Research Article

Inhibition of the prostaglandin-degrading enzyme 15-PGDH potentiates tissue regeneration

See allHide authors and affiliations

Science  12 Jun 2015:
Vol. 348, Issue 6240, aaa2340
DOI: 10.1126/science.aaa2340

A shot in the arm for damaged tissue

Tissue damage can be caused by injury, disease, and even certain medical treatments. There is great interest in identifying drugs that accelerate tissue regeneration and recovery, especially drugs that might benefit multiple organ systems. Zhang et al. describe a compound with this desired activity, at least in mice (see the Perspective by FitzGerald). SW033291 promotes recovery of the hematopoietic system after bone marrow transplantation, prevents the development of ulcerative colitis in the intestine, and accelerates liver regeneration after hepatic surgery. It acts by inhibiting an enzyme that degrades prostaglandins, lipid signaling molecules that have been implicated in tissue stem cell maintenance.

Science, this issue 10.1126/science.aaa2340; see also p. 1208

Structured Abstract

INTRODUCTION

Agents that promote tissue regeneration could be beneficial in a variety of clinical settings, such as stimulating recovery of the hematopoietic system after bone marrow transplantation. Prostaglandin PGE2, a lipid signaling molecule that supports expansion of several types of tissue stem cells, is a candidate therapeutic target for promoting tissue regeneration in vivo. To date, therapeutic interventions have largely focused on targeting two PGE2 biosynthetic enzymes, cyclooxygenase-1 and cyclooxygenase-2 (COX-1 and COX-2), with the aim of reducing PGE2 production. In this study, we take the converse approach: We examine the role of a prostaglandin-degrading enzyme, 15-hydroxyprostaglandin dehydrogenase (15-PGDH), as a negative regulator of tissue repair, and we explore whether inhibition of this enzyme can potentiate tissue regeneration in mouse models.

RATIONALE

We used 15-PGDH knockout mice to elucidate the role of 15-PGDH in regulating tissue levels of PGE2 and tissue repair capacity in multiple organs. We then developed SW033291, a potent small-molecule inhibitor of 15-PGDH with activity in vivo. We used SW033291 to investigate the therapeutic potential of 15-PGDH inhibitors in tissue regeneration and to identify a 15-PGDH–regulated hematopoietic pathway within the bone marrow niche.

RESULTS

We found that in comparison with wild-type mice, 15-PGDH–deficient mice display a twofold increase in PGE2 levels across multiple tissues—including bone marrow, colon, and liver—and that they show increased fitness of these tissues in response to damage. The mutant mice also show enhanced hematopoietic capacity, with increased neutrophils, increased bone marrow SKL (Sca-1+ C-kit+ Lin) cells (enriched for stem cells), and greater capacity to generate erythroid and myeloid colonies in cell culture. The 15-PGDH–deficient mice respond to colon injury from dextran sulfate sodium (DSS) with a twofold increase in cell proliferation in colon crypts, which confers resistance to DSS-induced colitis. The mutant mice also respond to partial hepatectomy with a greater than twofold increase in hepatocyte proliferation, which leads to accelerated and more extensive liver regeneration. SW033291, a potent small-molecule inhibitor of 15-PGDH (inhibitor dissociation constant Ki ~0.1 nM), recapitulates in mice the phenotypes of 15-PGDH gene knockout, inducing increased hematopoiesis, resistance to DSS colitis, and more rapid liver regeneration after partial hepatectomy. Moreover, SW033291-treated mice show a 6-day-faster reconstitution of hematopoiesis after bone marrow transplantation, with accelerated recovery of neutrophils, platelets, and erythrocytes, and greater recovery of bone marrow SKL cells. This effect is mediated by bone marrow CD45 cells, which respond to increased PGE2 with a fourfold increase in production of CXCL12 and SCF, two cytokines that play key roles in hematopoietic stem cell homing and maintenance.

CONCLUSIONS

Studying mouse models, we have shown that 15-PGDH negatively regulates tissue regeneration and repair in the bone marrow, colon, and liver. Of most direct utility, our observations identify 15-PGDH as a therapeutic target and provide a chemical formulation, SW033291, that is an active 15-PGDH inhibitor in vivo and that potentiates repair in multiple tissues. SW033291 or related compounds may merit clinical investigation as a strategy to accelerate recovery after bone marrow transplantation and other tissue injuries.

Inhibiting 15-PGDH accelerates tissue repair.

(A) The enzyme 15-PGDH degrades and negatively regulates PGE2. (B) SW033291 inhibits 15-PGDH, increases tissue levels of PGE2, and induces CXCL12 and SCF expression from CD45 bone marrow cells. This in turn accelerates homing of transplanted hematopoietic stem cells (HSC), generation of mature blood elements, and post-transplant recovery of normal blood counts. Inhibiting 15-PGDH similarly stimulates cell proliferation after injury to colon or liver, accelerating repair of these tissues.

Abstract

Agents that promote tissue regeneration could be beneficial in a variety of clinical settings, such as stimulating recovery of the hematopoietic system after bone marrow transplantation. Prostaglandin PGE2, a lipid signaling molecule that supports expansion of several types of tissue stem cells, is a candidate therapeutic target for promoting tissue regeneration in vivo. Here, we show that inhibition of 15-hydroxyprostaglandin dehydrogenase (15-PGDH), a prostaglandin-degrading enzyme, potentiates tissue regeneration in multiple organs in mice. In a chemical screen, we identify a small-molecule inhibitor of 15-PGDH (SW033291) that increases prostaglandin PGE2 levels in bone marrow and other tissues. SW033291 accelerates hematopoietic recovery in mice receiving a bone marrow transplant. The same compound also promotes tissue regeneration in mouse models of colon and liver injury. Tissues from 15-PGDH knockout mice demonstrate similar increased regenerative capacity. Thus, 15-PGDH inhibition may be a valuable therapeutic strategy for tissue regeneration in diverse clinical contexts.

