Profiling Lipoxygenase Metabolism in Specific Steps of Colorectal Tumorigenesis

Abstract

Lipoxygenases (LOXs) are key enzymes for the oxidative metabolism of polyunsaturated fatty acids into biologically active products. Clinical data on comparative levels of various LOX products in tumorigenesis are lacking. Therefore, we examined the profiles of several LOX (5- and 12-LOX and 15-LOX-1 and -2) products by liquid chromatography/tandem mass spectrometry in the major steps of colorectal tumorigenesis (normal, polyp, and cancer) in a clinical study of 125 subjects (49 with normal colon, 36 with colorectal polyps, and 40 with colorectal cancer) who underwent prospective colorectal biopsies to control for various potential confounding factors (e.g. diet, medications). Mean 13-hydroxyoctadecadienoic acid (13-HODE) levels were significantly higher in normal colons (mean = 36.11 ng/mg protein, 95% confidence interval [CI]: 31.56–40.67) than in paired colorectal cancer mucosa (mean = 27.01 ng/mg protein, 95% CI: 22.00–32.02; P = 0.0002) and in normal colons (mean = 37.15 ng/mg protein, 95% CI: 31.95–42.34) than in paired colorectal polyp mucosa (mean = 28.07 ng/mg protein, 95% CI: 23.66–32.48; P < 0.001). Mean 13-HODE levels, however, were similar between the left (mean = 37.15 ng/mg protein, 95% CI: 31.95–42.35) and right normal colons (mean = 32.46 ng/mg protein, 95% CI: 27.95–36.98; P = 0.09). No significant differences with regard to 12-hydroxyeicosatetraenoic acid (12-HETE), 15-HETE, or leukotriene B₄ (LTB₄) levels were detected between normal, polyp and cancer mucosa. 15-LOX-1 inhibited interleukin 1 beta expression. This study establishes that reduced 13-HODE levels are a specific alteration in the LOX product profile associated with human colorectal tumorigenesis.

Lipoxygenases (LOXs) are key enzymes in the oxidative metabolism of polyunsaturated fatty acids, particularly arachidonic and linoleic acids, into products that can influence cell signaling, structure, and metabolism (1). Preclinical and limited clinical data suggest that products of LOXs, especially of 5- and 12-LOX and 15-LOX-1 and -2, have differential roles in relation to human tumorigenesis (2-4). Up to the present, reported comparisons between levels of various LOX products in humans during different stages of multistep tumorigenesis have been limited to single LOX products in retrospectively collected surgical samples primarily from cancer patients (5-7). These studies were also limited by a lack of information on potential factors such as the LOX substrates linoleic and arachidonic acid, nutritional elements that modulate LOX activity (e.g., calcium, which is necessary for 15-LOX-1 activation; ref. (8), and medications (e.g., nonsteroidal anti-inflammatory drugs [NSAIDs]; refs. (7, 9, 10) that could have confounded LOX product measurements. Furthermore, tumorigenesis, especially colorectal tumorigenesis, is a multistep process (11), and no study reported to date has directly compared LOX product levels between the different steps of tumorigenesis.

Mass-spectrometry is an emerging technology that allows sensitive, specific, and simultaneous measurements of various LOX products and thus provides a LOX product profile of human tissues (12-14). The current clinical study examined the LOX product profile of each major step of colonic tumorigenesis, from normal colons-rectums to polyps to cancer, in prospectively collected biopsy samples of colonic mucosa, which allowed us to control for potential confounding factors.

Methods

Clinical samples

Colonic biopsy specimens were collected during colorectal endoscopic procedures after obtaining written informed consent from participating patients. Study patients were selected from among patients seen at outpatient gastrointestinal clinics at The University of Texas M. D. Anderson Cancer Center and other hospitals within the Texas Medical Center (the Gastroenterology section at Baylor College of Medicine, an outpatient gastrointestinal endoscopy unit affiliated with St. Luke's Hospital, and the Michael E. DeBakey VA Medical Center) for colorectal cancer screening and for the follow-up and management of colorectal cancers. This study was approved by the institutional review board at each participating institution.

