Abstract
Hyaluronic acid (HA), a constituent of the extracellular matrix, promotes colorectal cancer growth. CD44 is a relevant HA receptor in this context. However, HA is also a ligand for TLR4, a receptor of significance in colorectal cancer. In this study, we examine the relative contribution of HA interactions with CD44 and TLR4 in colon tumorigenesis. Colorectal cancer models included ApcMin/+ mice, azoxymethane/dextran sodium sulfate (AOM-DSS), and CT26 tumor isografts. We used knockout mice and CT26 colorectal cancer cells with CRISPR knockdown of CD44 and TLR4. HA activity was modulated by PEP1 (a 12-mer peptide that blocks HA from binding its receptors), hyaluronidase (which promotes HA degradation), or 4-MU (HA synthesis inhibitor). Blockade of HA binding via PEP1 decreased growth in all colorectal cancer models and in cell culture. The effects were significant in WT and with CD44 deletion, but not with TLR4 deletion. In the AOM-DSS model, mice deficient in CD44 or TLR4 had fewer tumors. CD44- and TLR4-deficient CT26 isografts grew more slowly, exhibiting decreased tumor cell proliferation and increased apoptosis. In vitro, endogenous HA blocked LPS binding to TLR4 suggesting that HA is a relevant TLR4 ligand in colon cancer. Finally, PEP1 enhanced tumor radiation sensitivity in the isograft model. Together, these results indicate that HA binding to TLR4, as well as CD44, plays a key role in colon tumorigenesis. These findings also raise the possibility that an agent that blocks HA binding, such as PEP1, may be useful as an adjuvant therapy in colon cancer.
Introduction
Hyaluronic acid (HA), a glycosaminoglycan polymer, is a constituent of the extracellular matrix. HA, which is secreted by many cell types, is assembled at the cellular plasma membrane by HA synthases (HAS) and extruded into the extracellular space (1). HA expression is increased in injury states including Crohn disease, dextran sulfate sodium–induced colitis in mice, and intestinal radiation injury (2, 3).
HA binds to CD44 that is expressed on the plasma membrane of many cells, including fibroblasts, smooth muscle cells, epithelial cells, and immune cells (4). Splice variants of CD44 have different biological effects. Although most of the biologic effects of HA are mediated by binding to CD44, HA also binds other receptors including toll like receptor 4 (TLR4), a component of the innate immune system (5). TLR4 is widely distributed in the gastrointestinal tract where it mediates the host response to gram-negative bacteria through binding lipopolysaccharide LPS (6). Although LPS is the ligand typically associated with TLR4, several host molecules, including HA, are also TLR4 ligands (5, 7, 8). HA binding to TLR4 promotes epithelial repair in the DSS-colitis and radiation injury models (2). HA binding to TLR4 mediates the normal growth of the intestine and colon including the proliferation of Lgr5+ epithelial stem cells (7, 8). PEP1 is a 12-mer peptide that binds HA and blocks the binding of HA to its receptors, (7–9). PEP1 administration to mice from 3–8 weeks of age impairs intestinal and colonic growth with a marked decrease in intestinal and colonic elongation (8). HA is also radioprotective of intestinal epithelial stem cells; intraperitoneal administration of HA, given prior to total body radiation, results in a decrease in radiation-induced apoptosis and an increase in crypt survival (3).
HA, TLR4, and CD44 are involved in the pathogenesis of colon cancer. HA is expressed in colon cancer where high levels of stromal HA are associated with poor prognosis (10, 11). HA is made by the tumor cells and by multiple cell types in the tumor microenvironment. HA enhances colorectal tumor cell proliferation and motility in vitro and in vivo (12, 13). Inhibition of HA production in SW620 colon cancer cells blocks Matrigel invasion (14). HA binding to CD44 stimulates multiple receptor kinases in HCT116 colon cancer cells leading to increased cell survival as well as cell proliferation, adhesion, and invasion (15). CD44 has been targeted for cancer therapy using DNA vaccines, anti-CD44 mAbs, and nanoparticle-mediated delivery of CD44 siRNA (16–18). APCmin/+ mice, which have a mutation in the APC gene as is seen in human colon cancers, develop small intestinal adenomas (19, 20). Knocking down CD44 with a hairpin RNA reduces the number of adenomas (11).
Many human colon cancers express TLR4, and TLR4 signaling drives tumorigenesis (21). TLR4 polymorphisms are associated with colon cancer (22). In colon cancer increased TLR4 expression is seen in both the tumor cells and in endothelial and stromal cells (23). In the APCmin/+ mouse model, mice that are also deficient in MyD88, a TLR4 adaptor molecule, have fewer adenomas (24). Administration of azoxymethane-dextran sulfate sodium (AOM-DSS) to mice models colitis-associated colon cancer as is seen in ulcerative colitis (18). In this model, mice that are deficient in TLR4 develop fewer and smaller tumors (25). Overexpression of TLR4 in intestinal epithelial cells results in increased epithelial proliferation, an expansion of Lgr5+ crypt epithelial cells and the development of spontaneous duodenal dysplasia (26).
Although both HA and TLR4 promote tumorigenesis in colon cancer, the possibility that HA promotes tumorigenesis through binding to TLR4 has not been addressed. In this study, we examine the relative contribution of HA interactions with CD44 and TLR4 in colon tumorigenesis using colorectal cancer cell lines and three mouse models of colon cancer (APCmin/+ mice, the AOM-DSS model, and CT26 syngeneic tumor isografts). We also examine the potential of blocking these interactions as an adjunctive colorectal cancer therapy.
