Research ArticleRadiation Toxicity

PHD Inhibition Mitigates and Protects Against Radiation-Induced Gastrointestinal Toxicity via HIF2

See allHide authors and affiliations

Science Translational Medicine  14 May 2014:
Vol. 6, Issue 236, pp. 236ra64
DOI: 10.1126/scitranslmed.3008523


Radiation-induced gastrointestinal (GI) toxicity can be a major source of morbidity and mortality after radiation exposure. There is an unmet need for effective preventative or mitigative treatments against the potentially fatal diarrhea and water loss induced by radiation damage to the GI tract. We report that prolyl hydroxylase inhibition by genetic knockout or pharmacologic inhibition of all PHD (prolyl hydroxylase domain) isoforms by the small-molecule dimethyloxallyl glycine (DMOG) increases hypoxia-inducible factor (HIF) expression, improves epithelial integrity, reduces apoptosis, and increases intestinal angiogenesis, all of which are essential for radioprotection. HIF2, but not HIF1, is both necessary and sufficient to prevent radiation-induced GI toxicity and death. Increased vascular endothelial growth factor (VEGF) expression contributes to the protective effects of HIF2, because inhibition of VEGF function reversed the radioprotection and radiomitigation afforded by DMOG. Additionally, mortality from abdominal or total body irradiation was reduced even when DMOG was given 24 hours after exposure. Thus, prolyl hydroxylase inhibition represents a treatment strategy to protect against and mitigate GI toxicity from both therapeutic radiation and potentially lethal radiation exposures.


Radiation exposure in a mass casualty setting is an ongoing threat that is a serious military and public health concern (1). Acute radiation syndrome, also known as radiation sickness, describes a constellation of symptoms that occur after total body exposure to radiation. At doses less than 8 Gy, fatal injuries are primarily hematopoietic in nature and can be treated with a bone marrow transplant and supportive care (2). Doses of more than 10 Gy universally lead to death, however, because of damage to the gastrointestinal (GI) tract (3). At these higher doses of radiation, a critical number of intestinal stem cells (ISCs) are believed to be irreparably killed, which impairs the regeneration of villi and compromises the epithelial integrity of the entire GI tract (4). The damaged and blunted villi cause malabsorption, fluid loss, and electrolyte imbalances, which can lead to death (5). Moreover, the loss of epithelial integrity can promote the direct access of enteric pathogens and flora into the bloodstream, which can lead to sepsis and death (6). These potentially lethal GI symptoms after radiation exposure are sometimes referred to collectively as the radiation-induced gastrointestinal syndrome (RIGS). Unfortunately, few effective treatments exist for radiation-induced GI toxicity. A handful of Food and Drug Administration (FDA)–approved radioprotectors work by eliminating internally ingested radiation (3), or through free radical scavenging, and have unfavorable side effect profiles (7) that would not be useful for treating patients on a large scale.

The biology that underlies RIGS has been studied extensively over many decades and is still subject to controversy. The seminal studies of Withers and Elkind (8) established the hypothesis that dose-dependent radiation damage to the ISCs located in the crypts of Lieberkühn was the primary cause of RIGS (9). Further molecular dissection of these crypt ISCs has demonstrated that although both Lgr5+ (10) and Bmi1+ cells (11) can repopulate the gut, it is the Bmi1+ population of cells that appears to be more critical in the injury response (12). The primacy of epithelial cells in the radiation response of the gut, however, has been challenged by data showing that genetic (13) or immunologic inhibition (14) of ceramide signaling in the endothelial cells also prevents death from RIGS. Consequently, there is ongoing debate regarding the importance of both epithelial and endothelial cell types in the radiation response of the gut.

Hypoxic signaling through hypoxia-inducible factors 1 and 2 (HIF1 and HIF2) is critical for many aspects of intestinal homeostasis. The intestine naturally exists in a steep physiologic hypoxia gradient, and HIF regulates several genes required for intestinal barrier function, such as intestinal trefoil factor (TFF3/ITF) and MDR1 (15). Augmenting HIF expression in the gut with an intestine-specific knockout of the von Hippel–Lindau (VHL) gene was shown to be protective against infectious or chemical stresses (16). However, the role of HIF in the radiation response of the gut remains unexplored.

The protein stability of the HIF family of transcription factors is regulated by prolyl hydroxylase domain (PHD)–containing proteins. During normoxia, PHD proteins hydroxylate HIF on critical proline residues that enable VHL to bind HIF and target it for proteasomal degradation (17). To date, three major oxygen-dependent prolyl hydroxylases (PHD1 to PHD3) have been identified in mammals (18), but their role in the radiation response of the gut is unknown. We posited that the inhibition of PHD function would stabilize HIF, improve epithelial integrity, and possibly reduce radiation toxicity. We show that genetic or pharmacologic inhibition of all three PHD isoforms robustly stabilizes HIF in normoxia, reduces morbidity and mortality from lethal radiation exposure, and is also an effective mitigation strategy. Thus, PHD inhibition may be an effective countermeasure for radiation exposure.


