Nox1 downstream of 12-lipoxygenase controls cell proliferation but not cell spreading of colon cancer cells


  • Daniela D. de Carvalho,

    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    Search for more papers by this author
  • Amine Sadok,

    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    Search for more papers by this author
  • Véronique Bourgarel-Rey,

    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    Search for more papers by this author
  • Florence Gattacceca,

    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    Search for more papers by this author
  • Claude Penel,

    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    Search for more papers by this author
  • Maxime Lehmann,

    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    Search for more papers by this author
  • Hervé Kovacic

    Corresponding author
    1. CNRS FRE 2737, Cytosquelette et intégration des signaux du microenvironnement tumoral (CISMET), Aix-Marseille Université, France
    • Faculté de Pharmacie, FRE CNRS 2737, Laboratoire de Biophysique, 27 Bd. Jean Moulin 13385, Marseille, France
    Search for more papers by this author
    • Fax: +33-4-91-83-55-06


The catalytic subunit of the NADPH oxidase complex, Nox1 (homologue of gp91phox/Nox2), expressed mainly in intestinal epithelial and vascular smooth muscle cells, functions in innate immune defense and cell proliferation. The molecular mechanisms underlying these functions, however, are not completely understood. We measured Nox1-dependent Omath image production during cell spreading on Collagen IV (Coll IV) in colon carcinoma cell lines. Knocking down Nox1 by shRNA, we showed that Nox1-dependent Omath image production is activated during cell spreading after 4 hr of adhesion on Collagen IV. Nox1 activation during cell spreading relies on Rac1 activation and arachidonic metabolism. Our results showed that manoalide (a secreted phospholipase A2 inhibitor) and cinnamyl-3,4-dihydroxy-α-cyanocinnamate (a 12-lipoxygenase inhibitor) inhibit Omath image production, cell spreading and cell proliferation in these colonic epithelial cells. 12-Lipoxygenase inhibition of ROS production and cell spreading can be reversed by adding 12-HETE, a 12-lipoxygenase product, supporting the specific effect observed with cinnamyl-3,4-dihydroxy-α-cyanocinnamate. In contrast, Nox1 shRNA and DPI (NADPH oxidase inhibitor) weakly affect cell spreading while inhibiting Omath image production and cell proliferation. These results suggest that the 12-lipoxygenase pathway is upstream of Nox1 activation and controls cell spreading and proliferation, while Nox1 specifically affects cell proliferation. © 2007 Wiley-Liss, Inc.

The phagocytic NADPH oxidase is composed of 5 subunits, including p22phox and gp91phox that constitute the membrane-bound flavocytochrome b558, and p40phox, p47phox and p67phox that translocate from the cytoplasm to the membrane upon cell activation.1 In addition, the small Rac GTPase have emerged as important functional components that are indispensable for NADPH oxidase activity. Since the initial identification of Nox1 (gp91phox homolog) in human Caco-2 colon carcinoma cells, the biological significance of superoxide production by NAPDH oxidase in nonphagocytic cells has garnered much interest.2 In contrast to gp91phox/Nox2, which catalyzes a massive and explosive production of superoxide involved in host defense, Nox1 controls a lower superoxide production that acts as mediator of cell signaling.2 Nox1 is expressed mainly in normal and cancerous intestinal epithelial cells and vascular smooth muscle cells. Although being the subject of intense investigation, the role and the endogenous regulation of Nox1 in colon epithelial cells is still unclear.3, 4 In colon carcinoma cells, homologs of p47phox and p67phox, respectively called Noxo1 (for Nox organizer 1) and Noxa1 (for Nox activator 1), have recently been identified.5, 6 In contrast to p47phox, Noxo1 appears to be constitutively associated with the NADPH oxidase complex.

Activation of NADPH oxidase is dependent upon lipid mediators, such as phosphatidic and arachidonic acids and phosphatidylinositol. Arachidonic acid (AA) is a well established activator of gp91phox/Nox2-dependent NADPH oxidase.7, 8 AA produced in cells by DAG-lipase and phospholipase A2 (PLA2) is further metabolized by lipoxygenase (Lox), cyclooxygenase (Cox) and cytochrome p450.9 In addition to the direct activating effect of AA, eicosanoids derivatives, leukotrienes or 12(S)-HETE or 15(S)-HETE, produced by 5-, 12-, and 15-Lox, also activate NADPH oxidase to stimulate superoxide production.8, 10, 11 To date, the involvement of AA metabolism in the regulation of Nox1 has not been addressed. In colonic epithelial cells, Lox- and Cox-deregulated activity are linked to inflammatory bowel diseases and tumor progression.12 Besides reactive oxygen species (ROS) activation, studies in different colonic tumor cell lines in vitro have indicated that increased production of leukotrienes through Lox activation is involved in the regulation of proliferation, adhesion and metastasis. For example, leukotriene D4 enhanced the binding of Caco-2 cells to Collagen I by increasing α2β1 integrin expression.13 AA metabolization by Lox, but not Cox, was required for subsequent PKC14, 15 and Rho-GTPase activation to properly direct remodeling of the actin cytoskeleton during cell spreading on collagen.16

