Reactive Oxygen Generated by NADPH Oxidase 1 (Nox1) Contributes to Cell Invasion by Regulating Matrix Metalloprotease-9 Production and Cell Migration^*

Cell Culture and Materials

Normal rat kidney fibroblast (NRK) cells, Kirstein-Ras-transformed NRK (KNRK) cells, and CaCO-2 cells were purchased from American Type Culture Collection (Manassas, VA). KNRK cell lines, which stably expressed scrambled siRNA (Neg-1) and Nox1 siRNA (N-7), were described in the previous study (3). Rabbit anti-IKKα antibodies were purchased from Cell Signaling Technology (Beverly, MA), and rabbit anti-RhoA antibodies were from Upstate Biotechnology, Inc. (Lake Placid, NY). Rabbit anti-IκBα antibodies and GST-IκBα-(1–100) were gifts from Dr. N. Rice and Dr. H. Nakano, respectively. Diphenyleneiodonium (DPI) was purchased from Calbiochem, and EGF, mouse anti-HA antibodies, N-acetyl-l-cysteine (NAC), and vitamin E were from Sigma. Rabbit anti-Nox1 antibodies were used as described previously (4). pSilencer-human Nox1 siRNAs were constructed as described previously (4).

Plasmids

The NheI/NcoI fragment (nucleotides −670 to +22) of the human MMP-9 promoter (17) was inserted into the reporter plasmid pGL-3 Basic (Promega, Madison, WI) to generate pGL-MMP-9-670. pGL-MMP-9-670kBmt (the NF-κB site AGTGGAATTCCCCA was mutated to AGTTCTCGAGCCCA) was constructed by PCR-based mutagenesis (7). HA-Rac1Q61L cDNA was subcloned into pEFBOS. pHBMn-SOD was provided by Dr. K. Hirose (18), and pcDNA3.0-Nox1, pEFBOS-HA-NOXO1, and pEFBOS-HA-NOXA1 were provided by Dr. H. Sumimoto. Adenoviral construct (Ad-HA-LMW-PTP^C12S) was described previously (19).

Immunoprecipitation and Immunoblotting

Cells were lysed in radioimmune precipitation assay buffer (8) unless specified, and lysates were subjected to immunoblotting or immunoprecipitation as described (8).

Transfection

Transfection was performed by using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Cells (5 × 10⁵∼10⁶) were transfected with 2∼4 μg of the indicated plasmid DNAs.

Luciferase Assay

Cells were transfected with 2 μg of luciferase reporters and 2 μg of various expression vectors with and without drug treatments. Cells were lysed after 48 h, and the promoter activity was assayed as described previously (7).

IKK Assay

The activity of IKK was measured as described previously (20). The cells were treated with chemical inhibitors for 6 h and lysed for 30 min. The lysates were immunoprecipitated with anti-IKKα antibodies, and the immunoprecipitates were incubated with 1 μg of GST-IkBα-(1–100) and 370 kBq of [γ-³²P]ATP(110 TBq/mmol; GE Healthcare) in kinase buffer for 1 h at 30 °C. The reaction products were analyzed by SDS-PAGE followed by autoradiography.

Reverse Transcription-PCR

Total RNAs were extracted from cells by using Trizol (Invitrogen). After reverse transcription, PCR for MMP-9 was performed by using the primers (forward 5′-CGGTCGGTATTGGAAGTTCTCG-3′ and reverse 5′-GCTGAAGCAAAAGAGGAGCCTTA-3′) as described previously (7).

Zymography

The conditioned medium was collected after treatment of cells with drugs and concentrated by an Ultracent-30 (Tosoh). Equal amounts of samples were loaded onto a SDS-PAGE gel containing gelatin (0.5 mg/ml), and the gels were incubated in a solution (50 mm Tris-HCl, pH 8.0, 0.5 mm CaCl₂, 1 μm ZnCl₂, and 1% Triton X-100) for 16 h at 30 °C according to a published method (21) with slight modifications. The areas representing for the gelatinolytic activity were visualized by negative staining with Coomassie Brilliant Blue.

