Research Article

URI is required to maintain intestinal architecture during ionizing radiation

See allHide authors and affiliations

Science  31 May 2019:
Vol. 364, Issue 6443, eaaq1165
DOI: 10.1126/science.aaq1165

Illuminating intestine irradiation

High-dose radiation exposure affects multiple body systems, including the blood and the neurovascular system. Radiation can also cause severe intestinal toxicity, known as gastrointestinal syndrome (GIS). Working in mice, Chaves-Pérez et al. focused on the mechanisms underlying GIS. The molecular chaperone URI (unconventional prefoldin RPB5 interactor) labeled slow-cycling label-retaining (LR) cells, which are essential for organ regeneration following ionizing radiation. Reduced URI levels rendered LR cells highly proliferative via the activation of the β-catenin/c-MYC axis. Consequently, LR cells became radiosensitive, increasing GIS severity. Thus, URI protects LR cells to promote tissue regeneration in response to high-dose irradiation.

Science, this issue p. eaaq1165

Structured Abstract

INTRODUCTION

Exposure to high-dose irradiation (>10 gray) from the uncontrolled release of radioactive materials or intensive radiotherapy for cancer treatment can cause gastrointestinal syndrome (GIS), a lethal disorder affecting the intestinal structure. Symptoms and signs include nausea, diarrhea, vomiting, bleeding, and intestinal perforation, leading to the death of the patient through bacterial infection and septic shock. Treatments are very limited and generally focused on reducing GIS symptoms. Natural products, dietary interventions, and antioxidants have also been tested preclinically but have shown very modest effects in mitigating GIS. To assess medical countermeasures, it is essential to develop specific and robust animal models in which the relationship between radiation doses, GIS incidence, and severity can be correlated with the histopathology of the intestine. Here, we aim to understand cell biology and molecular events of GIS after radiation exposure by using a genetic GIS mouse model generated in our laboratory. This information will help in identifying biomarkers that can predict degrees of intestinal toxicity after severe ionizing radiation and assist in developing procedures to protect against GIS.

RATIONALE

High levels of DNA damage and cell death are well-known clinical features of intestinal radiation toxicity. The molecular chaperone unconventional prefoldin RPB5 interactor (URI) reportedly controls and protects genome integrity in response to environmental insults. Therefore, we hypothesized that URI may play an important role in ensuring genome stability in the highly proliferative intestinal epithelium after exposure to ionizing radiation. By using genetically engineered mouse models for URI gain- and loss-of-function specifically in the intestinal epithelium, we studied the role of URI in GIS.

RESULTS

URI levels fluctuate in intestinal crypts dose- and time-dependently after different doses of ionizing radiation. Specifically, URI increases at early postirradiation stages (24 to 48 hours), correlating with activation of the DNA damage response, whereas URI levels decrease during the regenerative response (72 to 96 hours postirradiation). We altered the level of URI and found that overexpression in mouse intestinal epithelium protects against intensive ionizing radiation and GIS, whereas halving URI expression sensitizes to ionizing radiation. Moreover, genetic URI ablation in the intestine caused mouse death, mimicking features of human GIS. We show that URI labels the slow-dividing stem cell–like label-retaining (LR) cells and is essential for protecting them from death, allowing the regeneration of the injured organ after high-dose irradiation. Mechanistically, we demonstrate that URI loss activates the β−catenin–c-MYC axis specifically in LR cells. Reducing URI expression renders LR cells proliferative, thereby accumulating c-MYC–dependent replicative stress and hence making them radiosensitive. In line with these findings, complete URI deletion causes LR cell death and disruption of intestinal architecture, leading to GIS. Therefore, URI-positive (URIhigh) slow-cycling cells within the crypts constitute the facultative stem cell pool capable of repopulating the organ after ionizing radiation.

CONCLUSION

URI labels LR cells, which are essential for intestinal regeneration after high-dose irradiation. Reducing URI levels before ionizing radiation increases β−catenin–c-MYC–dependent proliferation and replicative stress-induced DNA damage, thereby rendering LR cells radiosensitive. On the basis of this work, selective c-MYC inhibition could be used as a countermeasure for humans at risk of developing radiation-induced GIS.

Schematic summarizing the role of URI in GIS.

(Top) URI is a marker of the slow-cycling LR cells. URI-positive (URIhigh) slow-cycling cells are radioresistant. URIhigh slow-cycling cells within the crypts represent the facultative stem cell pool essential for organ regeneration after high-dose irradiation. (Bottom) Reduced URI levels (URIlow) increase proliferation and DNA damage in LR cells, rendering them radiosensitive.

Abstract

Ionizing radiation (IR) can cause gastrointestinal syndrome (GIS), a lethal disorder, by means of unknown mechanisms. We show that high-dose irradiation increases unconventional prefoldin RPB5 interactor (URI) levels in mouse intestinal crypt, but organ regeneration correlates with URI reductions. URI overexpression in intestine protects mice from radiation-induced GIS, whereas halving URI expression sensitizes mice to IR. URI specifically inhibits β-catenin in stem cell–like label-retaining (LR) cells, which are essential for organ regeneration after IR. URI reduction activates β-catenin–induced c-MYC expression, causing proliferation of and DNA damage to LR cells, rendering them radiosensitive. Therefore, URI labels LR cells which promote tissue regeneration in response to high-dose irradiation, and c-MYC inhibitors could be countermeasures for humans at risk of developing GIS.

Gastrointestinal syndrome (GIS) is a lethal intestinal disorder that occurs after exposure to high-dose irradiation [>10 gray (Gy)] after uncontrolled release of radioactive materials or intensive radiotherapy for cancer treatment (1). Between 7 and 10 days after high-dose irradiation, intestinal mucosa is disrupted, leading to watery diarrhea, dehydration and electrolyte loss, gastrointestinal bleeding, and intestinal perforation. The breakdown of the epithelial barrier favors the translocation of bacteria into the bloodstream, leading to multisystem organ failure (1, 2).

Under homeostatic conditions, rapid turnover of the intestinal epithelium is driven by leucine-rich repeat–containing G protein–coupled receptor 5–positive (Lgr5+ or Lgr5high) intestinal stem cells (ISCs) located at the base of the crypts (3). As Lgr5high ISCs divide, they migrate upward and undergo 4 to 5 replications, becoming Lgr5low progenitors or transit-amplifying (TA) cells committed to differentiation. Differentiated TA cells generate the villi, which are composed of intestinal epithelial cells (IECs), namely goblet cells, enteroendocrine cells, and nutrient-absorbing enterocytes (4). IECs continuously move upward to villus tips, where they undergo apoptosis within 5 to 7 days of formation (4). At the +4 position between bases of the crypts and TA cells, quiescent or low-proliferating “label-retaining” (LR) cells provide a reserve pool of stem cells if ISCs are depleted (4, 5). LR cells are Lgr5+ and have a secretory phenotype like Paneth cells, also located at the base of the crypts (6).

