Inhibitory effects of olive oil phenolics on invasion in human colon adenocarcinoma cells in vitro
Article first published online: 17 OCT 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 122, Issue 3, pages 495–500, 1 February 2008
How to Cite
Hashim, Y. Z.H.-Y., Rowland, I. R., McGlynn, H., Servili, M., Selvaggini, R., Taticchi, A., Esposto, S., Montedoro, G., Kaisalo, L., Wähälä, K. and Gill, C. I.R. (2008), Inhibitory effects of olive oil phenolics on invasion in human colon adenocarcinoma cells in vitro. Int. J. Cancer, 122: 495–500. doi: 10.1002/ijc.23148
- Issue published online: 26 NOV 2007
- Article first published online: 17 OCT 2007
- Manuscript Accepted: 8 AUG 2007
- Manuscript Received: 25 OCT 2006
- olive oil phenolics;
- colorectal cancer;
- extracellular matrix
Studies in human, animal and cellular systems suggest that phenols from virgin olive oil are capable of inhibiting several stages in carcinogenesis, including metastasis. The invasion cascade comprises cell attachment to extracellular matrix components or basement membrane, degradation of basement membrane by proteolytic enzymes and migration of cells through the modified matrix. In the present study, we investigated the effect of phenolics extracted from virgin olive oil (OVP) and its main constituents: hydroxytyrosol (3,4-dihydroxyphenylethanol), tyrosol (p-hydroxyphenylethanol), pinoresinol and caffeic acid. The effects of these phenolics were tested on the invasion of HT115 human colon carcinoma cells in a Matrigel invasion assay. OVP and its compounds showed different dose-related anti-invasive effects. At 25 μg/ml OVP and equivalent doses of individual compounds, significant anti-invasive effects were seen in the range of 45–55% of control. Importantly, OVP, but not the isolated phenolics, significantly reduced total cell number in the Matrigel invasion assay. There were no significant effects shown on cell viability, indicating the reduction of cell number in the Matrigel invasion assay was not due to cytotoxicity. There were also no significant effects on cell attachment to plastic substrate, indicating the importance of extracellular matrix in modulating the anti-invasive effects of OVP. In conclusion, the results from this study indicate that phenols from virgin olive oil have the ability to inhibit invasion of colon cancer cells and the effects may be mediated at different levels of the invasion cascade. © 2007 Wiley-Liss, Inc.
In 2002, world cancer statistics for new colorectal cancer (CRC) cases were estimated to be more than 1 million, with more than 500,000 mortality cases.1 Surgical resection is an effective treatment for localized disease, achieving a 5-year survival rate of 90%, but other normal treatments for metastatic disease remain ineffective.2 Mediterranean countries have lower rates of CRC than other Western countries.3 This has been attributed to a number of factors, one of which is consumption of olive oil.3, 4
Recently, olive oil phenolics have been shown to exert anticancer effects in a number of studies (reviewed in Refs.5 and6). Our previous study demonstrated that olive oil phenolics have the ability to inhibit initiation, promotion and invasion events in in vitro models of carcinogenesis.7
Metastasis can be considered as a cascade of interrelated sequential steps. Cells must be able to disseminate from the primary tumor, invade the surrounding tissue, enter the circulatory system, evade immune responses, arrest at and colonize a distant site.8 A 3-step hypothesis has been proposed to describe invasion of tumor cells: attachment to basement membrane or extracellular matrices (ECM), protease activity that induces local degradation of the matrix, and migration of tumor cells through the modified matrix.8, 9 Integrins mediate tumor cell attachment to ECM components such as laminin, fibronectin, vitronectin and collagens, as well as modulating proteolytic enzymes of intracellular signalling pathways which govern the cytoskeletal organization and gene expression.10
Inhibition of invasion constitutes a new target for chemoprevention in which intervention could commence between the period of tumor proliferation and the onset of invasion, preventing the series of events leading to metastasis.11 In relation to the poor prognosis of CRC,12 a window of opportunity exists for a chemopreventive regimen to be implemented in line with other therapy for CRC. It is, therefore, the aim of this in vitro study to investigate the effects of an olive oil phenolics extract and its compounds on the invasion and metastasis of human CRC using a human colon tumor cell line as a model system.