Our study involved a total of 125 patients divided into three groups: 49 subjects with normal colon, 36 with colonic polyps, and 40 with colorectal cancer. The colorectal cancer group patients’ biopsies were obtained from the colorectal cancers and from normal-appearing mucosa at least 10 cm from the cancer. The colorectal polyp group included patients with no history of colorectal cancer. Biopsies were obtained from the colorectal polyps and from normal-appearing mucosa at least 10 cm form the polyp. The normal colon group included patients with no history of colorectal cancer or polyps and who had a normal colonoscopic examination at the time of biopsy. In this normal-colon group, two sets of biopsies of the colonic mucosa were obtained—one from the left and one from the right colon.

Subjects in all groups were between 45 and 85 year old, had no history of hereditary colon cancer (familial colorectal polyposis syndrome, hereditary nonpolyposis colon cancer syndrome, or family history of one or more first-degree relatives with colon cancer), and were U.S. citizens or permanent residents (to reduce the potential for large variability in risk factors such as dietary habits if international patients were included (15, 16)). Patients were excluded if they: had a history of inflammatory bowel disease; had received chemotherapy within 4 wk prior to the colonoscopy; had participated in a chemopreventive study during the month prior to the colonoscopy; had a history of bleeding diathesis; had a history of another active cancer within 5 years prior to enrollment (except for non-melanoma skin cancer); were taking warfarin; or were taking anti-inflammatory medications (e.g., nonsteroidal agents, aspirin, sulfasalazine) within 1 wk of the colonoscopies. Biopsies were collected between 2001 and 2006. All tissue samples were fresh frozen and stored at –80°C until the time of laboratory analyses.

LC/MS/MS measurements of levels of lipoxygenase products

Samples were subjected to extraction similar to procedures previously published (13, 17). Briefly, each frozen biopsy tissue sample was cut into approximately 1 × 1-2 mm stripes. Samples were transferred to sealed microcentrifuge tubes to which 500 μl of ice-cold tissue homogenization buffer was added (17). The sample was homogenized by an Ultrasonic Processor (Misonix, Farmingdale, NJ) at 0 °C for 3.5 min x2 with 1 minute rest in between and then centrifuged at 10,000rpm for 5 minutes at 4°C. A 400-μl aliquot of the supernatant was transferred to a glass tube; 600 μl phosphate-buffered saline (PBS) buffer (containing 1mM ethylenediaminetetraacetic acid (EDTA) and 1% butylated hydroxytoluene (BHT) and 10 μl of deuterated prostaglandin E₂ (PGE₂); 5-, 12-, or 15-hydroxyeicosatetraenoic acid (5-, 12-, 15-HETE); Leukotriene B₄ (LTB₄); or 13-hydroxyoctadecadienoic acid (13-HODE; 1 μg/ml) were added and samples were acidified with 0.5 N HCl to pH = 3.2-3.3. Lipid product extraction by adding 2 ml of ethyl acetate, vortexing for 30 seconds followed by centrifugation at 2000 rpm for 5 minutes at 4°C. The upper organic layer was collected, extraction was repeated for 2 more times and the organic phases from three extractions were pooled and then evaporated to dryness on ice under a stream of nitrogen. Samples were reconstituted in 100 μl of methanol: ammonium acetate buffer (10 mM at pH 8.5; 70:30, v:v) before liquid chromatography/tandem mass spectrometry (LC/MS/MS) analysis. The protein concentration was determined by a Bradford protein assay (Bio-Rad, Hercules, CA). LC/MS/MS analyses were performed using a Quattro Ultima tandem mass spectrometer (Micromass, Beverly, MA) equipped with an Agilent HP 1100 binary pump HPLC inlet as described previously (13).