Measurement of the Rho Activity

GST-Rho binding domain proteins derived from rhotekin were prepared as described previously (8). Cells were lysed in lysis buffer (50 mm Tris-HCl, pH 7.2, 100 mm NaCl, 5 mm MgCl₂, 1% Nonidet P-40, 1 mm dithiothreitol, 1% glycerol, 1 mm phenylmethylsulfonyl fluoride), and lysates were incubated with GST-Rho binding domain-coupled glutathione S-transferase resins for 90 min at 4 °C. The active Rho-GTP retained to the resins were analyzed by immunoblotting with anti-Rho antibodies.

5′-Fluorescein Iodoacetamide Labeling

Labeling was performed as described previously (8). Cell lysates were labeled with 10 μm 5′-fluorescein iodoacetamide (Molecular Probes, Eugene, OR) for 90 min at 4 °C, and the reaction products were dialyzed against phosphate-buffered saline to remove free 5′-fluorescein iodoacetamide.

Invasion and Migration Assays

The invasion assay was performed by using Matrigel-coated Boyden chambers (BD Biosciences) according to the company's protocol. Cells (5 × 10⁵) were inoculated into the top of the chamber, and the NRK serum-free-conditioned medium was added to the bottom of the chamber. After 20 h of incubation, the invaded cells were stained with trypan blue and counted. Migration was assayed by using Matrigel-uncoated Boyden chambers (Nalge Nunc, Rochester, NY). Cells (5 × 10⁵) were transfected with indicated vectors, and 24 h later, transfected cells were replated to the chambers in the Dulbecco's modified Eagle's medium plus 0.1% bovine serum albumin medium. In some experiments, cells were infected with Ad-HA-LMW-PTP^C12S after transfection as described previously (8). EGF (100 ng/ml) in the same medium was used as a cue. After 48 h of incubation, migrating cells were counted.

Nox1 Mediates Oncogenic Ras-induced Cell Invasion

To evaluate whether Nox1 is involved in cell invasion, we first examined a mediating role of Nox1 in Ras-induced cell invasion. To this end, NRK cells, KNRK cells, KNRK cells carrying Nox1 siRNAs (N-7), and KNRK cells carrying scrambled siRNAs (Neg-1) were compared for the invasion activity by using the Matrigel invasion assay. As shown in Fig. 1, the number of invading cells was markedly increased in KNRK cells compared with that in NRK cells, and the invasive activity of N-7 cells was reduced compared with that in Neg-1 cells. In addition, Nox1 siRNAs did not influence the low invasion activity of NRK cells. In our previous study, the Nox1 activity was enhanced due to K-RasV12-induced up-regulation of Nox1 in KNRK cells, and silencing of Nox1 siRNAs in KNRK cells suppressed the Nox1 activity (3). Thus, the alteration of cell invasiveness correlates well with that of the Nox1 activity, and these data support the notion that the Nox1 signaling mediates the activated Ras-induced cell invasion activity. Given that activation of both extracellular matrix protease production and cell motility contribute to increased cell invasiveness, we investigated the role of Nox1 in these two biological processes in the subsequent studies.

Nox1 Mediates Oncogenic Ras-induced MMP-9 Production

It is well documented that MMP-9 and MMP-2 are generated in response to oncogenic activation of Ras (21). To gain insight into the regulatory role of Nox1 in oncogenic Ras-induced metalloprotease production, we compared NRK cells with KNRK cells for the proteolytic enzyme secretion. Zymographic analysis showed that latent MMP-9 was predominantly secreted and, to a lesser extent, latent MMP-2 in the conditioned medium was isolated from KNRK cells, whereas little or no MMP-9 and MMP-2 were produced in NRK cells (Fig. 2A). Treatment of KNRK cells with a flavoprotein inhibitor, DPI, or anti-oxidants, NAC and vitamin E, resulted in an inhibition of the MMP-9 and MMP-2 activities. In addition, N-7 cells markedly reduced the level of MMP-9 and MMP-2, whereas Neg-1 cells maintained the activities, indicating the involvement of Nox1 in their expression (Fig. 2A). In contrast, the expression of Type 1 matrix metalloprotease (MT1-MMP) was not affected by Ras activation (data not shown). As MMP-9 was predominantly secreted in KNRK cells, we focused on the regulatory mechanism of MMP-9 expression in the subsequent study. Reverse transcription-PCR analysis was performed to confirm the above results at the mRNA level. The MMP-9 mRNA level was significantly increased in KNRK cells as compared with that in NRK cells (Fig. 2B). Furthermore, DPI and NAC treatment lowered the expression of MMP-9 mRNAs in KNRK cells. The synthesis of MMP-9 mRNAs was consistently suppressed in N-7 cells compared with that in Neg-1 cells (Fig. 2B). Thus, the data indicate that the levels of MMP-9 protein were changed upon Nox1 inhibition, correlating with those of MMP-9 mRNAs and that the regulation of MMP-9 production by Nox1 was through the transcriptional control.

