NPC1L1 knockout protects against colitis-associated tumorigenesis in mice

Colorectal cancer is strongly associated with lipid metabolism. NPC1L1, a sterol transporter, plays a key role in modulating lipid homeostasis in vivo. Its inhibitor, ezetimibe, began to be used clinically to lower cholesterol and this caused the great debate on its role in causing carcinogenesis. Here we explored the role of NPC1L1 in colorectal tumorigenesis. Wild-type mice and NPC1L1−/− (NPC1L1 knockout) mice were treated with azoxymethane (AOM)-dextran sodium sulfate (DSS) to induce colitis-associated colorectal tumorigenesis. Mice were sacrificed 10, 15, 18 or 20 weeks after AOM treatment, respectively. Colorectal tumors were counted and analyzed. Plasma lipid concentrations were measured using enzymatic reagent kit. Protein expression level was assayed by western blot. NPC1L1−/− mice significantly had fewer tumors than wild-type. The ratio of malignant/tumor in NPC1L1−/− mice was significantly lower than in wild-type 20 weeks after AOM-DSS treatment. NPC1L1 was highly expressed in the small intestine of wild-type mice but its expression was undetectable in colorectal mucous membranes or tumors in either group. NPC1L1 knockout decreased plasma total cholesterol and phospholipid. NPC1L1−/− mice had significant lower intestinal inflammation scores and expressed inflammatory markers p-c-Jun, p-ERK and Caspase-1 p20 lower than wild-type. NPC1L1 knockout also reduced lymphadenectasis what may be caused by inflammation. NPC1L1 knockout in mice decreased β-catenin in tumors and regulated TGF-β and p-gp in adjacent colons or tumors. There was not detectable change of p53 by NPC1L1 knockout. Our results provide the first evidence that NPC1L1 knockout protects against colitis-associated tumorigenesis. NPC1L1 knockout decreasing plasma lipid, especially cholesterol, to reduce inflammation and decreasing β-catenin, p-c-Jun and p-ERK may be involved in the mechanism.


Background
Cancer is a major public health problem worldwide, causing 1 in 4 deaths in the United States [1]. Colorectal cancer (CRC) ranks as the third highest incidence and mortality rates among men and women [1].
Evidence is accumulating that lipid metabolism is strongly associated with cancer, including CRC [1][2][3][4]. About 11% of CRC cases have been attributed to overweight and obesity in Europe. On the other hand, obesity is associated with a 30-70% increased risk of colon cancer in men [2]. High serum triglyceride (TG) and total cholesterol (TC), especially low-density lipoprotein cholesterol (LDL-C), are significantly and positively associated with cancer (including CRC) [1][2][3][4].
Serum lipid is modulated by many factors, including some genes, such as leptin, adipose triglyceride lipase and so on. Among these genes, Niemann-Pick C1 like 1 (NPC1L1) has been proven to be one of the most important sterol transporters [5]. It was reported that it plays a role in modulation of lipid homeostasis, including TG, phospholipid (PL), low-density lipoprotein, high-density lipoprotein, and most of all, cholesterol, in mice [6][7][8]. In clinic, NPC1L1 genotype was found to be significantly associated with plasma lipid concentration, especially TC and LDL-C [9]. The NPC1L1 inhibitor, ezetimibe, began to be used in clinic to lower cholesterol and has caused the great debate on its role in cancer. Dr. Rossebo reported that using ezetimibe with simvastatin to lower serum lipid caused a significant increase of new cancer incidents [6]. Dr. Peto disputed that analyses of cancer data from several ezetimibe trials did not provide credible evidence of adverse effect on rates of cancer [7].
Here, we explored the role of NPC1L1 in colorectal tumorigenesis in vivo by using transgenic mice. Our results testify that NPC1L1 knockout in mice protects against colitis-associated tumorigenesis.

Animals and diets
NPC1L1 −/− mice were described previously [8,9]. Wildtype (WT) mice and NPC1L1 −/− mice were derived from NPC1L1 +/− mice with a pure C57BL/6 background and housed in a specific pathogen-free animal facility in plastic cages in a temperature controlled room (22°C) with a daylight cycle from 6 AM to 6 PM. Mice had free access to a standard laboratory rodent chow diet (Rodent NIH-07 22.5-5; Zeigler Bros Inc., Gardners, USA) and water. All animal procedures were approved by the Institutional Animal Care and Use Committee at University of Maryland Health Sciences.

