Unlocking Aspirin's Chemopreventive Activity: Role of Irreversibly Inhibiting Platelet Cyclooxygenase-1

Ever since its discovery in 1897, in addition to its activity to inhibit fever/pain/inflammation, aspirin when used chronically has been linked to a reduced incidence of a number of diseases including thrombosis/heart disease, arthritis, and cancer activity (1). Insight into the cardiovascular preventive action of aspirin came with Vanes' discovery that aspirin and related NSAIDs act by inhibiting COX enzyme activity, that regulate the generation of eicosanoids, providing a mechanism for the class' anti-inflammatory/antiplatelet activity (1). Indeed, aspirin's unique ability to irreversibly inhibit platelet COX-1 (via acetylation), led to its widespread use in patients at the risk of heart disease/thrombosis (2, 3).

Interestingly, there is a long history connecting blood coagulation disorders with late-stage cancer that dates back to the pioneering observation of Armand Trousseau who, in 1865, made the observation linking later stage cancer and venous thrombosis (4–6). It is now well established that thrombocytosis and elevated circulating tissue factor levels are predictors of later stage cancer along with increased incidence (4- to 8-fold) of venous thrombosis. A body of work demonstrates that platelets have the capability of interacting with circulating tumor cells (CTC) either directly or as a fibrin-associated network with neutrophils, and in so doing, prolonging the CTC circulatory half-life (7). In addition, platelets have been reported to trigger epithelial–mesenchymal transition (EMT) of cancer cells (8–10).

The association between low-dose (75–325 mg) aspirin consumption and a significant (20%–40%) reduction in cancer incidence was shown for colorectal cancer and the development of nine different non-gastrointestinal cancers (11). It was also reported that aspirin use was significantly associated with a profound increase in patient survival, thereby reducing the risk of fatal adenocarcinoma (1, 11, 12). Of interest was the observation that the metastatic spread of cancer could also be reduced, even if aspirin consumption was initiated postdiagnosis (12, 13). Aspirin use has now been linked to a reduced cancer incidence in >20 cancers, including the major cancers afflicting our current population that are responsible for the preponderance of cancer-related hospitalizations/costs, morbidity, and mortality activity (1, 12, 14–20).

Because of this compelling evidence, we designed studies using in vitro and in vivo models of colon cancer to explore the fundamental question of whether aspirin's antineoplastic activity is linked to its established ability to irreversibly inactivate platelets via COX-1 inhibition. The experiments focus on whether aspirin blocks platelets from promoting EMT, cancer cell growth, and the metastatic spread of cancer. In the current study, we also evaluated the chemopreventive activity of a novel phosphatidylcholine (PC)-associated aspirin, which has been formulated on the basis of its reported improved gastrointestinal safety in both preclinical studies (21, 22), as well as a pilot clinical trial (23). There is an unmet need for a gastrointestinal-safe aspirin, as the toxicity of aspirin to cause peptic ulceration and excessive gastrointestinal bleeding in susceptible individuals has limited the recommendations of this commonly available drug to the public at-large as a chemopreventive agent. These issues are reflected by the September 2015 recommendations of the U.S. Preventive Services Task Force (USPSTF) where individuals aged 50–69 years are advised to take low-dose aspirin on a daily basis to prevent colorectal cancer, under the physician's consultation (24).

Mice were maintained in accordance and compliance with the policies approved by the Animal Welfare Committee (AWC), the Institutional Animal Care and Use Committee (IACUC) for The University of Texas Health Science Center at Houston (UTHealth, Houston, TX). Samples of human blood for platelet isolation were obtained from normal subjects after obtaining informed consent under Institutional Review Board (IRB)-approved protocols at the Baylor College of Medicine and UTHealth (Houston, TX).

Young adult (20–24 g) male BALB/c mice and young adult (22–24 g) female Swiss Webster mice were supplied by Harlan Laboratories, Inc. (now Envigo) and housed in the Center for Laboratory Animal Medicine & Care (CLAMC) facility at UTHealth (Houston, TX).