Tissue regeneration is often required during recovery from injury, disease, and certain medical treatments. For example, hematopoietic stem cell (HSC) transplantation, which includes bone marrow transplantation, is a potentially curative therapy for many hematologic malignancies (1). However, after HSC transplantation, individuals are at high risk of potentially lethal infections while awaiting regeneration of peripheral blood neutrophils and are also at risk of internal bleeding while awaiting regeneration of platelets (1). Therapeutically enhancing hematopoietic tissue regeneration has proved to be challenging. Experimental approaches have principally focused on strategies that use ex vivo treatments to expand the numbers or increase the efficacy of donor hematopoietic stem cells before transplantation (24). In a different disease, ulcerative colitis, tissue damage to the colonic epithelium, in part from immune cells, causes both gastrointestinal bleeding and diarrhea (5). Current treatments for ulcerative colitis primarily involve immune suppression; thus far, there are no therapies that potentiate healing and regeneration of the damaged colonic epithelium (5). Tissue regeneration is also required after partial hepatic resection for liver cancer or liver metastases, because the patients must regain adequate hepatic function to survive (68).

Prostaglandin PGE2 is a candidate molecule for potentiating regeneration in multiple tissues. PGE2 is a lipid signaling molecule produced by the enzyme activity of cyclooxygenase-1 or cyclooxygenase-2 (COX-1 or COX-2) followed sequentially by that of prostaglandin E synthase (9). PGE2 augments Wnt signaling (10, 11), a pathway that is involved in the maintenance of several types of tissue stem cells, including hematopoietic and colon stem cells (11, 12). PGE2, and its more stable analog 16, 16-dimethyl-PGE2 (dmPGE2), expand hematopoietic stem cell numbers in mice and in zebrafish (11, 13, 14). Murine bone marrow cells and human cord blood stem cells that are treated ex vivo with dmPGE2 show enhanced engraftment when these cells are injected back into recipient mice (4, 1417). dmPGE2 treatment of human cord blood stem cells before their administration in human HSC transplants is currently being tested in clinical trials (4). PGE2 similarly supports the expansion of human colon stem cells in cell culture (18), and, in a model of murine colitis, colon injury was exacerbated by a COX enzyme antagonist but was ameliorated by treatment with dmPGE2 (19).

We hypothesized that alternative potential strategies for increasing PGE2-mediated tissue repair in vivo could be either to increase the synthesis of PGE2 or to inhibit the normally rapid in vivo degradation of PGE2. The enzyme 15-hydroxyprostaglandin dehydrogenase (15-PGDH), which acts in vivo as a negative regulator of prostaglandin levels and activity (2022), provides a candidate target. Specifically, 15-PGDH catalyzes the first step in the degradation of prostanoid family molecules, oxidizing the prostanoid 15-hydroxyl group to a ketone and thereby abrogating binding to prostaglandin receptors (20). Here, we explore whether pharmacological inhibition of 15-PGDH can potentiate tissue repair in several mouse models of injury and disease.

Results

Genetic deletion or pharmacologic inhibition of 15-PGDH increases tissue PGE2 levels

To confirm that 15-PGDH broadly regulates PGE2 in vivo, we compared PGE2 levels in 15-PGDH knockout (21) and wild-type mice, retesting lung (21) and colon (22) and newly interrogating bone marrow and liver. Although basal PGE2 levels varied by a factor of 5 across these four tissues, the 15-PGDH knockout mice exhibited a consistent twofold increase in PGE2 levels (Fig. 1A). We hypothesized that a chemical inhibitor of 15-PGDH would have a similar effect and, further, would provide a tool to explore 15-PGDH as a therapeutic target for potentiating tissue regeneration.

Fig. 1 Biological effects of 15-PGDH inhibition in mice.

(A) PGE2 levels (ng PGE2/mg protein) in 15-PGDH knockout (KO) and wild-type (WT) mouse tissues. N = 5 mice per data point. (B) PGE2 levels in tissues of mice at 0 hour baseline and at 3 hours after intraperitoneal injections with either 10 mg/kg SW033291 (drug) or vehicle-control. N = 6 mice per data point. (C) Neutrophil counts from FVB strain 15-PGDH WT versus KO mice. y-axis scale, 103 cells per μl. *P = 0.031, Student’s t test. N = 16 mice per data point. (D) SKL cell numbers in bone marrow of FVB 15-PGDH WT versus KO mice. y-axis scale, 103 cells per femur. *P < 0.04, Student’s t test. N = 16 mice per data point. (E and F) Comparison of C57BL/6J mice treated with SW033291 [10 mg/kg intraperitoneally (IP) twice daily for five doses] versus vehicle-control. (E) Neutrophil counts. y-axis scale, 103 cells per μl. *P < 0.003, Student’s t test. N = 6 SW033291-treated and 9 control mice. (F) SKL cell numbers in bone marrow. y-axis scale, 103 cells per femur. *P = 0.039, Student’s t test. N = 9 SW033291-treated and 12 control mice. (G) Chemical structure of SW033291. All graphs show means ± SEM.