To understand whether the sole activation of the Nox1 activity can turn on the synthesis of MMP-9, NRK cells were co-transfected with Nox1 and its adaptors, NOXO1 and NOXA1, and the conditioned media were subjected to zymography. As shown in Fig. 2A, overexpression of Nox1 and its adaptor proteins stimulated ROS production (see Fig. 4A) and thereby enhanced MMP-9 production (Fig. 2A), indicating that at least the Nox1 activity alone is sufficient for induction of MMP-9. This is in sharp contrast to induction of VEGF that seemingly requires not only Nox1 but also an additional signaling pathway triggered by activated Ras (9).

NFκB Mediates Oncogenic Ras-induced MMP-9 Expression via Nox1

We next wished to dissect the biochemical signaling pathway linking Nox1 to the transcriptional regulatory system for MMP-9. Although the cis-acting element of the MMP-9 promoter encompasses several transcription factor binding motifs (17), we focused on the molecular pathway involved in NFκB-controlled MMP-9 transcription for the following reasons. Both NFκB (22) and Nox1 (3, 7) are critical downstream targets of the Ras signaling pathway in Ras-mediated oncogenesis, and NFκB is activated in colon adenocarcinoma cells overexpressing Nox1 (23). A reporter pGL-MMP-9-670 bearing a proximal 670-bp MMP-9 promoter fragment that harbors a NFκB binding site and a NFκB-site-mutated reporter pGL-MMP-9-670kBmt were constructed (supplemental Fig. 1) and transfected into NRK and KNRK cells. The luciferase activity assay showed that the MMP-9 promoter activity was markedly increased in KNRK cells, whereas it remained at the basal level in NRK cells (Fig. 3A). In contrast, mutagenic disruption of the NFκB binding site reduced the promoter activity by 60%, suggesting that NFκB is responsible, at least in part, for oncogenic Ras-induced transcriptional activation of the MMP-9 gene.

To determine whether Nox1 regulates the NFκB-dependent transcriptional activity of the MMP-9 promoter in response to Ras activation, KNRK cells were treated with DPI after transfection with a pGL-MMP-9-670 reporter and subjected to luciferase activity assay. The promoter activity was abrogated by DPI in an incubation time-dependent manner (Fig. 3B). Furthermore, a significant decrease in the MMP-9 promoter activity was detected in N-7 cells as compared with Neg-1 cells (Fig. 3C). As an additional means to demonstrate the regulatory effect of ROS on the MMP-9 promoter activity, we transfected Mn-SOD, a scavenger of superoxide into KNRK cells, together with the pGL-MMP-9-670 reporter. The promoter activity was significantly reduced upon transfection of Mn-SOD (Fig. 3D). Because our previous study shows that the superoxide generation is increased due to up-regulation of Nox1 in KNRK cells (3), these data collectively indicate that ROS generated by Nox1 are required for NFκB-mediated response of the MMP-9 expression to oncogenic activation of Ras. To understand whether up-regulation of the Nox1 activity alone can initiate transcription of MMP-9, NRK cells were co-transfected with Nox1, NOXO1, and NOXA1 and analyzed for the MMP-9 promoter activity assay. Overexpression of Nox1 and its adaptors enhanced superoxide production (Fig. 4A) and thereby induced the activity of a MMP-9-luciferase reporter (Fig. 4B). This correlated with the zymographic assay data (Fig. 2A), suggesting that the Nox1 activity per se may recapitulate the inducing effect of Ras on MMP-9 expression.