Colon cancer cells (human Caco-2; HTB37) were purchased from the ATCC and murine colon cancer cells (MC-26) were obtained from the NIH National Cancer Institute (Bethesda, MD). The cell lines were cultured in suggested growth medium with 10% FBS (Sigma-Aldrich). All the cell lines were regularly authenticated on the basis of viability, morphology, and growth. Tests for mycoplasma were negative and were conducted with the MycoAlert Mycoplasma Detection Kit from Lonza.

All chemicals, buffers, and cell culture media were purchased from Sigma-Aldrich, unless described separately. Matrigel Membrane Matrix (354234) was from Corning. Azoxymethane (AOM) and dextran sulfate sodium salt (DSS, 36,000–50,000 MW) were purchased from MP Biomedicals LLC.

For in vitro studies, pharmaceutical grade aspirin was purchased from Novacyl Inc. and was dissolved in RPMI1640 culture media before undergoing sterile filtration. Aspirin-PC was prepared from aspirin and soy PC (S100 from Lipoid LLC). Initially, the PC was dissolved in chloroform and dried in a sterile glass vial under nitrogen gas. Then solubilized aspirin (in culture media) was added to the PC vial at a 1:1 ratio by weight and the vial was bath sonicated for 30 minutes. For in vivo studies, aspirin was purchased from a local pharmacy (Walgreens) and was prepared by crushing a tablet and suspending in water with sonication to a concentration of 2.5 mg/mL for intragastric administration. Aspirin-PC was prepared by mixing Novacyl aspirin with Lipoid Phosal 35SB cGMP which is the PC in the commercial formulation of Aspirin-PC. This mixture was suspended by bath sonication in water for animal studies.

A mouse mAb of β-catenin (sc-7963), rabbit polyclonal antibody of vimentin (sc-5565), HRP-conjugated goat anti-rabbit IgG (sc-2054), and HRP-conjugated goat anti-mouse IgG (sc-2055) were from Santa Cruz Biotechnology. A rabbit polyclonal antibody of β-actin (PA1183) and was from Thermo Scientific.

Using a previously described technique to isolate platelets (25), human blood was drawn into an acid/citrate/dextrose (ACD) anticoagulant at a ratio of 1:9 (v/v) from healthy fasting donors. Platelet-rich plasma (PRP) was prepared by centrifuging blood at 200 × g for 15 minutes at room temperature. PRP supplemented with 75 nmol/L of PGE₁ was centrifuged at 1,000 × g for 10 minutes at room temperature to obtain a platelet pellet that was resuspended and washed with buffer and adjusted to 2.5 × 10⁸ platelets/mL. Mouse blood (BALB/c) was collected from the inferior vena cava of random isoflurane-anesthetized mice into 3.8% sodium citrate at the ratio of 1:10 (v/v) for studies with PRP, and into ACD at the ratio of 1:10 (v/v) for washed platelet studies. Blood was diluted 1:1 with PBS and PRP was isolated following centrifugation at 68 × g for 10 minutes. Washed platelets were obtained following additional centrifugation of PRP as described previously (26). Platelets were resuspended in RPMI1640 with 0.01 U/mL of apyrase and counts adjusted to 2.5 × 10⁸ platelets/mL.

At the time of drug testing, human and murine colon cancer cells were seeded onto the Transwell chamber (purchased from Corning) with 0.4 μmol/L porous membrane at 6.25 × 10⁴ in 250 μL of serum-free RPMI1640 and washed platelets that were freshly collected from human or mouse were added in the bottom well at 5 × 10⁷ in 500 μL of serum-free RPMI1640. Cells without platelet incubation served as a control. After 36-hour incubation, 3-(4,5-dimethylthiazolyl-2) 2,5-diphenyltetra-zolium bromide (MTT, Sigma-Aldrich) solution was added to the media of cells in the Transwell at a final concentration of 0.5 mg/mL for 4 hours and the cells extracted and stained for MTT as described previously (27). The membrane was cut and extracted into plate wells containing 200 μL of the formazan solvent (90% isopropanol, 0.2% SDS, and 0.01 N HCl) in a 48-well plate, and read at an absorbance of 570 nm.