Material and methods
Cell culture reagents were obtained from Invitrogen (Paisley, Scotland) and Sigma (Poole, UK) unless otherwise stated. Hydroxytyrosol (3,4-dihydroxyphenylethanol) was obtained from Cayman Chemical (Michigan, USA), tyrosol (p-hydroxyphenylethanol) from Extrasynthese (Lyon, France) and caffeic acid from Sigma, UK. Pinoresinol and olive oil phenolics extract (OVP) were prepared in the University of Helsinki, Finland and University of Perugia, Italy, respectively. Briefly, the virgin olive oil was extracted at an industrial plant in Italy. The oil was then subjected to a methanolic extraction. After a complete solvent removal, the final crude extract was reconstituted in dimethyl sulfoxide (DMSO) at a stock concentration of 100 mg/ml, aliquoted (20 μl) and stored at −80°C under a nitrogen atmosphere until use. The total phenolic content of the crude extract was evaluated using the Folin–Ciocalteau reagent while the concentrations of the individual phenolic compounds were evaluated using an HPLC system as described in Ref.7. Concentrations studied were 0–25 μM for the olive oil phenolic compounds hydroxytyrosol, tyrosol, pinoresinol and caffeic acid and 0–25 μg/ml for OVP. Hydroxytyrosol, tyrosol and caffeic acid were dissolved in ethanol while OVP and pinoresinol were dissolved in DMSO.
Cell line and culture conditions
Human adenocarcinoma cells (HT115) and human fetal lung cells (MRC-5) were obtained from the European Collection of Animal Cell Cultures (ECACC), Salisbury, UK. HT115 cells were cultured in Dulbecco's minimum essential medium containing 15% fetal bovine serum (FBS) and 2 mM L-glutamine. MRC-5 cells were cultured in minimum essential medium, 10% FBS, 2 mM L-glutamine and 1% nonessential amino acid. Cells were cultured for 5–7 days (70% confluent) at 37°C/5% CO2. Media were changed every 2–3 days.
Matrigel invasion assay
The assay was carried out as described in Ref.7. MRC-5 fetal lung cells that secrete hepatocyte growth factor are used as chemoattractant to enhance invasion rates of the HT115 cells.13, 14 HT115 cells were co-treated with isolated phenolics for 24 hr in Biocoat™ Matrigel invasion chambers. For investigation of effects of OVP on adhering and already adherent cells, 2 experimental conditions were involved and were performed simultaneously. In the first protocol (effects on adherent cells), cells were seeded in the Biocoat chambers and incubated for 24 hr. Then, OVP was added to the adherent cells and further incubated for 24 hr. In the second protocol (effects on adhering cells), cells were seeded in the Biocoat chambers, co-treated with OVP and incubated for 24 hr. To see parallel effects on adhering versus adherent cells, the co-treatment of OVP in the second protocol was set up to be at the same time point as the OVP treatment in the first protocol. Upon completion of fixing, staining and scrubbing procedures, the numbers of invasive and noninvasive cells were counted in 5 random fields of chambers using Kromascan software (Kinetic Imaging, UK 2000). Percent invasion was calculated [number of invasive cells/(number of invasive + noninvasive cells) × 100].
Cell attachment assay
HT115 cells were harvested and seeded into 25 cm2 flasks at 9 × 105 cells with media containing isolated olive oil phenolics or OVP, respectively. Cells were incubated for 24 hr before trypsinization and counting using hemocytometer and trypan blue dye.
Cell viability assay
HT115 cells were harvested and seeded into 25 cm2 flasks at 3× 105 cells. After 48 hr, media were removed and fresh media containing isolated olive oil phenolics or OVP were added. Cells were incubated for further 24 hr before trypsinization and counting as described above.
Cell adhesion assay to specific ECM
Assays for cell adhesion to specific ECM (fibronectin, vitronectin, laminin, collagen types 1 and 4) were performed using a cell adhesion kit (Chemicon, Temecula, USA) on OVP and hydroxytyrosol according to manufacturer's protocol. Briefly, the assay involves the seeding of single suspension HT115 cells in serum-free media onto ECM pre-coated wells, incubation (45 min) to allow adhesion, washing, fixing, staining and measurement of absorbance. Three experimental conditions were studied, i.e. co-treatment, pre-treatment and pre-treatment with co-treatment. In the co-treatment experiment, cells were seeded and co-treated with samples (45-min incubation) before measuring absorbance. In pre-treatment setting, cells were grown to 60–70% confluence in flasks before sample was added, then further incubated for 24 hr. Upon completion of incubation, cells were harvested and subjected to cell adhesion assay. In the pre-treatment with co-treatment setting, cells were pre-treated as above and samples were again added prior to 45 min incubation in the adhesion assay. Single suspension cells used in this assay were obtained from trypsinization using trypsin without EDTA. Cells harvested using mechanical dissociation showed an adhesion profile similar to that of cells obtained by the trypsinization method, but cells detached from substrate in large clumps, which were not suitable for use in the adhesion assay (data not shown).
