Register for Mechanobiology 2016!

Integrin α9β1 mediates enhanced cell migration through nitric oxide synthase activity regulated by Src tyrosine kinase
Shiv K. Gupta, Nicholas E. Vlahakis


Integrins are important mediators of cell adhesion and migration, which in turn are essential for diverse biological functions, including wound healing and cancer metastasis. The integrin α9β1 is expressed on numerous mammalian tissues and can mediate accelerated cell migration. As the molecular signaling mechanisms that transduce this effect are poorly defined, we investigated the pathways by which activated integrin α9β1 signals migration. We found for the first time that specific ligation of integrin α9β1 rapidly activates Src tyrosine kinase, with concomitant tyrosine phosphorylation of p130Cas and activation of Rac-1. Furthermore, activation of integrin α9β1 also enhanced NO production through activation of inducible nitric oxide synthase (iNOS). Inhibition of Src tyrosine kinase or NOS decreased integrin-α9β1-dependent cell migration. Src appeared to function most proximal in the signaling cascade, in a FAK-independent manner to facilitate iNOS activation and NO-dependent cell migration. The cytoplasmic domain of integrin α9 was crucial for integrin-α9β1-induced Src activation, subsequent signaling events and cell migration. When taken together, our results describe a novel and unique mechanism of coordinated interactions of the integrin α9 cytoplasmic domain, Src tyrosine kinase and iNOS to transduce integrin-α9β1-mediated cell migration.


Integrins play a central role in cell migration, which is essential for biological processes such as embryonic development, inflammation and tumor metastasis (Lauffenburger and Horwitz, 1996; Hood and Charesh, 2002; Guo and Giancotti, 2004). The signaling proteins, including Src, FAK and p130Cas, that transduce integrin functions are numerous and interact in a complex fashion. Rapid activation of Src tyrosine kinase and focal adhesion kinase (FAK) is often one of the first signaling events associated with ligand binding to integrins (Parsons and Parsons, 1997; Hynes, 2002); FAK undergoes auto-phosphorylation at Tyr397, which promotes binding to Src, and forms a potent FAK-Src tyrosine kinase complex (Schaller et al., 1994). Although integrins use common signaling pathways to transduce their effects, these pathways do not necessarily utilize or require the same signaling intermediates. For example, downregulation of FAK by genetic ablation or competitive inhibition prevents clustering of integrin α5β1 and cell migration (Ilic et al., 1995; Cary et al., 1996; Hsia et al., 2005). By contrast, integrin α4β1 can not only signal cell migration through activation of Src and FAK (Rabinowich et al., 1996; Yokosaki et al., 1996) but also by FAK-independent Src activation (Hsia et al., 2005).

Nitric oxide (NO), a byproduct of L-arginine catabolism, is a gaseous free radical that can mediate numerous and diverse biological effects, including control of blood vessel tone, renal sodium secretion and cell migration (Noiri et al., 1996; Murohara et al., 1999; Jadeski et al., 2003; Bove et al., 2007). The production and release of NO is the function of three NO synthases (NOS), including the inducible protein iNOS, which is expressed ubiquitously (Korhonen et al., 2005). The enzyme iNOS is known to interact with Rho GTPases, including Rac-1 (Kuncewicz et al., 2001), and NO itself can act as a heme cofactor for guanylyl cyclase (GC) generating cGMP, a second messenger acting coordinately to facilitate cell migration. The biological effects of NO are context- and cell-type-specific, resulting from either increased or decreased NOS activity (Pan et al., 1996; Fulton et al., 2005; Hausel et al., 2006; Milward et al., 2006). The enzymatic activity of NOS is often modulated by phosphorylation and/or its direct interactions with other intracellular proteins such as Src tyrosine kinases and Erk mitogen-activated protein (MAP) kinase (Kone et al., 2003; Korhonen et al., 2005; Zhang et al., 2007).

In this study, we explore the signaling mechanisms that are utilized by the integrin α9β1 to transduce cell migration. As components of the protein signaling pathways that transduce integrin and NO effects are shared, such as Src and Rac-1 noted above, we wished to determine whether there are interactions between integrins and iNOS and whether this interaction might modulate cell migration. Previous studies have reported that, when activated, β2 integrins expressed on neutrophils can stimulate NO production (Jenei et al., 2006). However, whether this effect occurs by means of β1 integrins and the nature by which NO might subsequently modulate integrin-associated cell migration are incompletely understood. We therefore chose to study the integrin α9β1 as its expression and function preferentially promote cell migration (Shang et al., 1999). We specifically wished to determine whether iNOS, NO and Src tyrosine kinase activity might contribute to integrin-α9β1-mediated cell migration and to identify the signaling intermediates involved and the requirements for activation of this signaling pathway.


Activation of integrin α9β1 induces iNOS phosphorylation and NO release to mediate cell migration

To study the role of NO in integrin-α9β1-mediated cell migration, we used SW480 cells, a stable cell line engineered to express integrin subunit α9. We confirmed by flow cytometry that integrin-α9-transfected SW480 cells (Fig. 1A, right panel), but not mock-transfected cells (left panel), express integrin α9β1. As expected, SW480α9 cells but not SW480-mock cells adhered to the surface-bound TnfnRAA, an integrin-α9β1-specific ligand (supplementary material Fig. S1A), confirming that the exogenously expressed integrin α9β1 is functionally active. We next detected the expression of iNOS in SW480 cells and found that, under standard growth conditions, it was highly expressed (Fig. 1B). We also tested other transfected cell lines that were used in our subsequent experiments and found that iNOS is also expressed in Chinese hamster ovary (CHO) cells and mouse embryonic fibroblasts (MEFs). By contrast, eNOS was detectable only in human umbilical vein endothelial cells (HUVECs) and human microvascular endothelial cell (HMVEC) lysates (data not shown). Furthermore, constitutive iNOS expression was not changed by the expression or activation of integrin α9β1 (data not shown).