As a first step in exploring the therapeutic potential of pharmacologically inhibiting 15-PGDH, we conducted a high-throughput screen to identify small-molecule modulators of 15-PGDH enzyme activity. We screened 230,000 synthetic compounds in the University of Texas Southwestern Medical Center chemical library using a cell-based assay designed to identify compounds that might bind to and/or alter the stability or amount of 15-PGDH. Positive hits were then further tested to identify functional enzyme inhibitors. This approach identified the small molecule SW033291 (molecular weight 412.59 daltons) as a potent inhibitor of 15-PGDH enzyme activity (Fig. 1G). Initial characterization showed that treating cells with SW033291 decreased cellular 15-PGDH enzyme activity by 85% (fig. S1A). In vitro incubations of this compound with recombinant 15-PGDH protein confirmed that SW033291 is a direct high-affinity inhibitor of 15-PGDH enzyme activity, with an inhibitor dissociation constant Kiapp = 0.1 nM (23). As expected, the IC50 (concentration to inhibit 50% of enzyme activity in vitro) of SW033291 closely approximated 50% of any 15-PGDH concentration tested (e.g., IC50 of 1.5 nM SW033291 against 6 nM 15-PGDH) (fig. S1, B to D). SW033291 inhibition of 15-PGDH was noncompetitive versus PGE2 over concentrations up to 40 μM PGE2 (fig. S1E). SW033291 also demonstrated selective high-affinity interaction with 15-PGDH in thermal denaturation assays. Specifically, SW033291 shifted the 15-PGDH melting temperature by 13.5°C (fig. S2) but had no impact on the melting temperature of two other closely related short-chain dehydrogenases (fig. S3). Treatment of A549 cells with SW033291 increased PGE2 levels by 3.5-fold at 500 nM, with 50% of maximal stimulation (EC50) at ~75 nM (fig. S4).

Importantly, mice injected with SW033291 closely phenocopied 15-PGDH knockout mice, with 10 mg per kg of weight (mg/kg) SW033291 inducing a twofold increase in PGE2 levels in bone marrow, colon, lung, and liver (Fig. 1B and fig. S5). Prostaglandins PGD2 and PGF2a are also candidate substrates of 15-PGDH. Basal PGD2 levels were lower than PGE2 levels in all four mouse tissues examined and showed a lesser, although still significant, induction with SW033291 (fig. S6A). PGF2a tissue levels, which were lower still, were mostly unaltered by SW033291 (fig. S6B). We detected no toxicity from chemical inhibition of 15-PGDH in adult mice. Mice injected with SW033291 at 20 mg/kg daily for 7 days showed equivalent daily weights and activity as mice injected with vehicle-control and showed no adverse changes in blood counts or serum chemistry.

15-PGDH inhibition promotes hematopoietic recovery after bone marrow transplantation

The ability of PGE2 to promote hematopoiesis has been documented by two types of experiments: (i) ex vivo treatment of hematopoietic stem cells with dmPGE2 improves stem cell engraftment and consequent survival of mice, and (ii) administration of PGE2 or dmPGE2 to sublethally irradiated mice accelerates hematopoietic recovery and increases numbers of the short-term subtype of hematopoietic stem cells (11, 1317, 2427). Previous work also revealed that dmPGE2-treated cells have increased levels of cyclic adenosine monophosphate (AMP), increased expression of CXCR4 and survivin, decreased apoptosis, and increased homing to the bone marrow stem cell niche (11, 1416, 24, 25, 27).

To evaluate whether 15-PGDH might regulate these responses, we examined 15-PGDH knockout mice and SW033291-treated wild-type mice. We observed a significant 43% increase in neutrophil counts in 15-PGDH knockout versus wild-type mice (Fig. 1C) and a significant 39% increase in bone marrow SKL (Sca-1+ C-kit+ Lin) cells (Fig. 1D), which are enriched for bone marrow stem cells. There were no significant differences in counts of other peripheral blood cells, in bone marrow cellularity, or in numbers of bone marrow SLAM (Sca-1+ C-kit+ Lin CD48 CD150+) cells (that are further enriched for bone marrow stem cells) (figs. S7 and S8, A to C). Mice injected with SW033291 for three consecutive days showed similar significant benefits, including a doubling of peripheral neutrophil counts (Fig. 1E and fig. S9), a 65% increase in marrow SKL cells (Fig. 1F), and a 71% increase in marrow SLAM cells (fig. S10, A to C).

In functional assays, bone marrow from 15-PGDH knockout mice and from SW033291-treated mice both showed significant 55% increases in the numbers of hematopoietic colonies generated after plating into methylcellulose (Fig. 2, A and B). Erythroid (BFU-E)– and myeloid (CFU-GM)–type colonies were each significantly increased in assays of marrow from both 15-PGDH knockout mice (fig. S8, D and E) and SW033291-treated mice (fig. S10, D and E). Likewise, ex vivo treatment of harvested bone marrow cells with SW033291 induced a significant 50% increase in hematopoietic colony formation (Fig. 2C), indicating that the compound acts directly on these cells. SW033291 did not increase colony formation of bone marrow from 15-PGDH knockout mice, confirming that the compound acts by directly targeting 15-PGDH (Fig. 2C).

Fig. 2 Effect of 15-PGDH inhibition on bone marrow.