To assess invasive activity, 2.5 × 10⁵ mouse MC-26 colon cancer cells and 1 × 10⁸ washed platelets were mixed in 500 μL of serum-free RPMI1640 and applied to the uppermost Transwell chamber with 8-μm porous membrane precoated with Matrigel (BD Biosciences), and 750 μL of serum-free RPMI1640 was added into the bottom chamber in a 48-well culture plate. Cells without platelets served as control. After 16–20 hours of incubation at 37°C and 5% CO₂, the noninvading cells and extracellular matrix were gently removed with a cotton swab and the number of invading cancer cells located on the lower side of the Transwell membrane was assessed by fixing and staining with Hema 3 stain kit (Fisher HealthCare Protocol), and counting under a microscope.

In studies to assess EMT activity, MC-26 cancer cells were cocultured with mouse washed platelets (mWP) at a ratio of 1:400 at 37°C, 5% CO₂ for 18–24 hours. The 1: 400 was ratio was modeled after published studies, where a clear platelet effect on cancer cell growth and invasive activity was reported (28). The cultured cells were washed one time with PBS and then lysed with RIPA buffer (Fisher Scientific) for Western blot of vimentin. To assess β-catenin translocation into the nucleus, subcellular fractionation was performed according to the Abcam subcellular fractionation protocol. Briefly, the above cocultured cells were collected in cold fractionation buffer, and cell membrane/cytosolic fraction and nucleus fraction were prepared by two rounds of differential centrifugation. The cell fraction samples were used for Western blot of β-catenin. For immunofluorescence staining, the MC-26 cells were cocultured with mWPs at ratio of 1:400 on 18-mm coverslips with or without the test drugs in a 12-well plate for 18–24 hours.

Protein concentration of the cell extracts (whole cell lysates or subcellular fractions) was first determined by the BCA assay (Pierce), and then extracts were resolved by SDS-PAGE, transferred to PVDF membrane, and analyzed by Western blot analysis with the indicated antibodies.

RNA was extracted from the MC-26 cells supplemented with aspirin test drugs in the presence or absence of platelets, and reverse transcribed using Quantitect Reverse Transcription kit (Qiagen). qRT-PCR was then performed using QuantiFast SYBR Green PCR Kit (Qiagen) with primers specific for mouse Snail (Snail forward: 5′-GGAAGCCCAACTATAGCGAGC-3′; reverse: 5′-CAGTTGAAGATCTTCCGCGAC-3′). Expression of GAPDH was used to normalize the data and the relative mRNA analysis was calculated using the comparative C_t method.

Based upon a previously described technique alone (29), young (40–50 days old) female Swiss Webster mice were randomly assigned to treatment groups. On the basis of our previous experience, a minimum of 5 animals per group were needed to detect significant differences. They were administered a single intraperitoneal injection of azoxymethane (AOM) at 7.5 mg/kg. One week after the AOM challenge, the mice were placed on drinking water containing dextran sulfate sodium (DSS) at 2%, 3%, or 4% for 7 days, and then provided normal diet and drinking water for the next 2 weeks. The DSS challenge was repeated two more times (in total 3-cycles of DSS) and the protocol took 10 weeks.

The test drugs, aspirin and Aspirin-PC, at a daily aspirin dose (20 mg/kg) were administered intragastrically starting a week before the AOM challenge. The AOM control group was daily intragastrically administered the same volume of saline. The absolute control group did not receive AOM, DSS, or test drugs. During the study, body weight was determined every week as a measure of general health and fecal samples were also collected weekly for hemoglobin analysis as a measure of gastrointestinal (GI) bleeding (potential side effect of aspirin). At the end of the 10-week study period, blood was collected, the mice were euthanized by overdose of isoflurane anesthesia followed by thoracotomy, and the colon measured in length and weighed and biopsies taken and snap frozen in liquid N₂ for biochemical analysis or fixed and stained for aberrant crypt foci (ACF) counts and immunohistochemical (IHC) staining; hematocrit was assessed to detect blood loss due to gastrointestinal bleeding. Whole blood from the four mice groups were used to isolate PRP, count platelets, and TXB₂ analysis. PRP was further challenged with 2.5 μg/mL of collagen or left untreated (basal). The degree of platelet activation was assessed by flow cytometry using anti-P-selectin FITC and isotype control antibodies, as published previously (26).