We recently reported a DPI-inhibitable ROS production occurring during cell adhesion and spreading of Caco-2 cells.17 The present study characterizes this transient ROS production in 2 colon cancer cells, Caco-2 and HT29-D4. Using shRNA and pharmacological tools, we showed that this effect is Nox1-dependent and regulated by at least 2 convergent signals transduced by Rac1 and 12-Lox. Interestingly, even if Rac1 is necessary for Nox1 activation, 12-Lox represents the limiting factor for Nox1 function. Furthermore, our data indicate that Nox1-induced ROS production during cell spreading has a mitogenic effect on colonic carcinoma cells. Together our results present a mechanism by which 12-Lox activation controls the extent of cell spreading and Nox1 activation, while Nox1-dependent superoxide production has no effect on cell spreading but regulates cell proliferation.


ROS, reactive oxygen species; Coll, collagen; PLA2, phospholipase A2; AA, arachidonic acid; Cox, cyclooxygenase; Lox, lipoxygenase; NADPH, nicotinamide adenine dinucleotide phosphate; DPI, diphenylene iodonium; CDC, cinnamyl-3,4-dihydroxy-α-cyanocinnamate; NAC, N-acetylcysteine; rlu, relative light unit; DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; HBSS, Hanks balanced salts solution.

Material and methods

Chemicals and reagents

Unless otherwise stated, all chemicals were obtained from Sigma-Aldrich (France). Fetal bovine serum (FBS), trypsin-EDTA, Dulbecco's modified Eagle's medium (DMEM), L-glutamine, and sodium pyruvate were obtained from Gibco-BRL (Invitrogen, Scotland-UK). Monoclonal anti-myc (clone E10) and anti-Rac1 antibodies were obtained from Upstate Biotechnology (Euromedex, France).

Cell culture and adenoviral infections

HT29-D4 cells, originating from the HT29 colon adenocarcinoma cell line,18 were cultured in DMEM with 25 mM of D-glucose supplemented with 10% FBS, sodium pyruvate (1% v/v) and L-glutamine (2 mM). The Caco-2 human colon adenocarcinoma cell lines (from ATCC) were cultured as described for HT29-D4, except for supplementing with 15% FBS and the addition of nonessential amino acids (1% v/v). Adenoviral infections were performed by diluting the viral vector in serum-deprived medium and infecting cells at 20 multiplicity of infection overnight, after which the virus-containing medium was removed and replaced with fresh medium. The myc-tagged Rac1V12 (constitutively active mutant), myc-tagged Rac1N17 (dominant-negative mutant) and the Ad-null control adenovirus dl312 (which lacks a cDNA insert) were kindly provided by Prof. Goldschmidt-Clermont and described elsewhere.19

shRNA transfection

After testing the efficiency of a set of 3 different hairpin cassettes with the H1 promoter directing expression of 3 different sequences targeting Nox1, we subcloned the most efficient construct in the pRNATH1.1/Neo expression vector (Genscript Corporation, NJ). The Nox1 hairpin shRNA sequence used here is as follows: H1 promoter-CATATAGGCCACCAGCTTGTTGATAT CCGCAAGCTGGTGGCCTATATG-termination signal (where italic letters represent antisense and sense sequences, and letters in bold indicate the loop). The empty plasmid was used as transfection control. Cells were treated with trypsin-EDTA, harvested in single cell suspension and transfected by amaxa nucleofector according to the manufacturer's optimized protocol. Efficiency and specificity of the transfection were evaluated 48 hr after transfection by immunoblotting and real-time PCR.

Measurement of ROS

Ninety-six-well plates were precoated with Type IV collagen (10 μg/ml) for 1 hr at 37°C. The nonspecific binding sites were blocked by incubation with bovine serum albumin (1%) for 30 min at 37°C, and each well was rinsed 3 times with phosphate saline buffer. Cells previously serum-depleted for 24 hr prior to experiments were detached with 0.25% trypsin, adjusted to a concentration of 5 × 105 cells/ml and seeded on coated plates. ROS generation was measured either by lucigenin chemiluminescence or H2-DCFDA fluorescence, detecting superoxide ions and H2O2, respectively. For the lucigenin assay, a final lucigenin concentration of 10 μM was used to suppress potential artifacts as previously described.19 Briefly, after preincubation of cells in regular culture medium without FBS either in suspension or adhered to 96-well plates (25 × 103 cells/well), media was replaced by measurement buffer containing lucigenin and NADPH. All measurements were performed at 37°C and the signal was measured each second over the course of 1 hr; the area under the curve was integrated to express the ROS production during the time of measurements. For H2-DCFDA ROS measurements, after adhesion in regular culture medium without FBS, medium was replaced with measurement buffer containing 10 μM of H2-DCFDA for 30 min. Cells were rinsed with measurement buffer without H2-DCFDA and fluorescence was measured at 490 nm excitation and 538 nm emission. Lucigenin and H2-DCFDA ROS measurements were both performed using a Fluoroskan Ascent FL plate reader (Labsystems, France). Results are expressed as total ROS measurements or as Nox1-dependent ROS production. Nox1-dependent ROS production represents the difference of ROS production measured in transfected control cells versus Nox1 shRNA transfected cells.

