Preparation and characterization of experimental oils
A major challenge to study the biological actions of un-oxidized LA vs. oxidized LA is that dietary LA is mainly consumed as vegetable oils; however, many commercial samples of vegetable oils, even being fresh, are already oxidized during production and storage and therefore always contain a mixture of un-oxidized LA and oxidized forms of LA 13,14.
To address this issue, we used silicic acid-activated charcoal chromatography, which is a well-established method to remove oxidized PUFA compounds in oils, to purify commercial samples of corn oil (see scheme of experiment in Fig. S1) 15. We used corn oil as a representative LA-rich vegetable oil, since it contains high levels of LA and low levels of other PUFA (the PUFA profile of corn oil constitutes ~55% LA and ~0.8% alpha-linolenic acid, see GC-MS analysis in Table S1). Peroxide assay showed that peroxide value (PV, a surrogate marker of the levels of oxidized LA) of the purified corn oil was as low as 0.12 ± 0.02 mEq/kg, validating that it contained little oxidized LA. To prepare oxidized corn oil, the purified corn oil was stored in a sealed bottle with headspace atmosphere and no light for 20 days (to mimic storage of vegetable oil in a supermarket or household). The PV of the oxidized corn oil was 9.98 ± 0.06 mEq/kg, which is higher than that of the purified corn oil, validating the presence of oxidized LA in the oxidized oil. In addition, the PV of the oxidized corn oil was lower or comparable to the current industrial standard for fresh vegetable oil, which states that the maximum PV of acceptable fresh vegetable oil is 10 mEq/kg 12,16. This result supports that the amounts of oxidized LA in the oxidized corn oil are in low human-consumption levels and it is feasible to use the oxidized corn oil to model human consumption of oxidized LA.
Oxidized LA has little impact on basal inflammation in mice
We treated mice with a completely defined isocaloric low-fat (10 wt/wt% of total fat) diet: the control diet (absence of oxidized LA) has a fat content of 10 wt/wt% of purified corn oil, and the oxidized LA diet has a fat content of 10 wt/wt% of oxidized corn oil (see diet composition and fatty acid profile in Table S1). To minimize further oxidation of LA in the diet during the animal feeding period, the diets were fortified with tocopherols as antioxidant, prepared freshly and changed every other day (see scheme of experiment in Fig. S1).
We treated mice with control diet or oxidized LA diet for 4- to 15- weeks. We found that compared with control diet, treatment with oxidized LA diet had little effect on basal inflammation in mice, as assessed by colon length, spleen weight, concentrations of pro-inflammatory cytokines in plasma, expression of pro-inflammatory genes in colon, and histology of colon tissue (Fig. S2-3). These results suggest that oxidized LA has little effects on inducing basal inflammation in mice.
Oxidized LA exacerbates chemically induced colitis in mice
We treated mice with control diet or oxidized LA diet, then stimulated the mice with DSS to induce colitis (see scheme of experiment in Fig. 1A). Compared with control diet, treatment with oxidized LA diet reduced colon length (P < 0.01, Fig. 1B), enlarged spleen tissue (P < 0.05, Fig. 1C), exaggerated crypt damage in the colon (P < 0.001, Fig. 1D), increased infiltration of leukocytes (CD45+) and macrophages (CD45+ F4/80+) into the colon (P < 0.05, Fig. 1E-F, see representative FACS images in Fig. S4), enhanced expression of pro-inflammatory genes Il-1β, Tnf-α, and Cox-2 in the colon (P < 0.05, Fig. 1G), and increased concentrations of TNF-α in both plasma and colonic explant (P < 0.05, Fig. 1H-I). Together, these results suggest that oxidized LA increased the severity of DSS-induced colitis in mice.
Besides corn oil, we also prepared un-oxidized and oxidized form of soybean oil, which is another LA-rich vegetable oil, and studied their effects on DSS-induced colitis in mice (see scheme of experiment in Fig. S5A). Compared with control diet (without oxidized soybean oil), treatment with a diet containing oxidized soybean oil exaggerated crypt damage of the colon tissue (P < 0.01, Fig. S5B) and enhanced colonic infiltration of CD45+ leukocytes and CD45+ F4/80+ macrophages (P < 0.05, Fig. S5C). This result demonstrates that beside oxidized corn oil, other types of oxidized vegetable oil also exacerbated colitis in mice, further supporting the colitis-enhancing effects of oxidized LA.