Serum or cell culture medium was measured by using the TXB₂ EIA kit (Cayman Chemical, catalog no. #501020) or a mouse VEGF-A ELISA Kit (Pierce, catalog no. #EMVEGFA) according to the manufacturer's specifications. Sera were collected from the blood samples of individual mice at the end of the experiment under terminal anesthesia following a protocol for cardiac puncture. Serum samples were separated from blood within 1 hour following blood collection by centrifugation at 500 × g for 10 minutes, and then aliquoted and stored at −80°C for subsequent testing. Cell culture media were collected at the end of the experiments and cleared by centrifugation at 17,000 × g for 10 minutes and then stored at −80°C for subsequent testing.

Aberrant crypt foci (ACF) were identified in colonic tissue using a previously described technique (30). The excised colonic tissue was opened and placed mucosal-side-up onto paper, fixed in 10% buffered formalin, and briefly stained with 0.3% methylene blue. A 40× magnifying lens was used for tissue examination. All measurements were performed by persons who were unaware of the treatment group.

The cultured cells on coverslips were fixed in 4% formaldehyde, permeabilized by incubation in 0.5% Triton X-100, blocked with 10% goat serum, and then stained with polyclonal anti-fibronectin (Sigma-Aldrich) followed by Alexa-647–conjugated goat anti-rabbit antibody (Molecular Probes/Invitrogen). Afterwards, the samples were stained for f-actin, g-actin by Alexa Fluor 488 phalloidin and the nuclei by DAPI (Molecular Probes/Invitrogen). To assess the presence of platelets in the colonic mucosa, sectioned colonic tissue mounted on the slides were stained with a rat anti-CD41 antibody, followed by a FITC-conjugated goat anti-rat polyclonal antibody, and the nuclei by DAPI, based upon a previously described technique (31, 32).

Finally, sections were mounted on slides with elvanol (DuPont). Specimens were then visualized either by Deconvolution Microscopy employing an Olympus IX-70 microscope connected to a DeltaVision imaging system (Applied Precision), or Nikon A1R confocal laser microscope system and the NIS-Elements C software. The quantitation of immunofluorescence density was done using ImageJ software by measuring pixel units.

Statistical analyses were performed using the statistics application StatView 5.01 (SAS Institute Inc.). Values are expressed as the mean ± SEM, and were evaluated by ANOVA followed by Fisher PLSD test. The association of colonic tumor mass with circulation platelet number was evaluated by percentiles by correlation/covariance. A P value < 0.05 with two-tailed testing was deemed statistically significant.

We initially assessed the growth-stimulatory effect of platelets on mouse (MC-26 cells, Fig. 1A) and human (Caco-2, Fig. 1B) colon cancer cell lines. The ability of both mouse and human platelets to promote the proliferation of cancer cells within 48 hours was, in turn, dose-dependently reduced by aspirin and Aspirin-PC, with evidence that the Aspirin-PC may possess greater growth-inhibitory efficacy, within a dose range of 0.1–1.0 mmol/L (Fig. 1A and B).

Using culture plates where the top and bottom chambers were separated by a porous membrane insert, we investigated the ability of platelets to promote translocation/invasion of the cancer cells to the bottom chamber. The results in Fig. 1C using MC-26 cells demonstrate that coculture with mouse platelets promoted invasive activity, as we measured an approximately 20-fold increase in cell number across the membrane. Interestingly, this platelet-induced invasion was dose dependently attenuated by addition of the aspirin test drugs to the medium at concentrations >0.3 mmol/L, with a tendency for Aspirin-PC to be more effective (Fig. 1C). Coculture of cancer cells and platelets promoted the generation of the COX-1 product, TXB₂, indicative of platelet activation, and both aspirin drugs significantly reduced TXB₂ levels by >80% (Fig. 1D). We also screened the culture medium for alterations in the concentration of other growth mediators and determined that the coculture of platelets with MC-26 cells resulted in a significant increase in levels of VEGF in the medium, which was blocked by both aspirin and Aspirin-PC (Fig. 1D). In contrast, platelet-derived TGFβ1 was not increased when cocultured with mouse or human cells, and its release was not affected by either of the aspirin test drugs (Supplementary Fig. S1).

