Mast Cell 5-Lipoxygenase Activity Promotes Intestinal Polyposis in APC^Δ468 Mice

The 5-Lipoxygenase (5LO), an essential enzyme in the arachidonic acid pathway, produces leukotriene B4 (LTB4) and the cysteinyl leukotrienes, a group composed of leukotriene C4 (LTC4), leukotriene D4 (LTD4), and leukotriene E4 (LTE4) (1–3). Several studies have implicated a role for 5LO in the development of lung, pancreatic, and in particular, colon cancer (4–12). In colon cancer, our previous studies and others have investigated the direct effects of 5LO byproducts in normal and tumor epithelial cells alone (10–12). However, there exists the distinct and as yet unexplored possibility that 5LO may exert its tumorigenic effect through the hematopoietic system.

The mast cell is a hematopoietic component that we have recently shown to have an important role in intestinal tumor development (13). In particular, we found that functional depletion of mast cells led to a regression of murine intestinal polyposis with decreased proliferative indices. We also showed that mast cells alter the tumor immune response (14). The immune system is believed to mediate the body's recognition and rejection of tumor cells. Conversely, dysfunction of the immune system allows for tumors to escape recognition and grow uninhibited. There are an abundance of candidate factors through which mast cells may exert their tumorigenic and immunomodulating effects, including TNF-α, IL-1, IL-6, and TGF-β (15).

Mast cells have also been shown to be carriers of the 5LO enzyme and produce significant amounts of 5LO byproducts (16). 5LO metabolism has been shown to be directly mitogenic to colon cancer cell types (10–12). LTB4 also chemoattracts multiple hematopoietic cell lineages involved in the tumor immune response (16–18). It is therefore possible that 5LO mediates the mast cell's direct effect on tumor growth as well as its effects on the immune response to tumor.

The myeloid-derived suppressor cell (MDSC) is a hematopoietic mediator of immunologic dysfunction that has been shown to contribute to tumorigenesis. MDSCs directly inhibit the tumor-specific response by CD8⁺ T cells (19–21). Furthermore, studies have shown that inhibition of MDSC tumor recruitment results in an increased tumor-specific CD8⁺ T cell-mediated response (22, 23). Much attention has thus been given to identifying the chemokines responsible for MDSC recruitment. However, the list of known chemotactic factors for MDSCs is limited and includes stem cell factor, S100, PGE2, and CCL-2 (22, 24–26). 5LO byproducts have been shown to recruit several myeloid lineages that share a similar derivation as MDSCs, including neutrophils and eosinophils (17, 18). It is therefore possible that 5LO byproducts also recruit MDSCs.

Utilizing a murine, APC-driven model of polyposis, we present the novel finding that genetic deletion of 5LO and more specifically hematopoietic 5LO, profoundly abrogates polyp burden. Mast cells isolated from 5LO-deficient mice, which are incapable of producing 5LO byproducts, have a significantly reduced capacity in promoting intestinal epithelial proliferation. We also show that LTB4 chemoattracts MDSCs and that 5LO-deficient mast cells have a diminished ability to recruit MDSCs when compared with mast cells with the 5LO enzyme. Correspondingly, we observed a significant decrease in MDSC polyp infiltration with 5LO deficiency. Furthermore, we found that MDSCs from 5LO-deficient mice had diminished immunosuppressive arginase-1 activity. Our study shows that hematopoietic 5LO, and particularly mast cell 5LO, promotes polyposis through its direct mitogenic effects on murine intestinal epithelium as well as through its recruitment of MDSCs. These results provide evidence that hematopoietic 5LO plays a key role in early tumorigenesis.

All animal work was approved and conducted under the guidelines of Northwestern University's Animal Care and Use Committee. APC^Δ468/5LO^+/− mice were crossed with 5LO^+/− mice to generate APC^Δ468/5LO^−/− mice. APC^Δ468 were used as positive controls. All mice were of the C57BL/B6 genetic background. Offspring were genotyped by PCR. Mice were sacrificed at 4 months of age.

APC^Δ468 mice were lethally irradiated at 6 weeks of age. Bone marrow-derived mononuclear cells were harvested from either B6 or 5LO^−/− donors and lineage-depleted using the Multenyi MS columns according to manufacturer's instruction. APC^Δ468 mice were then reconstituted with mononuclear cells from one of the 2 donor types. Mice were sacrificed at 4 months of age.