Gastrointestinal histopathological changes of chronic or high radiation doses include nuclear hyperchromasia and enlargement, mitotic arrest, death of rapidly proliferating crypt cells (ISCs and TA cells), epithelial surface area reductions, villous blunting, crypt hypoplasia, and fibrosis in submucosa, leading to ischemic lesions in the small bowel (1, 2). In less susceptible patients, intestinal tissue has a great regenerative potential when exposed to acute irradiation. The intestinal regenerative response involves the activation of slow proliferative cells participating in the initial and rapid formation of regenerative areas (RAs). Within several days, the intestine is completely restored. Why some patients are more susceptible than others remains an unsolved question, and thus, quantitative biomarkers are needed to predict GIS susceptibility and assess degrees of intestinal injury.

Highly proliferative Lgr5high ISCs are radiosensitive (79). However, several reports indicate that Lgr5+ cells are able to repopulate the crypt compartment after intense irradiation (1014), possibly by means of activation and proliferation of the LR cells, suggesting that LR cells are crucial for regeneration of the injured tissue. Additionally, after intestinal injury, LR cells give rise to the main IECs, suggesting that LR quiescent cells constitute a clonogenic reservoir of stem cells that proliferate after tissue injury (6). Further elucidation of the contributions of crypt cells and the mechanisms leading to intestinal regeneration after injury is required.

The DNA damage checkpoint protein p53 modulates signaling pathways involved in ionizing radiation (IR) responses in mammalian cells (8, 15). p53-deficient mice, which exhibit diminished crypt cell apoptosis, are still susceptible to IR injury, indicating that p53-induced apoptosis is not sufficient to cause GIS (16). Recent findings suggest that AIM2 (absent in melanoma 2) senses radiation-induced DNA damage in the nucleus, thereby inducing caspase-1–activating inflammasome formation, leading to pyroptosis and maturation of proinflammatory cytokine interleukin-1β (17). Clearly, mechanisms underlying GIS still remain poorly understood. There is thus a need to identify the genes required to maintain intestinal tissue architecture during IR and determine their function.

AbI affects URI levels in intestinal crypts

Previous studies suggest that unconventional prefoldin RPB5 interactor (URI) (18) controls and protects genome integrity in response to environmental insults (19, 20). To elucidate the role of URI in GIS, we constructed a chamber with lead plates to irradiate the abdominal cavities of mice, mimicking radiotherapy procedures (fig. S1, A to D, and materials and methods). Experiments with the device indicated that mice survived abdominal irradiation (AbI) at doses that induce death and bone marrow failure (hematopoietic syndrome) when administered as whole-body irradiation (WBI) (fig. S1, E and F).

Ketamine- and xylazine-anesthesized C57BL/6 mice placed in this device (fig. S2A) were exposed to 2- to 6-Gy AbI doses and sacrificed 24 to 182 hours later (fig. S2B), and URI levels were examined by immunofluorescence (IF), quantitative real-time polymerase chain reaction (qRT-PCR), and Western blot (WB) analyses. Low doses (2 to 6 Gy) did not affect URI expression in intestinal crypts (fig. S2, C to M). However, intense AbI (8 to 14 Gy), dose-dependently increased URI levels, mainly in upper crypts, 24 to 48 hours postirradiation (figs. S2, H and I, and S3, A to J), correlating with increases in DNA damage response markers (fig. S3, B, G, I, and J). We detected a drop in URI expression in RAs from C57BL/6 mice, 72 to 96 hours after intense AbI (8 to 14 Gy) (figs. S2, L and M, and S3, K to N). However, when intestine was fully regenerated 182 hours postirradiation, URI levels tended to normalize to homeostasis levels (fig. S3, A to G). These data demonstrate that URI levels fluctuate time- and dose-dependently in intestinal crypts and suggest a correlation between decreased URI expression, highly proliferative cells, and the formation of RAs after intense IR.

URI overexpression protects against GIS

Hypothesizing that early increases in URI levels may protect the intestine from high irradiation, we generated a knock-in (KI) mouse [designated hURI(+/KI)Int], expressing Flag-tagged human URI (hURI) via a tetracycline-dependent transactivator controlled by the villin promoter (fig. S4A). After 4 weeks on doxycycline treatment, hURI(+/KI)Int mice expressed hURI in the intestine from a single allele, at a level about six times greater than that of endogenous mouse URI (mURI). Notably, hURI(+/+)Int littermates lacked hURI expression (fig. S4, B to E).

hURI(+/+)Int and hURI(+/KI)Int mice had similar survival rates, normal body weights, and similar intestine lengths (fig. S4, F to H). However, longer villi were detected as hURI(+/KI)Int mice aged (fig. S4, I and J). Notably, hURI(+/KI)Int mice exhibited unchanged proliferative capacity, but increased differentiation, as they aged and when compared with URI(+/+)Int mice (fig. S4, K to Q).

To evaluate the responses of hURI(+/KI)Int mice to AbI, we determined the minimum AbI dose causing 70% lethality within 10 days (LD70/10), recapitulating GIS. About 70% of 22-Gy–abdominally irradiated C57BL/6 mice died within 10 days (fig. S5, A and B). hURI(+/KI)Int mice survived 22-Gy AbI for more than 25 days, whereas 70% of their control littermates died within 10 days (Fig. 1A). Moreover, losses of hURI(+/KI)Int mouse body weight and intestinal length were smaller, and, unlike hURI(+/+)Int mice, hURI(+/KI)Int mice showed no signs of diarrhea 10 days postirradiation (fig. S6, A to C). Bromodeoxyuridine-positive (BrdU+) surviving crypts were increased in hURI(+/KI)Int mice 96 hours post-AbI (fig. S6D). Additionally, 22-Gy–abdominally irradiated hURI(+/KI)Int mice had normal intestinal architecture with better regenerative capacity and increased differentiation than irradiated hURI(+/+)Int mice (Fig. 1B and fig. S6, E to I). Thus, hURI overexpression protects intestines from lethal IR doses. However, increasing radiation doses to 26 Gy caused similar body weight loss, intestinal length reduction, diarrhea, and death in hURI(+/+)Int and hURI(+/KI)Int mice (Fig. 1, A and B, and fig. S6, A to C).

Fig. 1 URI safeguards intestine after high-dose irradiation.

