Nox1^y/- mice were generated by Dr. K. H. Krause (Gavazzi et al., 2006) and backcrossed onto a C57Bl/6 background. The Institutional Animal Care and Use Committee of Emory University School of Medicine approved the animal protocol used in this study.

Transluminal mechanical injury of bilateral femoral arteries was induced by introducing a large wire, as previously reported (Sata et al., 2000; Lee et al., 2009). At 21 days, the mice were sacrificed and pressure-perfused at 100 mmHg with 0.9% sodium chloride, followed by pressure fixation with 10% formalin. Arteries were then carefully excised, embedded in paraffin, and processed for histological analysis.

Histology sections from WT and Nox1^y/- mice femoral arteries subjected to wire-induced injury were kindly provided by Dr. Kathy Griendling (Lee et al., 2009). Antigen retrieval was heat-induced in citrate buffer. Immunohistochemistry followed by DAB staining was performed using antibodies against phospho-Y307-PP2A-CA (Santa Cruz) and phospho-T410/403-PKCζ/λ (Cell Signaling) to determine the levels of inactivation of PP2A-CA and activation of aPKC, respectively. Images of the whole femoral artery were captured using a NanoZoomer SQ (Hamamatsu) using the 40X scanning mode. Approximately 3-5 femoral artery sequential slices were evaluated per animal, and a total of 5 animals were used per phenotype. In addition, the relative intensity of DAB staining was assessed using ImageJ.

Mouse embryonic fibroblasts (MEFs) were prepared from wild-type (WT) and Nox1^y/- mouse embryos as described before (Brown et al., 2014). Briefly, E13.5 embryos were isolated with their yolk sacs and dissected to remove and discard the head and internal organs. The yolk sac was removed and retained for genotyping. The dissected embryo was passed through an 18G needle and subsequently plated on gelatin-coated dishes in Dulbecco’s modified Eagle’s medium (DMEM) high glucose (Sigma) supplemented with 15% fetal bovine serum (Sigma). After two passages, MEFs were immortalized by expression of the SV40 large T-antigen (Addgene plasmid 13,970). MEFs were subcultured at a 1:10 ratio for nine passages upon reaching confluence. Immortalized MEFs were grown in DMEM high glucose supplemented with 10% fetal bovine serum (Atlanta Biological), 1% penicillin-streptomycin, and 1% Glutamax (Gibco) and used for experiments for additional 15 passages.

Mouse aortic smooth muscle cells (MASMCs) were isolated from wild-type (WT) and Nox1^y/- mice by enzymatic dissociation (Fernandez et al., 2015). Cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum), 1% penicillin-streptomycin, and 1% Glutamax (Gibco). All cultures were used between passages 6 and 10 for experiments. Cultures at 70%–80% confluence were made quiescent by incubation in serum-free media for 16 h before the experiments.

All the cell lines were grown in a 5% CO2 atmosphere at 37 °C.

Antibodies used in these experiments are described in Table 1. Rabbit polyclonal anti-Par3 antibodies were used for immunoprecipitation (Proteintech) and immunofluorescence (Millipore). Rabbit anti-pY307PP2A-CA (Upstate/Millipore and Santa Cruz), Rabbit anti-PKCζ/ι, anti-Cortactin, Rabbit anti-α/β Tubulin (Cell Signaling), rabbit anti-Tiam1, mouse PKCζ/ι (Santa Cruz), mouse anti-GST (Proteintech), anti-Actin (Sigma), anti-FITC (Invitrogen), PDGF-R (EMD Millipore). Primary antibodies were used in 1/1,000 dilution for immunoblotting and 1:200 for immunofluorescence and immunohistochemistry experiments. All secondary antibodies conjugated to HRP or to Alexa fluorophores were from Jackson ImmunoResearch and were used 1/3,000 for immunoblotting and 1/200 for immunofluorescence and immunohistochemistry experiments. Okadaic Acid (Cayman Chemical Company) was used at 1 μM for 30 min before PDGF-BB (R&D systems) stimulation. EZ-Link Sulfo-NHS-SS-Biotin (Life Technologies) and 5-Iodoacetoamido-fluorescein (Sigma) were used according to manufacture protocols.