To further establish that Nox1 is a mediator of Ras-induced NFκB signaling, we determined whether Nox1 controls the activity of IKKα kinase that phosphorylates IκBα, a key negative regulator of NFκB signaling. IKKα was immunoprecipitated from both NRK and KNRK cells, and the immunoprecipitates were subjected to in vitro kinase activity assay utilizing GST-IκBα as a substrate. The data indicate that the IKKα kinase activity was enhanced upon Ras transformation (Fig. 5A). Furthermore, treatment of KNRK cells with both DPI and NAC inhibited the IKKα activity (Fig. 5B). Similarly, ablation of Nox1 with Nox1 siRNAs blocked the IKKα activity (Fig. 5C). Because phosphorylation of IκBα by IKKα is thought to cause degradation of IκBα, resulting in NFκB activation (24), the levels of IκBα expression were examined. Immunoblotting analysis showed that the amount of IκBα protein was decreased in KNRK cells (Fig. 5D). In contrast, the normal IκBα level was regained after KNRK cells were exposed to DPI for 6 h (Fig. 5E). Simultaneously, the expression of IκBα was increased in N-7 cells compared with that in Neg-1 cells (Fig. 5F). Taken together, these results suggest that the signal initiated by Nox1-derived ROS induces the IKKα activation, leading to preferential degradation of IκBα.

Nox1 Regulates Cell Migration of Ras-transformed Cells

Oncogenic Ras is also known to potently stimulate cell migration and contribute to tissue invasiveness (25), and as shown in Fig. 1, it is evident that this Ras bioactivity is mediated by Nox1. Our previous study demonstrated that Nox1 mediated oncogenic-Ras-induced disruption of actin stress fiber formation and loss of cell adhesion by inactivating the LMW-PTP-p190RhoGAP-Rho axis (8). To test whether Nox1 regulates cell migration via a similar signaling pathway, the involvement of LMW-PTP in the cell migratory activity was examined. When transfected with the dominant negative LMW-PTP^C12S mutant, the migration activity of N-7 cells was readily restored, whereas that of Neg-1 cells was not changed (Fig. 6). This mutant was no longer accessed by Nox1 due to substitution of redox-sensitive cysteine-12 to serine, and hence, it maintained the p190RhoGAP activity, suppressing Rho (8). Thus, the available data support the idea that Nox1 can exert a critical regulatory role on cell migration through LMW-PTP.

Nox1 Down-regulates Rho Activity in CaCO-2 Cells

To further substantiate the involvement of Nox1 in cell motogenesis, we next addressed this Nox1 bioactivity in another biological system, EGF-induced migration of colon cancer cells. We first examined whether Nox1-generated ROS affect the activity of Rho, a key regulator of cytoskeletal contractility. Because overexpression of NOXO1 and NOXA1 is able to increase superoxide production in Nox1-abundant CaCO-2 cells (26), HA-NOXO1 and HA-NOXA1 were co-transfected into CaCO-2 cells, and cell lysates were subjected to the Rho activity assay. Overexpression of NOXO1 and NOXA1 increased ROS production that can be inhibited by DPI and NAC (supplemental Fig. 5). GST-Rho binding domain pulldown assays showed that the Rho activity was significantly down-regulated upon overexpression of NOXO1 and NOXA1 (Fig. 7A). When cells were treated with either DPI or NAC, the suppressive effect of overexpressed NOXO1 and NOXA1 on Rho was removed (Fig. 7A). Furthermore, we found that ablation of Nox1 by Nox1 siRNA prevented inhibition of the Rho activity by ectopic expression of NOXO1 and NOXA1 (Fig. 7B). Transfection of Nox1 siRNAs was demonstrated to effectively decrease the expression level of Nox1 proteins (supplemental Fig. 6). Taken together, these results indicate the suppression of the Rho activity by Nox1-generated ROS in CaCO-2 cells, similar to that in KNRK cells (8).