Measuring intake of nutritional, mineral and vitamin supplements and medications

Dietary intake and alcohol intake were assessed with a semi-quantitative food frequency questionnaire (FFQ). This 137 item self-administered FFQ elicited usual intake over the past 6 months. Martinez et al have described the details of this FFQ and its measurement characteristics (18). FFQ data were entered into the Food Frequency Data Entry and Analysis Program¹ and analyzed for 49 macro- and micronutrients as well as individual fatty acids using nutrient and gram weight information from the Food Intake Analysis System² USDA Survey Nutrient Data Base (US Department of Agriculture, Agricultural Research Service. 1997. ON: Nutrient Database for Individual Intake Surveys. 1994-96 Continuing Survey of Food Intakes by Individuals and 1994-96 Diet and Health Knowledge Survey. CD-ROM). Intake of hormonal replacement therapy, vitamins, and other nutritional supplements were measured by using a medication questionnaire.

RNA extraction and quantitative reverse transcription-polymerase chain reaction analyses

Total RNA was extracted from cells using TRI reagent (Molecular Research Center, Inc., Cincinnati, OH ref. (19). The integrity of total RNA was verified on an Agilent 2100 Bioanalyzer using the RNA 6000 Nano LabChip kit (Agilent Technologies, Palo Alto, CA). Extracted mRNA samples of adequate RNA quality (RNA integrity number [RIN] ≥ 8) from paired tissues were available from 11 patients with colorectal cancer. RNA was reverse transcribed and then measured quantitatively by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) using a comparative C_t method, as described previously (19). Primers and probes for human IL-1B (Assay ID: HS00174097_m1), 15-LOX-1 (HS00609608_m1), and human HPRT1 (4326321E; internal control for cell line expression studies) were purchased from Applied Biosystems (Foster City, CA). β-amyloid was the internal control for human colon tissue samples because of its similar expression levels in normal and cancer tissues of the colon (20). β-amyloid primers (forward primer: 5’-ctcatgccatctttgaccga-3’; reverse primer: 5’-gggcatcaacaggctcaact-3’) were purchased from Sigma (St. Louis, MO), and the β-amyloid 5' end FAM-labeled probe (5’-gttcagcctggacgatctccagc-3’) was purchased from Integrated DNA Technologies (Coralville, IA).

LoVo colon cancer cells transfection with adenoviral vectors

LoVo colon cancer cells [directly obtained from the American Type Culture Collection (Manassas, VA)] were cultured and transfected with modified 5/3 adenoviral vectors that express either 15-LOX-1 (Ad-15-LOX-1) or luciferase (Ad-luciferase) at 200 viral particle per cell as described previously (21). Transfected cells were harvested at 24 and 48 hours and processed for 15-LOX-1 and interleukin 1 beta (IL-1B) mRNA experession measurements by qRT-PCR.

Western Blot analyses

As previously described (21), cell lysate proteins were subjected to Western blot analyses using a solution of rabbit polyclonal antibody to human 15-LOX-1 (1:2,000 dilution) and IL-1B (1:200; Abcam, Cambridge, MA).

Statistical methods

Fisher's exact tests were used to determine the association between the disease status and categorical variables. Analysis of variance was used to compare age, BMI and each of the continuous energy adjusted nutritional variables among three disease statuses. 13-HODE, 12-HETE, and 15-HETE were compared within each group with paired t-test, while LTB₄ levels were compared with a sign test because of non-normal distributions secondary to undetectable levels in large numbers of subjects in all three disease categories. Continuous nutrient variables were energy adjusted using a regression method (22). Multinomial logistic regression analyses were used to determine the association of disease status and each of the medication intake and nutritional variables after adjusting gender effect. We performed a one-way analysis of variance for analyses involving single factors and more than two groups.