RNA isolation and quantitative RT-PCR analysis

Total cellular RNA was prepared with the High Pure RNA isolation kit according to the manufacturer's instructions (Roche Diagnostics, Mannheim, Germany). One to three micrograms of total RNA was used for reverse transcription with random primers and RevertAid M-MuLV reverse transcriptase (Fermentas, Euromedex, Mundolsheim, France). Quantitative PCR was performed on the Lightcycler® 480 System (Roche Diagnostics) using QuantiTect Primer Assays and QuantiTect SYBR Green PCR kit (Qiagen, Courtaboeuf, France) in a total volume of 20 μL, following the manufacturer's instructions. β-2-Microglobulin levels were used to normalize the results. References of the QuantiTect Primer Assays were as follows: Nox1, QT00025585; Nox2, QT00025533; Nox4, QT00057498; Nox5, QT00021924; NoxA1, QT00074438; NoxO1, QT00215789; p47, QT01004815; p67, QT00089341; β2m, QT00088935. Quantification and melting curve analysis was performed with the Lightcycler® software.

Preparation of whole cell lysates

Cells were lysed in 400 μL of MLB buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 10 mM MgCl2, 1 mM EDTA) with 60 mM N-octyl glucoside, supplemented with a mixture of protease inhibitors (1 mM PMSF, 500 units/ml aprotinin, 1 μg/ml leupeptin, 1μM pepstatin, 1 mM iodoacetamide and 1 mM ortho-phenantholine). The cells were harvested on ice, sonicated and debris was pelleted in a microcentrifuge. Protein concentration was determined using the bicinchoninic acid assay (BCA, Perbio Science, Brebieres, France) following the manufacturer's instructions.

Western blot analysis

Cell extracts (30–50 μg of protein) were separated on a 12% SDS-PAGE, transferred to Hybond ECL nitrocellulose membranes (Amersham Pharmacia Biotech, Piscataway, NJ) and processed with the SuperSignal chemiluminescent kit from Pierce on radiographic film. Nox1 was detected using goat anti-Nox1 antibody (SC-5821, Santa Cruz Biotechnologies, Heidelberg, Germany) and a horseradish peroxidase-conjugated anti-goat secondary antibody (HyClone Laboratories, Logan, UT). Anti-c-Myc antibody (clone 9E10) was used to confirm the expression of the adenovirally expressed Rac mutant.

Detection of cellular spreading

After the desired time of seeding on Coll IV, cells were washed in PBS 2 times at room temperature, fixed for 15 min with 4% paraformaldehyde, rinsed in PBS and permeabilized with saponin (0.1%). Cells were stained by successive baths in hematoxylin, aqueous ammonia and eosin for 5 min each, followed by washing in water at each step. Cells were mounted (Aquatex, Merk Darmstadt, Germany) and coverslips were sealed with nail polish. Images were obtained with an inverted Leica DMIRBE microscope (Leica, Bensheim, Germany) controlled by the Metamorph software (Molecular Devices Corporation, Downingtown, USA). Images were captured at original magnification, G10. Lamp voltage, camera integration time and image resolution were kept constant in the course of acquisitions. Gray level pictures were used to characterize each cell by morphometric parameters, such as area (A), perimeter (P) and shape factor (S = 4πA/P2). Cells were classified in 2 groups: no spreading cells with 0.85 < S ≤ 1, and spreading cells with 0 < S ≤ 0.85. Limits for S were determined “a priori” by identification of a subset of cells in both groups. Results were presented as the number of spread cells. In addition, cell surface variation in each condition was calculated using the normalized cell surface area. The area obtained after 1 hr of seeding was normalized as the surface corresponding to no spreading (0% variation); the area obtained after 4 hr of seeding was normalized as the surface corresponding to the optimal spreading (100% variation).