Methods
Reagents
HA-binding peptide PEP1 (H2N-GAHWQFNALTVR-OH; ref. 9) and scrambled peptide control (H2N-WRHGEALTAVNQ-OH) were from New England Peptide. Azoxymethane (AOM) and 4-methylumbelliferone (4Mu) were from Sigma-Aldrich. DSS was from TdB Consultancy AB.
Animals
Wild-type (WT), CD44−/−, and TLR4−/− mice on a C57BL/6J background, and BALB/cJ mice were from Jackson Laboratories and bred in-house. Animal procedures were carried out in accordance with the Washington University School of Medicine Animal Studies Committee, which approved the protocols.
Cell lines
The CT26 colon cancer cell line, developed from BALB/c mice (27, 28), was purchased from ATCC (product CRL-2638). Mycoplasma testing was performed by the Tissue Culture Support Center at the Washington University School of Medicine (St. Louis, MO). To assess the role of HA binding to CD44 and TLR4, CD44 knockdown (CD44-CRISPR) and TLR4 knockdown (TLR4-CRISPR) cell lines were generated from CT26 WT mouse colon cancer cells using the CRISPR/Cas9 system at the Washington University Center for Advanced Cellular Genetic Technology (St. Louis, MO). Briefly, sgRNAs were designed to target exon one of the genes and validated for cleavage activity in Neuro 2A cells. The wild-type CT26 cells were nucleofected with out-of frame in/dels in all alleles using next-generation sequencing by the Genome Engineering and iPSC Center (GEiC) at Washington University (St. Louis, MO). Clones 2C2 for TLR4-CRISPR and 1G9 for CD44-CRISPR were used in the in vivo and in vitro experiments (Supplementary Fig. S1). Hyaluronic acid in CT26 cells lysate and conditioned media was quantified by ELISA (cat no. LS-F28399, LifeSpan Biosciences).
APCmin/+ mice
APCmin/+ mice present a model of spontaneous intestinal tumorigenesis (19, 20). To determine the effect of blocking HA binding to its receptors on tumor formation and growth in small intestines of APCmin/+ mice, PEP1 was injected (intraperitoneally, i.p.) on alternate days starting at the time of weaning (age 21 days) until mice were sacrificed at age 103 days. Control mice received 0.9% sterile saline. Intestines were processed and photographed and the photographs analyzed as described previously (29).
AOM-DSS colon cancer
AOM-DSS treatment of WT C57BL/6 mice induces colonic tumor formation (25, 30). PEP1 was used to test whether blocking HA binding to its receptors CD44 and TLR4 would reduce AOM-DSS–induced colon cancer. PEP1, or scrambled PEP1 for controls, was given to mice on alternate days starting one day before the start of AOM-DSS treatment. On day one, 8-week-old WT C57BL/6J mice were given a single intraperitoneal injection of azoxymethane (10 mg/kg). One week after receiving azoxymethane, mice were given 2.5% DSS in their drinking water for 1 week, then 2 weeks normal drinking water, followed by a second 1-week course of 2.5% DSS, and finally 2 weeks normal drinking water, for a total of 7-week duration. Mice were sacrificed at the end of 7 weeks (31, 32).
CT26 tumors
CT26 WT, or CD44-CRISPR, or TLR4-CRISPR cells (100,000 cells/100 μL) were injected into the flanks of 8-week-old BALB/c mice to create syngeneic grafts. Syngeneic tumors developed and were allowed to grow to a volume not exceeding an average of 1,250 mm3 in a treatment group. Growth was monitored by measuring tumor volume with digital calipers.
IHC analyses
IHC and immunofluorescence (IF) analyses followed published protocols (7). Primary antibodies were goat anti-BrdU (1:2,000, a gift from Dr. Jeffery I. Gordon, Washington University), rat monoclonal anti-mouse CD44 (1:50; BD Biosciences), rabbit polyclonal anti-TLR4 (1:200; Novus Biologicals), and rabbit polyclonal anti-mouse caspase-3 (1:200; Cell Signaling Technology). Secondary antibodies purchased from Life Technologies were AF488 donkey anti-rat IgG, AF488 donkey anti-rabbit IgG, AF594 donkey anti-rabbit IgG, and AF594 donkey anti goat IgG, used at 1:200 dilution. IF detection of HA was done as described previously (2, 7).
Gene expression
Quantitative PCR was performed using SYBR Green Real Time PCR Master Mix (Thermo Fisher Scientific) in a CFX Real-Time System (Bio-Rad). Primer sequences are listed as follows: Bax forward: GTGGTTGCCCTCTTCTACTTT, and reverse: CTTCATCTCTCCCACCTCATAAC. Bcl2 forward: GTTTGGGTGTGGCCTTTATTC, and reverse: GTCCTGTGATTCTCCCTTCTTC.
Cell viability/proliferation and TUNEL assays
Proliferation was assessed using the WST proliferation assay (Cell Counting Kit-8, Dojindo Molecular Biotechnologies). Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) was performed using an In Situ Death Detection Kit from Roche Applied Sciences.