Statistical analysis for % invasion was carried out using Kruskal–Wallis nonparametric test, with Mann–Whitney post hoc analysis for OVP treatment, GLM Univariate with Dunnett t-test post hoc analysis for hydroxytyrosol, pinoresinol and caffeic acid and independent student's t-test for tyrosol (due to unequal variances between the tyrosol-treated groups). For total cell number in the Matrigel invasion assay, one-way ANOVA was used for OVP treatment, and GLM Univariate with Dunnett t-test post hoc analysis for isolated phenolics. Statistical significance for co-treatment adhesion assay was determined using GLM Univariate with Dunnett t-test (for OVP) and nonparametric Kruskal–Wallis with Mann–Whitney post hoc test (for hydroxytyrosol). For pre-treatment and co-treatment study, one-tailed independent samples t-test was used. Results from all assays are expressed as the mean of 3 independent experiments, except for the co-treatment adhesion assay (mean of 6 independent experiments). For each independent experiment, 3 replicate determinations were performed and a mean value calculated.
Concentration of OVP and isolated phenolics
The concentration (μg/ml) of the individual phenolics in the OVP was determined by HPLC as described in Ref.7. We have previously reported that there were 0.8 mg of hydroxytyrosol, 0.41 mg of tyrosol and 7.23 mg of pinoresinol in 100 mg of OVP.7 Based on these values, the equivalent micromolar (μM) concentration for each isolated phenolic compound studied was then calculated. At 25 μg/ml OVP, there was 1.30 μM of hydroxytyrosol, 0.75 μM tyrosol and 5.05 μM pinoresinol. Caffeic acid was not determined. Since individual compounds were tested at 0–25 μM, the nearest equivalent μM concentrations of individual compounds to 25 μg/ml OVP were as follows: 1.56 μM hydroxytyrosol, 1.56 μM tyrosol and 6.25 μM pinoresinol.
Effects of OVP and isolated phenolics on HT115 invasion through Matrigel basement membrane
As shown in Figure 1, OVP inhibited invasion of adhering and adherent cells through Matrigel. In the adhering cells, the greatest inhibition (∼60%) was shown by OVP at a concentration of 1.56 μg/ml (p = 0.025) while at 12.5 μg/ml and 25 μg/ml, inhibition was ∼45% (p < 0.05). In the already adherent cells, OVP inhibited invasion by 60% at 12.5 μg/ml and 25 μg/ml (p < 0.05).
In the Matrigel invasion assay, an estimate of total cell number was also taken by adding the mean number of invasive and noninvasive cells counted on the inserts. The total cell number denotes the effect of olive oil phenolics on attachment of cells to the Matrigel basement membrane. In the adhering cells, OVP showed significant total cell number reduction of ∼15–40% for all concentrations tested (p values between 0.002 and 0.013). No significant reduction of cell number was seen in the adherent cells.
Individual phenolic compounds (hydroxytyrosol, tyrosol and pinoresinol) inhibited invasion by ∼30–70% for the range of concentration tested (0–25 μM). Caffeic acid inhibited invasion by ∼40% at 3.13 and 6.25 μM (p = 0.003 and 0.006, respectively) but not at other concentrations. There were no significant differences in total cell count in the Matrigel assay for all isolated phenolics tested.
Comparison of anti-invasion effects of OVP and isolated phenolics
In Figure 2, effects of OVP on invasion and total cell number in the Matrigel system (effects on adhering cells) were co-plotted to the nearest equivalent doses of hydroxytyrosol, tyrosol and pinoresinol present in the extract at 25 μg/ml. The anti-invasion effects of individual compounds at their respective equivalent concentrations were comparable to the effects at 25 μg/ml OVP (45–55% significant inhibition). The reduction of total cell count in the Matrigel system was significant only for 25 μg/ml OVP but not the isolated phenolics.
As seen in Figure 3, there were fewer cells on the apical Matrigel membrane surface (noninvasive cells) after co-treatment with OVP than with isolated phenolics. The morphology of these noninvasive cells was also different: OVP-treated cells appeared more rounded while other treatments caused a normal spreading-like/flattened appearance.