(A to C) The y axis graphs comparison of hematopoietic colony-forming units (CFU) generated per femur on day 14 after plating mouse bone marrow cells into methylcellulose with cytokines. (A) Bone marrow from FVB 15-PGDH WT versus KO mice. N = 4 mice plated in duplicate per data point. *P < 0.003, Student’s t test. (B) Bone marrow from SW033291-treated (10 mg/kg IP twice daily for 5 doses) versus vehicle-control–treated C57BL/6J mice. N = 4 mice plated in duplicate per data point. *P < 0.009, Student’s t test. (C) Bone marrow from FVB WT or KO mice treated ex vivo with SW033291 (+) versus vehicle-control (–). N = 3 mice plated in duplicate per data point. *P < 0.025, Student’s t test. (D and E) Gene induction in CD45 bone marrow cells from mice treated with SW033291 (+) or with vehicle-control (–) in combination with the EP2 antagonist PF-04418948 (EP2), the EP4 antagonist L-161,982 (EP4), or control diluent (–). N = 4 pools of 3 mice each per data point. *P < 0.02 compared with mice treated with SW033291 but without an EP receptor antagonist, Student’s t test. (D) Relative CXCL12 expression. (E) Relative SCF expression. (F and G) Homing of carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled donor marrow cells to the bone marrow of irradiated mice, comparing SW033291-treated versus vehicle-control–treated recipient mice. The y axis graphs homing efficiency determined as the ratio at 16 hours of percentage of CFSE-labeled cells in the marrows of (SW033291-treated mice)/(vehicle-treated mice). N = 9 mice per data point. Graphs show comparison of recipient mice treated with SW033291 (+) or with vehicle-control (–) in combination with control diluent (–) or with the antagonists indicated. (F) EP2 antagonist PF-04418948 (EP2) or EP4 antagonist L-161,982 (EP4). *P < 0.013 versus mice receiving only SW033291, Student’s t test. (G) CXCL12 antagonist Plerixafor. *P < 0.019 versus mice receiving only SW033291, Student’s t test. All graphs show means ± SEM.

To determine the effects of SW033291 on gene expression, we sorted bone marrow from control and SW033291-treated mice into SKL marked and CD45 fractions. In CD45 cells, which are enriched for bone marrow stromal cells, SW033291 treatment induced a greater than fourfold increase in expression of CXCL12 and SCF (Fig. 2, D and E, and fig. S11A). These two cytokines in the hematopoietic niche play key roles in hematopoietic stem cell homing and maintenance (28). No changes were seen in expression of CXCR4 or survivin, and JAG1 expression was not detected in either treated or untreated fractions (fig. S11A). In SKL cells, SW033291 had no significant effects on expression of CXCL12, SCF, CXCR4, survivin, or JAG1 (fig. S11B). Induction of CXCL12 and SCF in CD45 cells was significantly blocked when SW033291 was coadministered to mice together with an antagonist of EP4- or EP2-type PGE2 receptors (Fig. 2, D and E). However, SW033291 fully induced CXCL12 and SCF when coadministered with antagonists of EP1 or EP3 receptors (fig. S11, C and D). As expected for activation of EP2 and EP4 receptors, SW033291 also increased cyclic AMP levels in CD45 cells (fig. S11E). We conclude that SW033291 increases PGE2 levels within bone marrow and that PGE2 functions through the EP2 and EP4 receptors to induce CXCL12 and SCF.

Having observed that SW033291 treatment increases the expression of cytokines from the hematopoietic niche, we next determined whether SW033291 would affect the ability of the mouse bone marrow to attract and support new hematopoietic stem cells. Accordingly, SW033291 or vehicle-control were administered to mice receiving a bone marrow transplant, and their respective abilities to attract new donor marrow cells to the transplant recipient’s bone marrow were compared. Transplant recipient mice treated with SW033291 showed a significant two- to threefold increase in the number of donor cells that homed to the recipient’s bone marrow (Fig. 2, F and G, and fig. S12). This effect of SW033291 could be blocked by cotreating the transplant recipient mice with an inhibitor of PGE2 generation, an inhibitor of PGE2 signaling, or an antagonist of CXCL12 (that is induced by PGE2). Thus, the increased homing of donor bone marrow cells was significantly blocked when transplant recipient mice were cotreated with SW033291 plus the COX inhibitor indomethacin (fig. S12), the EP4 antagonist L-161,982 (Fig. 2F), or the CXCL12 antagonist Plerixafor (Fig. 2G). An EP2 antagonist produced a weaker, although still significant, block of SW033291 effect (Fig. 2F). The SW033291-induced homing of transplanted donor bone marrow cells is thus driven, at least in part, by the induction of CXCL12 expression in CD45 nonmyeloid cells that are already resident in bone marrow of recipient mice. Together these results define a pathway in which 15-PGDH inhibition causes an increase in bone marrow PGE2, which in turn induces expression of bone marrow CXCL12 and SCF, cytokines that alter the bone marrow microenvironment to better support the homing of transplanted cells.

To investigate whether 15-PGDH inhibition promotes hematopoietic recovery after a bone marrow transplant, we studied lethally irradiated mice that had been infused with 200,000 strain-matched donor bone marrow cells. Under these conditions, the transplanted cells on their own are insufficient to reconstitute hematopoiesis. We further treated these transplanted mice twice daily with intraperitoneal injections of either SW033291 (5 mg/kg) or vehicle-control, starting immediately after irradiation and continuing for 21 days. The control mice all died between study days 6 and 14 (Fig. 3A). In contrast, all of the SW033291-treated mice survived for more than 30 days (Fig. 3A).