Results

Clinical characteristics of the study population

The three diseases groups had no significant differences in age or ethnic background (Table 1). Gender distribution was significantly different among the three groups with the proportion of male to female patients markedly higher in the polyp disease group as compared to the normal or cancer group (Table 1). This difference resulted from the high number of polyp patients recruited through the Michael E. DeBakey VA Medical Center. BMI was significantly higher in the polyp group as compared to the cancer group. A difference possibly explained by secondary weight loss because of cancer. Medication intake including NSAIDs was similar among the three groups (table 2). Nutrient intake with or without energy adjustment was not statistically different between the disease groups (Table 3). In the multinomial logistic regression analyses with gender adjustment, the only significant association found between medication and nutritional intake variables and disease status was having higher odd ratio of developing polyp than cancer for the subject taking NSAIDs (Odds Ratios =4.81, p=0.014).

Table 1.

Demographic characteristics and colon disease status groups

Variable	Colon Disease Status	n	mean	std	median	min	max	P-value*
Age	Normal	49	59.35	6.70	58.00	50.00	75.00	0.16
	Polyp	36	61.44	6.56	60.50	50.00	73.00
	Cancer	40	62.35	9.19	63.00	45.00	80.00
BMI	Normal	49	28.03	5.97	27.12	19.58	47.74	0.008
	Polyp	36	30.42	5.71	30.75	19.20	47.55
	Cancer	40	26.30	5.27	25.92	12.40	38.21
Race		White	African American	Asian	Hispanic	0.08
	Normal	43	5	1	0
	Polyp	30	2	0	4
	Cancer	29	4	2	5
Gender		Male	Female	<0.0001
	Normal	18	31
	Polyp	31	5
	Cancer	23	17

^*P-values are one way ANOVA for age and BMI and Fisher exact test for race and gender. All statistical tests were two-sided.

Table 2.

Comparison of medication intake among the colon-disease-status groups

Variables		Disease group
		Normal	Polyp	Cancer	P-value*
Cholesterol medication	No	40	27	35	0.37
	Yes	9	9	5
Cardiac medication	No	25	13	20	0.34
	Yes	24	23	20
Other medications	No	15	14	17	0.48
	Yes	34	22	23
NSAIDs	No	37	24	35	0.09
	Yes	12	12	5

^*P-values are Fisher exact test. All statistical tests were two-sided.

Table 3.

Energy-adjusted nutrient intake by disease group

Covariable	disease group	n	mean	std	median	min	max	*P-value
Adj_Saturated_Fat (g)	Normal	49	25.30	6.13	25.52	9.66	53.92	0.34
	Polyp	36	26.10	5.53	26.20	15.59	41.71
	Cancer	40	24.17	5.42	24.71	12.20	40.26
Adj_Polyunsat_Fat (g)	Normal	49	16.39	5.04	16.91	−6.90	28.13	0.59
	Polyp	36	17.34	5.14	16.35	0.05	29.78
	Cancer	40	16.40	3.84	16.80	3.45	24.14
Adj_Monounsat_Fat (g)	Normal	49	28.69	5.81	28.81	12.15	46.25	0.42
	Polyp	36	29.87	6.29	29.29	12.97	42.72
	Cancer	40	28.17	5.12	29.30	15.83	39.44
Adj_Dietary_Fiber (g)	Normal	49	19.96	6.29	19.39	4.17	37.50	0.98
	Polyp	36	20.16	5.68	18.70	10.79	35.60
	Cancer	40	19.86	7.00	18.57	3.28	38.04
Adj_Alcohol (g)	Normal	49	6.34	11.99	2.11	−3.93	66.50	0.22
	Polyp	36	6.56	15.92	1.89	−16.13	77.27
	Cancer	40	12.95	27.85	2.78	−4.83	153.45
Adj_Folate (mcg)	Normal	49	469.83	158.23	482.19	−133.07	829.41	0.57
	Polyp	36	506.39	163.85	488.07	151.22	971.62
	Cancer	40	483.58	149.11	497.22	41.88	861.31
Adj_Calcium (mg)	Normal	49	974.83	461.02	868.60	515.94	3153.33	0.37
	Polyp	36	852.13	312.79	834.35	301.46	2104.49
	Cancer	40	903.12	390.26	853.44	10.49	2118.69
Adj_Linoleic (g)	Normal	49	14.49	4.56	14.97	−7.07	24.66	0.61
	Polyp	36	15.36	4.68	14.52	−0.88	26.35
	Cancer	40	14.55	3.43	14.89	2.93	21.15
Adj_Arachidonic (g)	Normal	49	0.14	0.09	0.15	−0.11	0.32	0.43
	Polyp	36	0.16	0.10	0.15	−0.09	0.48
	Cancer	40	0.14	0.07	0.16	−0.04	0.28
Adj_Linolenic (g)	Normal	49	1.47	0.40	1.46	0.17	2.74	0.64
	Polyp	36	1.54	0.43	1.46	0.83	2.77
	Cancer	40	1.46	0.38	1.48	0.20	2.54
Adj_EPA (g)	Normal	49	0.03	0.03	0.03	−0.02	0.16	0.99
	Polyp	36	0.03	0.03	0.03	−0.01	0.12
	Cancer	40	0.03	0.02	0.03	−0.02	0.08
Adj_DHA (g)	Normal	49	0.09	0.07	0.10	−0.07	0.29	0.97
	Polyp	36	0.09	0.07	0.08	−0.06	0.27
	Cancer	40	0.09	0.05	0.09	−0.05	0.20