BrdU incorporation

For proliferation assays, cells were harvested, adjusted to a density of 5,000 cells/well and plated in regular medium in 96-well microtiter. After 24 hr, cells were treated with DPI (10 μM), and NAC (100 μM) or CDC (1 μM), and incubated for 48 hr at 37°C. DNA synthesis was quantified by the measurement of BrdU incorporation using an ELISA cell proliferation kit obtained from Roche Diagnostics (Manheim, Germany). After 48 hr, cells were fixed and treated as recommended by the manufacturer. A similar experiment was performed on Nox1 shRNA and control shRNA-transfected cells.

Statistical analysis

Results are expressed as means ± SD from at least 3 independent experiments. Statistical analysis was done using unpaired Student's test. The value of p < 0.05 was considered statistically significant.


ROS production during cell spreading is mediated by Nox1

We measured ROS production in 2 colonic epithelial cell lines during the process of cell adhesion and spreading on Collagen IV (Coll IV). Cells had completely adhered 2 hr after seeding, and optimal spreading was observed after 4 hr of adhesion. ROS production was detected in suspended cells and during cell spreading, and very weak ROS production was detected 1 and 3 hr after seeding (Fig. 1a). We next used a variety of inhibitors to identify the source of Omath image production occurring during cell spreading, and showed that DPI (10 μM) inhibited 90 and 75% of the Omath image production in HT29-D4 and Caco-2 cells, respectively. In contrast, rotenone (2 μM, mitochondrial respiratory chain inhibitor), allopurinol (1 mM, xanthine oxidase inhibitor) or L-name (100 μM, nitric oxide synthase inhibitor) did not affect Omath image production (Fig. 1b). These results suggest that the inhibitory effect of DPI, the flavoprotein inhibitor classically used to inhibit NADPH oxidase, does not affect our experimental conditions, mitochondrial source of superoxide, xanthine oxidase, or nitric oxide synthase.

Figure 1.

NADPH oxidase mediates Omath image production during cell spreading on Coll IV. (a) Serum-deprived Caco-2 cells (square) and HT29-D4 cells (circle) were plated on Collagen IV with (black) or without (white) DPI 10 μM. Omath image production was evaluated by lucigenin luminescence at different time points. (b) Three and half hours after seeding, cells were treated with different inhibitors for 0.5 hr and measurements were taken. Results represent mean + SD of triplicate samples; * indicates statistically significant results compared to control (p < 0.05).

To confirm the preliminary results obtained with DPI, we developed a system to down-regulate Nox1 expression in HT29-D4 cells using an shRNA approach. Transfection with Nox1 shRNA decreased Nox1 mRNA expression by 80% compared to the control shRNA as shown by real-time RT-PCR (Fig. 2a), and this corresponded to a strong down-regulation of Nox1 protein expression (Fig. 2b). The specificity of the shRNA was monitored by evaluating the protein expression of 2 unrelated genes, c-myc and β-actin, whose expression was unchanged in response to Nox1 shRNA expression (Fig. 2b). In addition, evaluation of the expression of the Nox homologs by quantitative RT-PCR shows that HT-29-D4 cells express only Nox1 and, to a lower extent, Nox2 (0.03% of Nox1 expression). Furthermore, Nox2 mRNA level was not affected by Nox1 shRNA (not shown).

Figure 2.

Knock-down of Nox1 by shRNA inhibits ROS production during cell spreading. (a) HT29-D4 cells were transfected with indicated shRNAs. Nox1 mRNA levels were assessed by quantitative RT-PCR, and normalized against β2m mRNA level. (b) Immunoblot for Nox1, actin and c-myc expression in HT29-D4 cells transfected with indicated shRNAs; blot is representative of 3 independent experiments. (c) Omath image production was evaluated by lucigenin luminescence at different times in HT29-D4 cells transfected with the control shRNA (black circle) or the Nox1 shRNA (white circle). (d) H2O2 production was evaluated by H2-DCFDA fluorescence at different time points in HT29-D4 cells transfected with the control shRNA (black circle) or the Nox1 shRNA (white circle). Results represent mean + SD of triplicate samples. * indicates statistically significant results compared to control (p < 0.05).

Kinetic measurement of Omath image production during HT29-D4 cell adhesion and spreading showed that decreasing Nox1 expression by shRNA compromised Omath image production similar to DPI treatment (Figs. 1a and 2c). Such an effect on superoxide production was confirmed by dihydro-ethidium and diogenes measurements (not shown). Measurement of H2O2 production by the DCF-DA assay showed a similar profile of ROS production during adhesion and spreading, which was also inhibited by Nox1 shRNA, although some residual H2O2 was detected (Fig. 2d). Thus, ROS detected during cell spreading in HT29-D4 cells are clearly synthesized through activation of Nox1.

