Assembly of fibronectin fibrils selectively attenuates platelet-derived growth factor–induced intracellular calcium release in fibroblasts

Cellular responses to platelet-derived growth factor (PDGF) are altered in a variety of pathological conditions, including cancers, fibroses, and vascular diseases, making PDGF-induced signaling pathways important therapeutic targets. The limited success of therapies designed to impact PDGF pathways may be overcome with a clearer understanding of how cells integrate signals from PDGF and the extracellular matrix (ECM). Here, we assessed the effects of fibronectin matrix assembly on the responsiveness of mesenchymal cells to PDGF. Our results indicate that fibroblast-mediated assembly of fibronectin fibrils attenuates intracellular calcium release in response to PDGF. The dose-dependent inhibition of PDGF-induced intracellular calcium release was specific to the ECM form of fibronectin. Further, a recombinant protein engineered to mimic ECM fibronectin similarly attenuated intracellular calcium release in response to PDGF. Of note, fibronectin attenuated the PDGF–calcium signaling axis at the level of phosphoinositide 3-kinase (PI3K) activation. Interestingly, ECM fibronectin did not alter other intracellular signals activated by PDGF, including activation of PDGF receptor β, AKT Ser/Thr kinase, phospholipase Cγ1, and extracellular signal–regulated kinase 1/2 (ERK1/2). Rather, fibronectin inhibited activation of the p55 regulatory subunit of PI3K in response to a variety of stimuli, indicating that ECM fibronectin selectively attenuates the intracellular calcium release cascade while leaving intact other PDGF signaling pathways. Selective regulation of calcium signaling by ECM fibronectin via the p55 regulatory subunit of PI3K represents a mechanism by which cells tune their response to PDGF and may therefore serve as a target to selectively regulate one branch of PDGF signaling.

wherein broadly up-or down-regulating PDGF signaling often produces off-target effects due to the number and impact of signaling pathways affected. A greater understanding of the mechanisms by which PDGF signaling is regulated and, in particular, how cells exert selective control over which PDGF-induced signaling pathways are activated will allow for the development of more targeted therapies.
PDGF-B binds to and activates PDGF-R␤, triggering a wellcharacterized intracellular signaling cascade that leads to release of calcium from intracellular stores (6). Intracellular calcium release requires concurrent binding of phospholipase C␥1 (PLC␥1) and phosphoinositide 3-kinase (PI3K) to phosphorylated tyrosine residues on activated PDGF-R␤ (6). Upon binding to the PDGF-R␤, PLC␥1 and PI3K become active via tyrosine phosphorylation (1). Activated PI3K catalyzes the formation of phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ) within the plasma membrane (9). In turn, PIP 3 serves as a docking site to recruit active PLC␥1 to the plasma membrane, where PLC␥1 hydrolyzes phosphatidylinositol 4,5-bisphosphate, resulting in generation of inositol trisphosphate (IP 3 ) (10,11). Free IP 3 then binds the IP 3 receptor on the endoplasmic reticulum (ER), resulting in release of calcium from the ER into the cytosol (6). PDGF-induced calcium signaling can contribute to a variety of cell behaviors, including cell differentiation, proliferation, and migration (1), and stimulates the expression of ECM-associated genes, including collagen A1 and fibronectin 1 (12).
ECM-derived signals can alter cell and tissue responsiveness to PDGF (13). Fibronectin is a principal component of the ECM, where it binds to and influences the activity of various cells, other ECM components, and numerous growth factors and cytokines (13,14). Cells secrete fibronectin in a soluble, protomeric form that is then assembled into insoluble ECM fibronectin fibrils by a tightly regulated, cell-mediated process termed fibronectin matrix assembly (14). Importantly, both the process of fibronectin fibril assembly and the resultant conformation of assembled ECM fibrils regulate tissue homeostasis. As examples, ECM fibronectin fibrils stimulate cell proliferation (15), migration (16), and differentiation (17), whereas transient changes in ECM fibronectin fibril conformation mediate local vasodilation in response to skeletal muscle contraction (18). Although there is evidence that fibronectin fragments and protomeric fibronectin can regulate cell responsiveness to PDGF, the effects of cell-assembled ECM fibronectin on PDGF signaling are not known.
The goal of this study was to determine whether and how the active assembly of a fibronectin matrix affects intracellular signals generated in response to PDGF. Intracellular calcium release in response to PDGF was assessed quantitatively, as this pathway mediates several of the downstream effects of PDGF (12,19). Results demonstrate that cell-assembled ECM fibronectin selectively attenuates PDGF-induced intracellular calcium release without affecting activation of PDGF-R␤. Moreover, ECM fibronectin attenuated PDGF-induced PI3K activation specifically, without altering PDGF-induced activation of other intracellular signaling molecules. Targeted inhibition of PDGF-PI3Kcalcium signaling by ECM fibronectin represents a previously unknown mechanism by which cells can tune specific responses to PDGF and thus may serve as a therapeutic target for specifically regulating one branch of the PDGF-signaling cascade.