(A) Kaplan-Meier curve of nonirradiated hURI(+/+)Int and hURI(+/KI)Int mice, and 22- and 26-Gy–abdominally irradiated hURI(+/+)Int and hURI(+/KI)Int mice. (B) Hematoxylin and eosin (H&E) staining and immunohistochemistry analysis in nonirradiated hURIInt and 22- and 26-Gy–abdominally irradiated hURIInt intestinal sections at time of death (8 to 10 days). (C) WB of intestines from URIInt mice after 15 days of tamoxifen treatment. Numbers to the left of the blot are in kDa. (D) Kaplan-Meier curve of URI(+/+)Int and URI(+/Δ)Int mice after 15 days of tamoxifen treatment, bone marrow–transplanted (BMT) URI(+/+)Int mice, URI(+/Δ)Int) mice, and 22-Gy–abdominally irradiated URI(+/+)Int and URI(+/Δ)Int mice. (E) H&E staining in intestinal sections from bone marrow–transplanted and 22-Gy–abdominally irradiated URIInt mice. Data represent mean ± SEM. In (A) and (D), **P < 0.01 and ***P < 0.001 by Mantel-Cox test analysis.

URI loss causes intestinal failure

Next, we crossed URI conditional knockout (URI lox) and Villin-creERT2 mice to explore the consequences of knocking out URI expression in intestinal epithelium (fig. S7, A to C). Activation of creERT2 recombinase under control of the Villin promoter was induced by dietary provision of tamoxifen to 8-week-old mice, yielding URI(+/+)Int, URI(+/Δ)Int, and URI(Δ/Δ)Int mice (where Δ indicates deletion). URI was specifically ablated in intestinal epithelium after 2 and 6 days in URI(Δ/Δ)Int mice and, after 15 days in URI(+/Δ)Int mice (fig. S7, D to F). URI(Δ/Δ)Int mice died within 7 to 9 days of tamoxifen treatment, like 22-Gy–abdominally irradiated C57BL/6 mice, mimicking GIS, but unlike their URI(+/+)Int or heterozygous URI(+/Δ)Int littermates (fig. S7, G to H). URI(Δ/Δ)Int mice also had reduced body weight, intestinal length, and blood glucose levels (similar to 22-Gy–abdominally irradiated C57BL/6 mice), whereas no clear phenotypic changes were observed in URI(+/+)Int and URI(+/Δ)Int mice at this stage (fig. S7, I to K). Both URI(Δ/Δ)Int and 22-Gy–abdominally irradiated C57BL/6 mice had dilated stomachs with accumulated food and grossly swollen small intestines with no luminal content at death (fig. S7L), suggesting normal food access but possible gastric outlet obstruction.

After 6 days of tamoxifen treatment, URI(Δ/Δ)Int mice had normal intestinal tissue architecture, but their crypts showed nuclear hyperchromasia, cell flattening, and signs of cell death (fig. S7, M and N). After 7 to 9 days, they showed drastic reductions in villus length, crypt hypoplasia, destruction of intestinal architecture, and loss of epithelial surface, as also seen in 22-Gy–abdominally irradiated C57BL/6 mice (fig. S7, M to O). Increased sirius red, Masson trichrome, and reticulin staining in both models revealed fibrotic intestines (fig. S8, A and B).

Moreover, compared with control littermates, URI(Δ/Δ)Int mice exhibited defects in terminally differentiated enterocytes, goblet cells, and enteroendocrine cells after 6 days of tamoxifen treatment (fig. S9, A to H). Lysozyme staining revealed no differences in Paneth cell number between URI(+/+) Int and URI(Δ/Δ)Int mice (fig. S9, I to K). Furthermore, local and systemic inflammation (fig. S10, A and B) and signs of bacteremia (fig. S10, C to F) were detected in URI(Δ/Δ)Int mice at death (but not at 2 days after tamoxifen treatment) and 22-Gy–abdominally irradiated C57BL/6 mice at 72 hours postirradiation. Finally, fluorescein isothiocyanate–dextran (FITC-dextran) assays and IF analysis of β-actin and zonula occludens (ZO-1) in URI(Δ/Δ)Int intestines (fig. S10, G and H) indicated loss of epithelial integrity in both models. These findings suggest that URI loss disrupts murine intestinal architecture, possibly leading to septic shock and death of mice. Thus, URI(Δ/Δ)Int mice seem to recapitulate GIS.

Halving URI sensitizes to GIS

Additional markers of stress were detected in intestinal epithelium of URI(Δ/Δ)Int and 22-Gy–abdominally irradiated C57BL/6 mice (fig. S11, A to J). URI(+/Δ)Int mice exhibited increased DNA damage response markers in the intestine after 15 days of tamoxifen treatment (fig. S12, A to E). Furthermore, proliferation markers were increased in URI(+/Δ)Int crypts (fig. S12, F to I).

To check whether URI(+/Δ)Int mice were sensitized to IR, we abdominally irradiated URI(+/+)Int and URI(+/Δ)Int mice with 6, 10, and 14 Gy (fig. S13A). At these doses, mouse survival was not affected after AbI (fig. S13B). However, high AbI doses (14 Gy) significantly reduced URI(+/Δ)Int mouse body weight, intestinal length, and villi length and affected intestinal architecture when compared with nonirradiated URI(+/Δ)Int mice and abdominally irradiated URI(+/+)Int littermate controls (fig. S13, C to G). Additionally, increased intestinal permeability was detected in URI(+/Δ)Int mice when compared with their littermate controls, but these effects were further amplified in abdominally irradiated URI(+/Δ)Int mice (fig. S13 H).

Irradiation with 6 and 10 Gy significantly increased DNA damage and proliferation in abdominally irradiated URI(+/Δ)Int mice when compared with nonirradiated URI(+/Δ)Int mice and irradiated URI(+/+)Int littermate controls, 96 hours postirradiation (fig. S14, A to D). Notably, 14-Gy AbI doses in URI(+/Δ)Int mice further amplified DNA damage increases when compared with 6- and 10-Gy doses (fig. S14, A and B). Consequently, intestinal regenerative capacity was impaired in 14-Gy–abdominally irradiated URI(+/Δ)Int mice (fig. S14, D and E), which also had lower differentiation capacity when compared with nonirradiated URI(+/Δ)Int mice (fig. S14, F to I). Thus, effects seen in URI(+/Δ)Int mice are further strengthened after AbI (fig. S14J), indicating that URI(+/Δ)Int mice are sensitized to IR.

Therefore, we checked effects of the LD70/10 dose in URI(+/Δ)Int mice. When exposed to 22-Gy AbI, these mice presented signs of severe diarrhea and lower survival rates, body weight, and intestinal RAs than URI(+/+)Int littermate controls (Fig. 1, C to E, and fig. S15, A and C). To check that 22-Gy–abdominally irradiated mice died from GIS rather than hematopoietic syndrome, bone marrow from Tg.CAG–EGFP (Tg., transgenic; CAG, CMV enhancer, chicken β-actin promoter and rabbit β-globin splice acceptor site; EGFP, enhanced green fluorescent protein) mice was transplanted into 22-Gy WBI URI(+/+)Int and URI(+/Δ)Int mice (fig. S15, D and E). Transplanted mice had similar phenotypes to 22-Gy–abdominally irradiated mice (Fig. 1, D and E, and fig. S15, B and C). Thus, URI(+/Δ)Int mice are sensitized to GIS and have a reduced life span.