We recently showed that compared with diet rich in saturated fatty acids (lard), administration of a diet rich in un-oxidized LA has little impact on development of colitis in Il-10-/- mice, suggesting that un-oxidized LA has a limited role in promoting colitis 15. To better understand the effects of dietary fats on colitis, we treated mice with diets rich in un-oxidized LA, oxidized LA, or saturated fatty acids (lard), then stimulated the mice with DSS to induce colitis (Fig. S6A). Compared with the diet rich in saturated fatty acids, treatment with the diet rich in un-oxidized LA has little effects on DSS-induced colitis, akin to our previous report in the Il-10-/- mouse model 15; in contrast, treatment with the diet rich in oxidized LA significantly increased the severity of DSS-induced colitis in mice, with increased infiltration of macrophages and enhanced crypt damage in the colon tissues (Fig. S6B-C). Together with our previous study 15, these findings support the conclusion that oxidized LA, rather than LA itself (un-oxidized LA), exacerbates the development of colitis.
Oxidized LA exacerbates spontaneous colitis in Il-10-/- mice
To further examine the colitis-enhancing effects of oxidized LA, we studied its effect on colitis in another colitis model, the Il-10-/- mice. The Il-10-/- mice were treated with control diet or oxidized LA diet for 15 weeks (see scheme of experiment in Fig. 2A). We found that treatment with oxidized LA diet reduced colon length (P < 0.05, Fig. 2B), increased infiltration of CD45+ cells into the colon and small intestine (Fig. 2C, see representative FACS images in Fig. S7), enhanced expression of pro-inflammatory cytokines (Tnf-α, Tlr-4, Mcp-1 and Ifn-γ) in the colon (P < 0.05, Fig. 2D), increased plasma concentrations of TNF-α and IFN-γ (P < 0.05, Fig. 2E), and exaggerated crypt damage in the colon (P < 0.05, Fig. 2F). Together, these results demonstrate that oxidized LA exaggerated spontaneous colitis in the Il-10-/- mouse model, further supporting the colitis-enhancing effects of oxidized LA.
Oxidized LA exacerbates colorectal tumorigenesis in mice
We treated mice with control diet or oxidized LA diet, then stimulated the mice with azoxymethane (AOM) and DSS to induce colorectal tumorigenesis (see scheme of experiment in Fig. 3A). Treatment with oxidized LA diet increased tumor size and total tumor burden (P < 0.05, Fig. 3B) and enhanced the expression of β-catenin and proliferating cell nuclear antigen (PCNA) in the colon (P ≤ 0.01, Fig. 3C). In addition, oxidized LA diet increased infiltration of CD45+ and CD45+ F4/80+ immune cells into the colon (P < 0.05, Fig. 3D, see representative FACS images in Fig. S8), enhanced expressions of pro-inflammatory genes in the colon (P < 0.05, Fig. 3E), and increased concentrations of pro-inflammatory cytokines in colonic explants (P < 0.05, Fig. 3F). Together, these results demonstrate that oxidized LA increased the development of colorectal tumorigenesis in mice.
Oxidized LA exacerbates colitis via gut microbiota-dependent mechanisms
To explore the mechanisms by which oxidized LA exacerbated colitis, we studied the roles of gut microbiota involved 17. First, we analyzed the effects of oxidized LA on the diversity and composition of gut microbiota. We treated mice with control diet or oxidized LA diet for 3 weeks, then performed 16S rRNA sequencing to analyze the microbiota. We found that treatment with oxidized LA diet reduced a diversity (P < 0.05, Fig. 4A) and modulated b diversity (P = 0.01, Fig. 4B) of the microbiota, and changed composition of the microbiota at both phylum and genus levels (Fig. 4C-D and Table S2-3). These results support that dietary intake of oxidized LA alters the microbiota.