Effect of ECM fibronectin on PDGF-induced intracellular calcium release
Exposure of mesenchymal cells to PDGF triggers calcium release from intracellular stores, resulting in a single, transient increase in cytosolic calcium concentration that typically becomes evident ϳ1 min post-PDGF treatment and lasts for ϳ3 min (19,20). Here, effects of ECM fibronectin on PDGF-induced intracellular calcium release were assessed using fibronectin-null mouse embryonic fibroblasts (FN-null MEFs). These cells do not produce fibronectin, but they are capable of assembling exogenously supplied fibronectin into ECM fibrils (15), allowing for the analysis of PDGF-induced intracellular signals in both the absence and presence of ECM fibronectin.
In the absence of fibronectin, the addition of PDGF to FNnull MEFs evoked a single transient increase in intracellular calcium that became apparent ϳ1 min post-PDGF treatment and peaked ϳ3.5 min post-PDGF treatment (Fig. 1A). This PDGF-induced intracellular calcium release was observed in the presence of the calcium chelator, EGTA, indicating release from intracellular stores (21). As expected (22), increasing concentrations of PDGF reduced the time to peak intracellular calcium concentration (Fig. 1A). Additionally, increasing concentrations of PDGF evoked a dose-dependent increase in fluorescence intensity that was statistically significant versus vehicle control at a PDGF concentration of 5 ng/ml (Fig. 1B). In agreement with previous studies of PDGF-R␤ activation (23) and PDGF-induced intracellular calcium release (19,20), the response to PDGF saturated at 20 ng/ml PDGF (Fig. 1B). These results serve to establish that the dynamics of PDGF-induced intracellular calcium release in FN-null MEFs is similar to that observed with other mesenchymal cell types.
We next asked whether the presence of a fibronectin matrix influences intracellular calcium release following PDGF exposure. To do so, FN-null MEFs were treated with increasing concentrations of fibronectin and cultured overnight to allow for assembly of fibronectin fibrils (24). No changes in intracellular calcium were detected following treatment with vehicle alone, in either the absence or presence of fibronectin (Fig. 1C, ϩVehicle). Pretreatment with fibronectin resulted in a dose-dependent attenuation of PDGF-induced intracellular calcium release (Fig. 1C) that was statistically significant at fibronectin concentrations of 12.5 nM and greater (Fig. 1D). To determine whether the matrix form of fibronectin was responsible for the attenuation of PDGF-induced intracellular calcium release, fibronectin matrix assembly was inhibited using two well-characterized approaches. First, a peptide derived from the adhesin F1 of Streptococcus pyogenes (termed FUD) was used to inhibit initial cell-fibronectin interactions and block fibronectin matrix assembly (25). Deletion of a single amino acid from the 49-amino acid sequence of FUD at position 29 (termed "del29") renders FUD inactive (26); thus, del29 was used as a negative control. In the second approach, the anti-fibronectin mAb, 9D2, was used to block fibronectin-fibronectin self-interac-