URI is expressed in Lgr5low cells

The above findings suggest that URI loss may specifically affect intestinal cells required for tissue regeneration after injury. IF analysis of intestines of Lgr5-EGFP-IRES-creERT2 (IRES, internal ribosome entry site; creERT2, Cre recombinase estrogen receptor type 2) and C57BL/6 mice showed that URI was predominantly expressed in the upper part of the crypts, colocalizing with its binding partner heat shock protein 90 (HSP90) (18), and not in Sox9+ cells, Paneth cells, or villi (fig. S16, A to G, and movie S1). Separation of Lgr5low (TA) cells and Lgr5high ISCs from Lgr5-EGFP-IRES-creERT2 mice by fluorescence-activated cell sorting confirmed enrichment of URI in Lgr5low cells (fig. S16, H to J).

To further check URI localization, we generated a URI-creERT2-IRES-EGFP knock-in mouse model carrying a creERT2-IRES-EGFP cassette under control of the Uri promoter (fig. S17, A to E). Heterozygous mice harbor a URI-creERT2-IRES-EGFP knock-in allele that abolishes Uri gene function and expresses creERT2 and EGFP proteins. We detected no EGFP signal in URI-creERT2-IRES-EGFP mice by standard IF (fig. S17F), possibly because the IRES element may affect EGFP expression. However, staining for cre recombinase confirmed that URI was mainly localized in the upper part of the crypts and not in Paneth cells (fig. S17G).

Slow-cycling cells are URI+

To check the role of URI+ cells during homeostasis, we crossed URI-creERT2-IRES-EGFP and reporter R26-stop–EYFP (enhanced yellow fluorescent protein) mice to generate URI-creERT2;R26-stop-EYFP mice (Fig. 2A). URI+ cells expressed an inducible cre recombinase that specifically deleted the loxP-flanked stop cassette in the Rosa26R reporter locus in the URI-expressing cells of the offspring. All descendants of the labeled cells thus inherited EYFP. IF analysis from mice injected with 4-hydroxytamoxifen (4-OHT) for five consecutive days showed that EYFP cells were present in the upper part of the crypts (Fig. 2, B and C). Moreover, villi originating from EYFP+ TA cells were also EYFP+ (Fig. 2, B and C), suggesting full recombination.

Fig. 2 URI labels intestinal LR cells.

(A) Schematic showing the cross between URI-creERT2-IRES-EGFP and R26-stop-EYFP mice. (B) Experimental design to express EYFP in URI+ cells. (C and D) IF (C) and quantification (D) of EYFP in intestinal sections from experiments described in (B). In (C), the orange arrow indicates URI-positive slow-cycling cells. Dapi, 4′,6-diamidino-2-phenylindole. (E and F) Co-IF (E) and quantification (F) of EYFP and Ki67 in intestinal sections from the labeling experiment in URI-creERT2;R26-stop-EYFP mice. (G) Schematic showing the cross between URI-creERT2-IRES-EGFP and CAG-stop-Katushka mice. (H) Experimental design to express Katushka in URI+ cells. (I) IF and quantification of Katushka (Kat) in intestinal sections from the labeling experiment in URI-creERT2;CAG-Katushka mice. (J) Co-IF and quantification of Katushka and Ki67 in intestinal sections from the labeling experiment in URI-creERT2;CAG-Katushka mice. (K) IF and quantification of Katushka in intestinal sections from the tracing experiment in URI-creERT2;CAG-Katushka mice. (L and M) qRT-PCR of Lgr5 (L) and proliferation markers (Mki67 and Aurkb) (M) in sorted Katushka+ and Katushka cells. Sorted Lgr5high and Lgr5low cells from Lgr5-GFP-IRES-creERT2 mice were used as controls (n = 3 per group). (N) Co-IF of H2B-GFP and URI in intestinal sections from H2B-GFPInt mice. The orange arrows indicate LR cells. (O) qRT-PCR of intestinal GFP+ and GFP cells sorted from H2B-GFPInt mice (n = 3 per group). (P) qRT-PCR of Lgr5high and Lgr5low cells and LR cells sorted from Lgr5-GFP-IRES-CreERT2 and H2B-GFPInt mice, respectively (n = 3 per group). (Q) Co-IF of Lgr5-GFP and URI in intestinal sections from Lgr5-GFP-IRES-CreERT2 mice. White and orange arrows show Lgr5+URI cells and Lgr5+URI+ cells, respectively. (R) Crypt schematic depicting URI location. Data represent mean ± SEM. In (L), (M), (O), and (P), *P < 0.05 and **P < 0.01 by Student’s t test analysis. Quantification was performed for more than 50 crypts per mouse in three independent mice. Dashed white lines represent crypts.

Next, we labeled URI+ cells in URI-creERT2;R26-stop-EYFP mice that were injected with 4-OHT for five consecutive days, and then tamoxifen treatment was stopped for 1 week to enable complete organ regeneration (Fig. 2, A and B). IF showed that 95% of EYFP+ cells in the crypt were slow-cycling cells (Fig. 2, C to F), most likely corresponding to LR cells. Notably, because Lgr5high ISCs were EYFP before tracing (Fig. 2C), all descendant cells from Lgr5high ISCs were also EYFP (Fig. 2C), suggesting that highly proliferative Lgr5high ISCs are the cells at the origin of organ regeneration, excluding the role of slow-cycling URI+ cells under homeostatic conditions.

We also crossed the URI-creERT2-IRES-EGFP mouse with the transgenic reporter CAG-Katushka mouse, which expresses the far-red fluorescent protein Katushka, driven by the hybrid CAG promoter upon cre-mediated recombination. To label URI+ cells, URI-creERT2;CAG-Katushka mice were injected with 4-OHT for five consecutive days (Fig. 2, G and H). Katushka+ cells were detected in the upper part of the crypts (Fig. 2I), and endogenous EGFP was detected after signal amplification colocalized with Katushka+ cells (fig. S17H). Moreover, slow-cycling cells were Katushka+ (Fig. 2J). Stopping tamoxifen treatment for 1 week to allow organ regeneration and qRT-PCR analysis of sorted Katushka+ cells indicated that these cells were URI+Lgr5medium and were slow-cycling cells, most likely corresponding to LR cells (Fig. 2, K to M, and fig. S17I).