As SW480 cells constitutively express iNOS, we next determined whether iNOS activation and subsequent NO production were influenced by integrin α9β1. Fig. 1C shows that basal levels of NO are similar in both SW480-mock and SW480α9 cells. However, compared with SW480-mock cells, stimulation with the integrin-α9β1-specific ligand TnfnRAA increases NO release by 1.5- to twofold in SW480α9 cells, which could be inhibited by an integrin-α9β1-specific blocking antibody. Specific activation of integrin α9β1 also resulted in a time-dependent increase in iNOS tyrosine phosphorylation, detectable as early as 5 minutes, with maximum activation at 10 minutes (Fig. 1D). There was no detectable iNOS phosphorylation in SW480-mock cells in the presence or absence of the integrin-α9-specific ligand, and there was no detectable serine phosphorylation of iNOS (data not shown). Furthermore, integrin-α9β1-dependent cell migration was significantly inhibited when iNOS activation was blocked, but not to the extent of directly blocking integrin α9β1 (Fig. 1E). Inhibition of guanylyl cyclase, a key intermediate for transduction of NO effects, also inhibited integrin-α9β1-induced migration, supporting the notion that NO can transduce the effects of activated integrin α9β1. Taken together, these data suggest that integrin-α9β1-dependent cell migration can be mediated through activation of iNOS and subsequent NO production.

Integrin-α9β1-induced iNOS activation is mediated by Src tyrosine kinase

Src tyrosine kinases are important mediators of integrin signaling (Calalb et al., 1995) and are also reported to mediate iNOS tyrosine phosphorylation, leading to its altered activity and/or subcellular distribution (Pan et al., 1996; Hausel et al., 2006). Thus, we next determined whether ligand binding to integrin α9β1 activates Src kinase and, if so, whether α9β1-induced Src activity is important for NO release. Fig. 2A shows that, in SW480α9 cells, specific ligation of integrin α9β1 induced Src tyrosine kinase activity by enhancing phosphorylation at Tyr416, a post-translational modification associated with Src tyrosine kinase activity. This activation of Src could be blocked by treatment with both a specific pharmacological inhibitor of Src and the integrin-α9-specific antibody Y9A2. By contrast, no Src activation was seen in SW480-mock cells. In addition, Fig. 2B shows that Src phosphorylation (top panel) and cell migration, measured using Transwell haptotaxis (bottom panel), were blocked by the Src inhibitor in a dose-dependent manner. Of note, treatment of SW480α9 cells with Src, iNOS or cGMP inhibitors also decreased their adhesion to integrin-α9β1-specific ligand (supplementary material Fig. S1B). Compared with inhibition of iNOS or cGMP, blocking Src activation resulted in the greatest inhibition of cell migration (data not shown), suggesting that Src might regulate several parallel integrin-α9β1-dependent migration signaling pathways.

Fig. 1.

Integrin-α9β1-dependent cell migration is mediated through iNOS activation and release of NO. (A) Flow cytometry analysis of integrin α9β1 expression (shaded area) in mock- (left panel) or α9-transfected (right panel) SW480 cells; the line represents the isotype control. (B) Western blot analysis of iNOS in various cell types (as indicated). (C) Total nitrite measured by Griess assay in mock- or α9-transfected SW480 cells grown on α9-specific agonist (TnfnRAA) and in SW480α9 cells in the presence of the integrin-α9β1 antibody (Y9A2); n=3. (D) Western blot to detect phosphorylated iNOS in mock- or α9-transfected SW480 cells stimulated with the integrin-α9β1 ligand TnfnRAA. (E) Migration of SW480α9 cells on the integrin-α9β1 ligand TnfnRAA. Left panel: photomicrographs of migrated cells in the absence (DMSO) or presence of the indicated inhibitors. Right panel: quantitative analysis of cell migration from three separate experiments; n=3, *P<0.05, **P<0.01; HPF, high power field.

Fig. 2.

Integrin-α9β1-induced iNOS phosphorylation, Rac-1 activation and migration rely on Src activity. (A) Western blot to detect activated pSrc(Y416) in SW480α9 or SW480-mock cells stimulated with integrin-α9β1 ligand (TnfnRAA), in the presence or absence of the Src inhibitor PP1 or the α9 antibody Y9A2. (B) Western blot showing dose-dependent inhibition of pSrc(Y416) (top panel) and migration (bottom panel) by inhibitor PP-1 in SW480α9 cells on TnfnRAA; **P<0.01, n=3. (C) Immunoprecipitation and western blot analyses of tyrosine phosphorylation of FAK (top panel), p130Cas (middle panel) and iNOS (bottom panel) in lysates of SW480α9 cells stimulated with TnfnRAA in the presence or absence of PP1. Blots were stripped and subjected to immunoblotting to detect total FAK, p130Cas or iNOS protein loading. (D) Western blot detection of phosphorylation of FAK using site-specific antibodies to pFAK(Y397) or pFAK(Y925) in SW480α9 cells stimulated on TnfnRAA in the presence or absence of PP1. (E) Western blot detection of the extent of siRNA-induced knockdown of FAK, iNOS, Src or p130Cas in SW480α9 cells transfected with specific siRNA to FAK, iNOS, Src or p130Cas. β-Actin was used as a loading control (left panel). Migration assay of SW480α9 cells on TnfnRAA following siRNA-mediated knockdown of FAK, iNOS, Src or p130Cas (right panel); n=3, *P<0.05, **P<0.01. (F) Immunoblot showing activation of Rac-1 in SW480α9 cells stimulated with the integrin-α9β1 ligand TnfnRAA. The left panel shows results from cells cultured in the absence (DMSO) or presence of Src inhibitor (PP1) or iNOS inhibitors (L-NAME or ODQ); the right panel shows activated Rac-1 in lysates from SW480α9 cells transfected with the indicated siRNAs. Immunoblotting was performed to determine the extent of knockdown of the targeted signaling protein. Total Rac-1 served as a loading control. Activated Rac-1 was analyzed using the GST-PAK1 binding, and equal Rac-1 expression was confirmed by immunoblotting for total Rac-1.