Pan-PHD knockout is required for high HIF2 expression and radioprotection of the gut

To determine the role of the PHD proteins in the radiation response of the intestinal tract, we created intestine-specific knockouts of all combinations of PHDs by backcrossing triple heterozygous PHD1/2/3 mice (PHD1/2/3fl/+) to create all possible combinations of PHD isoforms (PHD1fl/fl; PHD2fl/fl; PHD3fl/fl; PHD1/2fl/fl; PHD1/3fl/fl; PHD2/3fl/fl; PHD1/2/3fl/fl), as we have done previously (19). These individual homozygous floxed genotypes were then crossed with the Villin-Cre mouse (20) to knock out every possible combination of PHD genes within the GI epithelium (GI-PHD KO). The expression of PHD1, PHD2, or PHD3 in the small intestine and colon was diminished by 90 to 95% as determined by Western blot (Fig. 1A). HIF1 and HIF2 were not detected in single knockouts of PHD1, PHD2, or PHD3, nor was HIF stabilized in PHD1/2 knockout animals (Fig. 1A). Mice lacking both PHD1/3 and PHD2/3 in the intestines exhibited modest expression of both HIF1 and HIF2 proteins, whereas mice knockout for all three PHD isoforms in the gut (GI-PHD1/2/3KO) exhibited robust stabilization of HIF1 and HIF2 in both the small intestine and the colon (Fig. 1A).

Fig. 1. Loss of all PHD isoforms is required for radioprotection.

(A) Western blots from purified epithelial cells of the indicated genotypes and tissues. The wild-type (WT) lane is from a PHD1/2/3fl/fl animal without Villin-Cre. The various floxed PHD animals were of a mixed C57BL/6-FVB genetic background. (B) Kaplan-Meier analysis of GI-PHD1/2/3KO mice and controls after 18 Gy of TAI. n = 8 per genotype; P = 0.005 by log-rank test. (C) Kaplan-Meier analysis of GI-PHD1/2/3KO mice and controls after 14 Gy of TBI. n = 9 per genotype; P = 0.02 by log-rank test.

To assess the radioprotective effects of PHD proteins, we treated each combination of PHD knockout mice with 18 Gy of total abdominal irradiation (TAI), which is a supralethal dose for mice on a mixed genetic background (21). TAI was achieved with a custom irradiation jig (fig. S1), which protects the bone marrow of the upper body and reduces hematopoietic toxicity as a competing cause of death (22). GI-PHD1/2/3KO animals showed a 70% survival rate at 30 days after receiving 18 Gy of TAI, whereas all littermate controls died before 10 days (Fig. 1B and table S1). None of the other PHD knockout combinations showed a statistically significant survival advantage after TAI (fig. S2, A to F, and table S1).

Although TAI is useful in studying GI toxicity from radiation, accidental radiation exposure affects the entire body, and not just the lower half. To test whether the knockout of PHD proteins also protected against whole-body doses of lethal radiation, we performed total body irradiation (TBI) experiments on GI-PHD1/2/3KO animals at a lethal dose of 14 Gy (23). Indeed, GI-PHD1/2/3KO animals exhibited a 27% survival rate (3 of 11 mice, Fig. 1C and table S1) at 30 days, compared to a 0% survival rate in the littermate control group (0 of 9 mice, Fig. 1C and table S1), indicating that the loss of PHD proteins in the gut is sufficient to protect against radiation-induced GI mortality from abdominal and whole-body radiation.

Prolyl hydroxylase inhibition stabilizes HIF and protects against radiation-induced GI toxicity and death

To recapitulate the phenotype of the GI-PHD1/2/3KO mice, we used a small-molecule inhibitor of all PHD isoforms, called dimethyloxallyl glycine (DMOG), which is an oxoglutarate analog (24). Intraperitoneal administration of DMOG stabilized HIF1 and HIF2 in the small intestine and colon in a dose-dependent fashion (Fig. 2A). A time course of HIF stabilization after a single 8-mg intraperitoneal injection of DMOG revealed that both HIF1 and HIF2 were fully stabilized by 6 hours after DMOG injection (Fig. 2B) and that DMOG stabilized HIF1 and HIF2 in the gut for up to 24 hours after a single injection, although expression was diminished at the 24-hour time point. There was no gross morbidity or mortality from DMOG at the doses and time course used.

Fig. 2. Prolyl hydroxylase inhibition by DMOG protects against death from abdominal and total body irradiation.

All mice in these experiments were 8-week-old male C57BL/6 mice. (A) Dose response of HIF1 and HIF2 in the small intestine and colon after DMOG treatment. Epithelial cells were isolated 6 hours after injection, and each lane of the Western blot represents a different mouse. (B) Time course of HIF1 and HIF2 activation in the small intestine and colon after an 8-mg bolus of DMOG or saline control. (C) Scheme for radioprotection after TAI, where XRT = 18 or 20 Gy of TAI as indicated. (D) Regenerating jejunal crypts after 18 or 20 Gy of TAI. For both doses, n = 4 mice per treatment and 3 high-power field (HPF) per mouse, with P value versus saline by two-tailed t test. (E) TUNEL+ cells per HPF in the jejunum after 20 Gy of radiation, with n = 5 mice per treatment and 3 HPF per mouse; P = 0.001 versus saline by two-tailed t test. (F) Kaplan-Meier analysis of the mice treated in (C), with n = 10 for the saline group and n = 13 for the DMOG group; P = 0.005 by log-rank test. (G) Scheme for radioprotection after 16 Gy of TBI. (H) Kaplan-Meier analysis of the mice treated in (G), with n = 9 for the saline group and n = 10 for the DMOG group; P = 0.001 by log-rank test.