FlexiTube siRNA for mouse PP2A-CA sequences #4 and #7 were obtained from (Qiagen). Cells were seeded on coverslips and, after 16 h, were co-transfected with siGlo RNAi (GE Dharmacon) and siCtrl AllStars (Qiagen) or siRNA against PP2A-CA, using Lipofectamine RNAiMAX reagent (Thermo Fischer Scientific) according to the manufacturer recommendations. After 24 h of transfection, cells were serum starved and processed for immunofluorescence.

pAdEasy vector, which contains green fluorescent protein (GFP), was used to prepare viruses without an insert (pAdEasy-Ctrl) or hemagglutinin (HA)-tagged Nox1 (pAdEasy-Nox1-HA) as described before (Hanna et al., 2004). Cells were infected with the adenoviruses for 16 h in complete media. Subsequently, cells were trypsinized and seeded on coverslips for immunofluorescence assays.

Cells were seeded on 15 cm dishes, allowed to attach for 5-6 h, and then serum-starved overnight. The following day, the media was refreshed, and cells were stimulated with PDGF 10 ng/mL for 30 min. We analyzed phosphorylated proteins by affinity chromatography using PhosphoProtein Purification Kit (Qiagen) according to the manufacturer’s indications. Purified phosphoproteins were resolved by electrophoresis and blotted for Par3, Cortactin, and Actin.

In other experiments, to analyze phosphorylation of pY307-PP2A-CA and pThr410/403-PKCζ/ι, cells were lysed in boiling-2x Lamelli buffer, sonicated, and immediately subjected to electrophoresis.

Activation of Rac1 was determined by G-LISA kit (Cytoskeleton). Briefly, cells were seeded on 6-well plates, serum-starved overnight, and stimulated with PDGF-BB 10 ng/mL for the indicated times. Protocol was followed according to the manufacturer’s recommendations. Results were the average of three independent experiments in which each condition was performed in triplicate.

Cells were seeded on acid-washed coverslips coated with Collagen type I (Corning). After cell attachment (5–6 h), cells were serum starved for 16 h. The media was refreshed the following day. Cells were stimulated with PDGF-BB 10 ng/mL for 30 min. Media was removed, cells were immediately fixed in 4% Paraformaldehyde for 10 min, followed by permeabilization with 0.1% Triton X-100 for 10 min. Subsequently, coverslips were blocked for 10 min with 2% IgG-free BSA (Jackson Immuno Research), 0.1% Fish gelatin (Sigma) in UB buffer (50 mM Tris-HCl, pH 7.6; 0.15 N NaCl and 0.1% sodium azide). Primary antibodies were added for 1 h at room temperature and washed three times for 5 min each time with UB-0.1% Tween-20. Cells were blocked again as described above and then incubated with secondary antibodies conjugated to fluorophores, phalloidin, and DAPI for 1 h at room temperature. Coverslips were washed three times for 5 min in UB-0.1% Tween-20 and mounted in Mowiol solution. In some experiments, cells were seeded, and the next day they were co-transfected with siGlo RNAi and siRNA Control or siRNA PP2A-CA. During the Nox1 rescue experiments, cells were infected with the control virus (pAdEasy-Ctrl) or Nox1-HA (pAdEasy-Nox1-HA), and 24 h later, they were trypsinized and seeded on coverslips and processed as described above.

Lamellipodia formation was evaluated by double staining of cortactin and phalloidin using a Zeiss LSM 510 META or an LSM 800 Airyscan Laser Scanning Confocal Microscopes (Plan-Apo 63x NA 1.4 oil or Plan-Apo 20 × 0.8 NA). Pictures and analysis were performed blinded. Results are mean ± SEM of at least three independent experiments in which 10–100 cells were evaluated per condition.

MEFs were seeded on coverslips. After 24 h, cells were serum starved for 2 h and stimulated with 10 ng/mL PDGF-BB for 30 min. Then, cells were processed for a modified protocol of Far-immunofluorescence as previously described (Bustos et al., 2012; Pelletan et al., 2015; Valdivia et al., 2020). Briefly, cells were fixed, permeabilized, and blocked as described for immunofluorescence above. Subsequently, samples were incubated with 25 μg GST-Rac1G15A purified recombinant protein for 60 min and washed. Then, samples were blocked and incubated with anti-Tiam (Santa Cruz) and anti-GST antibodies for 60 min and washed. Coverslips were then blocked and incubated with anti-mouse AlexaFluor 568 and anti-rabbit AlexaFluor 633 (Jackson Immunological). Purified recombinant GST was used as a negative control. Both GST and GST-Rac1G15A were prepared as previously described (Garcia-Mata et al., 2006). Images were captured with LSM 800 Airyscan Laser Scanning Confocal Microscopes (Plan-Apo 63x NA 1.4 oil). Pictures and analysis were performed blinded. Results are mean ± SEM of at least three independent experiments in which 5–7 cells were evaluated in each condition.