Following integrin activation, Src tyrosine kinase is known to complex with other signaling intermediates to mediate its downstream functions, including two key proteins, FAK and p130Cas (Veracini et al., 2005). Consistent with a previous report (Yokosaki et al., 1996), Fig. 2C (top panel) shows that specific ligation of integrin α9β1 induces tyrosine phosphorylation of FAK. Integrin-α9β1-induced FAK phosphorylation was partially blocked by inhibition of Src, suggesting that integrin-α9β1-induced FAK tyrosine phosphorylation is only in part regulated by Src kinase activity. This appears to be because both Src-dependent (pY925) and Src-independent (pY397) FAK tyrosine phosphorylation sites are activated by integrin α9β1 (Fig. 2D). Furthermore, the multi-adaptor and scaffold protein p130Cas, which can complex and signal synergistically with Src and FAK, is phosphorylated following integrinα9β1 activation (Fig. 2C, middle panel). Similarly, integrin-α9β1-induced p130Cas phosphorylation was inhibited when Src tyrosine kinase signaling was blocked. Taken together, these data suggest that ligand-activated integrin α9β1 utilizes Src-FAK-p130Cas signaling in an integrated fashion to transduce cell migration. Furthermore, integrin-α9β1-induced iNOS tyrosine phosphorylation was inhibited by blocking Src activity (Fig. 2C, bottom panel), suggesting that Src is required early in the signaling pathway for integrin-α9β1-induced iNOS phosphorylation and activation. We also demonstrate that integrin-α9β1-induced NO and cGMP production was inhibited to a similar degree by blocking activity of either Src tyrosine kinase (using PP1) or iNOS (using L-NAME) (supplementary material Fig. S2A,B).

The nature of Src-FAK-p130CAS interactions has been variable but remains integral to canonical integrin signaling. However, the role of Src, FAK or p130Cas remains unclear with respect to integrin-α9β1 signaling to promote cell migration. Using a short interfering RNA (siRNA) approach, we suppressed expression of c-Src, FAK, p130CAS or iNOS individually in SW480α9 cells (Fig. 2E, left panel) to determine their relative importance for integrin-α9-mediated cell migration. As shown in Fig. 2E (right panel), inhibition of c-Src, p130Cas or iNOS expression in SW480α9 cells decreased migration to varying degrees. However, knockdown of FAK did not alter integrin-α9β1-mediated cell migration.

Integrin α9β1 induces Rac-1 GTPase activity through Src tyrosine kinase and p130Cas

A change in the cytoskeleton is an important mechanism for facilitating cell migration and is transduced by the Rho family GTPases (Ridley, 2001). In particular, activated Rac-1 has been implicated in integrin-induced migration, being rapidly exported to the cell membrane, where it facilitates lamellipodia formation at the leading edge of migrating cells (Kiyokawa et al., 1998; Brugnera et al., 2002; Gustavsson et al., 2004). Thus, we next determined whether Rac-1 played a role in integrin-α9β1-mediated cell migration. Fig. 2F (left panel) shows that SW480α9 cells display robust Rac-1 activation when exposed to a matrix of the integrin-α9-specific ligand TnfnRAA. In contrast to inhibitors of NO and cGMP (L-NAME and ODQ), pretreatment of SW480α9 cells with an inhibitor of Src tyrosine kinase suppressed integrin-α9β1-induced Rac-1 activation. Furthermore, using siRNA directed against cSrc, p130Cas or iNOS, we found that activation of Rac-1 was inhibited upon specific knockdown of cSrc or p130Cas but not iNOS (Fig. 2F, right panel). In support of these findings, we also found that ligation of integrinα9β1 induces membrane localization of Rac-1, which was suppressed following inhibition of Src but not NO (supplementary material Fig. S3). These results suggest that, following ligation of integrin α9β1, the pro-migratory effects of Rac-1 are initiated through Src activation and p130Cas phosphorylation, but the activation of iNOS and presumably the influence of NO is not required.

Integrin-α9β1-induced Src activation, p130Cas and iNOS phosphorylation is independent of FAK

As noted previously, FAK is considered to be an important mediator of integrin signaling for cell migration (Ilic et al., 1995). Our results described above showed that specific ligation of integrin α9β1 induced a robust tyrosine phosphorylation of FAK in part through Src tyrosine kinase activation. However, the necessity for FAK activation for integrin-mediated cell migration has been challenged (Yano et al., 2004; Hsia et al., 2005). In particular, it was shown that the integrin α4β1, which along with integrin α9β1 forms a subfamily of integrins based on their homology and shared function of accelerated cell migration, can signal cell migration through Src in a FAK-independent manner (Hsia et al., 2005).

To test the role of FAK in integrin-α9β1 signaling for enhanced cell migration, we knocked down the expression of FAK and assessed the effect on integrin-α9β1-induced Src activation and subsequent signaling through p130Cas and iNOS. Western blots of SW480α9 cell lysates showed that integrin-α9β1-induced Src phosphorylation (pY416) is not blocked in the absence of FAK (Fig. 3A,B). However, the extent of baseline Src phosphorylation is decreased in FAK-knockdown cells, and the temporal profile of Src phosphorylation was modified, with less robust early phosphorylation and more rapid decline by 15 minutes. FAK inhibition neither prevented integrin-α9β1-specific iNOS phosphorylation nor significantly inhibited NO release (supplementary material Fig. S4A,B). Fig. 3C shows that, following ligation of integrin α9β1, p130Cas complexes with FAK. However, this complex is not essential for integrin-α9β1-induced p130Cas activation as FAK knockdown affected neither its tyrosine phosphorylation nor Rac-1 activation (data not shown). Furthermore, in assessing cell migration, we found similar results: (1) Src activity was essential for integrin-α9β1-dependent cell migration and independent of FAK, as evidenced by a comparable reduction in cell migration by Src inhibitor in both FAK-proficient and FAK-knockdown cells; and (2) blocking NO production by L-NAME (a NOS inhibitor) or its function by ODQ (a cGMP inhibitor) also resulted in a decrease in integrin-α9β1-induced cell migration, independent of FAK (supplementary material Fig. S4C). Taken together, these findings suggest that Src-mediated iNOS phosphorylation and NO production can contribute to integrin-α9β1-induced cell migration in a FAK-independent manner.

Fig. 3.

Integrin-α9β1-induced Src activation is independent of FAK. (A) Western blot was used to detect activated pSrc(Y416) in SW480α9 cells transfected with FAK-specific or control siRNA and stimulated on TnfnRAA. β-Actin was used as a loading control (middle panel). Values below the Src blot present band density ratio for pSrc:Src for each lane; the extent of FAK knockdown (by siRNA) is demonstrated in the lower panel. (B) Fold increase in pSrc(Y416) as determined by densitometry for the pSrc:Src ratio. Values for control (no matrix) lanes were used to normalize other samples; n=3, *P<0.05. (C) Immunoblot to detect tyrosine phosphorylation and FAK binding of p130Cas in immunoprecipitates (of p130Cas antibody) from lysates of SW480α9 cells transfected with control or FAK-specific siRNA and stimulated with TnfnRAA; the blot was stripped and immunoblotted to detect total p130Cas loading.