Statistical analysis
All statistical analyses were performed and/or reviewed by statistician and author (Y. Yan). Student t test was used to determine significance between groups unless otherwise indicated. A P value of ≤0.05 was considered to be significant. The results are shown as mean ± SEM. A linear mixed-effects model was used to describe differential tumor growth between control and experimental groups (Fig. 3J). Kaplan–Meier methods and log-rank test were used for survival analyses.
Results
The HA-binding peptide PEP1 reduced adenoma development in the APCmin/+ mouse intestine
In the normal small intestine, a layer of extracellular HA was found below the basal surface of crypt epithelial cells (Fig. 1A and B). HA was also found in the muscularis mucosa. CD44 was expressed in lamina propria cells and in epithelial cells at the crypt base (Fig. 1C). TLR4 was found in lamina propria and epithelial cells. In the nonadenomatous areas of the APCmin/+ intestines, the distributions of HA, TLR4, and CD44 were similar to that in wild-type mice (Supplementary Fig. S2). In APCmin/+ adenomas, extracellular HA was distributed widely and the adenoma cells expressed abundant TLR4 and CD44 (Fig. 1D–F; Supplementary Fig. S3). To assess the role of HA binding to its receptors in adenoma development, APCmin/+ mice were treated with PEP1 or scrambled PEP1 on alternate days from 21 days to 103 days of age. Mice were sacrificed and the number of adenomas was counted. In each section of the small intestine, the number adenomas in the mice treated with PEP1 was reduced by about 50% (Fig. 1G). In the PEP1-treated mice the adenomas were smaller than in mice treated with scrambled PEP1 (Fig. 1H).
The HA-binding peptide, PEP1, reduced adenoma development in APCmin/+ mice. Hematoxylin and eosin photomicrographs (A, D), immunofluorescence for HA (B, E), and immunofluorescence for CD44 and TLR4 (C, F) in small intestine (A–C), and a well-developed adenoma (D–F) in an APCmin/+ mouse (original magnification, 200×). PEP1 reduces spontaneous adenoma development in the APCmin/+ mouse small intestine. Adenoma number (G) and size (H) was assessed. Data are means ± SEM for 12 mice per treatment group (***, P < 0.00001; **, P < 0.005 compared with controls).
The HA-binding peptide, PEP1, reduced adenoma development in APCmin/+ mice. Hematoxylin and eosin photomicrographs (A, D), immunofluorescence for HA (B, E), and immunofluorescence for CD44 and TLR4 (C, F) in small intestine (A–C), and a well-developed adenoma (D–F) in an APCmin/+ mouse (original magnification, 200×). PEP1 reduces spontaneous adenoma development in the APCmin/+ mouse small intestine. Adenoma number (G) and size (H) was assessed. Data are means ± SEM for 12 mice per treatment group (***, P < 0.00001; **, P < 0.005 compared with controls).
PEP1 reduced adenoma development in the colons of mice treated with AOM-DSS
The AOM-DSS model was used to assess the role of HA in the development of inflammation-associated colon cancer. AOM-DSS induces the development of adenomas (18, 27). Over time, the adenomas grow and some develop into carcinomas. In the normal colon extracellular HA was seen adjacent to epithelial cells in the lower half of the crypts and in the muscularis mucosa (Fig. 2A and B). CD44 was expressed in lamina propria cells and in epithelial cells at the base of the crypts (Fig. 2C). To assess the roles of CD44 signaling and TLR4 signaling on adenoma development, WT, CD44−/−, and TLR4−/− mice received AOM-DSS. In the nontumorous portions of the colon in AOM-DSS-treated WT mice there was shortening of the crypts, increased inflammation, and increased extracellular matrix with large amounts of HA (Fig. 2D and E). There was a wider distribution of HA compared with the normal colon with extension into the upper half of the crypts. The distribution of CD44 and TLR4 was similar to that in the normal colon (Fig. 2F). In a colon cancer in the AOM-DSS model, there was diffuse extracellular HA and the neoplastic cells expressed CD44 and TLR4 (Fig. 2G–I; Supplementary Fig. S3). In adenomas and cancers in both the APCmin/+ model and the AOM-DSS model, both CD44 and TLR4 are expressed more diffusely and in a less organized way than in normal intestine and colon. In this model, mice deficient in CD44 had a 50% reduction in the number of adenomas, whereas mice deficient in TLR4 had an almost 80% reduction (Fig. 2J).
PEP1 reduced adenoma development in the colons of mice treated with AOM-DSS. Hematoxylin and eosin photomicrographs (A, D, G), immunofluorescence for HA (B, E, H), and immunofluorescence for CD44 and TLR4 (C, F, I) in colons from untreated mice (A–C), inflamed areas of colons from AOM-DSS–treated mice (D–F), and adenomatous areas from AOM-DSS–treated mice (G–I). AOM-DSS–induced development of colonic adenomas is reduced by 50% in CD44−/− mice and by 80% in TLR4−/− mice compared with WT mice (J). PEP1 reduces AOM-DSS–induced adenoma development in WT C57BL/6 mouse colons. K–N, PEP1 significantly reduced the number (K, M) and average size (L, N) of adenomas in AOM-DSS–treated mice, both when PEP1 administration began from the start of AOM-DSS treatment (K, L) or when PEP1 treatment started only after AOM-DSS treatment ended, by which time there had developed a preexisting AOM-DSS injury (M, N). Data for J–N are means ± SEM for 10 mice per treatment group (***, P < 0.0001; **, P < 0.001; *, P < 0.02 compared with control).