We assessed whether hypoxia played a role in the normal radiation response of the intestines by quantitative polymerase chain reaction (PCR) for hypoxia-associated genes Glut1 (Slc2a1) and Pgk1 in the small intestine (fig. S3, A and B, and table S1) and the colon (fig. S3, C and D, and table S1). Similarly, HIF1 and HIF2 were detectable in the intestine after radiation alone (fig. S3E), but reached fourfold higher levels after treatment with DMOG (fig. S3E). Intraperitoneal administration of DMOG also increased HIF expression in the liver and kidneys, but not in the lung or peripheral blood (fig. S3F).

Although DMOG clearly modulates HIF expression in the gut, it is unknown whether HIF stabilization before radiation could reduce mortality after lethal radiation. Toward this end, C57BL/6 mice were injected with DMOG or saline control before and after 20 Gy of TAI, as depicted in Fig. 2C. C57BL/6 mice have a relatively high radiation tolerance, and 20 Gy was found to most reproducibly produce death from GI syndrome in saline-treated animals (fig. S4 and table S1). Mice were subjected to classical microcolony crypt survival analysis (for representative images, see fig. S5) to determine the survival and regeneration of crypts after radiation (25). DMOG caused a fourfold improvement in crypt survival after 18 Gy of TAI compared to saline controls (16.2 ± 2.1 versus 4.3 ± 0.9, DMOG versus saline; Fig. 2D and table S1) and a 22-fold increase in crypt survival after 20 Gy of TAI compared to controls (11.0 ± 2.1 versus 0.5 ± 0.5, DMOG versus saline; Fig. 2D and table S1). Crypt regeneration was also enhanced in the colon after 20 Gy of TAI (fig. S6A and table S1). Apoptosis, as determined by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining, was decreased in the small intestine (Fig. 2E and table S1) and colon (fig. S6B and table S1). There were no differences in γH2AX staining (18) between saline- and DMOG-treated animals, suggesting that PHD inhibition does not alter DNA repair pathways (fig. S7A, quantified in fig. S7B, and table S1). This improved crypt survival and decreased apoptosis correlated with increased survival because 67% of mice treated with DMOG survived beyond 60 days after 20 Gy of TAI, whereas none of the control mice survived beyond 10 days (Fig. 2F and table S1). Treatment with DMOG also improved mortality after 16 Gy of TBI (see scheme in Fig. 2G) because 40% of treated mice lived to 30 days, but none of the controls survived beyond 10 days (Fig. 2H and table S1).

PHD inhibition mildly increases hematocrit but does not protect tumors

We assessed how DMOG affected hematologic physiology because DMOG was given intraperitoneally and PHD proteins affect the bone marrow niche (19). DMOG mildly improved hematocrit (fig. S8A and table S1), but not hemoglobin levels (fig. S8B and table S1), at 8 days after TAI. Both hematocrit and hemoglobin levels did not change significantly with DMOG after TBI (fig. S8, C and D, and table S1). There were very similar levels of leukopenia and anemia (table S2), consistent with minimal protection of the bone marrow by DMOG.

Radioprotectors could also have use in clinical radiotherapy if they could protect normal tissue but not tumors. To determine if DMOG would also radioprotect tumors, we subcutaneously implanted human colorectal cancer cells (HCT116) or human lung carcinoma cells (A549) into nude mice. The xenografts were grown for 2 weeks before being treated with saline or DMOG for five consecutive days concurrent with sham XRT (0 Gy × 5) or a clinically relevant course of radiation treatments targeted to the lower abdomen and flank (5 Gy × 5) (26). DMOG treatment did not enhance the growth of HCT116 (fig. S9A and table S1) or A549 cells (fig. S9B and table S1) because the tumors grew at a similar rate after sham XRT. DMOG also did not decrease the tumoricidal effect of the 5 Gy × 5 radiation treatments when compared to controls (fig. S9, A and B, and table S1), indicating that pharmacologic inhibition of PHD proteins does not radioprotect tumors in these xenograft models.

DMOG improves epithelial integrity and GI tract function after radiation

The pathophysiology of the mechanism of RIGS is tightly linked to the loss of epithelial integrity of the GI tract, which leads to fluid loss, unfettered diarrhea, and electrolyte imbalances, all of which contribute to mortality (3, 27). An effective radioprotectant of the GI tract should thus reduce diarrhea and normalize the amount of formed stool. To determine how prolyl hydroxylase inhibition improved gut physiology after radiation, we treated mice with saline or DMOG according to the radioprotection protocol depicted previously in Fig. 2C.

Within 3 to 4 days after receiving 20 Gy of TAI, mice began having diarrhea, which was measured as a decrease in formed stool (figs. S10 and S11, A and B, and table S1). We quantified formed stool by manually removing droppings from the cage bedding (Fig. 3A) and by metabolic cage analysis (Fig. 3B, fig. S11A, and table S1). In both settings, control animals had almost no formed droppings by day 5 after radiation, whereas animals in the DMOG cohort exhibited only a 50% decrease in their dropping counts (Fig. 3B, fig. S11A, and table S1). These changes were not due to changes in food or water intake because intake did not differ between the saline and DMOG groups (fig. S11, B and C, and table S1). Radiation-induced diarrhea was associated with hypernatremia and hyperglycemia in the saline controls, but not in the DMOG-treated mice (table S3). Other electrolytes such as potassium, chloride, bicarbonate, BUN (blood urea nitrogen), and creatinine were not different between the treated and control groups (table S3).