Fig. 3 SW033291 potentiates hematopoietic recovery after bone marrow transplantation.

(A) Survival of mice lethally irradiated with 11 Gy and then transplanted with 200,000 bone marrow cells. Twelve hours before transplantation (i.e., immediately after irradiation), mice were divided into two groups and started either on treatment with intraperitoneal injections of SW033291, 5 mg/kg twice daily, or injections with vehicle-control. N = 8 mice in each cohort. *P < 0.0001 for increased survival in SW033291-treated mice, Log-rank test. (B to E) Recovery of peripheral blood counts in mice lethally irradiated with 11 Gy followed by transplantation with 500,000 bone marrow cells. Transplanted mice received treatments with twice daily intraperitoneal injection of SW033291 5 mg/kg (SW033291) or vehicle-control (Control). Parallel transplanted cohorts were assayed on posttransplant day 5 (group 1, N = 4 control and 4 treated mice), day 8 (group 2, N = 4 control and 4 treated mice), and days 12 and 18 (group 3, N = 8 control and 10 treated mice, plus two deceased control mice). (B) Neutrophil counts. y-axis scale, 103 cells per μl. *P < 0.014 on days 12 and 18, Student’s t test. On days 5 and 8, total white blood cell counts are substituted for neutrophil counts because values were too low for meaningful differential counts. (C) Platelet counts. y-axis scale, 103 cells per μl. *P < 0.003 on day 8, Student’s t test. **P = 0.005 for comparison of platelet recovery across days 5 to 18, paired Student’s t test. (D) Hemoglobin level. y-axis scale, g/dL. *P ≤ 0.03 on days 12 and 18, Student’s t test. **P < 0.01 for comparison of hemoglobin recovery across days 5 to 18, paired Student’s t test. (E) Number of SKL bone marrow cells per mouse on day 18 after transplant. Data are means from N = 8 control and 10 treated mice. *P < 0.0013, Student’s t test. All graphs show means ± SEM.

To characterize the hematopoietic recovery induced by SW033291, we examined the compound’s effect on mice that were transplanted with 500,000 bone marrow cells. Under these conditions, 63% of vehicle-control–injected mice and all of the SW033291-treated mice survived for 30 days and hence were available for examination. In this experiment, SW033291-treated mice showed significantly accelerated recovery of neutrophils, platelets, and red blood cells (Fig. 3, B to D). On day 12 after transplant, SW033291-treated mice had recovered 535 neutrophils/μl versus 182 neutrophils/μl in controls (Fig. 3B), differences that would likely be meaningful in a clinical setting. On day 18 after transplant, SW033291-treated mice reached 1380 neutrophils/μl versus 646 neutrophils/μl in controls (Fig. 3B). SW033291 thus accelerated neutrophil recovery in mice by nearly 6 days. SW033291-treated mice also showed a significantly accelerated platelet recovery and a significantly less severe platelet nadir (Fig. 3C). For example, on day 8 after transplant, SW033291-treated mice had 77,000 platelets versus 39,500 platelets in controls (Fig. 3C). Again, these differences would likely be meaningful in a clinical setting. SW033291-treated mice also showed accelerated red cell recovery (Fig. 3D). For example, hemoglobin levels on day 12 after transplant were 10.7 g/dL in SW033291-treated mice versus 8.0 g/dL in control mice (Fig. 3D). In humans, these differences would likely translate to lesser need for transfusion support. More rapid recovery of peripheral blood counts in SW033291-treated mice reflected an accelerated and more robust engraftment in the bone marrow, as evidenced by these mice having a 2.4-fold significant increase in SKL bone marrow cells on day 18 after transplant (Fig. 3E). SW033291 effectiveness in accelerating recovery of neutrophils, platelets, and red cells after bone marrow transplant is consistent with this compound increasing generation of both myeloid and erythroid colonies in cell culture assays.

Although systemic administration of PGE2 has been suggested to have potentially negative effects on myelopoiesis (29) and on long-term hematopoietic stem cells (24), no such adverse effects were seen with SW033291. In particular, in a repeat experiment, 8 of 10 mice transplanted with 500,000 donor bone marrow cells and receiving SW033291 injections for 21 days were alive 7 months later. This was comparable to 7 of 10 mice that were alive after receiving a similar transplant plus injections with vehicle-control. In a second assay to test for any long-term adverse effects of SW033291 on bone marrow stem cells, we harvested bone marrow from transplant recipient mice and challenged its capacity to reconstitute hematopoiesis when retransplanted into a new cohort of irradiated recipient mice. Marrow originating from mice treated with SW033291 during a first transplant could in this manner be serially harvested and repassaged for three successive rounds of harvest, retransplantation, marrow regeneration, and then repeat challenge by reharvest and retransplantation (table S1). In this assay, the marrow harvested from mice receiving SW033291 during the original transplant also showed equal long-term robustness to the marrow harvested from mice injected with vehicle-control during the original transplant, each achieving an 80% success rate in reconstituting hematopoiesis at the third round of serial marrow transfer (table S1). We suggest that the beneficial activity of SW033291 reflects its optimization of the physiology of both dose and route of delivery of PGE2—i.e., that inhibiting 15-PGDH generates increased PGE2 at physiologic levels and directly at tissue sites and avoids the extremes of concentration accompanying administration by exogenous routes.