Cysteinyl leukotriene and prostaglandin E metabolite levels were measured using Cayman Chemical enzyme immunoassay kits, according to manufacturer's instruction.

A 5-μm thick sections of frozen or formalin-fixed paraffin embedded intestine were cut. Formalin-fixed sections were deparaffinized and rehydrated in graded alcohol. Antigen retrieval was done using 1X Target retrieval solution (DAKO), whereas the frozen sections were fixed in ice-cold methanol for 15 minutes. Frozen and formalin-fixed sections were incubated with 1% bovine serum albumin for 30 minutes at room temperature. Subsequently, sections were incubated either with anti-mouse Mac3 (BD Pharmingen) and biotinylated anti-mouse Gr1 (BD Pharmingen) or anti-mouse CD4 (Abcam) and anti-mouse FoxP3 (eBioscience) or anti-mouse CD8 (Santa Cruz Biotechnology) and anti-mouse FoxP3 or anti-mouse CD11c (BD Pharmingen) and anti-mouse IA^b (Abcam) or anti-mouse CD11b (BD Pharmingen)/CD3 (BD Pharmingen) and 5LO (Abcam) overnight at 4°C. Primary antibodies were washed twice the next day with PBS and incubated with appropriate Alexafluor-488 and Alexaflour-594 secondary antibodies (Invitrogen) for 1 hour. Finally, sections were washed twice with PBS and incubated for 15 minutes with 4′,6-diamidino-2-phenylindole (Sigma), washed twice with PBS and mounted with antifade mounting medium. Images were acquired using TissueGnostics Tissue/Cell High Throughput Imaging and Analysis System.

Mice were injected with 400 μL of 10 mg/mL BrdU (Sigma) 2 hours prior to cull. Paraffin sections (4 μm) were stained for BrdU using the BD Pharmingen BrdU in situ detection kit.

The ApopTag Red in situ apoptosis detection kit from Chemicon International (catalog no. S7185) was used according to the manufacturer's instructions.

MNCs were prepared as previously described (27). Briefly, homogenized intestine was agitated in a solution containing collagenase-type 4 (Worthington Biochemical Corporation) in RPMI medium 1640 at 37°C for 20 minutes. Cells were isolated using a 40%–60% Percoll gradient and washed twice.

Frequencies of MCp in MNCs were measured as described previously (27). MNCs were suspended in 100 mL of RPMI medium consisting of 20 ng/mL of stem cell factor (SCF), 20 ng/mL of IL-3, and 100,000 irradiated spleen cells per well (3,000 rad). The 10,000 MNCs were plated per well in a 96-well plate and diluted serially in subsequent rows. After 15 days incubation at 37°C with 5% CO2, colonies of matured MC were scored.

Murine BMMCs were obtained by culturing femoral BM cells at 5 × 10⁵ cells/mL in complete RPMI 1640 with 5 ng/mL murine IL-3 and 12.5 ng/mL SCF. Media and flask were changed twice a week for 6 weeks and were maintained at 37°C in 5% CO2. At 6 weeks, conditioned media was obtained by replating cells at the same 5 × 10⁵ cells/mL density in serum-free RPMI with the same concentrations of IL-3 and SCF.

IMCE cells were kindly provided by Dr. Robert Whitehead of Vanderbilt University (28). IMCE cells were plated at a density of 5 × 10⁴ cells/mL in a 96-well plate (33° permissive temperature, 5% CO₂). Cells were treated with BMMC conditioned media from either APC^Δ468 or APC^Δ468/5LO^−/− mice. Both conditioned medias were supplemented with IFN-γ (5 U/mL). About 1 μCi ³H-thymidine was added for a 14-hour incubation at 24, 48, and 72-hour time points. Incorporation was measured using a Perklin-Elmer MicroBeta plate counter.

FACS acquisition was done with BD FACSCanto. Data analyses used FlowJo software (Tree Star). Gr-1-biotin, Gr-1-APC, CD11b-FITC, CD11b-PE, Ly6C-PerCP, and Ly6G-APC were purchased from BD Biosciences. L/D was purchased from Invitrogen.