PEP1 reduced adenoma development in the colons of mice treated with AOM-DSS. Hematoxylin and eosin photomicrographs (A, D, G), immunofluorescence for HA (B, E, H), and immunofluorescence for CD44 and TLR4 (C, F, I) in colons from untreated mice (A–C), inflamed areas of colons from AOM-DSS–treated mice (D–F), and adenomatous areas from AOM-DSS–treated mice (G–I). AOM-DSS–induced development of colonic adenomas is reduced by 50% in CD44−/− mice and by 80% in TLR4−/− mice compared with WT mice (J). PEP1 reduces AOM-DSS–induced adenoma development in WT C57BL/6 mouse colons. K–N, PEP1 significantly reduced the number (K, M) and average size (L, N) of adenomas in AOM-DSS–treated mice, both when PEP1 administration began from the start of AOM-DSS treatment (K, L) or when PEP1 treatment started only after AOM-DSS treatment ended, by which time there had developed a preexisting AOM-DSS injury (M, N). Data for J–N are means ± SEM for 10 mice per treatment group (***, P < 0.0001; **, P < 0.001; *, P < 0.02 compared with control).
To assess the role of HA binding in tumorigenesis, wild-type mice were given PEP1 or scrambled PEP1 intraperitoneally every other day during the 7 weeks of the AOM-DSS regimen. Administration of PEP1 to WT mice during the 7 weeks of the AOM-DSS regimen resulted in a greater than 50% decrease in the number of adenomas (Fig. 2K). There was also a 25% reduction in adenoma size (Fig. 2L). To assess the role of HA binding on tumor growth, wild-type mice were given either PEP1 or scrambled PEP1 on alternate days for 4 weeks beginning at the completion of the second 1-week course of 2.5% DSS. Mice were then sacrificed and the number of adenomas counted. In experiments in which PEP1 was given to WT mice after the completion of the AOM-DSS regimen, PEP1 reduced the number of adenomas by about 50% indicating that HA binding to its receptors decreases adenoma growth (Fig. 2M). There was also a small but significant decrease in adenoma size (Fig. 2N). Taken together these data are consistent with the interaction of HA with both CD44 and TLR4 driving the development of adenomas in the AOM-DSS model.
HA binding to CD44 and TLR4 promotes growth in CT26 syngeneic tumor grafts
CT26 cells, a colon cancer cell line developed in BALB/c mice, express HA (28). Other colon cancer cell lines including sw620 cells also release HA (12). Injection of CT26 cells into the flanks of BALB/c mice resulted in syngeneic tumor isografts, which have abundant extracellular HA (Fig. 3A and B). CT26 cells expressed both TLR4 and CD44 (Fig. 3C). To assess the relative contributions of HA binding to TLR4 and CD44 to tumor growth, we used CRISPR to generate CT26 tumor cell lines deficient in TLR4 or CD44 (Fig. 3D and G). Tumor cells were injected into the flank and the isografts were allowed to grow. Mice were sacrificed when the tumor reached a volume of 1,250 mm3 or at 35 days after implantation, whichever came first. As expected, CD44-CRISPR CT26 tumors expressed TLR4, but not CD44, and TLR4-CRISPR tumor expressed CD44, but not TLR4 (Fig. 3F and I). CT26 wild-type (WT) tumors, CD44-CRISPR CT26 tumors, and TLR4-CRISPR CT26 tumors had high levels of HA (Fig. 3E and H). CT26 cell tumors deficient in either TLR4 or CD44 grew more slowly than WT CT26 tumors (Fig. 3J). This suggests that it is the interaction of HA with CD44 and TLR4 on the tumor cells, rather than on a host cell, that promotes tumor growth.
PEP1 reduced growth of CT26 WT, CD44-CRISPR, and TLR4-CRISPR syngeneic tumors in BALB/c mice. Hematoxylin and eosin photomicrographs (A, D, G), immunofluorescence for HA (B, E, H), and immunofluorescence for CD44 and TLR4 (C, F, I) in CT26 WT (A–C), CD44-CRISPR (D–F), and TLR4-CRISPR (G–I) isografts (original magnification, 200×). J, Growth curves for CT26 WT, CD44-CRISPR, and TLR4-CRISPR tumors in host mice treated with PEP1 or scrambled PEP1 (control). CT26 WT, or CD44-CRISPR, or TLR4-CRISPR cells (1 × 105) were implanted into the flank of BALB/c mice and tumors allowed to grow to an average volume of 1,250 mm3 for each treatment group or to day 35, whichever came first. PEP1 or scrambled PEP1 (control) was given intraperitoneally on alternate days starting one day before cells were implanted until mice were sacrificed. Tumor volume was measured with digital calipers, with the thickness of the contralateral flank acting as a correction factor for the height measurement of the tumor. Data are means ± SEM for 10 mice per treatment group. A linear mixed-effects model is used to describe differential tumor growth between control and experimental groups. Tumor growth rate for WT control is 287 units per day, for CD44-CRISPR control it is 175 units per day (P < 0.037 compared with WT control) and for TLR4-CRISPR control it is 66 units per day (P < 0.0001 compared with WT control). Tumor growth rate in WT PEP1 is 158 units per day (P < 0.01 compared with WT control). Tumor growth rate in CD44-CRISPR PEP1 is 16.3 units per day (P < 0.01 compared with CD44-CRISPR control). Tumor growth rate in TLR4-CRISPR PEP1 is 47.7 units per day, and the difference compared with TLR4-CRISPR control is not statistically significant.