URI labels intestinal LR cells

To test whether LR cells express URI, we performed pulse-chase labeling with a tetracycline operator (TetO)–regulated H2B–GFP (histone 2B–green fluorescent protein) fusion protein. TetO-H2B-GFP (H2B-GFP) mice were crossed with Villin-rtTA2-M2 transgenic mice, generating H2B-GFPInt mice, in which H2B-GFP can be pulse-chased during aging (fig. S18, A and B). H2B-GFP protein expression was monitored after 2 weeks of doxycycline (Doxy ON) pulsing (fig. S18C) and a subsequent 3-week (Doxy OFF) chase in H2B-GFPInt mice aged 8 to 10 weeks, revealing that 74.67 ± 0.56% of H2B-GFP+ cells were URI-positive (URI+) (Fig. 2N and fig. S18D). Notably, some H2B-GFP+ cells were also URI, possibly Paneth cells, which are replaced every 20 to 23 days (fig. S18, D and E). Sorting of LR and non-LR cells confirmed that LR cells were URI+ (Fig. 2O and fig. S18F). Furthermore, Lgr5 expression was evaluated by qRT-PCR in sorted LR cells from H2B-GFP mice and compared with sorted Lgr5low (TA) cells and Lgr5high ISCs from Lgr5-EGFP-IRES-creERT2 mice, demonstrating that LR cells express a medium level of Lgr5 (Lgr5medium cells). So, URI+ LR cells might be considered as Lgr5+ or Lgr5medium cells (Fig. 2P). Additional studies on Lgr5-EGFP-IRES-creERT2 mice revealed a discrete URI-expressing subpopulation in Lgr5+ cells (Fig. 2, Q and R). Thus, URI marks LR cells under homeostatic conditions.

URI+ LR cells are radioresistant

To analyze the behavior of URI+Lgr5+ cells after IR, Lgr5-EGFP-IRES-creERT2 mice were exposed to 14- and 22-Gy irradiation and sacrificed 24 hours later (Fig. 3A). The GFP signal was significantly reduced in most intestinal cells, but co-IF analysis revealed some surviving GFP+ cells expressing URI after 14-Gy irradiation (Fig. 3B and fig. S19A), suggesting that 14-Gy-IR–resistant Lgr5+ cells are URI+. A lethal dose of 22-Gy AbI in Lgr5-EGFP-IRES-creERT2 mice blunted GFP+ signals (Fig. 3B and fig. S19A).

Fig. 3 URI+ cells repopulate the intestine after IR.

(A) Schematic showing the AbI procedure for Lgr5-GFP-IRES-CreERT2 mice. (B) Representative co-IF (left) for Lgr5-GFP and URI and quantifications (right) in intestinal sections from nonirradiated, 14- and 22-Gy–abdominally irradiated Lgr5-GFP-IRES-CreERT2 mice 24 hours postirradiation. (C) Schematic showing the experiment for pulse-chased and irradiated H2B-GFPInt mice. (D) Representative co-IF (left) of H2B-GFP and URI and quantifications (right) in intestinal sections from nonirradiated, 14- and 22-Gy–abdominally irradiated H2B-GFPInt mice at the indicated time points postirradiation. (E) Experimental design for tracing URI+ cells after AbI. (F) Representative IF of EYFP and quantification of URI+ cells in intestinal sections from labeled or 14-Gy–abdominally irradiated URI-creERT2;R26-stop-EYFP mice. (G) Representative IF of EYFP and Ki67 and quantification of URI+KI67+ cells. (H) Representative co-IF and quantification of EYFP, Sox9, and lysozyme in intestinal sections from nonirradiated and 14-Gy–abdominally irradiated URI-creERT2;R26-stop-EYFP mice 96 hours postirradiation. (I and J) Representative Alcian blue–periodic acid Schiff (PAS) (I) and alkaline phosphatase (J) staining and EYFP IF, as well as quantification, in consecutive intestinal sections from nonirradiated and 14-Gy–abdominally irradiated URI-creERT2;R26-stop-EYFP mice 96 hours postirradiation. Tam, tamoxifen. (K) Schematic showing the contribution of URI+ cells to intestinal regeneration after 14-Gy AbI. Data represent mean ± SEM. In (B) and (D), **P < 0.01 and ***P < 0.001 by Student’s t test analysis. Quantification was performed over 50 crypts or 100 villi per mouse in three independent mice. Dashed white lines represent RAs.

We next checked the effects of IR on LR cells in H2B-GFPInt mice (Fig. 3C). Numbers of H2B-GFP+ cells were unchanged in 14-Gy irradiated H2B-GFPInt mice 24 hours postirradiation, and surviving LR cells expressed URI (Fig. 3D and fig. S19B), suggesting that URI+ LR cells are radioresistant. However, 22-Gy irradiation killed all GFP signals (Fig. 3, C and D, and fig. S19B). Notably, at 48 to 72 hours after 14-Gy irradiation, reduced URI levels were detected specifically in highly proliferative LR cells on the basis of co-IF analysis (Fig. 3, C and D, and fig. S19C), suggesting that reduced URI expression in quiescent LR cells may render these cells highly proliferative.

URI+ cells repopulate the intestine after IR

To check whether URI+ cells (TA and slow-cycling URI+ cells) regenerated the intestinal epithelium after AbI, URI-creERT2;R26-stop-EYFP mice were injected with 4-OHT for five consecutive days before 14-Gy IR (fig. S19D). IF analysis indicated that 97.27 ± 1.42% of cells in RAs were EYFP+ 96 hours postirradiation (fig. S19D). Thus, URI+ cells (TA and/or slow-cycling URI+ cells) regenerate the intestine after 14-Gy IR.

To determine the contribution of slow-cycling URI+ cells in intestinal regeneration, URI-creERT2;R26-stop-EYFP mice were injected with 4-OHT for five consecutive days, tamoxifen treatment was stopped for 1 week, and then the mice were finally subjected to 14-Gy AbI (Fig. 3E). EYFP+ cells (93.7%) were detected in RAs and in incipient villi 96 hours postirradiation (Fig. 3, F and G). Further IF analysis in RAs indicated that URI+ (EYFP+) cells gave raise to different cell populations, representing cell lineages present in the crypts during homeostasis (Fig. 3H). Lgr5high ISCs may then emanate from the slow-cycling URI+ cells after IR. Additionally, EYFP+ cells stained for differentiation markers in the newly formed villi (Fig. 3, I and J). Similar results were observed with 14-Gy–abdominally irradiated URI-creERT2;CAG-Katushka mice (fig. S19, E to I). Thus, slow-cycling URI+ cells, most likely LR cells, constitute a predominant facultative stem cell pool capable of repopulating the organ after injury and when Lgr5high ISCs are damaged (Fig. 3K).

URI protects LR cells from IR

Next, to check whether reduced URI levels sensitized LR cells to IR, we crossed URI(+/Δ)Int and URI(Δ/Δ)Int mice with H2B-GFPInt mice, generating URI(+/Δ)Int-H2B-GFP and URI(Δ/Δ)Int-H2B-GFP mice, respectively. IF analysis of crypts and qRT-PCR of sorted LR cells from URI(+/Δ)Int-H2B-GFP mice after 15 days of tamoxifen treatment confirmed that halving URI levels rendered LR cells more proliferative, accumulating replicative stress and DNA damage in the absence of IR (Fig. 4, A to C, and fig. S20, A to C). When we administered 14-Gy AbI doses in URI(+/Δ)Int-H2B-GFP mice, we detected apoptotic LR cells within 24 hours, significantly reducing LR cell counts (Fig. 4, D to F), although this dose did not kill LR cells in URI(+/+)Int-H2B-GFP mice. Thus, reduced URI expression in LR cells increases proliferation and DNA damage, rendering these cells radiosensitive. IF analysis indicated that LR cells from URI(Δ/Δ)Int-H2B-GFP mice were apoptotic after 4 days of tamoxifen treatment (Fig. 4, G to I).