About 96-well chemotaxis plates (Neuroprobe) were used as previously described (29) with the following modifications: MDSCs were suspended at 2 × 10⁶ cells/mL of serum-free RPMI and 50 μL were applied to the top wells. 300 μL of serum-free RPMI at varying concentrations of LTB4, LTD4, and LTE4 were added to the lower wells. In some experiments, lower wells were plated with BMMC conditioned media from APC^Δ468 or APC^Δ468/5LO^−/− mice. For rescue experiments, 10 nmol/l LTB4 was added to APC^Δ468/5LO^−/− BMMC conditioned media. Plates were incubated for 3 hours at 37°C and migrated cells were counted using inverted microscopy. Assays were done in duplicate in 3 separate experiments.

About 10⁶ Gr1+ cells were plated for 18 hours in the presence of IFN-γ and LPS. Supernatant was used for NOS detection (Cayman Chemical) according to manufacturer's instruction. Cells were suspended in 500 μL PBS, lysed using sonication, and centrifuged. Supernatant was used for arginase assay (BioAssay Systems), according to manufacturer's direction.

Upon oxidization, hydroethidine intercalates with cellular DNA and emits at approximately 600 nm. MNCs were incubated with 1 μmol/L hydroethidine (Invitrogen) at room temperature for 30 minutes, then washed with PBS. Cells were then analyzed using flow cytometry.

Our lab has previously shown that 5LO is upregulated in human colon polyps and cancer as compared with normal colon tissue (10). In order to further investigate these findings, we utilized the APC^Δ468 mouse, which bears a truncated APC gene and develops a severe polyposis by 4 months (30). We hypothesized that 5LO would be increased in the intestines of APC^Δ468 mice as it is in human colon polyps. Analogous and supplemental to our previous findings in human colon tissue, we found both a local and systemic elevation of 5LO activity in the APC^Δ468 mouse.

5LO was upregulated in the intestinal lymphoid tissue of APC^Δ468 intestine compared to wild-type controls (Fig. 1A). Furthermore, enzyme immunoassays (EIA) were done for the 5LO byproduct group, cysteinyl leukotrienes. These were elevated in the sera of APC^Δ468 mice compared with wild-type controls (Fig. 1B v). In order to further delineate, which cells were responsible for systemic elevation of 5LO metabolites, we stained for the CD11b and CD3 markers to evaluate for the myeloid and lymphoid populations, respectively. We found that T-cells (CD3⁺) expressed 5LO whereas myeloid cells (CD11b⁺) did not colocalize for 5LO in the intestinal lymphoid tissue and spleens of APC^Δ468 mice (Fig. 1B –C).

Given the observed upregulation of 5LO in the APC^Δ468 mouse, we examined the effects of 5LO gene knockouts on this same model. 5LO^−/− mice were bred to APC^Δ468 mice and their phenotypes were evaluated at 4 months. 5LO null mutation led to a dramatic reduction in the number and size of intestinal polyps (Fig. 2A and B). The APC^Δ468 group developed 79 ± 4 polyps with a median polyp diameter of 2.7mm (Fig. 2I and J). In contrast, APC^Δ468/5LO^−/− mice had 46 ± 6 polyps with a median diameter of 1.2 mm, an approximately 40% (P < 0.005) and 44% (P < 0.0001) reduction, respectively (Fig. 2I and J). The APC^Δ468 group also had increased mitotic indices and decreased apoptosis compared with the APC^Δ468/5LO^−/− group (Fig. 2C – F, K, and L). Based on these observations, we concluded that 5LO significantly contributes to intestinal tumor development.

As 5LO was increased specifically in the intestinal lymphoid tissue of APC^Δ468 mice (Fig. 1A), we hypothesized that hematopoietic 5LO was primarily responsible for the observed tumorigenesis. To test our hypothesis, we conducted adoptive transfer experiments using APC^Δ468 littermates as recipients and reconstituted them with bone marrow from either healthy wild-type (wt BM) mice or healthy 5LO^−/− (5LO BM) mice. At 4 months of age, mice were evaluated for phenotype. The wt BM group had a similar degree of polyposis as the parental APC^Δ468 group with a mean of 112.8 ± 9.075 polyps (Fig. 2G and m). In contrast, 5LO BM mice had a profound reduction in their polyp burden by approximately 91%, with 9.8 ± 2.871 polyps (P < 0.001) (Fig. 3H and M). The degree to which polyposis was reduced in the 5LO BM mice precluded any statistically significant evaluation of polyp diameter (Supplementary Fig. S1), mitotic indices, or apoptosis.