AOM-DSS induced colitis-associated tumorigenesis
AOM-DSS induced colorectal tumorigenesis was described previously [10,11]. Six to seven week-old, male mice were given a single intraperitoneal injection of 10 mg/kg body weight AOM (Sigma-aldrich, USA). Starting at a week after injection, animals received DSS (MP Biomedicals, Molecular weight was 36,000-50,000 Da., USA) for 7 days via free access to drinking water containing 2% DSS. The DSS water was replaced every day. Then, mice were fed with regular water and there was not any further treatment after DSS treatment.
Mice were sacrificed at different time points and colorectums were excised. Colorectums were opened longitudinally, flushed with ice-cold PBS and fixed in 10% formalin/ PBS. Then macroscopic tumors were counted and measured with a caliper. Enlarged abdominal (from sciatic, lumbar, para-aortic to truncus coeliacus) lymph nodes were counted, collected and fixed in 10% formalin/PBS.

Histological analysis
10% formalin/PBS fixed colorectums and enlarged lymph nodes were sent to Histoserv INC. (Germantown, MD, USA) for paraffin embedding and hematoxylin and eosin (H&E) staining. In brief, paraffin embedded samples were cut into 6 μM and stained with H&E. A pathologist counted the malignant tumors (including high grade intraepithelial neoplasia and adenocarcinoma) and benign tumors in each slide. The malignant/tumor ratio of each slide was calculated and the average ratio of each group was presented. Inflammation was scored on a 0-4 scale (0, normal mucosa; 1, minimal inflammation (occasional scattered granulocytes and leukocytes); 2, mild inflammation (scattered granulocytes with occasional mild infiltrates); 3, moderate inflammation (scattered granulocytes with patchy moderate infiltrates); and 4, severe inflammation (multiple extensive areas with abundant granulocytes and marked infiltrates)) [12].

Plasma lipid concentrations measurement
Mice were sacrificed with isoflurane. Blood was collected by inferior vena cava injection in an anticoagulant tube and the tube was centrifuged at 800 g for 15 min at 4°C. Supernatant was collected as plasma. Plasma free cholesterol (FC), PL were measured using enzymatic reagent kit from Wako Diagnostics (USA), TC was measured using enzymatic reagent kit from Pointe Scientific (USA). TG was measured using enzymatic reagent kit from Sigma-Aldrich (USA).
To obtain membrane protein, colon mucous membrane or colorectal tumors in Membrane Buffer (20 mM Tris-HCl pH7.5, 2 mM MgCl 2 , 250 mM Sucrose) with protease inhibitors and phosphatase inhibitors were homogenized for 60 seconds on ice. Tissue homogenates were centrifuged at 2,000 rpm for 10 minutes at 4°C. The supernatant was transferred to a new tube and the tube was centrifuged at 110,000 g for 30 min at 4°C. The supernatant was removed and left the tube upside down to drain for 1 minute. Then, the pellet was resuspended in Sample Buffer (50 mM Tris-HCl pH8.0, 80 mM NaCl, 4 mM CaCl 2 , 1%(v/v) Triton X-100) with protease inhibitors and phosphatase inhibitors. This was membrane protein.

Statistics
The data shown represent the mean ± standard error. Statistical differences between groups were analyzed by Student's t-test, one-way ANOVA or chi-square test. P < 0.05 was considered statistically significant.