Rac1 involvement in the regulation of Nox1-dependent ROS production

Rac1 is a well known regulator of the actin cytoskeleton, involved in the control of cell adhesion, spreading and migration,3 and also activates NADPH oxidase. Therefore, we next characterized the role of Rac1 regulation in Nox1-mediated Omath image generation during HT29-D4 cell spreading by pull-down assays for activated Rac1, adenoviral expressions of Rac1 mutants, and pharmacological inhibition of Rac activation by NSC23766.20

Using the Pak binding domain of Rac1 to enrich and quantify the active form of Rac1 (Rac1-GTP) in cells lysates, we found that Rac1-GTP is undetectable in suspended cells. In contrast, Rac1-GTP increased with the time after adhesion to a maximal value 4 hr after seeding, maintained up to 8 hr after seeding (Fig. 3a). Treatment of cells with the Rac1 inhibitor NSC23766 (10 μM) or the RacN17 mutant significantly inhibited Nox1-dependent Omath image production during cell spreading after 4 hr of seeding, although it was ineffective in suspended cells (Fig. 3b). Although Rac1 is active in adherent cells after 2 hr of seeding (Fig. 3a), there was no Nox1-dependent ROS production (Fig. 3b, adhesion). As NADPH oxidase needs different regulatory subunits, some of them might not be present at the time of adhesion. This might explain that even in the presence of active Rac1 there is no Nox1 activation. To confirm the link between Rac1 activation and Nox1, we transduced HT29-D4 cells with RacV12, RacN17 or Ad-null adenovirus. Evaluation of Nox1-dependent Omath image production in transduced cells showed that RacV12 expression induced a doubling of Nox1-dependent ROS production in suspended cells, while it had no effect on cells spread after 4 hr of seeding on Coll IV (Fig. 3c). The expression of the myc-tagged Rac1V12 or RacN17 was confirmed by Western Blot analysis, and Nox1 mRNA level was assessed by quantitative RT-PCR in transduced cells. Results presented in Figure 3d showed that RacV12 induces a significant increase in Nox1 mRNA expression compared to RacN17, Ad-null and control cells. Collectively, these results showed that Rac1 induced Nox1-dependent Omath image production in suspension in the absence of endogenous activated Rac1. However, during cell spreading, although RacV12 induced a significant increase in Nox1 mRNA, it did not affect Nox1-dependent Omath image production. These results suggest that Rac1, although being a necessary component for Nox1-dependent NADPH oxidase activation, is not sufficient for full Omath image production and another factor is required, which might be limiting. mRNA levels for cytosolic component known to regulate Nox1 (i.e. Noxo1, Noxa1, p47phox, p67phox) as well as the level of p22phox were not affected during the time frame of cell spreading (Table I). Together this suggests that Nox1 activation during cell activation occurs via post-translational regulation.

Figure 3.

Rac1 controls Nox1-dependent Omath image production during cell spreading but not in suspended cells. (a) HT29-D4 were plated on Coll IV, Rac1-GTP was pulled down using the PAK binding domain at various time points and Rac-GTP level assessed by immunoblot. (b) Nox1-dependent Omath image production was measured, using Nox1 siRNA as described in Material and Methods section, in suspended cells, cells adhered for 2 hr and cells adhered for 4 hr with or without the Rac1 inhibitor (NSC23766), added 1 hr before the measurements. (c) Rac1V12- and Ad-null-transduced cells in suspension or after 4 hr of seeding on Coll IV were evaluated for Nox1-dependent Omath image production. Results represent mean + SD of triplicate experiments. (d) Nox1 mRNA level assessed by quantitative RT-PCR in HT29-D4 cells transduced with Rac1V12 or Ad-null adenovirus. Levels were normalized to β2m mRNA. Inset shows the immunoblot for Rac1 expression (myc) in transduced cells. * indicates statistically significant results compared to control (p < 0.05).

Table I. NADPH Oxidase Complex Subunits mRNA Variation during Adhesion and Spreading
  1. mRNA levels of NADPH oxidase subunits were evaluated by real-time RT-PCR as described in Material and Methods section. Prior to RNA extraction, cells were seeded on Coll IV for the time indicated (0 h = suspended cells). Results are expressed relative to the control in suspension. Results represents the mean + SEM of 3 independent experiments.

Control siRNA 
 0 hr1.0 ± 0.31.0 ± 0.41.0 ± 0.31.0 ± 0.31.0 ± 0.31.0 ± 0.31.0 ± 0.3
 2 hr1.1 ± 0.30.8 ± 0.30.9 ± 0.30.9 ± 0.31.2 ± 0.30.9 ± 0.31.1 ± 0.3
 4 hr0.9 ± 0.30.8 ± 0.50.9 ± 0.31.1 ± 0.31.1 ± 0.31.0 ± 0.31.0 ± 0.3
Nox1 siRNA
 0 hr0.2 ± 0.20.9 ± 0.41.1 ± 0.31.0 ± 0.31.0 ± 0.31.0 ± 0.31.0 ± 0.3
 2 hr0.1 ± 0.31.0 ± 0.51.0 ± 0.31.1 ± 0.31.1 ± 0.31.0 ± 0.30.9 ± 0.3
 4 hr0.1 ± 0.30.8 ± 0.31.0 ± 0.31.0 ± 0.31.1 ± 0.31.0 ± 0.31.0 ± 0.3