Fig. 4 URI protects LR cells from IR.

(A) Schematic showing the experiment for pulse-chased URI(+/Δ)Int-H2B-GFP mice. (B and C) Co-IF of H2B-GFP and Ki67 in intestinal sections (B) and qRT-PCR of Mki67 of sorted LR cells (C) from pulse-chased URIInt-H2B-GFP mice after 15 days of tamoxifen treatment (n = 3 per group). (D) Schematic showing the experiment for pulse-chase labeling followed by 14-Gy AbI in hURIH2B-GFP mice. (E and F) Co-IF of H2B-GFP and cleaved caspase-3 (Cl. Casp 3) (E) and quantification of H2B-GFP+ cells per crypt and apoptotic LR cells per crypt (F) in intestinal sections from nonirradiated and 14-Gy–abdominally irradiated hURIH2B-GFP mice. (G) Schematic showing the experiment for pulse-chased URI(Δ/Δ)Int-H2B-GFP mice. (H and I) Co-IF of H2B-GFP and Cl. Casp3 (H) and quantification (I) in intestinal sections from pulse-chased URIInt-H2B-GFP mice. (J) Schematic showing pulse-chase labeling followed by 22-Gy AbI in hURIInt-H2B-GFP mice. i.p., intraperitoneal administration. (K and L) Co-IF of H2B-GFP and Cl. Casp3 (K) and quantification of apoptotic LR cells (L) in intestinal sections from nonirradiated and 22-Gy–abdominally irradiated hURIInt-H2B-GFP mice. (M) WB of hURIInt-H2B-GFP intestines after intraperitoneal injections of doxycycline. (N) Schematic showing the experiment for 22-Gy–abdominally irradiated hURIInt-Lgr5-EGFP mice sacrificed 12 hours postirradiation. (O and P) Co-IF of H2B-GFP and Cl. Casp3 (O) and quantification of apoptotic Lgr5high cells (P) in intestinal sections from 22-Gy–abdominally irradiated hURIInt-Lgr5-EGFP mice 12 hours postirradiation. Data represent mean ± SEM. In (C), (F), (I), (L), and (P), *P < 0.05 and **P < 0.01 by Student’s t test analysis. Quantification was performed over 50 crypts per mouse in three independent mice. Dashed white lines represent RAs.

We therefore determined whether URI overexpression protects LR cells against IR by crossing hURI(+/KI)Int and H2B-GFP mice to obtain hURI(+/KI)Int-H2B-GFP mice. To this end, we delivered 22-Gy AbI doses because, at this dose, LR cells died (Fig. 3, C and D). LR cells survived 22-Gy AbI in hURI(+/KI)Int-H2B-GFP mice, but not in hURI(+/+)Int-H2B-GFP mice (Fig. 4, J to M), demonstrating that the increased regenerative capacity of 22-Gy–abdominally irradiated hURI(+/KI)Int mice described in Fig. 1C is due to an increased number of surviving LR cells. We therefore checked whether URI might protect against IR by interfering with DNA repair machinery by means of either nonhomologous end joining (NHEJ) or homologous recombination (HR). To test this possibility, HCT-116 (human colon cancer) cells were stably transfected with fluorescent reporter constructs containing a GFP gene with recognition sites for a rare-cutting I-SceI endonuclease to induce double-strand breaks (21). Successful repair of the I-SceI–induced breaks by NHEJ or HR restores the functional GFP gene, and numbers of GFP-positive cells detected by flow cytometry provide quantitative measures of NHEJ or HR efficiency. Applying this method to HCT-116 cells in which URI was silenced or overexpressed revealed that URI is involved in double-strand break repair by NHEJ (fig. S21, A to F).

We also analyzed the behavior of Lgr5high ISCs after IR in hURI(+/KI)Int mice. Hence, we crossed Lgr5-EGFP-IRES-creERT2 and hURI(+/KI)Int mice to generate hURI(+/KI)Int-Lgr5-EGFP mice. IF analysis indicated that Lgr5high ISCs died in 22-Gy–abdominally irradiated hURI(+/KI)Int-Lgr5-EGFP mice (Fig. 4, N to P). Thus, URI overexpression specifically radio-protects LR cells, but not highly proliferative Lgr5high ISCs.

To determine whether URI loss also affects the proliferation status of TA cells, we crossed URI(+/+)Int and URI(Δ/Δ)Int mice with Lgr5-EGFP-IRES-creERT2 mice, generating URI(+/+)Int-Lgr5-EGFP and URI(Δ/Δ)Int-Lgr5-EGFP offspring, respectively (fig. S22, A to C). Analysis of isolated Lgr5low TA cells from URI(Δ/Δ) Int-Lgr5-EGFP mice demonstrated that URI reductions increased TA proliferation (fig. S22, D and E) and reduced levels of lineage-specific transcription factors in TA cells after 6 days of tamoxifen treatment (fig. S22F). Moreover, despite p53 stabilization, Lgr5low TA cells from URI(Δ/Δ) Int-Lgr5-EGFP mice exhibited enhanced BrdU incorporation, cyclin D1 levels, and retinoblastoma protein phosphorylation after 6 days of tamoxifen treatment (fig. S22, G to I). Further IF analyses indicated that Lgr5low TA cells from URI(Δ/Δ) Int-Lgr5-EGFP mice entered apoptosis, but not Lgr5high ISCs (fig. S22, J to L). Thus, proliferative TA cells may accumulate DNA damage and die after URI loss. Notably, co-staining of p53 and Ki67 (a cell proliferation marker) in URI(Δ/Δ)Int mice indicated that some TA cells proliferate and may escape from p53-induced cell cycle arrest and cell death (fig. S22M). Consequently, Lgr5low cell numbers did not change in URI(Δ/Δ)Int-Lgr5-EGFP mice (fig. S22N), possibly because of cell death and proliferation, both of which normalized cell number after 6 days of tamoxifen treatment.