PEP1 reduced growth of CT26 WT, CD44-CRISPR, and TLR4-CRISPR syngeneic tumors in BALB/c mice. Hematoxylin and eosin photomicrographs (A, D, G), immunofluorescence for HA (B, E, H), and immunofluorescence for CD44 and TLR4 (C, F, I) in CT26 WT (A–C), CD44-CRISPR (D–F), and TLR4-CRISPR (G–I) isografts (original magnification, 200×). J, Growth curves for CT26 WT, CD44-CRISPR, and TLR4-CRISPR tumors in host mice treated with PEP1 or scrambled PEP1 (control). CT26 WT, or CD44-CRISPR, or TLR4-CRISPR cells (1 × 105) were implanted into the flank of BALB/c mice and tumors allowed to grow to an average volume of 1,250 mm3 for each treatment group or to day 35, whichever came first. PEP1 or scrambled PEP1 (control) was given intraperitoneally on alternate days starting one day before cells were implanted until mice were sacrificed. Tumor volume was measured with digital calipers, with the thickness of the contralateral flank acting as a correction factor for the height measurement of the tumor. Data are means ± SEM for 10 mice per treatment group. A linear mixed-effects model is used to describe differential tumor growth between control and experimental groups. Tumor growth rate for WT control is 287 units per day, for CD44-CRISPR control it is 175 units per day (P < 0.037 compared with WT control) and for TLR4-CRISPR control it is 66 units per day (P < 0.0001 compared with WT control). Tumor growth rate in WT PEP1 is 158 units per day (P < 0.01 compared with WT control). Tumor growth rate in CD44-CRISPR PEP1 is 16.3 units per day (P < 0.01 compared with CD44-CRISPR control). Tumor growth rate in TLR4-CRISPR PEP1 is 47.7 units per day, and the difference compared with TLR4-CRISPR control is not statistically significant.
Staining the isografts for CD3, CD45, CD11c, and MHCII demonstrated that all the isografts contain immune cells and there is no obvious difference between the isografts in the number or distribution of immune cells (Supplementary Fig. S4).
To further assess the role of HA in the growth of these tumor isografts, mice received intraperitoneal injections of PEP1 or scrambled PEP1 every other day beginning one day before CT26 cells were injected into the flank. Tumor volume was measured. Mice that received PEP1 had diminished tumor volumes compared with mice receiving scrambled PEP1 (Fig. 3J). Administration of PEP1 to mice with CD44-CRISPR isografts reduced tumor volumes, whereas PEP1 had no effect on tumor volumes in mice with TLR4-CRISPR isografts. This experiment demonstrates that HA activation of signaling through both CD44 and TLR4 promotes tumor growth and suggests that HA activation of TLR4 signaling has greater progrowth effects than HA activation of CD44 signaling.
Endogenous HA promotes CT26 proliferation in vitro and blocks LPS binding
Having found that PEP1 blocks the growth of CT26 isografts in vivo, we used in vitro culture to further establish the role of HA binding to TLR4 and CD44 in promoting CT26 proliferation (Fig. 4A). We performed experiments investigating the effects of PEP1, hyaluronidase (HYAL), which enhances HA degradation, and 4-methylumbelliferone (4MU), which inhibits HA synthesis (28, 29). Baseline proliferation was decreased in TLR4-CRISPR and CD44-CRISPR cells compared with WT. PEP1, HYAL, and 4MU decreased proliferation in WT and CD44 CRISPR cells. In TLR4-CRISPR cells, however, only 4MU and HYAL decreased proliferation, whereas PEP1 did not. These data support a role for endogenous HA in promoting CT26 proliferation as blocking HA binding to its receptors, inhibiting HA synthesis, or enhancing HA degradation all decrease CT26 proliferation.
Comparison of cell viability in WT, TLR4−/−, and CD44−/− CT26 cells. Viability assessed by Cell Counting Kit-8 in the presence or absence of PEP1, 4MU, or hyaluronidase. A, WT CT26, TLR4−/−, and CD44−/− were treated with PEP1 (100 μg/mL), 4MU (0.5 mmol/L), or hyaluronidase (300 μg/mL) for 72 hours. B–D, HA effect on cell viability. Cells were pretreated with 4-MU (0.5 mmol/L) at least 24 hours before administration of HA (10, 100 μg/mL) or LPS (2, 5, 10, 100 EU/mL). All the experiments were repeated at least three times, and for each experiment n ≥ 8. For *, statistical significance was assessed by comparing each treatment to the control of each cell type. *, P < 0.05; **, P < 0.01; ***, P < 0.001. The # denotes statistical significance as assessed by comparing the paradigms under the brackets.
Comparison of cell viability in WT, TLR4−/−, and CD44−/− CT26 cells. Viability assessed by Cell Counting Kit-8 in the presence or absence of PEP1, 4MU, or hyaluronidase. A, WT CT26, TLR4−/−, and CD44−/− were treated with PEP1 (100 μg/mL), 4MU (0.5 mmol/L), or hyaluronidase (300 μg/mL) for 72 hours. B–D, HA effect on cell viability. Cells were pretreated with 4-MU (0.5 mmol/L) at least 24 hours before administration of HA (10, 100 μg/mL) or LPS (2, 5, 10, 100 EU/mL). All the experiments were repeated at least three times, and for each experiment n ≥ 8. For *, statistical significance was assessed by comparing each treatment to the control of each cell type. *, P < 0.05; **, P < 0.01; ***, P < 0.001. The # denotes statistical significance as assessed by comparing the paradigms under the brackets.