Next, we determined the functional role of gut microbiota in the biological effects of oxidized LA. We tested the extent to which antibiotic-mediated suppression of gut microbiota modulated the colitis-enhancing effect of oxidized LA diet (see scheme of experiment in Fig. 4E). We used a well-established broad-spectrum antibiotic cocktail from previous studies 18,19, and found that treatment with the cocktail caused a dramatic reduction of total fecal bacteria in mice (Fig. S9), validating its suppressing effects on the microbiota. Without co-administration of the antibiotic cocktail, treatment with oxidized LA diet exacerbated DSS-induced colitis, with reduced colon length and exaggerated crypt damage in the colon; however, such effects were abolished by co-administration of the antibiotic cocktail (Fig. 4F-H). Two-way ANOVA analysis showed that there was a statistically significant interaction (P < 0.05) between diet (control diet vs. oxidized LA diet) and antibiotic (with or without co-administration of antibiotic cocktail) on the development of DSS-induced colitis (Fig. 4F-H). Overall, these results support that gut microbiota is required for the colitis-enhancing effect of oxidized LA.
Oxidized LA exacerbates colitis via Toll-like receptor 4 (TLR4)-dependent mechanisms
TLR4, an important mediator of innate immunity, contributes to host-microbiota interactions and plays critical roles in the pathogenesis of colitis and colorectal tumorigenesis 20. In the DSS-induced colitis model (Fig. 5A), we found that compared with the plasma from control diet-treated DSS mice, the plasma from oxidized LA diet-treated DSS mice showed enhanced activation of TLR4, as assessed by a TLR4 reporter assay (P < 0.05, Fig. 5B). To determine the mechanisms by which oxidized LA activates TLR4 signaling, we measured the concentrations of TLR4 ligands, including lipopolysaccharide (LPS) and bacteria, in the systemic circulation 20. We found that the plasma from the oxidized LA diet-treated DSS mice had a higher concentration of LPS, which is a well-known ligand of TLR4 (P < 0.05, Fig. 5C) 20. In addition, qRT-PCR analysis showed that the blood from the oxidized LA diet-treated DSS mice had higher expression of 16S rRNA gene, which is a surrogate marker of bacterial abundance 19 (P < 0.05, Fig. 5D). Together, these results showed that treatment with oxidized LA diet increased the levels of bacteria and/or bacterial products (e.g., LPS) in the systemic circulation, leading to enhanced activation of TLR4 signaling in vivo.
Impaired gut barrier function leads to increased translocation of bacteria and/or bacterial products from the gut into systemic circulation 21. After demonstrating that oxidized LA increased bacterial translocation, we tested its effects on intestinal barrier function. Using a fluorescein isothiocyanate (FITC)-dextran-based permeability assay, we found that compared with the DSS mice treated with control diet, the DSS mice treated with oxidized LA diet had enhanced leakage of FITC-dextran from the gut to the circulation (P < 0.001, Fig. 5E), suggesting that oxidized LA impaired intestinal barrier function. We further found that oxidized LA diet reduced colonic expressions of Muc3, Tff3, and Occludin, which are important mediators of intestinal barrier function (P ≤ 0.01, Fig. 5F) 21. We also analyzed expressions of intestinal barrier genes in the AOM/DSS tumor model and found that treatment with oxidized LA diet also reduced expression of Occludin and Tff3 in the colon tumor (P < 0.05, Fig. S10). Together, these results support that oxidized LA diet impaired intestinal barrier function and thus enhanced bacterial translocation from the gut into the circulation, leading to increased activation of TLR4 in vivo.
We examined the functional role of TLR4 in the colitis-enhancing effects of oxidized LA. We treated wild-type (WT) and Tlr4-/- mice with control diet or oxidized LA diet, then stimulated the mice with DSS to induce colitis. In the WT mice, treatment with oxidized LA diet exacerbated the development of DSS-induced colitis, with enhanced infiltration of CD45+ and CD45+ F4/80+ immune cells (P < 0.05, Fig. 5G-H, see representative FACS images in Fig. S11), increased expression of Cox-2 (P < 0.05, Fig. 5I), and exaggerated crypt damage (P < 0.0001, Fig. 5J) in the colon tissues; while such effects were abolished in the Tlr4-/- mice (Fig. 5G-J). Two-way ANOVA analysis showed that there was a significant interaction between mouse type (Tlr4-/- vs. WT mice) and diet (control diet vs. oxidized LA diet) on the severity of DSS-induced colitis (P < 0.05). These results demonstrate that TLR4 is required for the colitis-enhancing effect of oxidized LA.