Because Rac1 is believed to act as a critical molecular switch for the oxidase activity of Nox1 (27), we next examined whether Rac1 controls the Rho activity through Nox1. Transfection of the dominant active Rac1QL mutant decreased the amount of active Rho-GTP complexes, whereas treatment of Rac1QL-transfected cells with DPI or NAC restored the Rho activity (Fig. 8A). Similarly, silencing of Nox1 by Nox1 siRNAs blocked Rac1QL-induced down-regulation of the Rho activity (Fig. 8B). Moreover, transfection of Rac1QL increased the intracellular ROS level, whereas additional transfection of Nox1 siRNAs prevented Rac1QL-induced stimulation of ROS synthesis (Fig. 8C). This indicates that Rac1 stimulates Nox1-catalyzed ROS production and thereby suppresses the Rho activity. When cells transfected with both Rac1QL and Nox1 siRNAs were exposed to H₂O₂, Rho activity was reduced, implicating that exogenously added H₂O₂ mimics the repressive effect of Nox1-generated ROS on Rho (Fig. 8B). Collectively, these results support the notion that Nox1-derived ROS signals down-regulation of the Rho activity in CaCO-2 cells.

Nox1 Mediates EGF-induced Down-regulation of Rho in CaCO-2 Cells

The EGF receptor is expressed on the basolateral membrane of intestinal epithelial cells and plays a mediating role in the augmented migration of colon cancer cells (16). We, therefore, investigated whether the Nox1-Rho axis is engaged in EGF receptor-mediated cell movement. Although EGF treatment of CaCO-2 cells rapidly suppressed the Rho activity, transfection of Nox1 siRNA antagonized EGF-induced inhibition of Rho (Fig. 9A).

p190RhoGAP is activated after tyrosine phosphorylation by receptor or non-receptor tyrosine kinases, which in turn leads to down-regulation of Rho (29). We, therefore, reasoned that the observed negative regulation of Rho by Nox1 upon EGF stimulation could be mediated by EGF-induced activation of p190RhoGAP. Immunoblotting analysis indicated that EGF treatment increased tyrosine phosphorylation of p190RhoGAP, whereas Nox1 siRNAs markedly blocked the stimulatory effect of EGF on p190RhoGAP phosphorylation (Fig. 9B). From these results, it is conceivable that Nox1 mediates the EGF receptor-stimulated activation of p190RhoGAP and subsequently inhibits the Rho activity. To determine whether Nox1 is necessary for EGF-induced migration of cells, CaCO-2 cells were transfected with either Nox1 siRNAs or scrambled siRNAs and subjected to migration assay. Nox1 siRNAs impaired the ability of EGF receptor to stimulate cell migration (Fig. 10A), indicating the mediating role of Nox1 in EGF-induced motogenesis.

We next investigated whether LMW-PTP acts as a sensor for Nox1-generated ROS that transmits an activation signal to p190RhoGAP. By utilizing the 5′-iodoacetamide fluorescein-labeling approach, in which iodoacetamide derivative competes with intracellular H₂O₂ in attacking a redox-sensitive cysteine-SH residue (8), we analyzed a Nox1-induced oxidation state of LMW-PTP. When NOXO1 and NOXA1 were co-transfected into CaCO-2 cells, the labeling of exogenously expressed HA-LMW-PTP was suppressed as compared with that in control vector-transfected cells, and apocynin, a Nox inhibitor treatment, removed the suppressive effect (Fig. 10B). The data suggest that the cysteine-SH of LMW-PTP was oxidized by Nox1-generated ROS. Additionally, stimulation of migration with EGF was blocked in Nox1 siRNA-transfected cells, whereas forced expression of the LMW-PTP^C12S mutant restored EGF-mediated motogenesis in Nox1 siRNA-transfected cells (Fig. 10C). In contrast, overexpression of LMW-PTP^C12S did not alter the motogenic effect of EGF on scrambled siRNA-transfected cells because endogenous LMW-PTP was inactivated by EGF via Nox1. The data suggest that LMW-PTP mediates EGF receptor-Nox1-dependent motogenesis.