Fig. 4.

Integrin α-subunit determines the requirement for FAK, Src and iNOS-dependent cell migration. (A) Flow cytometry analysis of integrins α9β1 (solid line) and α5β1 (shaded area) in α9-transduced (left panel) or mock-transduced (right panel) MEFs; the dotted line represents the isotype control. (B) Western blot to detect FAK expression in mock- or α9-transduced MEFs transfected with control or FAK-specific siRNAs. (C) Migration of MEF-mock on fibronectin (black bars) or MEF-α9 on TnfnRAA (grey bars), following transfection with control or FAK-specific siRNA. (D) Migration of MEF-mock on the integrin-α5β1 ligand fibronectin (left panel) and MEF-α9 on the integrin-α9β1 ligand TnfnRAA (right panel) in the presence or absence of specific inhibitory antibodies [BMA5 (for α5) or Y9A2 (for α9)] or compounds PP1 (Src), L-NAME (iNOS) or ODQ (cGMP); n=3, *P<0.05, **P<0.01.

FAK- and iNOS-mediated cell migration is α-subunit-dependent

FAK phosphorylation is a crucial mediator of integrin-α5β1 signaling (Ilic et al., 1995), but our initial results suggest that it is not necessary for integrin-α9β1-induced cell migration. To determine the importance of the integrin α-subunit in FAK- or NOS-mediated cell migration, we used MEFs, which are known to express integrin α5β1 but not α9β1. Fig. 4A confirms that integrin-α9-transduced MEFs (left panel) express both integrin α9β1 (solid line) and integrin α5β1 (shaded area) and that mock-transduced MEFs (right panel) only express integrin α5β1. Following successful FAK protein inhibition by siRNA (Fig. 4B), we found that mock-MEF migration stimulated by the integrin α5β1 ligand fibronectin was significantly reduced (Fig. 4C, dark bars) in the absence of FAK. By contrast, TnfnRAA-induced migration of integrin-α9-MEFs was not affected by loss of FAK (Fig. 4C, grey bars). These results in MEFs further support the view that integrin-α9β1-mediated migration does not require FAK and, furthermore, that this FAK-independent migration might be α-subunit-dependent.

We next tested whether cell migration resulting from Src and iNOS activation following ligand activation of integrins was α-subunit dependent. As expected, mock-transduced MEFs showed no migration on the integrin-α9β1-specific ligand TnfnRAA (data not shown) but enhanced cell migration on fibronectin-coated Transwell inserts (Fig. 4D, left panel) that was significantly reduced by integrin-α5β1-specific antibody. Furthermore, migration was only partially blocked by Src inhibition and to a very subtle degree by iNOS inhibition. By contrast, Fig. 4D (right panel) shows that integrin-α9-MEFs underwent migration on TnfnRAA that could be significantly blocked by inhibition of Src (using PP1) or decreased by inhibition of iNOS (using L-NAME) or cGMP (using ODQ). Taken together, these results suggest that utilization of NO for cell migration might be integrin-specific and identifies integrin α9β1 as one such unique integrin.

The cytoplasmic domain of the integrin subunit α9 is required for iNOS phosphorylation and enhanced cell migration

It has previously been shown that the cytoplasmic domain (CD) of the α9 subunit is crucial for signaling through intermediates such as paxillin (Young et al., 2001) and transduces cell migration. To explore further the role of the α9 CD in integrin-α9β1-mediated migration, we next determined whether it was required for signaling through Src, p130Cas and iNOS. To investigate this, we used CHO cells that were mock transfected (Fig. 5A, first panel) or transfected with full-length α9 (second panel) or three CD mutants (Young et al., 2001): (1) α9DM3 (third panel), which retains only the proximal 11 CD amino acids of the CD, thus deleting amino acids IIEAEK, essential for cell migration; (2) α9α4 (fourth panel), a chimera in which the CD of α9 is replaced with the α4 CD; and (3) α9α5 (fifth panel), a chimera in which the CD of α9 is replaced with the α5 CD. We show that integrin-α9β1-specific activation of Src (Fig. 5B) and subsequent phosphorylation of FAK, p130Cas and iNOS (Fig. 5C) was not detectable in lysates of cells expressingα9-DM3 and α9α5. Fig. 5D shows that, in the presence of α9-specific ligand, migration of cells expressing α9DM3 and α9α5 did not increase significantly over the mock-transfected control cells, whereas migration of cells expressing α9α4 was similar to those expressing full-length α9. Furthermore, α9-induced production of cGMP (Fig. 5E) and NO (data not shown) was also inhibited in these cells. Taken together, these results suggest that the α9 CD is required for integrin-α9β1 signaling through Src, p130Cas and iNOS for enhanced cell migration.

Fig. 5.

The cytoplasmic domain of integrin subunit α9 is required for Src tyrosine kinase and iNOS transduction of integrin-α9β1-mediated cell migration. (A) Flow cytometry analysis to determine surface expression of integrin subunit α9 (shaded graph) in mock (top panel), full-length α9 (second panel), cytoplasmic domain mutant α9DM3 (third panel), α9 chimera with α4 or α5 cytoplasmic domains (α9α4, fourth panel and α9α5, bottom panel) in stably transfected CHO cells; line graph (empty) represents isotype control. (B) Western blot of lysates from transfected CHO cells, as indicated, to detect activation of pSrc(Y416) by TnfnRAA stimulation. The graph presents the relative induction of pSrc(Y416) as determined by densitometry for the pSrc:Src ratio, and values are normalized against mock treatment; n=3, **P<0.01. (C) Analyses of phosphorylation of FAK(Y925), p130Cas or iNOS in transfected CHO cells (as indicated) after TnfnRAA stimulation. (D) Migration assay of transfected CHO cells, as indicated, on the α9-specific ligand TnfnRAA; n=3, **P<0.01. (E) Concentration of cGMP in transfected CHO cells, as indicated, in response to stimulation by TnfnRAA; n=3, *P<0.05, **P<0.01 (compared with mock-transfected cells).