15-PGDH inhibition protects mice from colitis

To determine whether inhibiting 15-PGDH would promote regeneration in other tissues in which 15-PGDH inhibition increases PGE2, we examined the effects of 15-PGDH blockade in colons of mice treated for 7 days with oral dextran sodium sulfate (DSS), an agent that induces colonic ulcerations similar to those found in human ulcerative colitis (30) (Fig. 4, A and B). Mice treated daily with SW033291 and 15-PGDH knockout mice both showed marked resistance to DSS-induced colitis when compared with wild-type mice receiving vehicle-control. With either chemical or genetic ablation of 15-PGDH, there were 70% fewer colon ulcers counted on colonoscopy (Fig. 4C, days 11 and 15) and 87% less total area of ulcerated colon mucosa as measured by histomorphometry (Fig. 4D). SW033291-treated and 15-PGDH knockout mice showed marked suppression of mucosal inflammation by colonoscopic grading [murine endoscopic index of colitis severity (MEICS) score] (31) (Fig. 4F) and by histological cryptitis score (32) (Fig. 4, E and G). SW033291-treated (figs. S13A and S14A) and 15-PGDH knockout mice (fig. S14A) also show marked suppression of colitis symptoms, with significantly less diarrhea and rectal bleeding, as assessed by disease activity index (DAI) (32). SW033291-treated (figs. S13B and S14B) and 15-PGDH knockout mice (fig. S14B) also showed significantly less weight loss, with SW033291 protection increasing with increasing dose. SW033291-treated and 15-PGDH knockout mice also showed significant protection from colitis-induced colon shortening (fig. S15). Finally, SW033291-treated mice displayed a significant reduction in the levels of colitis-associated inflammatory cytokines, including tumor necrosis factor, interleukin-1 (IL-1) beta, interferon gamma, IL-2, chemokine (C-C motif) ligand 2 (CCL2), and CCL4 (fig. S16).

Fig. 4 Inhibition of 15-PGDH protects mice from DSS-induced colitis.

(A) Colonoscopic view of a DSS-induced colon ulcer (arrow). Scale bar, 1 mm. (B) Microscopic image of a DSS-induced colon ulcer (between vertical arrows). Scale bar, 2.3 mm. (C, D, F, and G) Effects of administering 7 days (study days 1 through 7) of DSS compared between experimental groups of: vehicle-control–treated mice (blue), SW033291-treated mice (5 mg/kg IP twice daily) (red), SW033291-treated mice (10 mg/kg IP twice daily) (gold), and 15-PGDH knockout mice (green). (C) Graph of mean number of colon ulcers per DSS-treated mouse counted by colonoscopy on study days 8, 11, and 15 (D8, D11, and D15) for vehicle-control–treated mice, SW033291-treated mice (5 mg/kg IP twice daily), SW033291-treated mice (10 mg/kg IP twice daily), and 15-PGDH knockout mice. N = 6 mice for each data point. Analysis of variance (ANOVA) P < 10−6. *P < 0.016 in daily subgroup comparisons to vehicle-control–treated mice, Student’s t test. (D) Graph depicts percentage of total colon length covered by ulcers on study day 15, determined by histomorphometric assessment [as in (B)] for DSS-treated mice in cohorts of: vehicle-control–treated mice (N = 7), SW033291-treated mice (5 mg/kg IP twice daily) (N = 8), SW033291-treated mice (10 mg/kg IP twice daily) (N = 8), and 15-PGDH knockout mice (N = 9). ANOVA P < 0.003. *P < 0.05 versus vehicle-control-treated mice, Student’s t test. (E) (Left) Representative photomicrograph of grade 2 cryptitis and mucosal thickening (on a 0 to 4 scale). (Right) Normal colon. Scale bars, 100 μm. (F) Graph of mean murine endoscopic index of colitis severity score (MEICS) in same mice as in (C). ANOVA P < 10−6. *P < 0.024 in daily comparisons to vehicle-control–treated mice, Student’s t test. (G) Graph of mean total colon cryptitis score (0 to 16 scale) on study day 8 for DSS-treated mice in cohorts of vehicle-control–treated mice (N = 7), SW033291-treated mice (5 mg/kg IP twice daily) (N = 8), SW033291-treated mice (10 mg/kg IP twice daily) (N = 8), and 15-PGDH knockout mice (N = 6). ANOVA P < 0.0005. *P < 0.022 versus vehicle-control–treated mice, Student’s t test. All graphs show means ± SEM.

To identify the cellular mediators responsible for the protective effect of 15-PGDH inhibition, we constructed chimeric mice from 15-PGDH wild-type mice that received bone marrow transplants from either 15-PGDH wild-type or 15-PGDH knockout marrow donors. The two groups of chimeric mice showed identical susceptibility to DSS-induced colitis as measured by disease activity, weight loss, and total percentage of ulcerated mucosa (figs. S17 and S18). These results suggest that colitis protection is due to inhibition of 15-PGDH in colonocytes, which are the primary site of 15-PGDH expression in the colon (33), rather than to inhibition of 15-PGDH in bone marrow–derived cells.