AA pathway involvement in Nox1 activation during cellular spreading on Collagen IV

Besides a direct activation of Nox1, an increase in Nox1 expression (this study), and in Noxo1 expression,21 Rac1 is also known to activate cytosolic phospholipase A2 (cPLA2) and subsequent AA synthesis. AA metabolism is a well established activator of Nox2.7 In addition, a recent report showed that lipoxygenase, a known catabolic pathway for AA, is associated with NADPH oxidase activation.22 To investigate the upstream regulator of Nox1 activation during cellular spreading on collagen, we used selective inhibitors of AA metabolism. As many inhibitors for AA metabolism are known antioxidants, we selected inhibitors and concentrations reported not to present any antioxidant properties,23 namely, mepacrine (a general PLA2 inhibitor), AACOF3 (a cytosolic-PLA2 inhibitor), manoalide (a secreted PLA2 inhibitor) and RHC-80267 (a DAG lipase inhibitor). These were used to assess pathways leading to AA synthesis, while aminobenzotriazole (a cytochrome p450 inhibitor), indomethacine (a cyclooxygenase inhibitor), MK-886 and diethylcarbamazine DEC (5-Lox inhibitors), cinnamyl-3,4-dihydroxy-α-cyanocinnamate CDC (a 12-Lox inhibitor) were chosen for pathways leading to AA metabolization. Results in Figure 4a showed that, besides DPI (NADPH oxidase inhibitor), mepacrine, manoalide, CDC and RHC-80267 significantly inhibited total Omath image production on HT29-D4 cells seeded for 4 hr on Coll IV by 33, 67, 78, and 82%, respectively, while other inhibitors were ineffective. Inhibition of enzymes producing endogenous AA, like PLA2 and DAG lipase, blocked Omath image production, suggesting that endogenous AA metabolism stimulated Omath image production during cell spreading on Coll IV. That manolide exhibited effective inhibition of Omath image while AACOF3 had little effect suggested that the activation of Omath image production is dependent on secreted and not cytosolic PLA2. A similar pattern of inhibition was observed in Caco-2 cells (not shown).

Figure 4.

Arachidonic acid pathway involvement in Omath image production during cellular spreading on Collagen IV. (a) HT29-D4 cells were serum-starved for 24 hr and replated on Collagen IV. Three and half hours after seeding, cells were treated for 0.5 hr with each inhibitor [DPI (10 μM), RHC-80267 (30 μM), mepacrine (100 μM), AACOF3 (10 μM), manoalide (10 μM), aminobenzotriazole (1 mM), MK-886 (10 μM), diethylcarbamazine (DEC 50 μM), CDC (1 μM) and indomethacin (25 μM)], and ROS production was evaluated by lucigenin chemiluminescence. (b) Omath image production was evaluated in control shRNA and Nox1 shRNA transfected cells with or without CDC (1 μM) and in presence or absence of 12-HETE (0.1 μM). Values represent the mean + SD of triplicate samples. * indicates statistically significant results compared to control (p < 0.05).

CDC inhibited Omath image to the same extent than DPI, suggesting a role for 12-Lox and downstream AA synthesis in Nox1-mediated production of Omath image (Fig. 4a). To confirm this, we examined the effect of exogenous 12-HETE (0.1μM) on Omath image production. While 12-Lox inhibition by CDC inhibited Omath image production, 12-HETE stimulated Omath image production even in presence of CDC (Fig. 4b). In addition, 12-HETE stimulation of Omath image production is blunted by Nox1 shRNA. These results lead to the conclusion that 12-Lox controls Omath image production through Nox1.

Secreted-PLA2/12-Lox/Nox1 pathway controls of cell spreading of colonic cells

As the Lox pathway has been shown to control epithelial cell spreading, we next sought to determine the contribution of the secreted-PLA2/12-lox/Nox1 pathway in the control of cell spreading on Coll IV. Involvement of Lox in the control of Omath image production during spreading might result from 12-Lox downstream product synthesis, which has been reported to activate NADPH oxidase in different cell types.8, 23 We next measured the effect of different inhibitors affecting PLA2, Lox and NADPH oxidase on cell spreading on Coll IV. Results in Figures 5a and 5b showed that manoalide and CDC completely inhibited HT29-D4 cell spreading on Coll IV, while DPI had a slight but significant effect. Together this implies that Nox1-dependent Omath image production has only a slight function in cell spreading, while, besides regulation of Nox1, 12-Lox functions in other signaling pathways to control cell spreading.