^*P-values are one way ANOVA. All statistical tests were two-sided.

Lipoxygenase metabolism in normal and cancer mucosa of colorectal cancer patients

13-HODE mean levels were significantly higher in normal (mean = 36.11 ng/mg protein 95% CI: 31.56 – 40.67 ng/mg protein) than cancer mucosa (mean = 27.01 ng/mg protein, 95% CI: 22.00-32.02 ng/mg protein; P = 0.0002; Fig. 1A). The ratio of normal to cancer mucosa was less than 1 in 31 of 40 (78%) subjects (mean cancer to normal ratio = 0.8, 95% CI = 0.68 – 0.91). In contrast, 15-HETE levels were similar between normal (mean = 6.21, 95% CI: 4.61 – 7.82) and cancerous mucosa (mean = 5.93, 95% CI: 4.18-7.68; P = 0.58; Fig. 1B). 12-HETE levels were similar between normal (mean = 2.71, 95% CI: 1.91– 3.52) and cancerous mucosa (mean = 2.79, 95% CI: 1.97 – 3.61; P = 0.82; Fig. 1C). LTB₄ levels were below detectable levels in normal mucosa in 27/39 (69%) and in cancerous mucosa in 24/39 (62%) of subjects. The levels in subjects with detectable LTB₄ levels were low and similar between normal (mean = 0.24, 95% CI: 0.06 – 0.43 ng/mg protein) and cancerous mucosa (mean = 0.4, 95% CI: 0.14 – 0.65 ng/mg protein; P = 0.5; Fig. 1D).

Fig. 1.

Lipoxygenase product levels in subjects with colorectal cancer. Biopsies of paired normal and cancer mucosa were analyzed for 13-hydroxyoctadecadienoic acid (13-HODE), 12- hydroxyeicosatetraenoic acid (12-HETE), 15-HETE, and leukotriene B₄ (LTB4) by chromatography/tandem mass spectrometry (LC/MS/MS). Values from each subject are depicted in the dot plots. Lines represent the mean values for groups.