Colocalization was quantified using Pearson’s R coefficients between two different channels within specific ROI located at the membrane of lamellipodium and lamellipodia-like protrusions using the Coloc2 macro in ImageJ. Values between 0.7 and 1 were considered true colocalization.

Migration was measured using a Boyden chamber. Cells were serum starved and allowed to migrate towards 10 ng/mL PDGF for 3 h. Migrated cells were stained with DAPI. Four random fields were visualized using Plan-Neo 20 × 0.5 NA in a Zeiss Axioskop2 wide-field microscope and quantified with ImageJ.

WT and Nox1^y/- MEFs were seeded in 2 well silicone inserts (Ibidi) for the wound healing assays and let grow to confluency overnight. Cells were serum starved for 2 h, and the external part of the dish was filled with starvation media containing 10 ng/mL PDGF-BB. Then, the silicone inserts were carefully removed, allowing the cells to migrate to the empty area for 30 min. In some experiments, cells were fixed and processed for immunofluorescence, as described above. The staining was performed using γ-Tubulin (Sigma) as a marker of MTOC and GM130 (ECM Bioscience) as a marker of Golgi. Cells were considered polarized when Golgi or MTOC were positioned within an angle of 120° in front of the nucleus facing the wound area and as not polarized when the signal was behind this angle or on top of the nucleus.

Data are expressed as the mean ± standard error of mean (mean ± SEM from at least three independent experiments (n = 3). In every independent experiment, for each experimental conditions we evaluated either 5–10 random field of view and counted 10–700 cells depending on the magnification used to capture the images. Data were compared using t-student test, 1-way or 2-way ANOVA analysis, accordingly. Significant differences were established at p < 0.05.

Similar to what has been reported in a variety of cell types (Schroder et al., 2007; Sadok et al., 2008; Lee et al., 2009; Shinohara et al., 2010; Khoshnevisan et al., 2020), Nox1 deficient (Nox1^y/-) MEFs have impaired PDGF-induced migration (Supplementary Figure S1) despite having no differences in PDGFR levels at the plasma membrane when compared to control cells (Supplementary Figure S2).

Previous work by our laboratory and other researchers have established that Nox1 activity is required for protrusion extension at the leading edge (San Martin et al., 2008; Maheswaranathan et al., 2011) and directional persistence (Sadok et al., 2009) during migration. Because the establishment of agonist-induced polarity precedes cell protrusion formation and is required for persistence during migration, we posited that Nox1-derived ROS are required for establishing agonist-dependent cell polarity. To examine this hypothesis, we assessed one of the first manifestations of cell polarization: the redistribution of the Golgi apparatus and the microtubule organization centers (MTOCs). Using a wound-healing assay, we found that in wild-type MEFs, PDGF induces the reorientation of the Golgi apparatus and the centrosome towards the wounded area at the front of the nucleus (Figure 1A). Interestingly, while correctly positioning their Golgi apparatus toward the wounded area, Nox1-deficient cells failed to redistribute the MTOC properly (Figure 1B).

It is believed that the orientation of the MTOC dictates the polarity of the microtubule (MT) network, which is critical for establishing the leading edge (Schutze et al., 1991; Palazzo et al., 2001). We observed that Nox1 deficient cells fail to relocate their MTOC properly, thus, prompting us to investigate if the formation of polarized lamellipodia was affected by Nox1 expression. We found that, in wild-type cells, 30 min of PDGF treatment induces a front-to-rear polarization characterized by a single polarized lamellipodium and a distinguishable tail (Figures 2A,B). In contrast, we observed that in Nox1 deficient cells, PDGF treatment does not produce a front-to-rear polarity axis, inducing multiple lamellipodia-like structures instead (Figures 2A,B). This aberrant phenotype was indeed due to a lack of Nox1 activity since cell polarization was recovered by re-expression of Nox1 (Figures 2C–E) or by exogenous addition of hydrogen peroxide (H₂O₂, 10μM, Figures 3A–C). These data demonstrate that Nox1-produced ROS is required to form a single polarized lamellipodium.