Figure 5.

12-Lox inhibition induces a complete inhibition of HT29-D4 cell spreading on Collagen IV. Serum-deprived (24 hr) HT29-D4 cells were plated on Coll IV for 1 or 4 hr in the presence or absence of DPI (10 μM), CDC (1 μM) or manoalide (10 μM). Cells were fixed, stained and mounted for image acquisition. Five different and independent microscopic fields for each condition were analyzed using the Metamorph software. Significant differences (p < 0.05) were determined using Student t test and identified by *. (a) The percentage of spread cells was determined using morphometric parameters as reported in Material and Methods section. (b) Cellular spreading was evaluated by morphometric parameters as reported in Material and Methods section, and normalized using mean cell surface of control measured 1 hr after seeding as reference for no spreading (0%) and the mean cell surface measured of control after 4 hr as the reference for full spreading (100%). (c) HT29-D4 spreading in the presence or absence of DPI or CDC at original magnification.

Secreted-PLA2/12-Lox/Nox1 pathway controls of cell proliferation of colonic cells

AA metabolism and Lox-derived products are involved in the control of cell proliferation; therefore, we next checked if Nox1 downstream of 12-Lox might participate in the control of cell proliferation. Inhibition of Omath image production by DPI inhibits DNA synthesis of HT29-D4 cells by 77 ± 3% as determined by BrDu incorporation (Fig. 6). Nox1 shRNA, CDC and NAC inhibit DNA synthesis to similar levels as DPI. These results suggest that while Nox1 activation has a weak effect on cell spreading, it has a major function in the control of cell proliferation downstream of 12-Lox. In addition, Nox1 and 12-Lox inhibition also control residual cell proliferation in serum deprived conditions (not shown).

Figure 6.

Nox1 inhibition blocks BrdU incorporation in HT29-D4 cells. Cells were plated in FBS containing medium in 6-well plates 24 hr prior to treatment with DPI (10 μM), NAC (10 mM) or CDC (1 μM). Forty-eight hours after treatment, DNA synthesis was measured with BrdU. Values are mean + SD of triplicate samples. * indicates statistically significant results compared to control (p < 0.05).


The initial step of cellular adhesion and spreading on the extracellular matrix involves the engagement of integrins, which have been reported as redox-sensitive components that can reversibly modulate ROS production. We previously showed that α2β1 integrin engagement on Collagen I or IV leads to major activation of superoxide production compared to vitronectine or laminin in the Caco-2 cell line.17 Initial investigations on the involvement of ROS production during integrin engagement suggested that NADPH oxidase activation occurs directly after integrin ligation during adhesion.17, 22, 24, 25 Recent reports, however, showed that NADPH oxidase activation is not directly linked to integrin engagement, but secondarily to cytoskeletal reorganization, which occurs during cell spreading.26

Our current study show that suspended cells produce Nox1-dependent superoxide and that the initial step of adhesion leads to a decreased superoxide production. A second phase of superoxide production is observed later during cell spreading (4 hr). This behavior is supported by other previous data showing that integrin engagement inhibits NADPH oxidase activation and delays Rac activation during the first step of cell adhesion.27, 28 Rac is a well-known regulator of actin cytoskeleton controlling adhesion and spreading, but its initially identified physiological function was NADPH oxidase activation. While the direct regulation of Nox2 by Rac2 through p67phox binding was evidenced nearly 30 years ago, the interaction between Rac1 with Nox1 is only a recent finding. Initial attempts to stimulate Nox1 by Rac1-GTPγS in broken cell preparation has been unsuccessful.2 Other experiments showed that Rac1 binds to the COOH-terminal region of Nox1 in a growth factor-dependent manner.29 Furthermore, recent finding confirm a direct interaction of Rac1 with Nox1, but this association seems highly dependent on the cell type and the Rac mutant used.30–32 Our data using the Rac inhibitor NSC23766 and Rac1N17 argue against an involvement of Rac1 in Nox1-dependent superoxide in suspended cells. Rac1V12 increases Nox1-dependent superoxide production, but this effect seems to be mainly mediated through the increase in Nox1 mRNA. In contrast, Nox1-dependent superoxide production in adherent cells during cell spreading is clearly dependent on Rac1 activation. These results support the idea that depending on the physiological condition Rac1 might not be necessary for Nox1 superoxide-dependent production. Recently, in a reconstituted system, Miyano and Sumimoto33 reported that although Rac1 was absolutely required for the function of a Nox1-based oxidase containing p47phox and either p67phox or Noxa1, it was not required with Noxo1 as the organizer. Overall, these findings suggest that in our experimental settings, different regulating components of NADPH oxidase are recruited whether cell are adherent or in suspension.