Having established that endogenous HA promotes CT26 proliferation, we next determined whether HA binding to TLR4 promotes proliferation. To address this, we incubated exogenous TLR4 ligands, either HA or LPS, with CT26 cells and found they did not affect proliferation (Fig. 4B). This raised the possibility that endogenous HA occupies all the TLR4 receptors preventing exogenous HA or LPS from binding to TLR4. To test this hypothesis, we blocked endogenous HA synthesis with 4MU and then added exogenous HA or LPS. Now both exogenous HA and LPS induced proliferation in a dose-dependent manner in WT and CD44-CRISPR cells. However, when a similar experiment was done in TLR4-CRISPR cells, exogenous HA or LPS had no effect on proliferation even when the synthesis of endogenous HA was inhibited. Thus, endogenous HA binding to TLR4 on CT26 cells promotes proliferation and blocks the stimulation of further proliferation by the TLR4 agonist LPS. Blocking the synthesis of endogenous HA frees up TLR4 permitting stimulation of proliferation by TLR4 agonists. In conditions in which exogenous HA was added (Fig. 4B–D), the assumption was that the concentration of exogenous HA exceeded the concentration of endogenous HA made by the CT26 cells. To confirm this, we used an ELISA to measure the HA concentration in the cultured media of CT26 cells. The concentration at 48 hours was 35 ng/mL, far less than the concentration of exogenous HA added.
HA binding to TLR4 and CD44 increased proliferation and decreased spontaneous apoptosis in CT26 tumor isografts
In mice with CT26 WT isografts, treatment with PEP1 diminished isograft growth. Similarly, CD44-CRISPR and TLR4-CRISPR tumor isografts had diminished growth compared with CT26 WT isografts. To determine whether these effects were due to decreased proliferation or increased spontaneous apoptosis, we performed immunofluorescence for BrdU and caspase-3, assessed TUNEL positivity (Fig. 5A and B), and counted the number of BrdU-positive, caspase-3–positive, and TUNEL-positive cells. Both CD44-CRISPR and TLR4-CRISPR tumor isografts were associated with decreased proliferation and increased spontaneous apoptosis compared with CT26 WT tumor isografts (Fig. 5C and D). PEP1 treatment of mice with CT26 WT or CD44-CRISPR tumor isografts resulted in decreased proliferation and increased apoptosis. In mice with TLR4-CRISPR tumors, PEP1 treatment did not affect proliferation or apoptosis. This experiment demonstrates that the inhibition of isograft growth by PEP1 is a product of diminished proliferation and increased apoptosis. Moreover, both the proproliferative and antiapoptotic effects of HA are mediated by activation of both TLR4 signaling and CD44 signaling, but in each case, activation of TLR4 signaling has the dominant effect.
PEP1 treatment of host BALB/c mice inhibits the growth of CT26 WT, CD44- CRISPR, and TLR4 CRISPR tumors by reducing tumor cell proliferation and increasing apoptosis. Immunofluorescence for BrdU and caspase-3 (A, B) in CT26 WT tumors grown in mice treated with scrambled PEP1 (A) or PEP1 (B; original magnification, 200×). Quantitative data for BrdU-labeled proliferating cells (C), caspase-3 expressing apoptotic cells (D), and TUNEL-positive cells (E). n = 25 photographs from 5 tumors for each treatment group (***, P < 0.0001 comparing PEP1 treated with control; +++, P < 0.0001 compared with WT control).
PEP1 treatment of host BALB/c mice inhibits the growth of CT26 WT, CD44- CRISPR, and TLR4 CRISPR tumors by reducing tumor cell proliferation and increasing apoptosis. Immunofluorescence for BrdU and caspase-3 (A, B) in CT26 WT tumors grown in mice treated with scrambled PEP1 (A) or PEP1 (B; original magnification, 200×). Quantitative data for BrdU-labeled proliferating cells (C), caspase-3 expressing apoptotic cells (D), and TUNEL-positive cells (E). n = 25 photographs from 5 tumors for each treatment group (***, P < 0.0001 comparing PEP1 treated with control; +++, P < 0.0001 compared with WT control).
Blocking HA binding sensitized CT26 tumors to radiation
We demonstrated that endogenous HA binding to CD44 and TLR4 blocks spontaneous apoptosis in CT26 WT tumor isografts (Fig. 5; exogenous HA blocks radiation-induced apoptosis in the normal mouse intestine; ref. 3). Given this, we next sought to determine whether blocking HA binding to its receptors would affect the radiosensitivity of CT26 tumors. To address this question, CT26 tumors were grown in the flanks of BALB/c mice. At 10 days after the implantation of the tumors, all the mice received fractionated radiation, 2 Gy of total body irradiation daily for 5 days. One treatment group received PEP1 prior to each dose of fractionated irradiation. The other group received scrambled PEP1. The isografts in the mice that received PEP1 grew more slowly (Fig. 6A). We used the same protocol to study the effect of PEP1 on survival in mice with CT26 isografts that received fractionated irradiation. All of the mice that received radiation and scrambled PEP1 were moribund by 18 days after irradiation (Fig. 6B). The mice that received PEP1 prior to each dose of radiation survived longer with 20% surviving longer than 30 days.