To further characterize the mechanism by which 15-PGDH inhibition protects the colon epithelium, we assessed the effects of SW033291 on cell proliferation in the colon crypts. On day 8, after 7 days of oral DSS treatment, wild-type mice receiving concurrent injections with vehicle-control showed significant suppression of bromodeoxyuridine (BrdU) incorporation in colonic crypts (Fig. 5 and fig. S19). In contrast, day 8 BrdU incorporation was increased 2.5-fold in SW033291-treated animals (Fig. 5, A and B, 10 mg/kg SW033291) and increased 2.4-fold in 15-PGDH knockout mice (Fig. 5, A and C). Immunofluorescence staining for cleaved caspase 3 showed only scattered apoptotic cells, and these were present in similar amounts in control and SW033291-treated mice. Thus, 15-PGDH inhibition appears to confer protection primarily by helping maintain colonocyte proliferation in the DSS-damaged mucosa.

Fig. 5 Inhibition of 15-PGDH increases BrdU incorporation in colonic crypts.

BrdU incorporation assessed in crypts of DSS-treated mice. (A) Immunofluorescence visualization of BrdU incorporation (red), E-cadherin-stained crypts (green), and 4′,6-diamidino-2-phenylindole–stained nuclei (blue) from distal colon. (Top) Photomicrographs of 15-PGDH wild-type mice (15-PGDH+/+) at pre-DSS baseline day 1 (D1) and then at day 8 (D8), after 7 days administration of DSS concurrent with twice daily intraperitoneal injections with either 0, 5, or 10 mg/kg SW033291. (Bottom) Photomicrographs of 15-PGDH knockout (15-PGDH−/−) mice at D1, and D8 upon completion of DSS. Scale bars, 50 μm. (B) Graphical plots of average number of BrdU-positive cells per crypt in mice from top panel of (A). N = 6 mice for day 1 and N = 8 mice for each day 8 data point. ANOVA P < 0.0001. *P < 0.013 versus day 8 mice injected with 0 mg/kg SW033291 (i.e., vehicle-control), Student’s t test. (C) Graphical plots of average number of BrdU-positive cells per crypt in day 1 (D1) baseline and day 8 (D8) DSS-treated mice that are 15-PGDH WT versus 15-PGDH (KO). N = 6 to 8 mice per group. ANOVA P < 0.0001. *P < 3 × 10−5, Student’s t test. All graphs show means ± SEM.

15-PGDH inhibition promotes liver regeneration

We next examined the effects of 15-PGDH inhibition on regeneration in the liver. Inhibiting 15-PGDH increases PGE2 levels in this organ (Fig. 1, A and B) and, provocatively, COX-2 inhibitors antagonize regrowth of the liver after surgery (11, 34, 35). We found that after resection of two-thirds of the liver mass (35), both 15-PGDH knockout mice and SW033291-treated wild-type mice showed a markedly increased rate and extent of liver regeneration as compared with wild-type mice treated with vehicle-control (Fig. 6, A to C). Although the 15-PGDH knockout and wild-type mice had equal preoperative liver weights, the knockout mice attained 28% and 40% greater total liver weight than controls on postoperative days 3 and 4, respectively (Fig. 6, A and B). Liver regeneration was similarly greater in SW033291-treated mice, which attained higher liver weights than controls on each of postoperative days 2 to 7 (Fig. 6C). Between postoperative days 1 to 7, SW033291-treated mice regenerated 77% greater new liver weight (0.48 g increase) than did controls (0.27 g increase) (Fig. 6C). Hepatocytes in SW033291-treated mice displayed a burst of proliferation on postoperative day 2: 40% of hepatocytes showed labeling with BrdU in treated mice versus 16% in vehicle-control–treated mice (Fig. 6, D to F).

Fig. 6

Liver regeneration in mice after partial hepatectomy. (A and B) Box plots of regenerated liver weight (grams) attained after partial hepatectomy in control and 15-PGDH knockout mice. (A) Postoperative day 3 (D3). N = 10 control mice and 3 knockout mice. *P < 0.0014, Student’s t test. (B) Postoperative day 4 (D4) N = 10 control mice and 6 knockout mice. *P < 0.0005, Student’s t test. (C) Total liver weight (grams) attained after partial hepatectomy (y axis) in mice injected intraperitoneally twice daily with vehicle-control or SW033291 at 5 mg/kg across postoperative days 1 to 7 (D1 to D7). Each value is mean of N = 10 mice. *P < 0.011, Student’s t test. (D to F) BrdU incorporation after partial hepatectomy. (D) Percentage of BrdU-positive cells (y axis) in vehicle-control–treated versus SW033291-treated mice on postoperative days 1 to 7 (D1 to D7) (x axis). Each value is the mean of from 6 to 10 mice. Error bars are smaller than symbols at nearly all data points. *P < 1.0 × 10−17 for postoperative day 2 comparison of SW033291-treated and control mice, Student’s t test. [(E and (F)] Representative high-powered micrograph fields of BrdU staining hepatocytes on postoperative day 2 in SW033291-treated mice (E) and vehicle-control–treated mice (F). Scale bar, 100 μm. All graphs show means ± SEM.

Consistent with SW033291 up-regulating prostaglandin signaling, regenerating livers of SW033291-treated mice showed 36 to 55% higher levels of cyclic AMP from postoperative days 1 to 3 (P < 0.002, Student’s t test) (fig. S20). However, we did not find differences between livers of SW033291-treated versus control mice in other downstream signaling targets, including levels of beta-catenin protein; nuclear localization of beta-catenin; and levels of phosphorylation at cAMP-dependent protein kinase target sites on CREB serine 133, beta-catenin serine 675, and beta-catenin serine 552 (10, 11).