Lipoxygenase metabolism in normal and polyp mucosa of colorectal polyp patients

13-HODE mean levels were significantly higher in normal (mean = 37.15, 95% CI: 31.95-42.34 ng/mg protein) than polyp mucosa (mean = 28.07, 95% CI: 23.66 – 32.48; P < 0.001; Fig. 2A). The ratio of normal to polyp mucosa was less than 1 in 28 of 36 (78%) subjects (mean polyp to normal ratio = 0.79, 95% CI = 0.69 – 0.90). In contrast, 15-HETE levels were similar between normal (mean = 6.75, 95% CI: 4.98 – 8.53) and polyp mucosa (mean = 6.00, 95% CI: 4.72- 7.29; P = 0.35; Fig. 2B). 12-HETE levels were similar between normal (mean = 2.44, 95% CI: 1.8 – 3.09) and polyp mucosa (mean = 2.58, 95% CI: 1.86 – 3.29; P = 0.73; Fig. 2C). LTB₄ levels were below detectable levels in normal mucosa in 25/37 (68%) and in polyp mucosa in 24/37 (65%) of subjects. The levels in subjects with detectable LTB₄ levels were low and similar between normal (mean = 0.34, 95% CI: 0.14 – 0.55 ng/mg protein) and polyp mucosa (mean = 0.26, 95% CI: 0.11 – 0.41 ng/mg protein; P = 0.3; Fig. 2D).

Fig. 2.

Lipoxygenase product levels in subjects with colorectal polyps. Biopsies of paired normal and polyp mucosa were analyzed for 13-hydroxyoctadecadienoic acid (13-HODE), 12- hydroxyeicosatetraenoic acid (12-HETE), 15-HETE, and leukotriene B₄ (LTB4) by liquid chromatography/tandem mass spectrometry (LC/MS/MS). Values from each subject are depicted in the dot plots. Lines represent the mean values for groups.

Lipoxygenase metabolism in the colonic mucosa of subjects with normal colons

13-HODE mean levels were not significantly different between the left (mean = 37.15, 95% CI: 31.95-42.35) and right normal colonic mucosa (mean = 32.46, 95% CI: 27.95-36.98; P = 0.09; Fig. 3A). 15-HETE levels were also similar between left (mean = 8.23, 95% CI: 6.85 -9.61) and right normal colonic mucosa (mean = 7.87, 95% CI: 6.6 -9.15; P = 0.6053; Fig. 3B). 12-HETE levels were similar between left (mean = 3.37, 95% CI: 2.32-4.43) and right normal colonic mucosa (mean = 3.06, 95% CI: 2.23- 3.9; P = 0.6154; Fig. 3C). LTB₄ levels were below detectable levels in left normal colonic mucosa in 29/49 (59%) and in right normal colonic mucosa in 30/49 (61%) of subjects. The levels in subjects with detectable LTB₄ levels were low and similar between left (mean = 0.53, 95% CI: 0.29 – 0.77 ng/mg protein) and cancerous mucosa (mean = 0.64, 95% CI: 0.07 – 0.1.21 ng/mg protein; P = 0.44; Fig. 3D).

Fig. 3.

Lipoxygenase product levels in subjects with a normal colon. Paired left and right colonic mucosal biopsies were analyzed for 13-hydroxyoctadecadienoic acid (13-HODE), 12- hydroxyeicosatetraenoic acid (12-HETE), 15-HETE, and leukotriene B₄ (LTB4) by liquid chromatography/tandem mass spectrometry (LC/MS/MS). Values from each subject are depicted in the dot plots. Lines represent the mean values for groups.

IL-1B and 15-LOX-1 in human colonic tumorigenesis

Among the 11 patients with evaluable mRNA, 15-LOX-1 relative mRNA expression levels were lower in cancer than paired normal mucosa in 10 patients and equal in one (mean cancer-to-normal ratio = 0.18, 95% CI = 0.09 – 0.39; Fig. 4A; P = 0.0014). In contrast, IL-1B relative mRNA levels were higher in 9 of these 11 patients (mean cancer-to-normal ratio = 4.34, 95% CI = 1.98 – 9.48) (Fig. 4B; P = 0.0043); the ratios were 0.94 and 0.85 in the other two cases. In LoVo colon cancer cells, Ad-15-LOX-1 viral vector induced 15-LOX-1 expression (P < 0.0001; Fig. 4C), which reduced IL-1B mRNA expression by > 50% (versus control [Ad-luciferase]; P = 0.014; Fig. 4D) and IL-1B protein expression (Fig. 4E).