Concerning the superoxide production during cell spreading, the overall level of Nox1 mRNA is increased by Rac1V12, though no further increase in superoxide production was observed. These results suggest that the endogenous activation of Rac1 might be sufficient for Nox1 activation or that another NADPH oxidase component is a limiting factor. In support of the latter, the level of Rac1-GTP is still elevated after 6 hr of adhesion, though no more superoxide production is observed, supporting the involvement of another factor leading to NADPH oxidase activity termination.

A putative limiting factor for Nox1 activation might involve AA metabolism, which, besides being a Nox2-dependent NADPH oxidase activator, is involved in the control of cell adhesion, spreading and proliferation. PLA2 and downstream effectors like lipoxygenase (Lox) and cyclooxygenase (Cox) have been shown to control cell spreading and are of particular importance in colorectal cancer.12 Nox2 may be activated directly by AA and indirectly by the subsequent metabolization of AA by Lox.8 The present study using pharmacological inhibitors approached the involvement of arachidonic metabolism on Nox1-dependent superoxide production. Results show that an endogenous AA source, such as phospholipase A2 and DAG lipase, are involved in Nox1 dependent Omath image production during cell spreading in colon epithelial cancer cells. Rac is a known activator of cytosolic-PLA2 and the 5-Lox pathway in different cell types,34–36 and the 5-Lox pathway leading to the production of AA metabolites had been described in several intestinal epithelial cell lines, including Caco-2 and HT29.37 However, in HT29-D4 cells, that inhibitors of cytosolic-PLA2 (AACOF3) and 5-Lox (MK886) do not inhibit Omath image production during spreading discards a possible involvement of this pathway in Nox1 activation. It should be noted that pharmacological inhibition of Lox in the context of ROS production should be performed with caution, since many Lox inhibitors, such as NDGA and MK-861, possess intrinsic antioxidant properties. In our study, all Lox inhibitors used were reported to be free of antioxidant properties at the concentration used8 and have been tested in vitro using Cyt C assay in our laboratory (not shown). Our data evidenced that secreted PLA2 (inhibited by manoalide) and 12-Lox (inhibited by CDC 1μM) control activation of Nox1 in HT29-D4 cells. The specificity of CDC for 12-Lox was confirmed by 12-HETE, a downstream product of 12-Lox, which reversed the effect of CDC on Nox1-dependent Omath image production (Fig. 4b).

Different reports showed that Lox downstream products are involved in NADPH activation in fibroblast or vascular smooth muscle cells.8, 22 In addition, Block et al. showed that a p22phox-dependent NADPH oxidase is activated by PLA2.38 However, ours is the first report showing an involvement of lipoxygenase in Nox1 activation in colonic epithelial cell. While the functional significance of Nox1-dependent ROS production in colonic cell remains to be fully elucidated, our data clearly address the involvement of the 12-Lox/Nox1 pathway on cell spreading and proliferation. Results in Figure 5 show that manoalide and CDC fully inhibit cell spreading, confirming that secreted-PLA2 and 12-Lox are involved in the control of cell spreading. In contrast, DPI weakly but significantly affects cell spreading on Coll IV while proliferation was equally affected by manoalide, CDC, DPI or Nox1 shRNA. Together these results suggest a pathway by which activation of secreted-PLA2 and 12-Lox control cell spreading in colorectal cancer cells induces ROS production through the activation of Nox1, which ultimately regulates cell proliferation. Consistent with these results, downstream products of Lox are known to activate Rho GTPases and different PKC through the initial activation of heterotrimeric G proteins and are known to control the extent of cell spreading in epithelial cells.39 The precise mechanism of Nox1 activation by 12-Lox and 12-HETE warrants further investigation.

In summary, in this study, we showed that the Nox1-dependent NADPH oxidase is activated during colorectal cancer cell spreading on Collagen IV. This activation is dependent on Rac1-GTPase and 12-Lox pathways upstream of Nox1. Although Rac1 is necessary for Nox1 activation during cell spreading, the extent of Nox1 activation seems limited by the 12-Lox pathways. 12-Lox pathways regulate both cell spreading and Nox1 activation concomitantly, but the cell spreading is regulated in a Nox1-independent way. Thus, downstream of 12-Lox is a crossroad, with one side leading to the control of cell spreading, and the other side to the control of cell proliferation. Further studies will be necessary to delineate which of the numerous compounds downstream of 12-Lox controls cell spreading or cell proliferation. Different subsets of PKC activated by leukotrienes and derivatives are known to control spreading, proliferation as well as Nox1 activation, and represent good candidates as intermediates between 12-Lox and Nox1.


This work was supported by a grant from GEFLUC Marseille, Provence (to Hervé Kovacic). The authors would like to thank CNPq (Conselho Nacional do Desenvolvimento Científico e Tecnológico) Brazil, for supporting the post doctoral fellow Daniela D. de Carvalho.