Effects of OVP and hydroxytyrosol on HT115 cell adhesion to specific ECM
In the co-treatment study, OVP produced a significant decrease in adhesion to collagen type 4 (p = 0.004). In the pre-treatment with co-treatment study, OVP resulted in a significant decrease in adhesion to collagen type 4 (p = 0.0465). There was no significant effect of OVP in the pre-treatment only study. Hydroxytyrosol did not show any significant effects in either study (Figs. 4a and b).5
The major phenolic compounds identified in olive oils are hydroxytyrosol (3,4-dihydroxyphenylethanol) and tyrosol (p-hydroxyphenylethanol), secoiridoids (linked forms of hydroxytyrosoland tyrosol) and lignans [(+)-pinoresinol and (+)-1-acetoxypinoresinol].15 The total phenolic content in the OVP used in this study was 34% (w/w) with 7.2% (w/w) lignans (both forms), 0.8% (w/w) hydroxytyrosol and 0.4% (w/w) tyrosol.7 Caffeic acid was not determined in this extract. However, it has been reported to be present in virgin olive oil phenolic fractions16, 17 and its concentration in virgin olive oil is very low (500 μg/kg of oil) (Servili M, unpublished data, 2006).
In the present study, the Matrigel invasion assay was employed to evaluate tumor cell invasiveness in the presence or absence of OVP and its individual compounds. At 25 μg/ml, OVP showed ∼45% inhibition of invasion, similar to that reported from our previous study.7 This inhibition of invasion by OVP showed a parabolic dose response effect. It was apparent that hydroxytyrosol showed a similar parabolic trend but tyrosol, pinoresinol and caffeic acid did not. This parabolic trend may be due to autooxidation of the compounds at concentrations ≥10 μM, as reported in a similar study with lycopene.18 Further, in a bleomycin assay, hydroxytyrosol has been reported to have pro-oxidant effects at a concentration of 0.01% (w/v),19 which may account for the parabolic dose respond effect in our invasion assay.
Phenolic antioxidant actions have been mostly ascribed to their free radical scavenging activity, metal chelating properties and ability to modulate cell signalling pathways and gene expression.20 At the invasion and metastasis level, reactive oxygen species are associated with induction of invasion-related genes and may also act as intracellular signalling molecules favoring invasion and metastasis.21, 22 Antioxidant properties in tea compounds23 and lipophilic ascorbic acid derivatives24 have been shown to suppress invasion in vitro.
It has been suggested that the antioxidative capacity of olive oil phenolics is conferred by their ortho-diphenolic (catecholic) structure, which is present in hydroxytyrosol and caffeic acid but not in tyrosol.25, 26 Quiles et al.27 reported that olive oil phenolics showed the most significant antioxidative effects in the order of hydroxytyrosol > caffeic acid > tyrosol at 10–250 μM in PC3 human prostate cancer cell line. However, in our study, only hydroxytyrosol and pinoresinol showed significant anti-invasive effects at 25 μM, while at the lowest concentration tested (1.56 μM), pinoresinol showed the highest anti-invasive effects (65% of control).
To make comparisons between OVP and its individual phenolic components, the anti-invasive effects of the equivalent doses of individual compounds present in 25 μg/ml OVP were co-plotted (Fig. 2). All respective compounds caused an inhibition of invasion around 45–55%. This suggests that the anti-invasive effects shown by OVP may be attributed to the specific phenolics. In a study by Owen et al.,15 pinoresinol, a type of lignan, showed the highest antioxidative activity as compared with other phenolics present in olive oil. Thus, our findings seem to support the antioxidant mediated inhibition of invasion.
Interestingly, apart from the antioxidant mediated inhibition of invasion, our results clearly show that OVP, but not individual compounds, reduced total cell number in the Matrigel invasion system. This indicates that the anti-invasion effects of OVP could be due, at least in part, to anti-attachment properties whereas anti-invasion effects shown by the individual compounds are most likely mediated at other levels of the invasion cascade. It also suggests that the apparent anti-attachment properties of OVP may result from synergistic properties of its individual compounds.
Further, OVP reduced the number of the adhering cells to a constant level throughout the concentration range tested, but had no such effect in the adherent cells. Interestingly, the same concentration range showed a parabolic inhibition of invasion through the Matrigel in the adhering cells but a linear inhibition of invasion in the adherent cells. Together, this suggests that in the adhering cells, the anti-invasive effects of OVP are cumulative of its effects at the anti-attachment, antiproteolytic and antimigration levels. In the adherent cells, by contrast, the anti-invasive effects are mediated by the antiproteolytic activity and/or antimigration activities but not at the attachment level.