Fig. 3. DMOG improves the epithelial integrity of the lower GI tract.

Mice were treated with DMOG before and everyday after receiving 20 Gy of abdominal XRT (day 0). (A) Dropping counts removed from cage bedding per day from representative cages. (B) Formed stool counts from individual mice in metabolic cages from day 0 and day 5 after radiation. n = 6 mice per treatment; P values are versus day 5 saline. (C) Measurement of epithelial integrity by FITC-dextran (4 kD) gavaged at a dose of 0.6 mg/kg, 5 days after treatment as in Fig. 2C. n = 4 to 6 per group; P values are versus saline + 20 Gy. (D) Body weight at the indicated time points of mice treated with DMOG/saline and irradiated. Mice were sacrificed if they lost more than 25% of body weight or exhibited any signs of distress. (E) Surviving mice from the DMOG cohort treated with TAI (right) and age-matched controls (left). Note the change in fur color in the irradiated lower body. (F) Quantitative PCR for relative mRNA levels of epithelial barrier genes Tff3 and Mdr1 in the jejunum. n = 6 per treatment; P values are versus saline controls.

Death from GI radiotoxicity may also stem from compromised epithelial integrity and barrier function, which facilitates both electrolyte disturbances and possible parenteral access of enteric pathogens (3). We investigated the epithelial integrity of the GI tract with a fluorescein isothiocyanate (FITC)–dextran assay, wherein mice are gavaged with dextran covalently coupled to FITC, which cannot cross the GI epithelia unless the epithelial barrier is compromised (28). Four hours after gavage, FITC-dextran levels were measured in the blood. Treatment with DMOG decreased FITC-dextran uptake in the bloodstream of XRT-treated mice by fourfold over saline-treated controls (Fig. 3C and table S1). In contrast, there was almost no uptake in wild-type mice that did not receive radiation (WT + 0 Gy, Fig. 3C and table S1).

The improved epithelial integrity afforded by DMOG reduced weight loss after radiation (Fig. 3D and table S1). The surviving mice from the DMOG cohort gained back most of the weight lost after radiation treatment. By 2 months after radiation, the DMOG-protected mice that survived 20 Gy of ionizing radiation to the abdomen and lower body were physically indistinguishable from the controls in size and weight, except for radiation-induced alopecia of the lower body (Fig. 3E). These data indicate that the mice treated with DMOG not only survived but also had enough functional recovery of their GI tract to display catch-up in growth compared to unirradiated age-matched controls.

The expression of intestinal trefoil factor (TFF3/ITF1) and multidrug resistance protein 1 (MDR1) is critical to maintaining epithelial integrity after toxic stimuli (29, 30). DMOG treatment increased TFF3 and MDR1 in the jejunal epithelium by seven- and threefold, respectively (Fig. 3F and table S1). Thus, both physiologic and molecular data strongly indicate that increased activation of epithelial barrier function in the small intestine and colon is critical to the radioprotective phenotype.

Enhanced intestinal vascular endothelial growth factor expression may play a role in radioprotection

Damage to the GI endothelium has also been reported to play a role in the lethality of abdominal radiation because loss of the vascular endothelium can lead to intestinal ischemia and hypoxia (13). The endothelial radiation response can be modified by cell signaling mediators such as ceramide (13) or growth factors such as fibroblast growth factor (31) and vascular endothelial growth factor (VEGF) (32). Pharmacologic inhibition of PHD1 to PHD3 increased Vegf expression in isolated epithelia from the jejunum (Fig. 4A and table S1) and colon (fig. S12A and table S1), and this correlated with increased serum levels of VEGF (Fig. 4B and table S1). DMOG increased microvessel density in the jejunal crypts after radiation as determined by Meca32 immunohistochemistry (Fig. 4C, quantified in Fig. 4D, and table S1). In addition, there were more CD105+ cells in the intestinal crypts after radiation (Fig. 4E and table S1; representative images in fig. S12B), which may be an indication of increased angiogenesis (33). Genetic knockout of PHD1/2/3 also increased Vegf expression in jejunal epithelia (Fig. 4F and table S1) and the amount of VEGF in the serum (Fig. 4G and table S1). Serum VEGF was not increased in the other GI-PHD KO animals (fig. S12C and table S1).

Fig. 4. Prolyl hydroxylase inhibition increases Vegf.

(A and B) After treatment with radioprotective protocol as in Fig. 2C, jejunal epithelium was isolated and assessed by quantitative PCR for relative mRNA levels of Vegf (A; n = 6 per treatment; P value versus saline) and by enzyme-linked immunosorbent assay for serum VEGF levels (B; n = 12 per treatment; P value versus saline). (C) Meca32 staining for endothelial cells after 20 Gy of TAI +/− DMOG. Yellow arrows indicate microvessels. Scale bars, 50 μm. (D and E) Microvessel density (D; n = 6 mice per treatment and 4 HPF per mouse; P value versus saline) and CD105+ cells per HPF (E; n = 5 mice per treatment with 3 HPF per mouse; P value versus saline) in saline or DMOG controls. (F and G) Jejunal Vegf expression (F; n = 6 per treatment; P value versus WT) and serum VEGF levels (G; n = 6 per treatment; P value versus WT) in the indicated knockout animals and littermate controls. (H) Kaplan-Meier analysis of mice treated with the indicated adenoviruses and with intraperitoneal saline or DMOG as in Fig. 2C. P = 0.002 by log-rank test, Fc DMOG versus Flt1 DMOG. (I) Kaplan-Meier analysis of HIF1LSL Tie2Cre or HIF2LSL Tie2Cre mice and their littermate controls after 18 Gy of TAI.