Azoxymethane (AOM)-dextran sodium sulfate (DSS) treatment induces tumors only in colorectum
To explore colorectal tumorigenesis in vivo, the widely used AOM-DSS model of colitis-associated colorectal tumorigenesis was employed [10,11]. Mice were sacrificed 10, 15, 18 or 20 weeks after AOM injection, respectively ( Figure 1B). Macroscopic tumors were counted and measured. The detail information was summarized in Figures 1, 2 and Table 1. Tumors were mainly distributed on the distal part of colorectum, with more distal tumors increasing in density (Figures 2A,B). Tumors were not found in cecum or small intestine. With increased time post-injection, tumors increased in numbers and size ( Figures 1C-F).
The liver, lung, spleen and kidneys of each mouse were examined and neither tumor nor metastatic cancer was found (Additional file 1: Figure S1E).
The difference of susceptibility to tumorigenesis between WT mice and NPC1L1 −/− mice does not appear to be caused by colon length because no significant difference in colon length between the two was detected ( Figure 1G).
Mice at 12 weeks and 24 months of age (including WT and NPC1L1 −/− ) without any treatment were sacrificed and did not show tumors in the colorectum, cecum, small intestine, lung, liver, spleen or kidneys. This suggests NPC1L1 knockout alone should not cause tumorigenesis in those organs within 24 months. NPC1L1 −/− mice have a lower ratio of malignant tumor/tumor than WT Eight colorectums of 20 W mice from each group were used to do H&E staining. All slides were examined by a pathologist. Squamous metaplasia, infiltration of inflammatory cells, adenoma, high grade intraepithelial neoplasia and adenocarcinoma were found in colorectums of both WT group and NPC1L1 −/− group ( Figure 2C). NPC1L1 −/− mice had a significant lower ratio of malignant tumor/tumor than WT (p < 0.05) ( Figure 2D). Malignant tumor/tumor ratio of WT was 33.9 ± 6.1% while that of NPC1L1 −/− was 14.2 ± 2.8%.

NPC1L1 in mice colorectal mucous membranes or colorectal tumors is undetectable by western blot
It was reported that NPC1L1 mRNA expression in both mice colon and human colon was very low [8,14]. Western blot was employed to assay NPC1L1 protein in intestine and colorectal tumors. At first, total proteins from adjacent colorectal mucous membranes and tumors were used and NPC1L1 signal was not detectable. Because NPC1L1 is mainly expressed in cell membrane [14,15], membrane proteins from adjacent colorectal mucous membranes and tumors were blotted at last. Small intestines were divided into 5 segments and total protein from the second segment was used as a control. As shown in Figure 3A, NPC1L1 was highly expressed in the jejunum of WT mice and was not detectable in that of the NPC1L1 −/− mice. NPC1L1 signal was not detectable in tumors or adjacent colorectal mucous membranes in either group. To confirm this result, western blot membrane was over exposed and the result was the same.
To confirm the quality of membrane protein, membrane protein loading control α-tubulin and membrane protein E-cadherin were both detected on the same membrane. As shown in Figure 3A, α-tubulin and E-cadherin were easily detected by western blot. N-cadherin was also assayed. It was easily detected in total protein of the jejunum and was undetectable in tumors or adjacent colorectal mucous membranes ( Figure 3A).

NPC1L1 knockout decreases lymphadenectasis
Inflammation and cancer metastasis are two main causes in lymphadenectasis [23,24]. In this study, enlarged abdominal lymph nodes were examined (Figures 3E,F and Table 2). Lymph nodes H&E staining showed that no metastatic cancer was found ( Figure 3F). This led to the hypothesis that enlarged abdominal lymph nodes may be caused by inflammation. The number of enlarged abdominal lymph nodes in WT mice was higher than that found in NPC1L1 −/− mice ( Figure 3E and Table 2). At week 10, all seven WT mice had enlarged abdominal lymph node and 5 mice had more than one. Three of the 7 NPC1L1 −/− mice had enlarged lymph nodes and none of them had more than one. The WT group had 1.9 ± 0.3 enlarged lymph nodes per mouse and the NPC1L1 −/− group had 0.4 ± 0.2 (p < 0.01). At week 20, all WT mice had enlarged lymph node and 9 of the 10 mice had more than one while 8 of the 10 NPC1L1 −/− mice had enlarged lymph nodes and only 2 mice had more than one. The WT group had 2.6 ± 0.3 enlarged  The effect of NPC1L1 knockout on β-catenin/p53/TGF-β/ p-gp in colitis-associated tumorigenesis was evaluated To explore the possible involved pathways, western blot was employed to assay β-catenin/p53/TGF-β/p-gp. The specific protein expression level in adjacent colons and tumors was measured on the same membrane. β-catenin was much higher in tumors than in adjacent colons in both groups. NPC1L1 knockout significantly decreased β-catenin expression level in tumors but no changes were detected in adjacent colons (Figures 4A,B). NPC1L1 knockout in mice didn't detectably change p53 expression in tumors or adjacent colons (Figures 4A,B).  (A) Western blot. Total proteins from small intestines and membrane proteins from adjacent colon mucous membranes or tumors were blotted. NPC1L1 in mice colorectal mucous membranes or colorectal tumors was undetectable by western blot. Membrane protein loading control α-tubulin and membrane protein E-cadherin were easily detected in tumors or adjacent colorectal mucous membranes. N-cadherin was easily detected in small intestine tissues while it was undetectable in tumors or adjacent colorectal mucous membranes. (B) Plasma samples from 12 W old mice without any treatment (control), 10 W or 20 W after AOM injection were assayed for FC, TC, TG and PL. *p < 0.05, **p < 0.01. (C) The intestinal inflammation scores of NPC1L1−/− mice were significantly lower than those of WT mice. *p < 0.05. (D) Western blot. NPC1L1 ablation decreased inflammatory markers p-c-Jun, p-ERK and Caspase-1 p20 in tumors. (E) Fresh enlarged (sciatic and lumbar) lymph nodes. Mice were sacrificed 20 weeks after AOM injection. (F) H&E staining of enlarged lymph nodes showed that metastasis was not found. (Scale bars represent 400 μm).
Interestingly, NPC1L1 knockout dramatically increased active TGF-β in tumors but dramatically decreased it in adjacent colons. On the contrary, NPC1L1 knockout decreased TGF-β precursor in tumors and increased it in adjacent colons (Figures 4 C, D).
NPC1L1 knockout in mice dramatically increased p-gp in tumors and the opposite in adjacent colons (Figures 4C,D).