As platelets have been reported to induce EMT in cancer cells, we performed in vitro experiments to assess whether aspirin interferes with the EMT process. We initially observed that, in addition to a marked increase in cell number, MC-26 cells underwent a profound morphologic transformation within 48 hours after being cocultured with platelets. Both a change in shape to a more mesenchymal cell type and the formation of fibronectin-associated microfilamentous structures that formed an extracellular network connecting adjacent cells were seen (Fig. 2A and B). These platelet-induced morphologic changes were less evident when aspirin or Aspirin-PC was added to the culture medium (Fig. 2C). We also demonstrated that platelets promoted the translocation of β-catenin from the membrane/cytoplasm to the nucleus as shown by Western blot analysis of subcellular fractions (Fig. 3A), which was reduced when the aspirin test drugs were added to the culture medium, with Aspirin-PC being a more effective EMT inhibitor (Fig. 3B). Furthermore, platelet coculture induced an increase in vimentin expression of whole-cell lysates (Fig. 3C), and similar to the effects on β-catenin, this platelet-induced precancerous change was reduced if the cells were incubated with the aspirin test drugs with Aspirin-PC being consistently more effective (Fig. 3D). Finally, platelets increased mRNA expression of Snail (Snai1) in MC-26 cells, an EMT-inducing transcription factor, which was dose dependently inhibited by both aspirin and Aspirin-PC (Supplementary Fig. S2).

The in vivo model system chosen to study antineoplastic drug efficacy was an established mouse model of colon cancer, in which a chemical carcinogen that induces colonic dysplasia/neoplasia, azoxymethane (AOM), is used in combination with a chemical, dextran sodium sulfate (DSS), which induces colonic inflammation/colitis (schematically depicted in Fig. 4A). This combination induces the formation of aberrant colonic crypts which progresses to adenoma at an accelerated rate in comparison with AOM alone (29).

To optimize the DSS concentration, a study was performed to test cyclic use of 2%, 3%, or 4% DSS in the drinking water after a single AOM injection. At the end of the study, it was found that the circulating platelet number was increased dose dependently with 2%–4% DSS, with significant differences observed in AOM/3% DSS and AOM/4% DSS groups compared with that of the mice in the saline control group (Fig. 4B). This is in contrast to the lack of effect of AOM, in the absence of DSS, on platelet number (Fig. 4B), emphasizing the importance of colonic inflammation as the driver of thrombocytosis. Most interestingly, the ratio of colon weight to body weight, which is a measure of colonic tumor mass, was tightly associated (R = 0.82, P < 0.001) with the number of circulating platelets (Fig. 4C).

Subsequently, we used 3% DSS after AOM to evaluate the chemopreventive activity of the aspirin test drugs. Mice were intragastrically administered aspirin or Aspirin-PC commencing 1-week prior to AOM injection, at a daily aspirin dose (20 mg/kg) that is comparable with low-dose aspirin (75–150 mg/day) in humans. As shown in Fig. 4D, the circulating platelet number increased in the AOM/DSS/saline group (P < 0.01 vs. controls), and the aspirin test drugs reversed the thrombocytosis induced by AOM/DSS treatment (P < 0.05). We also determined that serum TXB₂ levels were decreased 80%–90% (P < 0.01) in the two aspirin treatment groups (Fig. 4E), indicating that the drugs' ability to inhibit platelet COX-1 activity, as well as platelet number. Compared with platelets from the control mice, AOM/DSS/saline treatment appeared to increase basal platelet activation as observed by an increase in P-selectin staining. Importantly, platelets from the mice treated with aspirin and Aspirin-PC also failed to demonstrate collagen-induced platelet activation, as measured by P-selectin expression (Supplementary Fig. S3).