Our next goal was to identify a hematopoietic lineage through which 5LO was tumorigenic. We first quantified total intestinal mast cell presence in the 5LO-deficient groups compared with their controls. We found that mast cells were significantly reduced in the polyps of the APC^Δ468/5LO^−/− and 5LO BM groups compared with their respective APC^Δ468 and wt BM controls (Fig. 3). We also observed a decrease in polyp infiltration by Mac3⁺ and Gr1⁺ cells (Fig. 3). As our lab had previously found the mast cell to have a critical role in polyposis, we further explored its potential tumorigenic role in the context of 5LO deficiency.

In addition to a quantitative difference of mast cells being present between the 4 mouse groups, we also explored the possibility of a qualitative difference. We harvested conditioned media from mast cells derived from APC^Δ468 and APC^Δ468/5LO^−/− mice, and compared the effects of media treatment from these respective groups on the proliferation of Immorto-Min colonic epithelial (IMCE) cells. Harboring an APC min mutation, IMCE is an immortalized intestinal epithelial cell line (28). Measuring thymidine incorporation, APC^Δ468 mast cells potentiated IMCE proliferation to a significant degree over APC^Δ468/5LO^−/− mast cells at 24, 48, and 72 hours (Fig. 4A). Furthermore, APC^Δ468 mast cells increased proliferation compared with standard nonconditioned media at 48 and 72 hours (Fig. 4A).

We next evaluated if 5LO modulates the tumor immune response. The 5LO byproduct LTB4 has been specifically shown to chemoattract a variety of hematopoietic lineages, including neutrophils and eosinophils, which share similar myeloid derivation as MDSCs (16, 17). We hypothesized that MDSCs might be similarly recruited to tumor by LTB4. MDSCs were isolated from APC^Δ468 bone marrow using Gr-1 column selection, which yielded >90% purity for Gr-1⁺CD11b⁺ MDSCs. 96-well chemotaxis plates were used with MDSCs plated in the top wells. LTB4, LTD4, and LTE4 were applied to the lower wells in concentrations of 1 nmol/L, 10 nmol/L, and 100 nmol/L. LTB4 induced migration of MDSCs in a concentration-dependent manner up to 10 nmol/L (Fig. 4B). LTD4 and LTE4 did not chemoattract MDSCs in any concentration (Fig. 4B).

Mast cells have been shown to produce LTB4 in a significant concentration of 10.6 nmol/L ex vivo (16). We therefore evaluated if the mast cell affects the tumor immune response through an ability to chemoattract MDSCs via LTB4. We compared mast cells with and without the 5LO enzyme and evaluated their respective capacities to chemoattract MDSCs. Bone marrow-derived mast cells (BMMCs) were isolated from APC^Δ468 and APC^Δ468/5LO^−/− mice and grown in culture for 8 weeks with SCF and IL-3. Conditioned media was harvested and replated into the lower wells of chemotaxis plates. MDSCs were plated in the top wells. Indeed, conditioned media from APC^Δ468/5LO^−/− BMMCs chemoattracted MDSCs to a significantly lower degree than did conditioned media from APC^Δ468 BMMCs (Fig. 4C). Furthermore, addition of LTB4 (10 nmol/L) to the APC^Δ468/5LO^−/− media resulted in a rescue of the chemoattraction of MDSCs (Fig. 4C).

If LTB4 recruits MDSCs, deficiency in 5LO would therefore mitigate MDSC recruitment, thus enhancing the antitumor response and decreasing polyposis. To evaluate the capacity of 5LO to recruit MDSCs to tumor, in vivo, we analyzed Gr-1⁺CD11b⁺ MDSC presence in the polyp, mesenteric lymph nodes, spleen, and bone marrow of APC^Δ468/5LO^−/− and APC^Δ468 mice. To ensure that any potential reduction in MDSC recruitment in 5LO deficient mice was not solely a function of their diminished tumor burden, polyps were microdissected from both groups and compared. With polyp-to-polyp comparison, Gr-1⁺CD11b⁺ MDSCs were still significantly reduced by approximately 4 fold (Fig. 5A and B).