Several Partitioning-defective (PAR) proteins control polarization during cell migration in fibroblasts, astrocytes, and T-cell (Ludford-Menting et al., 2005; Wang et al., 2012). The Par3 homolog (Par3) polarity protein has been directly implicated in regulating centrosome re-localization during cell polarization and lamellipodium formation at the leading edge (Schmoranzer et al., 2009; Hong et al., 2010). The active Par3/aPKC complex couples with the guanine nucleotide exchange factor (GEFs) Tiam1 at the leading edge of migrating cells to activate Rac, which induces lamellipodia formation and cell migration (Etienne-Manneville and Hall, 2001; 2003).

Since Nox1-deficient cells display altered lamellipodia and centrosome polarization, we posited that Nox1 is implicated in the Par3/aPKC/Tiam/Rac complex activation. We first quantified Rac activity before and after PDGF treatment in WT and Nox1^y/- cells to test this hypothesis. We found that in basal the levels of active Rac (Rac-GTP) are similar in both genotypes. However, after 1 and 2 min with PDGF, Nox1^y/- cells showed significantly higher levels of Rac-GTP compared with WT cells (Figure 4A). The differential Rac activation observed between WT and Nox1y/- cells lead us to look for active Tiam in the lamellipodia. We visualized the localization of active Tiam1 through its colocalization with the Rac nucleotide-free mutant GST-Rac1G15A, known to bind with high affinity to the active GEFs (Valdivia et al., 2020). We observed the highest levels of colocalization between Tiam1 and GST-Rac1G15A in the multiple lamellipodia-like protrusions of Nox1^y/- cells, while in WT cells, the colocalization was limited to the main lamellipodium (Figures 4B,C). We used recombinant GST protein (Supplementary Figure S3) and a non-immune IgG (data not shown) as specificity controls.

Numerous studies have shown that Par3 binding capacity is regulated by phosphorylation. Par3 phosphorylation can block or facilitate the recruitment of binding partners. Par3’s binding to Tiam and the ability to localize to the cells’ edge are associated with its phosphorylation status (Hurd et al., 2003; Goldstein and Macara, 2007; Ling et al., 2010). Therefore, we evaluated Par3 phosphorylation by selectively purifying phosphorylated proteins from wild-type and Nox1^y/- cells before and after PDGF stimulation. The isolated phosphoproteins and phosphoprotein-containing complexes were analyzed by immunoblots using specific antibodies. In wild-type cells, PDGF induces the phosphorylation of the two spliced variants of Par3 (180kDa and 100 kDa), while in Nox1 deficient cells, we observed exacerbated phosphorylation of Par3 after PDGF stimulation (Figure 4D). In contrast, PDGF-induced phosphorylation of cortactin reached a similar level in both genotypes (Figure 4D), indicating that the increase in Par3 phosphorylation is unlikely to result from a widespread increase in protein phosphorylation.

Par3 is phosphorylated by the aPKC (Lin et al., 2000; Li et al., 2010; Morais-de-Sa et al., 2010). Thus, we investigated whether Nox1 expression affects aPKC activity. We found the phosphorylation of the activation loop of the aPKC (PKCζ/λ) decreased after PDGF stimulation in WT whereas it increased in Nox1^y/- cells (Figure 4E).

Together, these data indicate that Nox1^y/- cells display a higher activation of the Par3/aPKC complex, probably driven by hyperphosphorylation.

Several protein phosphatases have been associated with the de-phosphorylation of the Par3/aPKC complex (Nam et al., 2007; Traweger et al., 2008; Krahn et al., 2009; Ogawa et al., 2009; Schumann et al., 2012). In particular, it is well established that the aPKC is a substrate for the protein phosphatase 2A (PP2A) (Nunbhakdi-Craig et al., 2002). Since aPKC shows increased phosphorylation in Nox1^y/- cells, we tested the hypothesis that PP2A undergoes inactivation in Nox1-deficient cells.