To test the role of VEGF in the radioprotective effects of DMOG, we injected mice with an adenovirus that encodes Flt1 (Ad-Flt1), a soluble VEGF receptor that binds and inhibits the action of VEGF, or Fc (Ad-Fc) control (34). Five days after treatment with adenovirus, mice were subjected to DMOG treatments and TAI in accordance with the protocol in Fig. 2C. The inhibition of VEGF function by Ad-Flt1 eliminated the survival effects of DMOG (Fig. 4H and table S1) because DMOG demonstrated a survival advantage in Ad-Fc control animals but did not protect animals treated with Ad-Flt1 (P = 0.002, Fc DMOG versus Flt1 DMOG). Saline controls in both groups perished within 10 days. Ad-Flt1 had no effect on white blood cell counts (fig. S13A), but increased hematocrit and hemoglobin (fig. S13, B and C) via known effects on hepatic erythropoietin production (35), but this had no effect on survival, which suggests that the prominent effect of Ad-Flt1 on intestinal radiation sensitivity stems from local effects in the gut rather than modulation of hematopoietic factors.

Because a pharmacologic PHD inhibitor could potentially affect both epithelial and endothelial cells, we expressed HIF1 or HIF2 in the endothelial cells to determine if this was sufficient to protect the gut from RIGS. We used transgenic lox-stop-lox (LSL) mice that express either HIF1 or HIF2 only in the presence of a tissue-specific Cre, which removes the stop cassette (36). We bred these LSL mice with Tie2-Cre to create HIF1LSL Tie2Cre and HIF2LSL Tie2Cre mice, which activate either HIF transgene only in endothelial cells (37). Endothelial-specific expression of either HIF1 or HIF2 was not sufficient to protect mice from mortality after 18 Gy of TAI (Fig. 4I and table S1).

HIF2, but not HIF1, mediates the radioprotective effects of PHD inhibition in the GI epithelium

Prolyl hydroxylase inhibition increases the levels of both HIF1 and HIF2, but it is unknown whether one or both of these factors are required for radioprotection. To determine if HIF1 or HIF2 is more important in the radioprotective effect of DMOG in the GI epithelium, we conditionally activated a stabilized HIF1 (HIF1LSL VillCre) or HIF2 (HIF2LSL VillCre) transgene in the intestinal and colonic epithelia via an LSL cassette and Villin-Cre (20). We verified the expression of HIF1 and HIF2 transgenes by Western blot in the small intestine (Fig. 5A) and colon (fig. S14). The expression of HIF1 and HIF2 was comparable to their induction by DMOG treatment (rightmost lane in Fig. 5A and fig. S14).

Fig. 5. Intestine-specific HIF2 expression is sufficient for GI radioprotection and survival.

(A) Western blots of purified epithelial cells from the intestinal tracts of mice receiving the indicated treatments. (B and C) Kaplan-Meier analysis of mice after 18 Gy of TAI after intestine-specific overexpression of stabilized HIF2 (B; n = 10 per group; P = 0.002 by log-rank test) or HIF1 (C; n = 11 per group; P = 0.2 by log-rank test). (D and E) Total feces per cage from HIF2LSL (D) or HIF1LSL (E) mice on the indicated days after radiation. n = 6 per group; P values are versus day 5 saline. (F and G) Jejunal expression of Vegf (F; n = 6 per group; P values are versus day 5 saline) and serum levels of VEGF (G; n = 6 per group; P values are versus WT controls) in HIF2LSL mice bred with Villin-Cre and littermate controls. (H and I) Jejunal Vegf expression (H) and serum levels of VEGF (I) in HIF1LSL mice bred with Villin-Cre and littermate controls.

HIF1LSL mice, HIF2LSL mice, and littermate Cre-negative controls were treated with 18 Gy of TAI and subjected to survival analysis. Mice expressing stabilized HIF2 in the gut demonstrated improved survival after lethal TAI (Fig. 5B and table S1), but the intestinal-specific overexpression of HIF1 had no effect on survival (Fig. 5C and table S1). Accordingly, HIF2LSL VillCre mice had threefold more formed stools than controls 5 days after receiving TAI (Fig. 5D and table S1), but HIF1LSL VillCre mice showed no changes in formed stools, suggesting that increased epithelial integrity and function contributed to improved survival (Fig. 5E and table S1). HIF2LSL VillCre mice also showed increased expression of Vegf within the gut epithelia (Fig. 5F and table S1) and higher concentration of serum VEGF (Fig. 5G and table S1). Mice with elevated HIF1 in the intestines showed no significant increase in epithelial Vegf (Fig. 5H and table S1) or serum levels of VEGF protein (Fig. 5I and table S1).