The subfamily of integrins α9β1 and α4β1 promotes cell migration through Src and NO

A striking finding of Fig. 5B-E is that, following ligand binding to integrin-α9β1-specific ligand, CHO cells expressing the α9α4 chimera (where the α9 CD is replaced by that of α4) demonstrated similar Src, iNOS, cGMP and cell migration responses to those of full-length α9. Based on these findings, we utilized α9-, α4- or mock-transfected CHO cells to determine whether the Src and iNOS signaling is important for integrin-α4β1-mediated cell migration. Fig. 6A shows the constitutive expression of integrinα5β1 on CHO cells (grey-shaded area in all panels) and robust expression of integrin α9β1 (solid line, middle panel) and integrin α4β1 (broken line, right panel) in α9- and α4-transfected CHO cells, respectively. Fig. 6B shows that both integrin-α9β1-mediated (TnfnRAA, middle panel) and integrin-α4β1-mediated (VCAM1, right panel) cell migration is decreased by inhibition of Src (using PP1) and iNOS (using L-NAME and ODQ) without any significant impact on adhesion to the ligand (supplementary material Fig. S1C). By contrast, integrin-α5β1-mediated migration (CHO-mock cells) on plasma fibronectin was not suppressed by iNOS inhibition and was suppressed only minimally by Src inhibition (Fig. 6B, left panel). Furthermore, specific activation of integrins α4β1 and α9β1 resulted in: (1) Src phosphorylation (pY416) that was blocked by specific Src inhibitor (supplementary material Fig. S5); (2) FAK, p130Cas and iNOS phosphorylation (Fig. 6C); and (3) cGMP production as a functional readout of NO production (Fig. 6D). Taken together, these data and previously published work (Hsia et al., 2005) provide further evidence that the signaling mechanism mediating integrin α9β1 function is similar to that of integrin α4β1 but distinct from that of integrin-α5β1.

Endogenously expressed integrin α9β1 in RD cells utilizes Src and iNOS to transduce migration

Our earlier results have shown that, in genetically modified cell lines, integrins α9β1 and α5β1 utilize differing signaling mechanisms for inducing cell migration. To determine whether endogenously expressed integrin α9β1 requires Src and iNOS activation for enhanced migration, we studied RD cells, which simultaneously express integrin α9β1 as well as α5β1 (Fig. 7A). As shown in Fig. 7B, integrin-α5β1-dependent migration of RD cells on fibronectin is decreased significantly by integrin-α5β1-blocking antibody and minimally by Src inhibition (using PP1) but not following iNOS inhibition (using L-NAME, ODQ); by contrast, integrin-α9β1-dependent migration (TnfnRAA) was decreased significantly by inhibition of both Src and iNOS (Fig. 7C). Furthermore, we show that knockdown of iNOS, Src or p130Cas in RD cells (Fig. 7D) results in decreased integrin-α9β1-mediated cell migration (TnfnRAA) but not integrin-α5β1-mediated migration (fibronectin) (Fig. 7E). Consistent with our previous results, knockdown of FAK did block integrin-α5β1-mediated migration but had no significant impact on integrin-α9β1-induced motility (Fig. 7D,E).

Fig. 6.

Similar to integrin α9β1, integrin-α4β1 also induces Src-dependent iNOS activation to facilitate cell migration. (A) Flow cytometry analysis of integrin α9β1, α4β1 or α5β1 expression in transfected CHO cells, as indicated. (B) Migration of CHO-mock cells on the integrin-α5β1-ligand fibronectin (left panel) or CHO-α9 cells on the α9β1 ligand TnfnRAA (middle panel) or CHO-α4 cells on the integrin-α4β1 ligand VCAM-1 (right panel) in the presence or absence of specified antibodies or inhibitors; n=3, *P<0.05, **P<0.01, compared with DMSO-treated cells. (C) Immunoprecipitation and/or immunoblots to demonstrate phosphorylation of Src(Y416), FAK-(Y925), p130Cas or iNOS in CHO-mock, CHOα9 or CHOα4 cells stimulated on the ligands fibronectin, TnfnRAA or VCAM-1, respectively. (D) cGMP concentration in response to stimulation with the ligands fibronectin, VCAM-1 or TnfnRAA in CHO cells transfected with mock, α4 or α9 constructs, respectively; n=3, **P<0.01.

Ligation of endogenously expressed α9β1 in RD cells promotes Src and iNOS activity

Our preceding data showed that, in transfected cell lines, integrin-α9β1-mediated cell migration involves Src, p130Cas, iNOS and NO. To confirm these findings in cells endogenously expressing integrin α9β1, we suppressed expression of integrin α9β1 by transfecting α9-specific siRNA into RD cells (Fig. 8A). Knockdown of integrin α9β1 in RD cells diminished their ability to adhere (Fig. 8B) and migrate (Fig. 8C) in response to the integrin-α9β1-specific ligand TnfnRAA. Knockdown of integrin α9 also suppressed activation of Src tyrosine kinase and its downstream phosphorylation of FAK (pY925), p130Cas and iNOS (Fig. 8D), seen in RD cells transfected with negative control siRNA when stimulated on TnfnRAA. Furthermore, there was a significant decrease in cGMP levels in cells transfected with integrin-α9 siRNA compared with negative control siRNA (Fig. 8E). Taken together with our previous results, these data in cells endogenously expressing integrin α9β1 further establish that integrin α9β1 utilizes Src and iNOS activation to transduce cell motility.


The cellular mechanisms that induce cell migration are complex and involve the interplay of numerous receptors, including integrins, and their associated signaling intermediates (Hynes, 2002). The gaseous molecule NO is one of these intermediates and likewise can transduce signals from a diverse array of stimuli and activated cell-surface receptors (Shizukuda et al., 1999; Noiri et al., 1997). An adhesion-dependent increase in iNOS expression in neutrophils suggests that adhesion receptors such as integrins might indeed regulate NO production (Webb et al., 2001). In this study, we have extended these findings and shown for the first time that activation of the integrin α9β1 can transduce cell migration through production of NO.

Fig. 7.

Endogenously expressed integrin α9β1 promotes Src- and iNOS-dependent cell migration. (A) Flow cytometry analysis showing expression of integrin α9β1 and α5β1 in RD cells. (B) Migration assay of RD cells on the integrin-α5β1 ligand fibronectin and integrin-α9β1 ligand TnfnRAA in the presence or absence of inhibitors; n=3, *P<0.05, **P<0.01 (compared with DMSO-treated cells). (C) Immunoblot performed to detect iNOS, Src, FAK or p130Cas in RD cells transfected with the respective specific siRNAs. β-Actin was used as a loading control. (D) Migration assay of RD cells on the indicated matrices following siRNA-mediated knockdown of FAK, iNOS, Src or p130Cas; n=3, *P<0.05, **P<0.01.