We also found that 5LO expands MDSCs in peripheral lymphoid organs. There was a significant reduction in Gr-1⁺CD11b⁺ MDSCs isolated from the mesenteric lymph nodes and spleens of APC/5LO^−/− mice, by approximately 3 fold, respectively (Fig. 5A and B). Interestingly, there was a trend of near significance (P = 0.09) toward increased Gr-1⁺CD11b⁺ MDSCs in the bone marrow compartment of APC/5LO^−/− mice. This may suggest a buildup of unrecruited MDSCs (Fig. 5A and B) that remain at their point of origin, the bone marrow. 5LO BM mice showed a similar reduction in Gr-1⁺CD11b⁺ MDSC recruitment in the gut (66%), mesenteric lymph node (53%), and spleen (47%) (Fig. 5C). These findings suggest that MDSC recruitment to the tumor site, mesenteric lymph nodes, and spleen, is 5LO-dependent.

Two distinct subsets of MDSCs were found during the analysis of Gr-1⁺CD11b⁺ cells, consistent with previously published literature (31, 32). The predominant subpopulation was the granulocytic MDSC that is identified by its CD11b⁺Ly6G⁺Ly6C^low phenotype (Fig. 6A). Monocytic MDSCs, with the epitopes CD11b⁺Ly6G⁻Ly6C^hi, composed the second subtype and was generally present to a lesser degree than granulocytic MDSCs from the same mouse (Fig. 6A). Only the granulocytic subpopulation was reduced in APC^Δ468/5LO^−/− mice and 5LO BM mice compared with their APC^Δ468 and wt BM controls (Fig. 6B and C). Monocytic MDSCs, on the other hand, were not reduced to a significant degree, except in the gut of 5LO BM mice. As there were not enough polyps in the 5LO BM intestine to analyze their MDSCs, the majority of intestine evaluated was grossly normal from this group. We therefore compared these MDSCs with MDSCs isolated from adjacent normal intestine in the wt BM group. In the gut, we found a distinct subpopulation with the epitopes, CD11b⁺Ly6G⁺Ly6C^hi (Fig. 6A). This population was reduced in the 5LO BM group in similar fashion as the granulocytic MDSCs from the spleen and mesenteric lymph nodes.

We further analyzed the activity of arginase, ROS, and NOS by the wt BM and 5LO BM groups. The wt BM group had a statistically significant greater degree of arginase-1 activity than did the 5LO BM group (Fig. 7A). However, we did not find a significant difference in ROS or NOS production (Fig. 7B and C). 5LO seems to play a significant role in the recruitment of granulocytic MDSCs and in the activity of arginase-1 in MDSCs.

At the time of this report, a study implicated the IL-17 pathway in the mobilization of MDSCs (33). We have previously reported finding elevated IL-17 in polyposis and suggested that polyposis is a Th17-driven disease (14, 34). Both CD4+ T-cells and antigen presenting cells contribute to long-term Th17 inflammation. Therefore, in order to understand the impact of 5LO deficiency on the T-cell compartment and on antigen presenting cells, we evaluated the infiltration of polyps by CD4⁺, CD4⁺FoxP3⁺, CD11c⁺IA^b+, and CD8⁺ cells in the APC^Δ468 and APC^Δ468/5LO^−/− experimental groups. There was no statistically significant difference observed in the infiltration by CD8⁺ cells between these 2 experimental groups (Supplementary Fig. S3). However, CD4⁺, CD4⁺FoxP3⁺, and CD11c⁺IA^b+ cell infiltration was significantly reduced in the polyps of 5LO-deficient mice (Supplementary Fig. S3).

There are a limited number of studies that have evaluated the role of 5LO in intestinal tumorigenesis. Of these studies, all have focused on the effects of 5LO activity on the epithelial cell alone. The 5LO byproduct, LTD₄, was shown to increase proliferation and cell survival in human intestinal epithelial cells (35). We have previously shown that 5LO inhibitor treatment of mice xenografted with human colon cancer significantly reduced tumor volume (10). In the current study, we have observed diminished tumor burden specifically through 5LO genetic deletion and have provided evidence for a hematopoietic mechanism.