We tested the importance of endogenous HIF1 and HIF2 by generating mice that lacked HIF1 (HIF1KO) or HIF2 (HIF2KO) in the intestines by crossing homozygous HIF1fl/fl or HIF2fl/fl mice with Villin-Cre transgenic animals. These knockout animals and their littermate Cre-negative controls (labeled WT) were then treated with saline or DMOG (as in Fig. 2C) to determine if either HIF isoform was necessary for radioprotection. Mice lacking HIF1 demonstrated a survival response to DMOG treatment (Fig. 6A and table S1), which implied that HIF1 was not required for radioprotection. Although HIF1KO mice treated with DMOG showed a lower survival rate compared to HIF1 wild-type mice treated with DMOG (40 versus 60% at 30 days), the difference between these curves was not statistically significant (P = 0.7). Conversely, the loss of HIF2 in the intestines nullified the survival effects of DMOG after 18 Gy of TAI (Fig. 6B and table S1), suggesting that HIF2 was necessary for radioprotection afforded by DMOG. Thus, HIF2 is both necessary and sufficient for the intestinal radioprotection by DMOG, and HIF1 appears to have a minimal role in the radioprotection promoted by PHD inhibition.

Fig. 6. Intestinal HIF2 expression is necessary for radioprotection.

(A and B) Kaplan-Meier analysis of HIF1fl/fl (HIF1WT) or HIF1fl/fl-VillCre (HIF1KO) mice (A; P = 0.2 by log-rank test between HIF1WT/DMOG and HIF1KO/DMOG; n = 8 per group) and HIF2fl/fl (HIF2WT) or HIF2fl/fl-VillCre (HIF2KO) mice (B; P = 0.002 by log-rank test between HIF2WT/DMOG and HIF2KO/DMOG; n = 8 per group) treated with either saline or DMOG and (n = 8 per group) after 18 Gy of TAI. Both HIF1fl/fl and HIF2fl/fl animals were of a mixed C57BL/6-FVB background. (C and D) Kaplan-Meier analysis of HIF1LSL Lgr5Cre or HIF2LSL Lgr5Cre (C) or HIF1LSL Bmi1Cre or HIF2LSL Bmi1Cre (D) mice and their littermate controls after 18 Gy of TAI.

Because Villin-Cre is also expressed in the ISC niche, we assessed whether expression of HIF1 or HIF2 in ISC populations could recapitulate the effects of expression in all epithelial cells of the gut. We again used the HIF-LSL system and Lgr5-CreERT2 (10) and Bmi1-Cre-ER (38). Lgr5+ cells are the crypt base columnar cells that are interspersed along with Paneth cells throughout the GI tract and are considered the primary stem cells of the GI tract (10). The expression of either HIF1 or HIF2 in Lgr5+ cells, however, was not sufficient to protect mice from lethal TAI (Fig. 6C and table S1). Bmi1+ cells in the gut have recently been shown to function as a reserve pool of ISCs during intestinal injury (39). Nevertheless, the expression of HIF1 or HIF2 in Bmi1+ cells was also not sufficient to protect mice from lethal TAI (Fig. 6D and table S1).

Animals that survived lethal irradiation live normal life spans and exhibit minimal morbidity

We followed a DMOG-treated cohort of mice that survived 18 Gy of TAI to determine the extent of chronic morbidity. Kaplan-Meier analysis in Fig. 7A revealed that all control mice (10 of 10) who received saline died within 10 days (median survival, 8.5 days), whereas 20% (3 of 15) of animals who were protected with DMOG lived beyond 20 months of age (median survival, 215 days). Sixty-three percent (7 of 11) of age-matched unirradiated controls lived up to and beyond 20 months (Fig. 7A and table S1). Thus, a few mice that would have otherwise perished without DMOG treatment lived a normal life span.

Fig. 7. Long-term survivors of lethal TAI have lower body weights and microscopic intestinal changes.

(A) Kaplan-Meier analysis of mice treated with saline, DMOG with XRT, or no XRT, 18 months after 18 Gy of TAI. P = 0.01, XRT versus no XRT; P = 0.008, XRT versus saline (log-rank test). (B) Surviving mouse from the DMOG cohort treated with TAI (right) and an age-matched control (left). Note the change in fur color in the irradiated lower body. (C) Body weight of mice at about 20 months of age. n = 8 per group; P value versus no XRT. (D and E) Trichrome (D) and hematoxylin and eosin (H&E) staining (E) of intestines of mice 18 months after receiving TAI or no radiation. Arrows indicate microscopic fibrotic bands. Scale bars, 50 μm.

We saw little morbidity at the age 20 months in the DMOG cohort. The surviving mice appeared smaller (Fig. 7B) and weighed less than age-matched unirradiated controls (Fig. 7C and table S1, *P = 0.02 versus no XRT), but had no other gross abnormalities other than hypopigmented fur on their lower bodies. Necropsies of sacrificed animals revealed no fistulas, palpable fibrosis, or any evidence of malignancy within the abdomen (see example in fig. S15). Microscopic analysis of the gut did not reveal significant fibrosis with trichrome staining (Fig. 7D) but did show slightly reduced villi density (Fig. 7E). Blood counts revealed that the survivors had a mild anemia, with a hematocrit of 31.0 ± 1.1% in the surviving irradiated animals and 39.7 ± 1.8% in the unirradiated controls (table S4). This anemia may be partially explained by hypocellularity and fatty marrow replacement in the long bones of the lower body, which were not shielded in TAI (fig. S16). The bone marrow of the humerus exhibited normal morphology because these limbs were shielded from radiation during TAI (fig. S17).