Fig.4.

Effect of 15-LOX-1 on IL-1B expression in colorectal cancer. A and B. Expression of 15-LOX-1 and interleukin 1 beta (IL-1B) in human colonic tumor and normal mucosa. 15-LOX-1 (panel A) and IL-1B mRNA expression levels (panel B) were measured by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) in paired colonic normal and tumor mucosa samples from colorectal cancer patients. Each value depicts the mean of triplicate measurements from each subject. Lines represent group mean values. C and D. Effects of 15-LOX-1 expression on IL-1B expression in colon cancer cells. LoVo colon cancer cells were transfected with either 15-LOX-1 adenoviral vector that expresses 15-LOX-1 (Ad-15-LOX-1) or with the same vector except for replacing15-LOX-1 cDNA by luciferase cDNA (Ad-luciferase) (control vector). C. Cells were harvested at 24 hours post transfection and 15-LOX-1 was measured by qRT-PCR. D. Cells were harvested 48 hr after transfection, and IL-1B mRNA was measured by quantitative real-time PCR. Values are the means ± SDs of triplicate experiments. (* P < 0.0001, ** P = 0.014, ANOVA). E. Cell lysates were collected 48 hours after transfection and were processed for 15-LOX-1 and IL-1B protein expression. Plus sign (+), a positive control for 15-LOX-1.

Discussion

This study indicates 1) that a reduced level of 13-HODE is a specific alteration in the LOX product profile of human colorectal polyps and cancer (versus normal colorectal mucosa) and 2) that 13-HODE is the predominant component of the LOX product profile of colorectal normal, polyp, and cancer mucosa.

The reduced 13-HODE level in polyps or cancer was associated with levels of 12-HETE, 15-HETE, or LTB₄, which did not differ significantly between polyps or cancer and normal mucosa. 13-HODE was the predominant LOX product in colonic mucosa, with levels that were several-fold higher in normal, polyp, and cancer patients than were levels of the other LOX products (15- and 12-HETE, LTB₄). We selected 13-HODE, 12-HETE, 15-HETE, and LTB₄ for measurements in the current study because prior reports support their key roles in tumorigenesis (2). This reduction in 13-HODE establishes the clinical relevance of prior preclinical data showing that the expression of 15-LOX-1, the key enzyme for 13-HODE production, is lost in colorectal cancer cells (6, 23, 24) and 15-LOX-1 re-expression by antitumorigenic agents such as NSAIDs and histone deacetylase inhibitors, or by adenoviral delivery vectors inhibit colorectal tumorigenesis (9, 10, 12, 21, 25-29). The findings of reduced 13-HODE in colorectal polyps and cancer are consistent with prior reports of 15-LOX-1 downregulation in clinical samples of surgically resected colorectal cancer (6, 7, 30, 31), colorectal polyps (7, 31), and adenomas from a small cohort of familial adenomatous polyposis (FAP) patients (n=5; ref. (12). More important, however, the current study is the first to comprehensively examine the full profile of LOX pathways (5- and 12-LOX and 15-LOX-1 and -2) in sporadic colorectal polyps and cancer and to demonstrate the relative significance of 15-LOX-1 downregulation to colonic tumorigenesis compared with other LOX pathways. Our very sensitive and specific mass-spectrometry methods (12, 13) allowed these simultaneous LOX-product measurements and thus direct comparisons between them. Prior studies were mostly limited to measuring individual LOX pathways (5, 7, 12, 32), and it is difficult to compare various product levels between different studies because of important interstudy differences. The one exception to these individual-LOX studies was a very small retrospective study of various LOX profiles in 5 FAP patients (12). Although the present results agree with those of the prior FAP study findings (12), they provide indispensable confirmation from a much larger study in patients with far more-common sporadic colorectal polyps and cancer. Furthermore, the LOX product profiles, including 13-HODE levels, of normal mucosa in subjects without colonic polyps or cancer were unknown prior to the current study.