Recent studies show that signals from myeloid cells, mast cells in particular, are critical determinants of tumor behavior regulating growth and the tumor immune response (13, 14). APC^Δ468 mice reconstituted with 5LO-deficient bone marrow experienced a profound abrogation in their polyposis, even though the intestinal epithelial compartment of these mice had a constitutive APC mutation and were 5LO-proficient. We found that the mast cell is an important effector of the tumorigenic properties of hematopoietic 5LO. Previous studies have shown that LTB4 is a potent chemoattractant of mast cells (16). In line with these prior findings, we observed a significant decrease in mast cell polyp infiltration in the APC^Δ468/5LO^−/− and 5LO BM groups, compared with their respective controls. 5LO deficiency also resulted in a decrease in the polyp infiltration by Mac3⁺ cells, which have also been shown to utilize 5LO for many of their effector functions (36, 37). Our findings further indicate that homing and expansion of mast cells and macrophages are 5LO-dependent.

We showed that the mast cell population from APC^Δ468 mice potentiated proliferation of IMCE murine enterocytes to a significantly greater degree at 3 different time points than mast cells from the APC^Δ468/5LO^−/− group. When LTD4 was originally shown to increase intestinal epithelial cell proliferation, it was thought to be part of an autocrine system. The epithelial cells were believed to produce their own LTD4, providing for their own survival and proliferation. Our results show that mast cells may alternatively have a paracrine growth effect, wherein mast cells produce 5LO byproducts that potentiate the proliferation of the intestinal epithelium.

The mast cell is also tumorigenic through its effect on the tumor immune response. By being a hematopoietic source for LTB4 in the APC^Δ468 model, mast cells recruit MDSCs to the tumor site. At the time of this report, a separate group described a capacity of mast cells to potentiate the accumulation of MDSCs in a xenograft model for hepatocellular carcinoma (33). They suggested that this regulation was IL-17-mediated. We herein show that mast cells specifically chemoattract MDSCs, and that a significant portion of this recruitment is 5LO-mediated. BMMCs produce LTB4 at 10 nmol/L ex vivo, the concentration that we found was optimal for chemoattracting MDSCs (16). In fact, MDSCs seem to become less reactive to LTB4 with concentrations higher than 10 nmol/l. This exquisite sensitivity separates MDSCs from other myeloid lineages like neutrophils and mast cells, which require LTB4 at a peak concentration of 100 nmol/L (16, 38). Furthermore, 5LO deficiency in mast cells significantly reduced their ability to chemoattract MDSCs.

In vivo, 5LO deficiency markedly reduced MDSC presence in the spleen, mesenteric lymph nodes, and gut. In response to cancer, murine MDSCs are mobilized from the bone marrow to the peripheral lymphoid organs and tumor site (39). We found that 5LO contributes to an expanded presence of MDSCs in the secondary lymphoid organs and polyp. 5LO seems to play a role in the chemokinesis specifically of granulocytic MDSCs, which were significantly reduced in the spleen and mesenteric lymph nodes of 5LO-deficient mice. Monocytic MDSCs, on the other hand, were not changed to a statistically significant degree, with the exception of monocytes infiltrating the intestine of APC^Δ468 mice with 5LO-deficient BM.

We also found that within the gut, granulocytic MDSCs seemed to shift toward a CD11b⁺Ly6G⁺Ly6C^med phenotype, similar to that of inflammatory peritoneal neutrophils (Fig. 6A) (40). This population was present more in the polyp than in adjacent normal intestine in APC^Δ468 mice. Furthermore, 5LO-deficient APC^Δ468 mouse intestine had a lower percentage of CD11b⁺Ly6G⁺Ly6C^med cells when compared with the APC^Δ468 group, consistent with the reductions observed with splenic and mesenteric lymph node granulocytic MDSCs. This epitope shift reflects increasing evidence that MDSCs have phenotypic and functional overlap with other established myeloid lineages and emphasizes the plasticity of MDSCs. It is also important to acknowledge that eicosanoids and MDSC physiology have been previously linked in the context of COX-2 metabolism (41). This prior study showed that the COX-2 metabolite, PGE2, induced the differentiation of MDSCs. However, we did not find any evidence that the COX-2 pathway was affected in our model, confirming that 5LO affects MDSC physiology in a mechanism distinctly separate from COX-2 metabolism (Supplementary Fig. S4). In summary, 5LO seems to play a significant role in the mast cell's recruitment of granulocytic and CD11b⁺Ly6G⁺Ly6C^med MDSCs.