DMOG mitigates mortality from the GI syndrome

Radioprotectors such as amifostine are effective only when administered before exposure to lethal radiation (40), but not afterward. On the other hand, a radiation mitigator would reduce the toxicity from radiation after an exposure, and would thus be useful as a medical countermeasure for a nuclear incident. Currently, there are no FDA-approved radiation mitigators available to treat radiation injury. Toward this end, we tested whether DMOG administered 4 hours after radiation would mitigate mortality from 20 Gy of TAI, and found that, indeed, DMOG improved survival versus saline controls (fig. S18A and table S1, 45 versus 0% at 10 days, log-rank P = 0.002).

In the event of a potentially lethal radiation exposure, it could take many hours before a patient is diagnosed and properly treated. Thus, we tested whether DMOG could mitigate death from the GI syndrome when administered 24 hours after radiation exposure (40). C57BL/6 mice were treated with 17 or 19 Gy of TAI, then the first dose of DMOG or saline was given 24 hours after the radiation exposure as depicted in the scheme in Fig. 8A. Kaplan-Meier plots shown in Fig. 8B reveal that at 19 Gy, DMOG exhibits no mitigative properties. At the 17-Gy dose, however, DMOG treatment mitigates death from GI syndrome because 75% of mice (6 of 8 animals) in the DMOG cohort survived, compared to only 18.2% of the saline cohort (2 of 11 animals, Fig. 8B and table S1, log-rank P = 0.02).

Fig. 8. DMOG mitigates GI radiotoxicity but requires intact bone marrow.

(A) Scheme of radiation mitigation experiments after TAI. (B) Kaplan-Meier analysis of C57BL/6 mice treated with TAI at the indicated doses and then given daily DMOG beginning at 24 hours after XRT (P = 0.002 by log-rank test, 17 Gy of TAI DMOG versus saline; n = 8 to 11). (C) Scheme of radiation mitigation experiments after TBI and a bone marrow transplant (BMT). (D) Kaplan-Meier analysis of C57BL/6 mice given saline or DMOG along with a BMT or mock BMT 24 hours after 16 Gy of TBI. P = 0.0007 by log-rank test, DMOG + BMT versus DMOG + mock.

To determine if VEGF also played a role in the radiomitigative properties of DMOG, mice were given the VEGF inhibitor Ad-Flt1 or control Ad-Fc 5 days before receiving 17 Gy of TAI. DMOG or saline was then injected intraperitoneally 24 hours after radiation. The inhibition of VEGF function through Ad-Flt1 completely abolished the survival effects of DMOG at this dose (fig. S18B and table S1), indicating that VEGF signaling may play a role in the mitigative properties of DMOG.

We also tested the ability of DMOG to mitigate GI toxicity 24 hours after TBI. Because TBI ablates all bone marrow, we also gave a bone marrow transplant at the 24-hour time point to reduce hematopoietic death. The scheme for this treatment is outlined in Fig. 8C. Three of eight mice (37.5%) survived beyond 30 days when given DMOG and a bone marrow transplant 24 hours after radiation (Fig. 8D and table S1). DMOG alone without a bone marrow transplant, however, was not sufficient to mitigate death from 16 Gy of TBI (Fig. 8D and table S1).


We demonstrate through both genetic and pharmacologic methods that the PHD proteins are critical regulators of radiation sensitivity of the intestinal tract. PHD inhibitors such as DMOG have been shown to promote intestinal healing after chemical stresses (41) and sublethal TBI (42), but our study shows that DMOG is capable of robust and reproducible protection against supralethal doses of ionizing radiation to the abdomen or whole body.

We demonstrate that DMOG mitigates radiation injury when given 24 hours after initial exposure to TAI or TBI, which distinguishes prolyl hydroxylase inhibition from many other types of radioprotectors such as free radical scavengers (43) or growth hormones (44), which require administration before receiving radiation. These data support the general strategy of PHD inhibition as a potential countermeasure to radiation exposure to mitigate toxicity (1). A caveat to radiomitigation afforded by PHD inhibition is that the protective effects diminish substantially when the drug is given after radiation exposure by TAI, and are abrogated completely in lethal TBI exposure, unless a bone marrow transplant is also administered. Treatments with radiomitigative properties in the bone marrow (45) may thus be complementary to radiomitigators of GI toxicity.

The cellular mechanisms of radiation protection by PHD inhibition are complex, but likely stem from improved epithelial integrity of the GI tract. The increased integrity of the GI tract allows the gut to maintain proper fluid homeostasis and barrier functions, which reduce death by the two most common means: electrolyte disturbances and sepsis (5, 6). By reducing this initial wave of morbidity and mortality, we posit that the intestinal tract and the animal are afforded enough time to heal and recover from injury.

