by 2.8-fold, but this was not statistically significant ( Figure 5B ). Interestingly, CBRLuc-mCherry organoids that continued receiving rWnt5a treatment after irradiation did not display improved cryptogenic capacity ( Supplementary Figure 3A ). Moreover, initiating rWnt5a treatments only after irradiation also did not improve the cryptogenic capacity of CBRLuc-mCherry organoids ( Supplementary Figure 3B ). These results suggest that Wnt5a could be radioprotective to small intestinal crypts but does not mitigate radiation damage after it has occurred. A possible explanation for these results is that Wnt5a suppresses intestinal organoid proliferation by inducing TGF-β signaling (26), which would protect cycling ISCs from radiation, but would dampen their capacity to regenerate crypts if Wnt5a treatment was continued after the radiation injury was incited.

To understand whether Wnt5a is necessary for intestinal crypt regeneration following radiation injury, we generated conditional Wnt5a knockout (Wnt5a^CKO ) duodenal organoids from Wnt5a^fl/fl mice and transduced them with Adeno-Cre-GFP or control Adeno-GFP vectors. In agreeance with published work demonstrating that Wnt5a is dispensable for homeostasis in the gut postnatally (40), deletion of Wnt5a did not affect organoid growth or morphology ( Figure 5C ). We then performed a modified spheroid formation assay using the Wnt5a^CKO duodenal organoids and z-stack imaging, rather than bioluminescence, to quantify regenerating organoids ( Figure 5D ). Loss of Wnt5a did not affect the cryptogenic capacity of unirradiated duodenal organoids ( Figure 5E and Supplementary Figure 4 ), again confirming that Wnt5a is dispensable for crypt homeostasis ( Figure 5C ) (40). On the other hand, deletion of Wnt5a significantly reduced the fraction of surviving duodenal organoids after 6 Gy of radiation ( Figure 5E and Supplementary Figure 4 ), indicating that Wnt5a is necessary for crypt regeneration following radiation. Furthermore, treatment with recombinant Wnt5a (rWnt5a) rescued Wnt5a-depleted duodenal organoids from radiation-induced ISC death ( Figure 6 ). Taken together, these results suggest that Wnt5a increases ISC survival and is both necessary and sufficient for ISC radioprotection.