A previous report demonstrated that intracellular calcium increased, in a NO-dependent fashion, when rat neonatal cardiomyocytes were exposed to a pentapeptide with the common integrin-activating amino acid sequence RGD (Van der Wees et al., 2006). However, the authors could not implicate specific integrins as RGD is a nonspecific integrin ligand. Our finding that activated integrin α5β1 does not induce iNOS phosphorylation (Figs 5, 6, 7) suggests that not all integrins that bind to RGD utilize NO for cell migration. It also appears that the RGD sequence itself is not essential for activation as integrin α9β1 does not use RGD as a binding motif (Yokasaki and Sheppard, 2000). Furthermore, specific β-subunits do not appear to be essential because β2 integrins also induce NO production in neutrophils (Jenei et al., 2006). Instead, our results suggest that the α-subunit, and specifically its cytoplasmic domain, is the key determinant for integrin-induced iNOS activation. Of note, our findings suggest a role for intracellular NO in cell migration, but whether the cell responds in a similar fashion or through similar integrin-induced mechanisms to extracellular NO donors remains to be determined. Previous data suggest that these mechanisms might be different and dependent on the NO concentration within the extracellular versus intracellular compartments (Jenei et al., 2006).

The extent to which NO plays a role in biological functions varies across cell types and their organ of origin (Milward et al., 2006); therefore, the role of integrin-α9β1- and NO-mediated cell migration might vary considerably for any given organ system. Our findings suggest that integrin-α9β1-dependent NO might be important in the case of neutrophils, in which integrin α9β1 is highly expressed and plays an important functional role during granulopoiesis and neutrophil migration (Shang et al., 1999; Ross et al., 2006; Chen et al., 2006). Furthermore, integrin-induced NO was shown to promote neutrophil adhesion (Webb et al., 2001; Jenei et al., 2006). Thus, taken together with our current results, we speculate that activated integrin α9β1 might utilize NO to transduce its effects in neutrophils; but this remains to be determined.

Integrins and iNOS can each utilize numerous interacting downstream signaling intermediates that act coordinately to transduce cell migration (Giancotti, 2000; Schwartz and Ginsberg, 2002; Yaroslavskiy et al., 2005). In the case of integrin α9β1, these signaling intermediates include Src, p130Cas and Rac-1. Although not unique to integrin α9β1 (Hsia et al., 2005), our results do demonstrate a novel interaction between these proteins and iNOS. First, Src activation occurs early and proximal in the signaling pathway and is required for overall coordination of integrin-α9β1-induced cell migration; second, it appears that p130Cas and iNOS can both be phosphorylated by Src tyrosine kinase in a FAK-independent manner; third, integrin-α9β1-induced NO production is associated with the tyrosine phosphorylation, but whether p130Cas and iNOS interactions are important for NO production or cell migration is not yet clear; and fourth, Rac-1 is activated in an Src-dependent, iNOS-independent fashion. Although we showed the importance of p130Cas and NO for cell migration induced by integrin-α9β1, it appears that NO function was not required for Rac-1 activation. As the adaptor protein p130Cas is known to activate Rac-1 (Gustavsson et al., 2004), our results support the concept that integrin α9β1, like other integrins, utilizes p130Cas to activate Rac-1 downstream of Src. Although iNOS does not regulate Rac-1 activation, we speculate that it might affect integrin-α9β1-induced lamellipodia formation because Rac-1 and iNOS can interact directly (Kuncewicz et al., 2001), but this remains to be determined.

Fig. 8.

Ligation of endogenously expressed integrin-α9β1 promotes Src and iNOS activity. (A) Flow cytometry analysis of RD cells for endogenous integrin-α9β1 and following transfection of α9-specific or control siRNA. (B) Integrin-α9β1-dependent adhesion of RD cells to TnfnRAA in the absence or presence of integrin-α9β1 antibody (Y9A2), adhesion of RD cells transfected with control (siControl) or α9-specific siRNA (si-alpha9); n=6, **P<0.01. (C) Migration of RD cells on TnfnRAA in the presence or absence of α9-blocking antibody (Y9A2) or in cells transfected with control or α9-specific siRNA; n=3, **P<0.01. (D) Immunoprecipitation and/or immunoblotting of lysates from RD cells, in the presence or absence of TnfnRAA and transfected with control (siCont) or α9-specific siRNA (si-Intα9), for activated Src (pY416), pFAK-(pY925), phospho-p130Cas or iNOS. (E) Concentration of cGMP in RD cells transfected (or non-transfected) with control or α9-specific siRNA and following stimulation on TnfnRAA; n=3, **P<0.01 (comparisons between untreated and Y9A2-treated RD cells; control siRNA versus si-alpha9-transfected cells).

Activation of FAK and Src is one of the first events associated with integrin signal transduction (Parsons and Parsons, 1997), the two proteins forming a complex that can itself act as a potent tyrosine kinase, potentiating multiprotein complex formation and subsequent signal amplification (Schaller et al., 1994; Schlaepfer and Hunter, 1996; Mitra and Schlaepfer, 2006). Similarly, we found that ligand binding to integrin-α9β1 induces tyrosine phosphorylation of FAK and Src, and that FAK phosphorylation was in part regulated by Src tyrosine kinase activity. However, it appears that FAK is not required for integrin-α9β1-induced phosphorylation of Src, p130Cas, iNOS or, ultimately, cell migration, which might appear contradictory to previous reports suggesting that FAK acts as a key regulator of cell migration (Ilic et al., 1995) and plays a crucial role in RGD-induced NOS phosphorylation (Van der Wees et al., 2006). However, others have reported that cell migration can be stimulated in cells expressing kinase-dead FAK or FAK-targeting siRNA (Cary et al., 1996; Yano et al., 2004), suggesting that, in some cases, FAK might play a redundant role. In the case of integrin α9β1, the redundancy of FAK can be explained by considering its similarities to integrin α4β1, with which it forms a unique subfamily (Young et al., 2001), and that integrin α4β1 can also promote migration in a FAK-independent manner (Hsia et al., 2005).