We also found that 5LO deficiency directly decreases arginase-1 activity in MDSCs. Utilizing various mechanisms, MDSCs suppress the CD8⁺ T-cell-mediated response to tumor. Granulocytic MDSCs produce large amounts of reactive oxygen species (ROS) whereas the other subtype, monocytic MDSCs, produce nitric-oxide synthase (NOS) (32, 42). Both subtypes produce arginase-1 that catabolizes L-arginine, depleting it from the microenvironment thereby inhibiting T-cell proliferation through several known mechanisms (43). We did not observe a difference in ROS or NOS production with 5LO deficiency. Thus, 5LO deficiency inhibits the mast cell-mediated recruitment of MDSCs as well as the immunosuppressive activity of arginase in MDSCs.

Interestingly, 5LO BM mice experienced a greater degree of polyp suppression than did the APC^Δ468/5LO^−/− mice, which are deleted for 5LO in both the epithelial and hematopoietic compartments. There is the possibility that 5LO serves a protective role at the intestinal epithelium, whereby deleting 5LO would remove that protective mechanism. A corollary example may be found in the gastroduodenal mucosa, where arachidonic acid metabolism has a protective role against helicobacter-induced preneoplasia (44). In other words, baseline 5LO activity may be protective in healthy mice whereas too much activity in cancer is pathogenic. In our APC^Δ468/5LO^−/− cross, we may have the benefit of eliminating 5LO overactivity but then also suffer from loss of baseline protection. Nonetheless, polyposis was still significantly suppressed in the APC^Δ468/5LO^−/− mouse as reflected by polyp number, median polyp diameter, decreased mitotic activity, and increased apoptosis.

We did not observe a difference in CD8⁺ cell infiltration in the intestines of the APC^Δ468 and APC^Δ468/5LO^−/− experimental groups. It is important to note, however, that CD8⁺ is upregulated in gut-infiltrating intraepithelial lymphocytes, which are in a tolerized state (45, 46). The infiltration by tumor-infiltrating CD4⁺FoxP3⁻ and CD4⁺FoxP3⁺ cells were significantly decreased in APC^Δ468/5LO^−/− mice compared with APC^Δ468 controls. Although CD4⁺FoxP3⁺ cells have been shown to recruit mucosal mast cells (47) and promote inflammation, regulatory T-cells (Tregs) normally have antiinflammatory properties. However, we recently reported that in tumor-bearing mice, Tregs can switch functions and contribute to the escalation in cancer-associated inflammation (34). Indeed, our data suggest that T-cells are a major source of 5LO activity. This is in line with a proinflammatory function of gut-infiltrating lymphocytes in our model. Thus, along with the observed decrease in MDSCs, the drop in CD4⁺ cells could imply a second mechanism through which 5LO contributes to tumor growth.

Our findings indicate that hematopoietic 5LO significantly contributes to polyp development. In particular, the mast cell utilizes 5LO to potentiate the growth of the intestinal epithelium as well as recruit MDSCs to the polyp site. 5LO also increases the activity of the immune suppressive enzyme arginase-1 in the MDSC. Based on these findings, we propose that targeting 5LO may be an effective therapeutic strategy for polyposis and potentially colorectal cancer.

No potential conflicts of interest were disclosed.

We thank Dr. Steven Rosen for his support. Dr. Guang-Yu Yang is thankfully acknowledged for his advice. We thank Dr. Robert Whitehead of Vanderbilt University for graciously providing our lab with the IMCE cell line.

This work was supported by the Society for Surgery of the Alimentary Tract Career Development Award and the American Medical Association Seed Grant (E.C. Cheon). K. Khazaie is supported by the Circle of Service Award of the Robert H. Lurie Comprehensive Cancer Center and American Cancer Society Research Scholar Grant, 113422 RSG. D.J. Bentrem is supported by a career development award from the Department of Veterans Affairs. H.G. Munshi and D.J. Bentrem are supported by the National Institutes of Health Grant 1RO1CA126888.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.