In the case of the antiproteolytic activity, phenolics compounds may chelate the metal ions required by matrix metalloproteinases (MMPs) to exert their proteolytic activity, thus inhibiting the overall invasion.28 Inhibition of MMPs has also been reported in several studies of phenolics compounds from green tea.29, 30, 31
The reduction in total cell number in the Matrigel invasion assay was not due to cytotoxicity, as demonstrated in cell viability study (data not shown). In the assay of cell attachment to plastic substrate, there were no significant effects of either OVP or the isolated phenolics (data not shown). On plastic surfaces, tumor cells have to secrete their own ECM to attach, spread and grow.32 This phenomenon is possibly inadequate for olive oil phenolics to be able to show the anti-attachment effect.
It is possible that other compounds present in the OVP conferred the anti-attachment effects. For instance, high antioxidative properties in virgin olive oil have been attributed to hydroxytyrosol linked to the dialdehydic form of elenolic acid, which is present at higher concentrations than free hydroxytyrosol. The dialdehydic form of elenolic acid, linked to hydroxytyrosol or to tyrosol, has been shown to account, in part, for the antiproliferative and apoptotic effects of virgin olive oil phenol extract in human promyelocytic leukemia cells.33 Strong antioxidant activity of 3,4-dihydroxyphenylethanol-elenolic acid was also reported.26 The higher concentration of the linked forms of hydroxytyrosol and tyrosol present in our OVP extract7 may account for the effects shown by OVP as compared with the isolated compounds studied.
In the Matrigel invasion assay, treatment of OVP on already adherent cells, as opposed to the adhering cells, did not reduce total cell number significantly, further suggesting that the OVP may affect the integrin–ECM interaction of adhering cells but not cause detachment of adherent cells, at least in the 24-hr incubation period. Treatment of adhering cells with OVP extract showed significant decrease of adhesion to collagen type 4 both in the co-treatment (only) study and in the pre-treatment with co-treatment study. No significant effect was observed in pre-treatment (only) study. Hydroxytyrosol showed no significant effects on adhesion to specific ECM, confirming the absence of detachment shown in the invasion assay. Other isolated phenolics (tyrosol, pinoresinol and caffeic acid) tested at 25 μM (co-treatment) showed no effects on cell adhesion to ECM (data not shown).
Based on this observation, a possible mechanism by which OVP exerts its anti-invasion and anti-attachment effect may be by interrupting integrin–ECM interaction. In this context, caffeic acid phenethyl ester was shown to regulate integrin-mediated signalling in human colon carcinoma cells.34
Olive oil phenolics may exert biological effects systemically or directly in the gut. It has been shown from ileostomy studies that 55–73% of ingested olive oil phenols are absorbed.35 Another study reported that 100% of hydroxytyrosol was absorbed in the small intestine36 while hydroxytyrosol and tyrosol were shown to be dose-dependently absorbed in humans and excreted in urine as glucuronide conjugates.37
Owen et al.38 suggest that the unabsorbed fraction of OVPs may reach the colon and confer chemopreventive effects against CRC. The unabsorbed fraction of pinoresinol in the olive oil phenols may be used by intestinal microflora to produce the mammalian lignans enterodiol and enterolactone, which have been shown to reduce invasion in breast cancer cell lines.39, 40
Dietary olive oil intake has been reported to reach 50 g/day, which corresponds to ∼25 mg of total phenols.41 Our extract (OVP) contains ∼34% total phenols.7 Based on this amount, at the highest concentration of OVP tested in this study (25 μg/ml), there were ∼20 μg of total phenols present. Therefore, the concentration used was of physiological range.
Previous studies on the biological activity of olive oil phenolics in relation to cancer have focused on antioxidant effects. Our study indicates that they have other potential mechanisms of anticancer activity, notably relating to invasion. Such effects may be mediated by a variety of molecular mechanisms but our results suggest that prevention of attachment of cancer cells to the ECM may be particularly important.
Yumi Z.H-Y. Hashim is a recipient of an award from the Ministry of High Education, Malaysia.
- 1GLOBOCAN 2002: cancer incidence, mortality and prevalence worldwide. IARC CancerBase no.5 version 2.0. Lyon, France: IARC Press, 2004., , , .
- 8Molecular basis of tumor metastasis. In: LemoineNR,WrightNA, eds. The molecular pathology of cancer. Plainview, NY: Cold Spring Harbor Laboratory, 1993; 175–99., .
- 11General mechanisms of metastasis. Cancer 1997; 80(Suppl): 529–37., , .
- 29Molecular targets for green tea in prostate cancer prevention. J Nutr 2003; 133(7S; Health Module): S2417–S2424., , .
- 32Culture of animal cells. A manual basic technique, 4th edn. New York: Wiley-Liss, 2000..
- 41The Mediterranean diet: an attempt to define its present and past composition. Eur J Clin Nutr 1989; 43 (Suppl 2): 13–29., .