URI loss activates the β-catenin–c-MYC axis in LR cells

Increased proliferation and decreased differentiation suggested that reducing URI activates WNT–β-catenin signaling in TA and LR cells. Accordingly, analysis of gene expression in LR cells sorted from URI(+/+)Int-H2B-GFP and URI(+/Δ)Int-H2B-GFP mice revealed significant increases in ISC markers (Olfm4 and Lgr5) and activation of WNT–β-catenin target genes (c-Myc and Ascl2) in URI(+/Δ)Int-H2B-GFP mice after 15 days of tamoxifen treatment (Fig. 5, A and B). IF analysis confirmed expression of c-MYC and Axin 2 in URI(+/Δ)Int-H2B-GFP LR cells (Fig. 5, C to F). Similar patterns were observed in Lgr5low (TA) cells sorted from URI(+/+)Int-Lgr5-EGFP and URI(Δ/Δ)Int-Lgr5-EGFP mice (fig. S23, A and B). WNT–β-catenin activation was corroborated by WB and immunohistochemistry analyses from URI(Δ/Δ)Int crypts (fig. S23, C and D). Notably, c-Myc expression was increased in the early stages of URI deletion (2 days), but apoptosis was not detected, correlating with increased proliferation and cell number (figs. S11D and S23, E and F). Additionally, in HCT-116 cells, URI down-regulation (using URI-silencing RNA) increased Axin 2 and c-MYC expression, whereas HA (hemagglutinin)–URI overexpression reduced their levels (fig. S23, G to J). Thus, URI loss activates WNT–β-catenin signaling pathway in TA-LR cells. Accordingly, URI overexpression in Lgr5high ISCs [hURI(+/KI)Int-Lgr5-EGFP] decreased expression of WNT–β-catenin target genes and stemness (fig. S23, K and L) without affecting numbers of cells per crypt (fig. S23M).

Fig. 5 URI loss activates β-catenin–c-MYC axis in LR cells.

(A and B) qRT-PCR of Lgr5 and Olfm4 (A) and Ascl2 and c-Myc (B) mRNA levels in LR cells sorted from URIInt-H2B-GFP mice after 15 days of tamoxifen treatment (n = 3 per group). (C and D) Co-IF (C) and quantification (D) of H2B-GFP and c-MYC in intestinal sections from nonirradiated and 14-Gy–abdominally irradiated URIInt-H2B-GFP mice after 15 days of tamoxifen treatment. Insets in (C) are magnified views. nd, not detected. (E and F) Co-IF (E) and quantification (F) of H2B-GFP and Axin2 in intestinal sections from nonirradiated and 14-Gy–abdominally irradiated URIInt-H2B-GFP mice after 15 days of tamoxifen treatment. Insets in (E) are magnified views. Data represent mean ± SEM. In (A), (B), (D), and (F), *P < 0.05 and **P < 0.01 by Student’s t test analysis. Quantification was performed over 50 crypts per mouse in three independent mice. Dashed white lines represent crypts in nonirradiated mice and RAs in 14-Gy-abdominally irradiated mice.

URI binds and inhibits β-catenin nuclear translocation

Cytosolic biochemical fractionation of HCT-116 cells and crypts isolated from URI(+/+)Int and URI(Δ/Δ)Int mice after 6 days of tamoxifen treatment demonstrated that URI depletion increased nuclear localization of β-catenin (Fig. 6A and fig. S24A). Notably, β-catenin stability was not affected in URI(Δ/Δ)Int crypts (fig. S24B). TOP-FOP reporter assays of β-catenin transcriptional activity in HCT-116 cells demonstrated that URI down-regulation increased luciferase activity, whereas overexpression of HA-URI diminished it (Fig. 6B and fig. S24C). Thus, URI loss in TA-LR cells may selectively activate the canonical WNT–β-catenin pathway, possibly inducing TA-LR cell proliferation and stemness.

Fig. 6 URI binds and inhibits β-catenin nuclear translocation.

(A) WB of nuclear and cytosolic fractions of crypts from URIInt-H2B-GFP mice after 6 days of tamoxifen treatment. Numbers to the left of the blot are in kDa. Tubulin and lamin were used as loading controls. (B) TOP-FOP reporter assay in HCT-116 cells 48 hours after transfection with either small interfering RNA (siRNA) control (siCtrl) or siRNA against URI (siURI) and treated with or without R-spondin 1 (n = 3). RLU, relative luminescence units. (C and D) β-catenin (C) and URI (D) immunoprecipitation in crypts from C57BL/6 mice. WCE, whole-cell extract; IgG, immunoglobulin G. Numbers to the left of the blots are in kDa. (E) Pull-down of IVT full-length HA-URI (residues 1 to 534) and HA-URI fragments (residue ranges indicated) using GST or GST–β-catenin. Numbers to the left of the blots are in kDa. (F) Pull-down of IVT full-length β-catenin (residues 1 to 781), β-catenin mutants (S33Y) and (deltaN47) using full-length GST-URI (residues 1 to 534) or GST-URI fragments (residues 1 to 158 and 156 to 534). Numbers to the left of the blots are in kDa. (G) Schematic showing β-catenin binding to URI and vice versa. N, N terminus; C, C terminus. (H and I) Schematics showing the cross between the URI-creERT2-IRES-EGFP mouse and the Catnb+/lox(ex3) mouse (H) and between the URI lox;Villin-creERT2 mouse and the Catnb+/lox(ex3) mouse (I). (J) Kaplan-Meier curve of Catnb(+/+)TA/LR mice, Catnb(+/Δ)TA/LR mice, Catnb(+/Δ)Int;URI(+/+)Int mice, and Catnb(+/Δ)Int;URI(+/Δ)Int mice. (K) H&E staining from Catnb(+/+)TA/LR, Catnb(+/Δ)TA/LR , Catnb(+/Δ)Int;URI(+/+)Int, and Catnb(+/Δ)Int;URI(+/Δ)Int mice at indicated time points of tamoxifen treatment. Yellow dashed lines and black arrows represent crypt and villi, respectively. Data represent mean ± SEM. In (B) and (J), *P < 0.05, **P < 0.01, and ***P < 0.001 by Student’s t test analysis (B) and Mantel-Cox test analysis (J).

URI is part of a complex that inhibits several enzymes and transcription factors (18, 22, 23). We thus checked whether URI inhibits β-catenin activity by means of direct physical interaction that abolishes its nuclear translocation. In reciprocal immunoprecipitation assays, endogenous URI and β-catenin interacted in C57BL/6 crypts and HCT-116 cells (Fig. 6, C and D, and fig. S24, D and E). We therefore evaluated the direct binding of URI and β-catenin in vitro by incubating recombinant glutathione S-transferase (GST) or GST-tagged β-catenin with in vitro translated (IVT) full-length HA-URI or selected HA-URI fragments. Pull-down assays revealed that URI’s prefoldin N-terminal (Nt) domain (residues 1 to 193) interacted with β-catenin (Fig. 6E). This interaction was verified by incubating GST-URI or GST-URI fragments with IVT full-length β-catenin or various Flag-tagged β-catenin mutants. Full-length β-catenin bound to URI’s Nt prefoldin domain (residues 1 to 158) (fig. S24, F to H). An Nt-truncated β-catenin mutant (deltaN47) with high T cell factor (TCF) activity did not interact with URI. Additionally, URI bound less strongly to the β-catenin mutant S33Y (Ser33→Tyr), which exhibits enhanced β-catenin activity (Fig. 6F and fig. S24, I and J). Thus, β-catenin binds to URI’s Nt and vice versa (Fig. 6G), suggesting that direct binding to URI inhibits the nuclear translocation of β-catenin.