The current clinical study is the first prospective analysis of colonic biopsy samples of normal, polyp, or colorectal-cancer mucosa collected via the same methods, and the first analysis to adjust for the effects of factors including dietary intake (e.g., of linoleic and arachidonic acid and calcium) and other factors (e.g., NSAID use) known to be potential confounders of measurements of the LOX product profile. In contrast, prior studies were limited by a lack of information regarding important confounding factors that can influence 13-HODE levels (e.g., NSAIDs and intake of linoleic acid) and questions concerning the generalizability of data from FAP and the generally advanced stage of the surgically resected sporadic polyps (average diameter of 3 cm; ref.(7). After we adjusted for these potential confounding factors, the 13-HODE level was still lower in polyps and cancer than in paired normal mucosa and colorectal mucosa from subjects with normal colons (supplementary Fig. 1). Therefore, the reduction in 13-HODE levels was unlikely secondary to lower availability of substrate levels of linoleic acid and more likely was secondary to downregulated 15-LOX-1 expression [(6, 7, 24, 30, 31) and current report (Fig. 4A)]. Reduced 13-HODE in the present study also was unrelated to a familial or hereditary syndrome since such patients were excluded. Furthermore, the reduced 13-HODE level was unrelated to variability in colonic biopsy sites since 13-HODE levels were similar in the right and left sides of normal colons in this study.

Linoleic acid is thought to be the most abundant polyunsaturated fatty acid in western diets (33). The current study confirmed this notion in finding that the dietary intake of linoleic acid was higher than that of arachidonic and other polyunsaturated fatty acids. Therefore, the predominance of 13-HODE, which is the main oxidative product of linoleic acid, over 12- and 15-HETE and LTB₄, which are products of arachidonic acid, is possibly related to substrate availability. Further studies in other human organs will help to determine whether the profile of LOX products is organ-specific. The current study utilized undissected colorectal biopsies, which likely contained variable proportions of epithelial and subepithelial tissues. Although variable proportions of epithelial and subepithelial tissues may have influenced LOX product measurements, both subepithelial and epithelial tissues are thought to contribute to polyunsaturated fatty acid oxidative metabolism (34). Future studies potentially should examine the contribution of each compartment to LOX product levels.

Emerging data increasingly support a mechanistic link between inflammation and cancer (35), especially in the case of colonic tumorigenesis (36). IL-1B is a major proinflammatory cytokines that contributes to the pathogenesis of human colitis (37) and tumorigenesis (38-41). Based on prior reports suggesting an antiinflammatory effects for 15-LOX-1 (42) (43), we examined the relationship between 15-LOX-1 and IL-1B. Our results showed for the first time the relationship between 15-LOX-1 downregulation and IL-1B upregulation in human colon cancer. Our in vitro studies of 15-LOX-1 expression in human LoVo colon cancer cells demonstrate the mechanistic significance of this association, as 15-LOX-1 expression downregulated IL-1B expression in human colon cancer cells in vitro. These findings further support the proposed 15-LOX-1 antiinflammatory and antitumorigenic role in colorectal carcinogenesis.

The current study demonstrates that the LOX-product profile can be detected in biopsies of colonic mucosa and that a reduced 13-HODE level is a specific alteration in the LOX-product profile of colorectal polyp and cancer mucosa. We believe that these results support future study of the utility of altered 13-HODE as a biomarker of colorectal tumorigenesis and of the effects of molecular-targeted approaches to preserve 15-LOX-1 expression for colorectal cancer chemoprevention.