Mice treated with AOM/DSS appeared to gain less body weight than controls, with only the AOM/DSS/saline group being significantly lower than that of the control group at the study's end, (Supplementary Figs. S4 and S5), suggesting that the animals developed systemic disease. The measurement of colonic aberrant crypts presented in Fig. 5A demonstrates that AOM/DSS treatment increased this dysplastic change in the colonic mucosa (P < 0.001), and both aspirin and Aspirin-PC treatments were effective (P < 0.01) in reducing aberrant crypt number toward control values. In addition, it was found that the colon weight/length in the AOM/DSS groups was higher than that of the control group (P < 0.01), which is reflective of mucosal edema/inflammation and that this ratio was lower in animals in both aspirin treatment groups (Fig. 5B, P < 0.05). To assess colonic tumor mass, the ratio of colon weight to body weight was calculated, as shown in Fig. 5C. Consistent with the above observations, there was an increase in this measure of colonic tumor mass over control values in the AOM/DSS/saline group (P < 0.01), which was reduced (P < 0.05) by both aspirin test drugs.

Changes in fecal hemoglobin (Supplementary Fig. S6) reflect gastrointestinal bleeding at the times of DSS administration, which was resolved by the end of the study period (week 10). Furthermore, at euthanasia, the hematocrit did not show any differences among groups (Supplementary Fig. S7), indicating the transient DSS-induced gastrointestinal blood loss was not lasting and there was no detectable gastrointestinal bleeding from any of the drug treatments. This supports the gastrointestinal safety of chronic dosing with the relatively low aspirin dose of 20 mg/kg dose of aspirin and Aspirin-PC in mice.

Using an immunofluorescent chemical (IFC) stain for CD41 platelet surface integrin, we assessed whether platelets seed into colonic tissue undergoing dysplasia/neoplasia, and if so, whether this pattern was affected by treatment with aspirin or Aspirin-PC. Representative IFC results shown in Fig. 6A and B, provides evidence that platelets are generally present at a low density and restricted to the large vessels present in the submucosa of control colonic mucosal tissue, whereas their number is noticeably increased within the major blood vessels of the dysplastic/hypertrophic colonic mucosa of AOM/DSS–challenged mice with evidence of translocation across the endothelium and traffic to adjacent submucosal/subepithelial tissues. This morphologic pattern was less evident in the aspirin and Aspirin-PC–treated mice, where infiltration of the platelets into the submucosa of the colonic mucosa was rarely observed, which was quantitatively supported by a reduction in fluorescent intensity of colonic tissue, with Aspirin-PC tissues having a substantial inhibitory effect, similar to values seen in normal (noncancerous) tissue (Fig. 6C).

The mechanism by which aspirin use is linked to a reduction in cancer incidence and metastases has yet to be fully elucidated. As aspirin consumption has been associated with a reduced incidence of numerous cancers, it suggests that one way the drug may act is at a stage which is common for most tissues undergoing dysplasia (1, 11, 12). As tissue inflammation is tightly linked to many chronic disease states including cancer, and aspirin has clear anti-inflammatory actions, this drug action may be an important component in its chemopreventive activity (1, 33). Indeed, a number of laboratories (34, 35) reported that colorectal cancer tissue overexpresses COX-2, suggesting it may be the primary target enzyme by which NSAIDs and notably COX-2–selective inhibitors (coxibs) reduce cancer incidence and recurrence. Although this indeed is an attractive theory, it has a number of limitations. These include evidence that coxibs have chemopreventive activity that may be limited to familial adenomatous polyposis patients. In comparison, there is more compelling evidence for aspirin protection in more common colorectal cancers such as in Lynch Syndrome (hereditary nonpolyposis) patients (36, 37). Also, not all cancers overexpress this COX isoform, and Rothwell reported that cancers affecting the proximal colon, which generally do not overexpress COX-2, are more responsive to aspirin use than those of the distal colon (38). Finally, for aspirin to target COX-2 in most cancers and affect the metastatic spread of tumor cells would require that aspirin remain stable in circulation. However, pharmacokinetic data demonstrates that aspirin rapidly hydrolyzes to salicylic acid within 10–20 minutes after it enters the blood (39), and salicylic acid has a much lower (100–1,000 fold) COX-2 inhibitory activity versus aspirin, especially in tissues undergoing inflammation (40).