Discussion
NPC1L1 protein, a protein of more than 1,300 amino acids, has 13 transmembrane domains and 5 of them constitute a sterol-sensing domain [5,25]. Its major function is a sterol transporter [5,8,15]. The NPC1L1 inhibitor, ezetimibe, began to be used to lower serum cholesterol in clinic but this has caused the great debate [6,7]. Biological scientists and clinical doctors raised the concern about its role in cancer recently.
Here, we testified that NPC1L1 −/− mice were resistant to colitis-associated tumorigenesis. At different time points, NPC1L1 −/− mice consistently had much fewer colorectal tumors than WT mice ( Figures 1C-G). The ratio of malignant tumor/tumor in NPC1L1 −/− at week 20 was also significantly lower than in WT ( Figure 2D).
To explore clues to possible mechanisms of NPC1L1 knockout protecting mice against colitis-associated tumorigenesis, NPC1L1 protein in colorectal mucous membranes and tumors was assayed. Western blot showed that NPC1L1 protein level was high in small intestines while it was undetectable in colorectal mucous membranes or colorectal tumors ( Figure 3A). It is reported that NPC1L1 mRNA was highly expressed in liver and small intestines of humans and only in small intestines of mice [8,14]. In the colon, its mRNA is very low [8,14]. Based on these results, NPC1L1 knockout reducing colitis-associated tumorigenesis is unlikely due to NPC1L1 ablation in colorectal mucous membranes.
To explore clues to the detail mechanisms of NPC1L1 knockout reducing colitis-associated tumorigenesis, βcatenin, p-c-Jun, p-ERK, TGF-β, p-gp and p53 were assayed.
NPC1L1 knockout significantly decreased β-catenin, p-c-Jun and p-ERK in tumors. Increasing β-catenin, c-Jun phosphorylation and ERK activation were reported to promote colon tumorigenesis [10,11,21,22]. It is logical to consider that β-catenin, p-c-Jun and p-ERK may be involved in the mechanism of NPC1L1 knockout protecting mice from colitis-associated tumorigenesis.
TGF-β and p-gp also play a positive role during intestinal tumorigenesis [26,27]. Interestingly, NPC1L1 knockout dramatically increased active TGF-β and p-gp in tumor but dramatically decreased them in adjacent colon. On the contrary, NPC1L1 knockout decreased TGF-β precursor in tumor and increased it in adjacent colon ( Figure 4B). It seems that TGF-β and p-gp might play complicated roles in mechanisms.
Protein p53, the most famous tumor suppressor protein, is involved in colorectal tumorigenesis and lipid metabolism [28,29]. Beyond our expectation, NPC1L1 knockout did not significantly change its expression either in adjacent colon or in tumor.