Discussion

In summary, we have shown that 15-PGDH negatively regulates tissue regeneration and repair in the bone marrow, the colon, and the liver. Although bone marrow, colon, and liver each demonstrate very different tissue architectures, in each of these tissues this function of 15-PGDH is conserved, suggesting a biological generality of this pathway. PGE2 and dmPGE2 show activities that favor tissue regeneration when used to treat cells in culture and when administered to animals in pharmacological doses (11, 1316, 18, 2427). Our findings further imply that in normal in vivo physiology, prostaglandins play a widespread homeostatic role in the maintenance and repair of injury across multiple body tissues. Indeed, a doubling of tissue PGE2 levels can markedly potentiate tissue repair. Of most direct utility, our observations identify 15-PGDH as a therapeutic target and provide a chemical formulation, SW033291, that is an active 15-PGDH inhibitor in vivo and that potentiates repair in multiple tissues.

Inhibitors of 15-PGDH, such as SW033291, may have applicability to human HSC transplantation by potentially increasing transplant safety and reducing the required time of hospitalization (1). The mechanism of SW033291, which activates the bone marrow niche of transplant recipients, is distinct from the effect of treating donor hematopoietic stem cells ex vivo with dmPGE2. Ex vivo treatment with dmPGE2 instead activates the donor stem cells, for example by up-regulating donor cell expression of CXCR4 (the CXCL12 receptor) (15, 16). These previous and current studies together thus identify, respectively, two independent PGE2-responsive pathways for potentiating HSC transplants, one targeted by the ex vivo dmPGE2 treatment of donor hematopoietic stem cells and the other by in vivo SW033291 administration to HSC transplant recipients. The use of granulocyte colony-stimulating factor to assist neutrophil recovery after HSC transplant provides one precedent for the therapeutic utility of tissue regeneration approaches (36, 37). Distinguishing features of SW033291 are that in addition to accelerating neutrophil recovery (by a different approach), this 15-PGDH inhibitor accelerates regeneration of platelets and red blood cells and also expands the numbers of more primitive bone marrow SKL cells. We anticipate that 15-PGDH inhibitors would potentiate bone marrow regeneration after HSC transplants that employ either autologous or allogeneic derived stem cells, which are most often harvested from the donor’s peripheral blood (1). Because SW033291 also enables successful murine bone marrow reconstitution from a reduced inoculum of marrow cells, SW033291 may additionally have applicability to HSC transplants that use human umbilical cord blood stem cells, in which numbers of donor cells are limited by supply and expense (2, 3).

Inhibitors of 15-PGDH, such as SW033291, may also have applicability to treatment of human ulcerative colitis. Mucosal healing is increasingly recognized as an important therapeutic goal in the treatment of this disease (5, 38). The activity of SW033291 in stimulating colon epithelial regeneration in the mouse DSS colitis model suggests potential applicability to this treatment need. Inhibitors of 15-PGDH, such as SW033291, may also have therapeutic applicability to patients undergoing surgical resection of primary liver tumors or of colon cancers metastatic to the liver. In both these cases, the patient’s eligibility for surgery is limited by the requirement that the postoperative liver remnant be sufficient to enable regenerating an adequate liver mass (68).

Previous studies from our group indicated that 15-PGDH shows tumor suppressor activity in colon cancer but that many individuals have low 15-PGDH expression levels in their colonic mucosa (22, 3941). The current observations suggest that one explanation for this paradox may be that in certain biological contexts, such as in bowel injury, low levels of colonic 15-PGDH may provide a biological advantage by potentiating faster organ recovery. More generally, this study provides an example in which the regulated and short-term inhibition of a tumor suppressor gene may provide therapeutic utility without adverse effect, particularly in a disease context in which tissue repair is the immediate medical goal.

Supplementary Materials

www.sciencemag.org/content/348/6240/aaa2340/suppl/DC1

Materials and Methods

Figs. S1 to S20

Table S1

References (4246)

References and Notes

  1. Acknowledgments: We thank the members of the University of Texas Southwestern High-Throughput Screening (HTS) Core facility for their support of the high-throughput screen; A Dannenberg, D. Montrose, and M. Datt for helpful scientific discussions; and K. Lingas, B. Maitra, and J. Ren for helpful technical support. This work was funded by NIH grants P50 CA150964 (S.D.M.), P30 CA043703 (S.L.G.), R25CA148052 (Y.Z.), U54HL119810 (Paul DiCorleto), 1P01CA95471-09 (Steven McKnight and J.M.R.), and 5P30 CA142543-03 [The Simmons Cancer Center (J.K.V.W.)], which support the University of Texas Southwestern HTS Core Facility; P50 CA130810 (Dean Brenner); P30 DK097948 (F.C.); a Scholar Innovator Award from the Harrington Discovery Institute (S.D.M.); grants from the Marguerite Wilson Foundation (S.D.M.), Inje University (S.Y.Y. and K.B.B.), and National Research Foundation, Ministry of Science, ICT and Future Planning of Korea 2014002213 (K.B.B); the Welch Foundation (I-1612, J.M.R.); Cancer Prevention Research Institute of Texas (RP110708); a medical research fellowship from the Howard Hughes Medical Institute (L.H.); and gifts from the Leonard and Joan Horvitz Foundation (S.D.M) and the Richard Horvitz and Erica Hartman-Horvitz Foundation (S.D.M.). Case Western Reserve University, the University of Texas, and the University of Kentucky have filed patent applications (PCT/US2013/036790 and PCT/US2013/036790) related to SW033291 and other 15-PGDH inhibitors and their use in the treatment of disease.
View Abstract

Navigate This Article