Because of these aforementioned reasons, we felt an alternative mechanistic basis of aspirin's antineoplastic action needed to be explored. Indeed, as indicated earlier, platelets are a primary target of aspirin and are involved in thrombosis of late-stage cancer patients. Platelets are known to promote EMT of cells in the early stage of dysplasia thereby playing a role in cancer cell invasiveness, attachment, and growth, processes that are thought to be mediated in part by the release of mediators, notably platelet-derived growth factor (PDGF), VEGF, and TGFβ (8–10). Thus, as speculated by a number of investigators (33, 41), the platelet is a likely target of aspirin, especially as the irreversible inhibition of platelet COX-1 via acetylation, will inactivate the platelet for its circulatory half-life (7–10 days in humans and 5–7 days in mice), until a new generation of platelets enters the circulation via megakaryopoiesis.

Based upon this background, we performed in vitro and in vivo studies to directly evaluate the role of aspirin's antiplatelet activity in its ability to inhibit colon cancer growth and invasive potential. We also evaluated the chemopreventive activity of a novel phospholipid-associated aspirin, which was the result of our laboratory studies that showed surface-active lipids, such as phosphatidycholine (PC) play an intrinsic role in protecting the gastrointestinal mucosa from surface injury, as may be caused by NSAIDs (42, 43). Subsequently, a family of PC-NSAIDs was developed that possesses reduced gastrointestinal toxicity, primarily by attenuating the surface-injurious action of NSAIDs, while maintaining the therapeutic activity of the drugs (44). These observations were the impetus for the commercial development of PL2200 Aspirin (PLx Pharma Inc) which combines aspirin with a PC-enriched soy lecithin. We have previously reported that PL2200 Aspirin induces significantly fewer gastroduodenal endoscopic ulcers than normal aspirin in a 7-day trial of healthy subjects, aged >55 years, who are most susceptible to the gastrointestinal side-effects of NSAIDs (23). PL2200 Aspirin also appears to have profound antiplatelet activity in both healthy and diabetic subjects, supporting its potential use for both cardiovascular disease and possibly cancer. Results described above demonstrate that platelets promote the growth and invasive potential of both mouse (MC-26) and human (Caco-2) colon cancer cells, and that these actions can be dose dependently blocked by both aspirin and Aspirin-PC. Similar to Hynes and colleagues' studies (10), we confirmed that platelets promote EMT in colon cancer cells, and that these robust platelet-dependent cellular/biochemical changes can also be blocked by our aspirin test drugs.

We performed in vivo studies in mice with an established model of colon cancer that uses a regimen of AOM and DSS (29). It should be noted that we used this particular model, as it simulates the colonic inflammation and adenoma formation that occurs clinically in colorectal cancer, and has been demonstrated to be responsive to NSAIDs and other chemopreventive agents, whereas another colorectal cancer model, APC (Min/+) mice demonstrated little or no response to aspirin, possibly because the polypoid lesions occur primarily in the small intestine (45). Accordingly, using the AOM/DSS model we demonstrated that daily treatment of mice with 20 mg/kg of aspirin or the equivalent dose of Aspirin-PC, significantly reduced both the formation of aberrant crypts in the colon, which is a sign of dysplasia as well as colonic tumor mass in comparison with saline-treated controls. Interestingly, the AOM/DSS–challenged animals also had elevated platelet counts, which were reduced in the aspirin-treated groups, with little difference in the anticancer action observed between Aspirin-PC and the unmodified NSAIDs. This result thereby differs from our in vitro results where Aspirin-PC consistently possessed more pronounced potency/efficacy to inhibit cancer cell growth and invasive activity. Future experiments will be designed to provide an explanation for these results. Using the mouse colon cancer model system, we demonstrated that platelets migrated into the inflamed colonic mucosa and adjacent tumor tissue, appeared to be associated with the vasculature as well as surrounding areas of mucosal inflammation/dysplasia and this response was reduced in mice treated with aspirin and Aspirin-PC. These observations are consistent with studies by Granger's laboratory (46) that reported increased platelet reactivity in colonic mucosa undergoing DSS-induced inflammation, and a recent study by Qi and colleagues (31) that reported the presence of platelets in both intestinal tumors from APC Min mice and from clinical biopsies of patients with colorectal cancer, and a study from Miyashita and colleagues (32) of platelet extravasation into the invasive front of pancreatic ductal adenocarcinomas. We extended these very interesting observations, to demonstrate that the ability of platelets to seed into areas of colonic dysplasia/neoplasia. Using a similar approach, Guillem-Llobat and colleagues (47) recently reported that platelet-induced biochemical changes associated with EMT of human HT-29 colon carcinoma cells was blocked by COX-1 inhibitors including aspirin.