To check whether β-catenin activation in TA-LR cells disrupts intestinal architecture, recapitulating features of URI(Δ/Δ)Int mice and GIS, we crossed URI-creERT2-IRES-EGFP mice with a conditional β-catenin–targeted mouse strain, Catnb+/lox(ex3), generating Catnb(+/Δ)TA/LR mice with constitutive activation of β-catenin in TA-LR cells, owing to the absence of glycogen synthase kinase 3 (GSK3) phosphorylation sites in β-catenin’s Nt domain and URI binding site after tamoxifen treatment (Fig. 6H). Catnb(+/Δ)TA/LR mice died after 8 days of tamoxifen treatment, and Catnb(+/Δ)TA/LR intestines showed crypt hypoplasia with evidence of DNA damage, fibrosis, and reduced numbers of villi (Fig. 6, J and K, and fig. S24K), recapitulating features of URI(Δ/Δ)Int mice. Mice expressing activated β-catenin throughout their intestines, obtained by crossing Catnb+/lox(ex3) and Villin-creERT2 mice [Catn(+/Δ)Int], survived and exhibited mild crypt hyperplasia, but no defects in intestinal structure, after 15 days of tamoxifen treatment, possibly because of β-catenin overactivation in the mitotically active Lgr5high ISCs (Fig. 6, I to K). Shortened villi with loss of absorptive enterocytes and more pronounced crypt hyperplasia were detected in Catn(+/Δ)Int;URI(+/Δ)Int mice [generated by crossing Catn(+/lox)Int and URI lox mice] after 15 days of tamoxifen treatment (Fig. 6, I to K). The presence of URI thus modulates compartment-dependent differential activation of β-catenin, leading to different phenotypes.

Discussion

We show here that URI labels quiescent LR cells to maintain their identity and organ architecture during IR. Halving URI expression renders LR cells highly proliferative, and hence radiosensitive (Fig. 7), whereas complete URI reductions induce LR cell death and impairment of intestinal regenerative capacity. In line with these findings, URI overexpression in mice [hURI(+/KI)Int mice] protects LR cells from lethal irradiation doses, allowing a better regenerative capacity after intense irradiation. Although URI+ LR cells survive IR in hURI(+/KI)Int mice, they should also proliferate to allow organ regeneration. URI levels in RAs of irradiated hURI(+/KI)Int mice were similar to URI levels in RAs of hURI(+/+)Int mice (not shown). Thus, URI might be degraded by means of unknown mechanisms in hURI(+/KI)Int mice after IR, allowing active proliferation in RAs. We also do not exclude that IR-induced paracrine signals in hURI(+/+)Int mice can affect LR cell fate in irradiated hURI(+/KI)Int mice.

Fig. 7 Mechanisms of GIS.

Schematics depicting the mechanisms of GIS after URI deletion in intestinal epithelium.

Our data argue that TA-LR cells in URI(+/Δ)Int mice have stem cell–like traits, with high (potentially tumorigenic) DNA damage. URI binds less to the β-catenin mutants (deltaN47 and SY33), which are found in neoplastic intestinal lesions and which display high TCF activation (24), suggesting that URI loss may have a carcinogenic potential through β-catenin–c-MYC activation. It is also tempting to speculate that low URI expression in Lgr5high ISCs may be responsible for these cells’ high β-catenin activity and stemness. This is consistent with our findings demonstrating that URI overexpression in Lgr5high ISCs reduces WNT–β-catenin signaling.

On the basis of our findings, we can speculate that selective inhibition of c-MYC may be a prime strategy for safeguarding intestinal tissue architecture during high-dose irradiation. c-MYC inhibitors may thus allow increases of radiation doses during cancer radiotherapy, decreasing off-target effects and, hence, providing a more effective way of treating tumors. Moreover, c-MYC inactivation may cause tumor shrinkage in these patients. Finally, further work is required to determine whether reduced URI levels could be valuable predictive indicators of patients’ responses to high-dose irradiation.

Methods summary

To study the role of URI in radiation-induced GIS, conditional Uri knockout mice were generated by deletion of exon 4 of the Uri gene. URI was specifically ablated in intestinal epithelium by crossing conditional URI lox mice with Villin-creERT2 mice (25). Moreover, to study whether URI+ cells repopulate the intestine after high irradiation doses, URI knock-in mice were generated by inserting a creERT2-IRES-EGFP cassette under the control of the Uri promoter in exon 1 of the Uri gene and crossing with R26-stop-EYFP or URI-creERT2;CAG-Katushka mice.

To avoid bone marrow failure and specifically irradiate in the abdominal cavity, we designed a device using 3-cm lead chambers. Mice were anesthetized by using ketamine (100 mg/kg) and xylazine (10 mg/kg) and placed in a methacrylate chamber to hold them. Mice were fully protected except for a 3-cm open window where the abdominal cavity was located, allowing for absorption of 100% of the irradiation. The irradiator Mark I 30A system was used for irradiation of 8-week-old mice. The source used in this case was Cs-137 (662 KeV E.max). Doses ranging between 2 and 22 Gy were delivered at a dose rate of 1.54 Gy/min, and mice were sacrificed at different time points.

Supplementary Materials

science.sciencemag.org/content/364/6443/eaaq1165/suppl/DC1

Materials and Methods

Figs. S1 to S24

Tables S1 to S4

References (2637)

Movie S1

References and Notes

Acknowledgments: We are very thankful to all the mouse providers as described in materials and methods. We also thank the CNIO Transgenic Mice Core Unit for technical support. Funding: A.C.-P. is a recipient of the La Caixa Ph.D. fellowship. N.D. is a recipient of the Spanish Ramón y Cajal fellowship. This work was supported by the Spanish Ministry of Economy and Competitiveness and co-funded by the European Regional Development Fund (ERDF-EU, SAF2016-76598-R) and the National Institute of Health Carlos III. Author contributions: A.C.-P. designed and performed most of the experiments. A.C.-P. and N.D. generated the URI-creERT2-IRES-EGFP mouse. M.Y. initiated the project, designed experiments, and engineered the URI lox allele with N.D. C.P. performed all histopathological analyses of murine tissues. S.d.l.R. characterized the URI-creERT2-IRES-EGFP mouse and helped with studying NHEJ repair in vitro. All authors analyzed the data. N.D. designed the experiments and conceived, developed, and wrote the project and study. N.D. wrote the manuscript and secured all funding. Competing interests: The authors declare no competing financial interests. Data and materials availability: All data are available in the main text or the supplementary materials. Materials are available upon request to N.D., and the sharing of materials described in this work will be subject to standard material transfer agreements.
View Abstract

Stay Connected to Science

Navigate This Article