PEP1 slows postirradiation tumor growth (A) and improves postirradiation mortality (B) of host BALB/c mice carrying syngeneic CT 26 tumors. CT26 cells (1 × 105) implanted into the flanks of mice grew for 10 days, followed by fractionated total body irradiation (TBI) at 2 Gy per day for 5 consecutive days. PEP1 or scrambled PEP1 (control) was also injected intraperitoneally at 1 hour before each radiation dose. Mice were followed postirradiation for tumor growth (A) until the faster growing irradiated control tumors reached an average volume of 1,250 mm3 at 25 days postimplantation, or for survival (B) for 30 days postirradiation. *, P < 0.01 for tumor volume at days 20 and 25 compared with irradiated control. P = 0.0017 by the log-rank Mantel–Cox test comparing survival of 20 PEP1-treated irradiated mice with irradiated controls. PEP1 enhances radiosensitivity of CT26 tumors. Tumors were grown in mice for 18 days to an average 750 mm3. Mice were given intraperitoneal injections of PEP1 or scrambled PEP1 (control) on 3 consecutive days (days 19, 20, 21), followed by a single dose of 12 Gy TBI on day 21. Mice were sacrificed 24 hours after 12 Gy, and given BrdU 90 minutes before sacrifice. Tumors were resected and assessed for the number of apoptotic cells (C and D) and for the expression of Bax and Bcl2 by qRT-PCR (E). Data are means ± SEM for 20 photomicrographs of caspase-3 IHC staining apoptotic cells. ***, P < 0.0003 comparing PEP1 treated vs. controls. C and D, Data are means ± SEM for 5 tumors. ***, P < 0.0005 comparing gene expressions of Bax and Bcl2 by qRT-PCR in PEP1 (E).
PEP1 slows postirradiation tumor growth (A) and improves postirradiation mortality (B) of host BALB/c mice carrying syngeneic CT 26 tumors. CT26 cells (1 × 105) implanted into the flanks of mice grew for 10 days, followed by fractionated total body irradiation (TBI) at 2 Gy per day for 5 consecutive days. PEP1 or scrambled PEP1 (control) was also injected intraperitoneally at 1 hour before each radiation dose. Mice were followed postirradiation for tumor growth (A) until the faster growing irradiated control tumors reached an average volume of 1,250 mm3 at 25 days postimplantation, or for survival (B) for 30 days postirradiation. *, P < 0.01 for tumor volume at days 20 and 25 compared with irradiated control. P = 0.0017 by the log-rank Mantel–Cox test comparing survival of 20 PEP1-treated irradiated mice with irradiated controls. PEP1 enhances radiosensitivity of CT26 tumors. Tumors were grown in mice for 18 days to an average 750 mm3. Mice were given intraperitoneal injections of PEP1 or scrambled PEP1 (control) on 3 consecutive days (days 19, 20, 21), followed by a single dose of 12 Gy TBI on day 21. Mice were sacrificed 24 hours after 12 Gy, and given BrdU 90 minutes before sacrifice. Tumors were resected and assessed for the number of apoptotic cells (C and D) and for the expression of Bax and Bcl2 by qRT-PCR (E). Data are means ± SEM for 20 photomicrographs of caspase-3 IHC staining apoptotic cells. ***, P < 0.0003 comparing PEP1 treated vs. controls. C and D, Data are means ± SEM for 5 tumors. ***, P < 0.0005 comparing gene expressions of Bax and Bcl2 by qRT-PCR in PEP1 (E).
To determine whether the improved survival in the PEP1-treated mice was due to increased radiation-induced apoptosis, we performed an experiment in which CT26 cells were injected into the flanks of BALB/c mice. When the tumor reached an average size of 1,500 mm3, the mice were given PEP1 or scrambled PEP1 intraperitoneally daily for three days. After the third dose of PEP1, the mice received a single 12 Gy dose of total body radiation. Twenty-four hours later, the mice were sacrificed and the number of apoptotic cells counted using both caspase-3 and TUNEL. Administration of PEP1 increased the number of apoptotic cells by more than 2-fold indicating that HA binding to its receptors decreases radiation-induced apoptosis (Fig. 6C and D). Thus, the decreased tumor growth (Fig. 6A), and the increased survival seen in the PEP1-treated mice (Fig. 5B), were both due to an increase in radiation-induced cancer cell killing. qRT-PCR analysis of tumor mRNA demonstrated that pretreatment with PEP1 resulted in an increase in Bax mRNA and a decrease in Bcl2 mRNA (Fig. 6D).