These findings further highlight the shared α9-α4 signaling mechanisms showing that, like integrin α9β1: first, integrin-α4β1 signaling for cell migration can be transduced through Src tyrosine kinase and iNOS activation; and second, the cytoplasmic domain of α4 does induce activation of Src and iNOS. That both the integrins utilize Src and iNOS through the α-subunit CDs suggests that there are specific conserved amino acid sequences that are necessary for NO-induced cell migration following their activation. The exact region within the CDs of the integrins that is required for NO induction remains to be determined. Of note, not all signaling pathways for cell migration are shared by these two integrins because paxillin is required for migration mediated by integrin α4β1 but not by integrin α9β1 (Liu et al., 1999; Young et al., 2001; Nishiya et al., 2005), and SSAT (Chen et al., 2004; deHart et al., 2008) is exclusively utilized by integrin α9β1. In the light of such differences in signaling mechanism, the minimal effect of iNOS inhibition on integrin-α4β1-induced migration (Fig. 6D) might presumably be due to some unknown signaling pathway, acting as a compensatory mechanism.

Although much is known about the various ligands that can bind to integrin α9β1 to induce cell migration (Yokosaki et al., 1998; Yokasaki and Sheppard, 2000; Vlahakis et al., 2005; Vlahakis et al., 2007; Shinde et al., 2008; Staniszewska et al., 2008), the extent and nature of signaling pathways distal to the activated integrin remain unclear. Fig. 9 is a cartoon summary of the proposed signaling pathways that transduce cell migration following activation of integrin α9β1. The figure demonstrates that there are a number of interacting pathways (shown by different coloured arrows) that are initiated by activated Src. Src, as the most proximal signaling intermediate, activates p130Cas and iNOS, with subsequent Rac-1 activation and NO production, respectively. Although not required, FAK might also complex with p130Cas-iNOS, which we presume might amplify the biological effects of NO. Furthermore, the cytoskeletal changes required for cell migration appear to be transduced through Rac-1 activation, which is Src- and p130Cas-dependent but NO-independent.

Fig. 9.

Proposed signaling mechanisms for integrin-α9β1-induced cell migration (see main text for details).

Clearly, the pathways that we have outlined are simple, and more complex interactions of signaling proteins and pathways are certainly involved. However, as the role of integrin α9β1 in normal and disease states become clearer (Huang et al., 2000; Vlahakis et al., 2005; Chen et al., 2006; Vlahakis et al., 2007), these initial findings serve as a basis for ongoing investigations to determine the intracellular signaling mechanisms of integrin α9β1 and, in turn, better understand its therapeutic potential.

Materials and Methods


Mouse monoclonal antibody (Y9A2) raised against human integrin α9β1 was prepared as described previously (Wang et al., 1996). Rabbit polyclonal to p130Cas, iNOS and eNOS were from Santa Cruz Biotechnology; anti-cSrc, phosphoSrc (Y416) and phospho-FAK (pY925) from Cell Signaling; phospho-FAK (pY397) from BD Biosciences; monoclonal anti-phospho-tyrosine (clone 4G10) and anti-FAK (clone 4.47) from Upstate; mouse monoclonal anti-β-Actin from Sigma; mouse monoclonal antibodies to human integrin α4: MAB16983, α5: NKI-SAM-1 (cat#CBL497) and antibody to mouse integrin α5: BMA5 (cat#MAB1984) from Chemicon; Phycoerythrin (PE)-conjugated goat anti-mouse, goat anti-rat were from Jackson ImmunoResearch Laboratories; horseradish-peroxidase-conjugated anti-mouse, goat anti-rabbit antibodies were from Amersham Biosciences. The integrin α9β1-specific ligand, TNfn3RAA, which is a recombinant form of the third fibronectin type III repeat of chicken tenascin-C containing alanine (A) substituted for glycine (G) and aspartate (D) residues of the common integrin-binding domain RGD was produced as published previously (Yokosaki et al., 1994). Integrin α5β1 ligand: plasma fibronectin (FN), and α4β1 ligand: vascular adhesion molecule-1 (VCAM-1), were procured from Chemicon and R&D, respectively. PP1, a Src inhibitor and L-NAME, a NOS inhibitor, were from BioMol International; ODQ, a cGMP inhibitor, was obtained from Sigma. Optimum concentrations of the pharmacological inhibitors were empirically determined to avoid nonspecific toxic effects.

Cells and cell culture

Integrin subunit α9 or mock-transduced SW480 or MEFs were grown as described previously (Vlahakis et al., 2007). CHO cells expressing integrins α4, or α9 or CD-deficient α9 (DM3) or chimeras of α9α5 or α9α4 (kind gift from Dean Sheppard, UCSF, CA) were grown as described previously (Young et al., 2001). α9-DM3 cells do not undergo integrin-α9β1-induced cell migration; α9α4 cells, but not α9α5, transduce α9-ligand-induced cell migration (Young et al., 2001). Human rhabdomyosarcoma (RD) cells obtained from the ATCC were grown in DMEM containing 10% fetal bovine serum.

siRNA and cell transfection

Silencer Negative Control #1siRNA and siRNA targeted against human protein tyrosine kinase (PTK2 or FAK) (from Ambion), c-Src and iNOS (from Santa Cruz Biotechnology) or p130CAS (BCAR1) (from Dharmacon) were used for inhibition of FAK, cSrc, iNOS or p130Cas expression, respectively. Integrin subunit-α9-specific siRNAs (Vlahakis et al., 2005) were obtained from Ambion.

For transfection, SW480α9, MEFs or RD cells were grown to 60-80% confluency in DMEM (10% FBS) medium without antibiotics in six-well plates. Cells were transfected with control or specific siRNA to FAK, c-Src, iNOS, p130CAS or integrin subunit α9 [80nM in Opti-MEM (Gibco BRL)] using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions and were kept at least for 96 hours in growth medium before any further experiments.