It is very important to emphasize that the findings presented above that aspirin's chemopreventive activity is attributable to the NSAID's ability to irreversibly inhibit platelet COX-1 and their subsequent activation, has limitations, being generally correlative in nature, and future experiments are planned to study aspirin sensitivity in animals where we experimentally manipulate platelet number and COX expression (e.g., transgenic mice where platelet COX-1 is disrupted). Also, it has been reported that aspirin and related NSAIDs can inhibit cancer cell growth and/or promote apoptotic activity of cancer cells by COX-independent pathways, that involve activation of the tumor suppressor AMP kinase (AMPK), or inhibition of the transcription factors, NFκB and AP-1 or directly affecting mitochondrial function (48, 49).

In summary, our findings suggest that aspirin's antineoplastic actions can be explained, in part, by the drug's ability to rapidly inhibit the number/activity of circulating platelets via irreversible COX-1 inhibition. Thus, our results support the use of low-dose aspirin for chemoprevention that is sufficient to block platelet–cancer cell interactions, which play a fundamental role in cancer formation and progression, while having limited anti-inflammatory activity. These findings address the initial actions of aspirin on cancer development and do not dismiss COX-2 and its downstream proinflammatory eicosanoid products as being important in certain cancers, which remains an attractive target in explaining the chemopreventive action of coxibs/NSAIDs when used at anti-inflammatory doses. Finally, we present evidence that a novel lipid-based aspirin, Aspirin-PC/PL2200 Aspirin, possesses similar chemopreventive actions to low-dose aspirin and may be more effective. Future experiments are planned to determine the efficacy and safety of PL2200 Aspirin in subjects that are at high risk for the development of colorectal cancer.

L.M. Lichtenberger has ownership interest (including patents) in PLx Pharma Inc and is a consultant/advisory board member for PLx Pharma Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: L.M. Lichtenberger, R.J. Bick, E.J. Dial, K.V. Vijayan

Development of methodology: L.M. Lichtenberger, D. Fang, R.J. Bick, B.J. Poindexter, E.J. Dial, K.V. Vijayan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.M. Lichtenberger, D. Fang, R.J. Bick, B.J. Poindexter, T. Phan, A.L. Bergeron, K.V. Vijayan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L.M. Lichtenberger, D. Fang, R.J. Bick, B.J. Poindexter, S. Pradhan, K.V. Vijayan

Writing, review, and/or revision of the manuscript: L.M. Lichtenberger, D. Fang, R.J. Bick, B.J. Poindexter, E.J. Dial, K.V. Vijayan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.M. Lichtenberger, R.J. Bick, B.J. Poindexter, A.L. Bergeron

Study supervision: L.M. Lichtenberger, K.V. Vijayan

Other (image acquisitions-core lab director): R.J. Bick

Other (conducted experiment and analysis of data): S. Pradhan

This work was supported by NIH grants R41 CA171408 awarded to PLx Pharma Inc and L.M. Lichtenberger and R21 CA182798 awarded to L. Lichtenberger and K.V. Vijayan. This material is the result of work supported in part with resources and the use of facilities of the Michael E. DeBakey VA Medical Center and the membership of Dan. L. Duncan Comprehensive Cancer at Baylor College of Medicine.

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.