Discussion
In the normal mouse intestine and colon, endogenous HA binding to CD44, and especially to TLR4, promotes Lgr5+ stem cell proliferation, crypt fission, and intestinal and colonic elongation (7, 8). Moreover, binding of endogenous HA to TLR4 promotes epithelial proliferation in mouse intestinal and colonic wound repair models (2). Here, we studied the effects of HA binding to its receptors in three in vivo colon cancer models and in vitro. In all these models, tumors express TLR4 and CD44 and there are large amounts of extracellular HA. In all of these models, extracellular HA is positioned next to the tumor cells where it can interact with CD44 and TLR4 on the tumor cells. Endogenous HA binding to its receptors promotes tumor growth in these models as demonstrated by the effects of PEP1 on the number of adenomas in the APCmin/+ and the AOM-DSS models and on the growth of CT26 syngeneic tumor isografts. Studies with CD44 deficiency and TLR4 deficiency in the AOM-DSS model and with CD44-CRISPR and TLR4-CRISPR CT26 cells indicate that HA binding to both CD44 and TLR4 is important in promoting tumor growth. In the AOM-DSS model, TLR4 deficiency resulted in a greater decrease in adenoma numbers than did CD44 deficiency. Similarly, in the CT26 syngeneic tumor isografts knockdown of TLR4 resulted in a greater decrease in tumor growth than knockdown of CD44, suggesting that HA signaling through TLR4 is of greater importance than HA signaling through CD44 in promoting tumor growth. The CRISPR studies also indicate that the TLR4 and CD44 signaling relevant to tumor growth are mediated by TLR4 and CD44 expression on the tumor cells rather than on host cells in the tumor environment.
There was a decrease in the number of adenomas in the APCmin/+ mice treated with PEP1 similar to the decrease in adenoma numbers seen in APCmin/+ mice that were also deficient in MyD88 (24). These data suggest that at least some of the effects of PEP1 on adenoma number in this model are mediated by blocking HA binding to TLR4, thus blocking TLR4 signaling. In the AOM-DSS model, the effects of PEP1 on adenoma number in wild-type mice are similar to the effects of TLR4 deficiency suggesting that the effects of PEP1 in wild-type mice are mediated by blocking HA binding to TLR4.
In CT26 isografts, HA binding to TLR4 promotes proliferation and blocks spontaneous apoptosis. TLR4 signaling in intestinal epithelial cells and in colon cancer promotes proliferation through activation of the β-catenin pathway (26). TLR4 activation enhances Akt phosphorylation, which, in turn, activates β1 integrin. Drugs that block this pathway ameliorate colon cancer (33–35). This indicates that TLR4 activation of the Akt pathway plays a central role in the growth and progression of colon cancer. The likely mechanism for the decrease in spontaneous apoptosis induced in colon cancers by HA binding to TLR4 is NFκB activation. Diminished apoptosis resulting from NFκB activation induced by binding of HA to TLR4 was described in bleomycin-induced apoptosis in mouse alveolar epithelial cells (36).
Although a role for TLR4 signaling in promoting colon cancer growth is well documented, the relevant TLR4 ligand in promoting tumor growth has not been established (36). LPS, the most widely studied TLR4 ligand, has been proposed as the relevant ligand (34). However, APCmin/+ mice raised in a germ-free environment have only a slight decrease in tumor numbers compared with conventionally raised mice, suggesting that a TLR4 ligand other than LPS is driving tumor growth (20). LPS is a large molecule with limited tissue distribution in vivo. LPS should only have access to colon cancer cells that abut the colon lumen. TLR4-CRISPR tumor isografts grow much more slowly than wild-type CT26 tumors. Thus, TLR4 signaling promotes the growth of colon cancers in a site where the cancer cells are in contact with extracellular HA, but have no exposure to LPS. In CT26 cells in vitro, endogenous HA binds to TLR4 blocking LPS binding. LPS can bind to TLR4 in CT26 cells only when the synthesis of endogenous HA is inhibited. These experiments suggest that in colon cancer, HA is the relevant TLR4 ligand driving cancer growth.
PEP1 is a radiosensitizer for CT26 syngeneic tumor isografts. TLR4 signaling promotes radioresistance in the normal mouse intestine (37). The likely mechanism for PEP1-induced radiosensitization is that PEP1 promotes radiation-induced apoptosis by blocking HA binding to TLR4. HA binding to TLR4 activates NFκB resulting in antiapoptotic signaling (5). Although PEP1 increased radiation killing in CT26 cells, administration of PEP1 prior to radiation in wild-type mice did not decrease the number of surviving crypts (3). These findings raise the possibility that administration of PEP1 prior to radiation would sensitize colon cancer cells to radiation induced cell death without increasing injury to normal intestinal epithelium.
In this study, we found that HA binding to TLR4 promotes proliferation and inhibits spontaneous apoptosis in colon cancer. This suggests that an agent like PEP1 that blocks HA binding to TLR4 may have a role as adjuvant therapy in colon cancer or other cancers marked by TLR4 expression and HA production. Moreover, PEP1 or a similar agent may be useful as a radiosensitizing agent in patients receiving radiotherapy for rectal cancer or other malignancies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: S. Makkar, T.E. Riehl, B. Chen, M.A. Ciorba, W.F. Stenson
Development of methodology: S. Makkar, T.E. Riehl, B. Chen, D.M. Alvarado, M.A. Ciorba, W.F. Stenson
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Makkar, T.E. Riehl, B. Chen
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Makkar, T.E. Riehl, B. Chen, Y. Yan, M.A. Ciorba, W.F. Stenson
Writing, review, and/or revision of the manuscript: T.E. Riehl, B. Chen, Y. Yan, D.M. Alvarado, M.A. Ciorba, W.F. Stenson
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T.E. Riehl, B. Chen
Study supervision: S. Makkar, T.E. Riehl, B. Chen, W.F. Stenson
Acknowledgments
This work was supported by NIH grants R01KD33165, P30DK052574, RO1DK109384, T32DK077653, and R21CA206039 and Crohn's and Colitis Senior Fellowship Award (ref #370863).
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.