Immunoprecipitation, co-immunoprecipitation, SDS-PAGE and western blot analysis

For immunoprecipitation of iNOS, FAK or p130Cas, SW480α9 or CHO cell lines were grown in full growth medium to achieve 70-80% confluency and subsequently in serum-free medium for 24-36 hours. Cells were trypsinized, washed with PBS and suspended in DMEM at a density of 107 cells per ml. For treatment, cells were divided into aliquots of 1 ml, treated with pharmacological inhibitors as required and transferred to 12-well plates precoated with 5 μg/ml TnfnRAA, FN or VCAM-1. Cells were incubated for the required time ranging from 5 to 30 minutes, washed with PBS and lysed in RIPA buffer (50 mM Tris-HCl, pH 7.6, 500 mM NaCl, 1% Triton X-100, 0.5 mM MgCl2, 1 mM PMSF, 10 μg/ml leupeptin, 10 μg/ml aprotinin). Immunoprecipitation and subsequent western blotting were performed as described previously (Vlahakis et al., 2007). To detect FAK binding to p130Cas, immunoprecipitates were subjected to western blot analysis using a FAK-specific antibody.

For densitometry, images of original blots (X-ray films) were captured using a BioRad GS-800 scanner linked to the calibrated densitometer, and bands were then analyzed using Quantity One 1-D analysis software (BioRad).

Rac-1 activation assay

SW480α9 and CHO cells were grown to ∼80% confluency, serum starved overnight, trypsinized and washed with PBS. Cells suspended in serum-free DMEM were incubated for 4 hours on ice and subsequently treated with pharmacological inhibitors, as indicated. Cells were then seeded on 12-well plates coated with TnfnRAA (5 μg/ml), incubated for 4 hours at 37°C and lysates obtained with lysis buffer. Cleared lysates equivalent to 200 μg protein were incubated with 50 μg GST fusion protein of the CRIB-binding domain of PAK (GST-PBD) bound to glutathione-Sepharose beads (Upstate) for 45 minutes at 4°C. Active Rac-1 bound to GST-PBD was precipitated, washed three times in lysis buffer and the associated proteins were eluted with Laemmli sample buffer. Active, total Rac-1 (in lysates) was then assessed by immunoblotting.

Fluorescence microscopy

Cells were suspended in serum-free medium, plated on glass cover slips pre-coated with 5 μg/ml TnfnRAA and allowed to adhere for 4 hours. Cells were washed with PBS, fixed and incubated with anti Rac-1 monoclonal antibody diluted in 5% normal goat serum (Upstate) at 4°C overnight. Cells were washed, incubated with anti-mouse IgG Alexa Fluor-488 antibody (BD Biosciences). Images were acquired using a fluorescence microscope (AX70; Olympus) equipped with a CCD camera (C4742; Hamamatsu). Magnification bars were calculated by using a calibration slide (Carl Zeiss) using the same 100× oil-immersion objective lens used to capture the images.

Measurement of NO

Production of NO was determined by measuring the metabolites nitrite and nitrate in the culture media using a nitrite-nitrate assay kit (Fluka, Buchs, Switzerland) as per the manufacturer's instructions. Briefly, 180 μl samples were mixed with 10 μl NADP and 10 μl of enzyme nitrate reductase, incubated for 2 hours at room temperature to convert nitrate into nitrite, 20 minutes after addition of Griess reagent (50 μl). Nitrite and nitrate-nitrite concentrations were calculated by measuring absorbance at 560 nm.

Determination of cellular cGMP

CHO or RD cells were suspended in serum-free DMEM, seeded on 12-well culture plates precoated with 5 μg/ml TnfnRAA. After 6 hours, the cells were washed with PBS, lysed and cellular cGMP was quantified in cell lysates by competitive ELISA, using a cGMP high-throughput screening (HTS) chemiluminescent immunoassay kit (Upstate).

Cell migration assays

Cell migration assays were performed as described previously (Vlahakis et al., 2007) using Transwell plates with an 8 μm pore size (Corning Costar). Briefly, to assess TnfnRAA-, VCAM-1- or FN-stimulated haptotaxis, the underside of the membrane was coated with 5 μg/ml of the indicated protein. For experiments involving pharmacological inhibitors, cells were pretreated for 30 minutes with 50 μM PP1, 1 mM L-NAME or 100 μM ODQ before adding the cells to the transwells. Cells were treated with specific antibodies to integrin α4, α5 or α9 (20 μg/ml) for 30 minutes on ice to block cell migration mediated by each integrin. SW480, RD or CHO cells were allowed to migrate for 12 hours (or MEFs for 4-6 hours) at 37°C. Non-migrated cells were wiped from the upper surface of the membrane, and cells that migrated to the underside of the membrane were washed with PBS, fixed, stained (DiffQuik, Pierce) and counted in five different high-power (100×) fields.

Cell-adhesion assays

Cell-adhesion assays were performed in 96-well flat-bottom microtiter plates (ICN, Linbro/Titertek, Aurora, OH). Plates were coated with 5 μg/ml TnfnRAA, VCAM-1 or FN at 37°C for 1 hour, blocked with 1% BSA (Sigma) in PBS for 30 minutes at 37°C. Cells were trypsinized, washed and suspended in serum-free DMEM at a density of 5×105 cells/ml, incubated with specific anti-integrin antibodies (20 μg/ml), 50 μM PP1 (Src inhibitor), 1 mM L-NAME (NO inhibitor) or 100 μM ODQ (cGMP inhibitor) for 30 minutes at room temperature (on ice for antibodies), and DMSO was used as vehicle control; assays were subsequently performed as described previously (Vlahakis et al., 2005).

Flow cytometry

Cells were trypsinized, washed and suspended in PBS at a concentration of 0.5×106 to 1×106 per ml, and aliquots were incubated with appropriate primary antibodies to specified integrins for 30 minutes, washed, resuspended in PBS and incubated with a phycoerythrin-conjugated secondary antibody. Finally, cells were assessed using a FACS-Scan flow cytometer and the data analyzed using Cell Quest software (Becton Dickinson).

Statistical methods

Unless otherwise indicated, data are presented as mean ± s.d. from three or more experiments; P values were determined using paired Student's t-tests. Western blots were performed at least three times and a representative example presented.


  • Supplementary material available online at

  • We thank Dean Sheppard (Lung Biology Center, UCSF, CA) for critical reading of this manuscript, the generous gift of cell lines and antibody (Y9A2) used in this study. We also thank Amha Atakilit for helpful suggestions and technical help. This work was supported by the NIH grant 1K08 HL076455-01A2 and Mayo Foundation Scholar Award to N.E.V. Deposited in PMC for release after 12 months.

  • Accepted March 4, 2009.


View Abstract