The critical role of epithelial cells in the radiation response of the GI tract is further highlighted by the fact that HIF2 is radioprotective when expressed in GI epithelial cells. Although endothelial (13), Lgr5+ (46), and Bmi1+ (39) cells have been shown to play various roles in intestinal regeneration after radiation injury, the expression of HIF in these specific cell lineages was not sufficient for radioprotection. Notably, however, our data do not rule out involvement of these other cell types in the radiation response, but only support the notion that epithelial cells are critical to radioprotection by PHD inhibition.

Epithelial VEGF expression likely mediates some of the radioprotective effects of PHD inhibition by promoting angiogenesis and has previously been demonstrated to have powerful radioprotective properties (32). The growth of new vessels after radiation may further contribute to the improved fluid and nutrient transport after radiation. Although the increased number of CD105+ cells supports a role for angiogenesis, the possibility of reduced endothelial apoptosis or the recruitment of endothelial progenitors cannot be ruled out. Further studies will be needed to understand the precise mechanism involved and whether the endothelium indeed supports the epithelium in recovery after radiation, as has been posited (13), or is merely a marker of HIF activation.

PHD proteins are a pharmacologically tractable target for radioprotection and radiomitigation of toxicity to the GI tract. Pan-PHD inhibition is likely necessary for radioprotection because no other combination of PHD knockouts showed this effect. With their distinct mechanism of action, PHD inhibitors may complement the activity of other published radioprotectors, such as Toll-like receptor 5 agonists (47) and anti-ceramide antibodies (14), and may also benefit from being administered along with a radiation mitigator for the hematopoietic system (48).

The major limitation of this preclinical study is that the results are not immediately applicable to human treatment. The amount of radiation that evokes toxicity in a mouse is often more than what is required in human patients (49). Moreover, although DMOG does not radioprotect tumors in a xenograft model, further study is required to determine if PHD inhibition is also protective against conventionally fractionated radiation therapy used for spontaneously growing solid tumors. Our experiments are a proof of principle for radioprotection by PHD inhibition, but DMOG may not have the proper pharmacological profile for use in humans (50). With further study and development, however, PHD inhibitors could be rapidly translated to human use because they are already being developed to treat anemia (51). A pharmacologically active PHD inhibitor could conceivably be exploited as a medical countermeasure after a large-scale accidental radiation exposure.


Materials and Methods

Fig. S1. TAI requires a custom jig.

Fig. S2. Single and double PHD knockout combinations are not radioprotective.

Fig. S3. Radiation and DMOG both induce HIF in normal tissues.

Fig. S4. DMOG radioprotects over a range of doses.

Fig. S5. Jejunal cross sections are used for crypt counting.

Fig. S6. DMOG increased regenerating crypts and decreased apoptosis in the colon after TAI.

Fig. S7. γH2AX staining in the intestine is not different between saline and DMOG mice after TAI.

Fig. S8. DMOG mildly increases hematocrit after TAI.

Fig. S9. Tumors are not radioprotected by PHD inhibition.

Fig. S10. Mouse diarrhea is recognized as loose stools and lack of formed stools.

Fig. S11. Metabolic cage measurements quantified intake and waste products from each mouse.

Fig. S12. VEGF and angiogenesis are involved in radioprotection.

Fig. S13. VEGF inhibition by Flt1 increases hematocrit.

Fig. S14. LSL–Villin-Cre animals express HIF in the colon.

Fig. S15. Necropsies of long-term survivors reveal no major abnormalities.

Fig. S16. Radiation induced fatty change in tibia bone marrow of long-term survivors.

Fig. S17. Humerus sections from aged survivors are indistinguishable from age-matched controls.

Fig. S18. DMOG mitigates radiation toxicity when administered 4 hours after radiation exposure.

Table S1. Original data (provided as an Excel file).

Table S2. Complete blood count measurements at 5 days after radiation showing no differences between saline and DMOG treatments.

Table S3. TAI-induced hypernatremia and hyperglycemia in saline controls.

Table S4. Anemia in long-term survivors of TAI compared to unirradiated controls.

References (52, 53)


  1. Acknowledgments: We would like to thank K. Takeda and G.-H. Fong (University of Connecticut) for their gift of the PHD1fl/fl, PHD2fl/fl, and PHD3fl/fl mice. We also thank C. J. Kuo (Stanford) for providing the AdFlt1 and AdFc adenoviruses. Funding: C.M.T. was supported by RSNA (Radiological Society of North America) Resident Research Grants 1018 and 1111. C.W. was supported by a training grant from the Canadian Institutes of Health and Research. A.N.D. was supported by a T32 training grant in Comparative Animal Medicine at Stanford University. A.J.G. was supported by grants from NIH (CA 67166 and 88480), the Silicon Valley Foundation, and the Sydney Frank Foundation. Author contributions: C.M.T. and A.J.G. designed all the experiments, wrote and revised the manuscript, and shared oversight over this project. C.M.T. performed and analyzed data for most experiments, except as follows: Y.R.M. performed the TUNEL assays and crypt survival experiments. A.N.D. performed all necropsies and generated crypt survival data. C.W. generated the knockout animals and contributed to the design of all animal experiments. C.W. and E.B.R. performed the bone marrow transplant experiments. T.F.A. and L.X. helped to design the radiation experiments and performed the dosimetry for the mouse jigs. Competing interests: A patent application was filed on the basis of the results of this study, titled “Use of prolyl hydroxylase inhibitors as a radioprotective drug for the lower gastrointestinal tract” (International Application No. PCT/US2012/052232).
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article