Several post-translational modifications regulate PP2A: one is the Tyr 307 phosphorylation within the catalytic subunit of PP2A (PP2A-CA), which is associated with loss of phosphatase activity (Chen et al., 1992). Figure 5A shows that PP2A-CA Tyr 307 phosphorylation decreases shortly after PDGF stimulation in WT while increasing in Nox1^y/- cells. This result indicates that after PDGF stimulation, PP2A is activated in the wild type while it is inhibited in Nox1^y/- cells. Furthermore, the pattern of PP2A inactivation mirrors the phosphorylation in the activation loop of the aPKC (Figure 4E), suggesting that increased aPKC phosphorylation in Nox1-deficient cells is the result of PP2A inactivation.

Consistent with a role for PP2A phosphatase activity in agonist-mediated polarization downstream of Nox1, wild-type cells treated with siRNA against PP2A-CA exhibit the same phenotype of aberrant polarization (Figures 5B–E) and increased Tiam localization to lamellipodia-like protrusions (Figure 6) as Nox1-deficient cells. Regarding the localization of the polarity complex, we observed that, in control cells, Par3 is distributed to the edge of the lamellipodium while amassing at the edge of multiple small lamellipodia-like structures in Nox1^y/- cells and in wild-type cells when PP2A-CA expression is silenced (Figure 7).

Finally, we observed that in the presence of the PP2A inhibitor okadaic acid (OKA, 1 μM), PDGF treatment in wild-type cells accurately recapitulates the phenotype of aberrant polarization (Figures 8A–C) and Par3 accumulation observed in Nox1 deficient cells (Figure 8D). Because OKA has no effect in Nox1 deficient cells, this experiment suggests that protein phosphatase activity is already inhibited and cannot be inhibited further in Nox1^y/- cells. This result is consistent with the idea that PP2A is a downstream effector of Nox1-induced cell polarization after PDGF treatment.

To further confirm the role of PP2A activity downstream of Nox1 in lamellipodia formation, we observed that endogenous PP2A-CA localized at the edge of lamellipodium in WT cells while getting excluded from the multiple lamellipodia in Nox1^y/- cells (Figure 9A). Although H₂O₂ can quickly diffuse in the cell, our results imply a proximity-based relationship between Nox1 and PP2A. Since Nox1 is expressed in low amounts in MEFs, making it hard to detect, we overexpressed Nox1 with an HA-tag. By staining for the HA tag, we observed that Nox1-HA localizes to the lamellipodia and to lateral zones at the edge of the cell (Figure 9B). Additionally, to rule out the possibility that reactive oxygen species (ROS) derived from Nox1 could affect the activity of PP2A, we used the thiol-reactive probe 5-IAF (5-iodoacetoamidofluorescein). Nox1 expression levels did not affect the redox status of PP2A-CA (Supplementary Figure S4).

These experiments demonstrate that PP2A is not localized at the proximity of the leading edge and that its activity is inhibited in Nox1-deficient cells. Furthermore, PP2A activity is responsible for the agonist-induced aberrant polarity in these cells.

After vascular injury, PDGFR beta activation by PDGF induces vascular smooth muscle migration and contributes to vascular diseases (Banai et al., 1998; Kohno et al., 2013; Guan et al., 2014; Wu et al., 2021). In addition, PDGF also increases ROS production by activating Nox1 (Marumo et al., 1997; Lassegue et al., 2001; Lavigne et al., 2001). Because Nox1 expression increases after injury and is required for the injury-induced formation of neointima in small animal models (Lee et al., 2009; Xu et al., 2012), we speculate that our signaling pathway is conserved in this cell type.

Indeed, as we observed in mesenchymal cells (MEFs), Nox1 depletion affected the polarization of primary mouse aortic smooth muscle cells (MASMCs) in response to PDGF, also inducing multiple lamellipodia in these cells (Figures 10A,B). Accordingly, Nox1^y/- MASMC also showed an aberrant inactivation of PP2A after PDGF treatment, consistent with exacerbated atypical PKCζ/λ activation (Figures 10C,D).

It is well-accepted that vascular smooth muscle cell migration contributes to the injury-induced neointimal formation (Inoue and Node, 2009). Therefore, to evaluate the physiological relevance of our pathway in vivo, we performed the murine model of femoral artery wire injury followed by histological analysis. Our results showed that the inhibition of neointimal hyperplasia in Nox1^y/- mice was accompanied by increased staining for the inactive form of PP2A-CA and an increased signal for active aPKCζ/λ (Figures 11A–C). These results suggest that the inactivation of PP2A and the aberrant activation of aPKC may contribute to reducing neointima formation observed in the Nox1^y/- mice arteries.