Fibronectin attenuates intracellular calcium release
tions and, thus, inhibit fibronectin fibril formation; 9D2 mAb does not affect initial cell-fibronectin interactions (27). FNnull MEFs were treated overnight with fibronectin in the presence of either FUD or 9D2 mAb or their respective controls. The addition of FUD peptides or 9D2 mAb blocked the ability of fibronectin to attenuate PDGF-induced intracellular calcium release ( Fig. 2A). In contrast, the addition of the corresponding controls, del29 or IgG, had no effect on the attenuation of PDGF-induced intracellular calcium release by fibronectin ( Fig.  2A).
Fibronectin matrix assembly did not alter PDGF-R␤ expression or the ability of PDGF to stimulate phosphorylation of PDGF-R␤ (Fig. 2B), indicating that fibronectin does not block ligation and activation of the PDGF-R␤. The addition of the FUD peptides to normal human dermal fibroblasts (NHDFs; Fig. S1) or WT MEFs significantly enhanced PDGF-induced intracellular calcium release compared with treatment with the control peptide, del29 (Fig. 2, C and D), demonstrating that ECM fibronectin attenuates PDGF-induced intracellular calcium release in other mesenchymal cell types. Taken together, these data indicate that the ability of fibronectin to attenuate PDGF-induced intracellular calcium release requires assembly of soluble fibronectin into ECM fibrils and that it is not due to sequestration of PDGF from cells by ECM fibronectin fibrils.
During fibronectin matrix assembly, fibronectin undergoes a series of conformational changes that result in exposure of a matricryptic site within the first type III repeat (FNIII1) (28). To mimic the structure and signaling capacity of ECM fibronectin, a recombinant fibronectin fragment was developed previously by directly coupling an "open" FNIII1 fragment having constitutive heparin-binding activity (FNIII1H) to the cell-binding FNIII8 -10 domain (FNIII1H,8 -10) (28). FNIII1H,8 -10 triggers a number of cell and tissue responses to a similar extent as ECM fibronectin (18,28). Thus, studies were conducted to compare effects of ECM fibronectin and FNIII1H,8 -10 on PDGF-induced intracellular calcium release. Treatment of FNnull MEFs for 4 h with FNIII1H,8 -10 attenuated PDGF-induced intracellular calcium release compared with pretreatment with the control protein, glutathione S-transferase (GST; Fig. 2E). In contrast to FNIII1H,8 -10, the ability of fibronectin to attenuate PDGF-induced intracellular calcium release was not evident at 4 h but, rather, was observed 20 h after fibronectin addition (Fig. 2E). These effects are consistent with the additional time required for soluble fibronectin to undergo assembly into ECM fibrils (24). Taken together, these data indicate that it is the ECM form of fibronectin that attenuates PDGFinduced intracellular calcium release.

Fibronectin does not affect PDGF-induced PDGF-R␤, ERK1/2, or PLC␥1 activation
To determine whether fibronectin alters the time course of PDGF-R␤ phosphorylation in response to PDGF, FN-null

Fibronectin attenuates intracellular calcium release
MEFs were incubated overnight in the absence or presence of fibronectin and then stimulated with PDGF for 2, 5, 10, or 20 min. Total PDGF-R␤ expression and PDGF-R␤ Tyr 857 phosphorylation were assessed at various times after PDGF addition via immunoblot analysis. Fibronectin had no effect on either PDGF-R␤ expression or PDGF-induced PDGF-R␤ Tyr 857 phosphorylation for any of the PDGF treatment durations tested (Fig. 3, A-C).
PDGF-induced PDGF-R␤ phosphorylation activates ERK1/2, a member of the mitogen-activated protein kinase family (6). In the absence of PDGF, ERK1/2 phosphorylation was not observed in either the absence or presence of fibronectin (Fig. 3A, 0 min). Two min after PDGF addition, ERK1/2 phosphorylation was detected and remained elevated for at least 20 min (Fig. 3A). Fibronectin did not affect PDGF-induced ERK1/2 phosphorylation at any of the time points tested (Fig. 3A). Together, these results indicate that fibronectin does not alter the rate or extent of PDGF-induced PDGF-R␤ autophosphorylation and, moreover, that fibronectin is not a general inhibitor of PDGF-R␤ signaling.
Intracellular calcium release in response to PDGF or ATP requires IP 3 binding to IP 3 receptors located on the ER (29). Intracellular calcium release in response to PDGF depends upon PLC␥1-generated IP 3 (6), whereas ATP-induced intracellular calcium release utilizes PLC␤-generated IP 3 (30). Thus, to determine whether fibronectin attenuates intracellular calcium release by globally blocking IP 3 receptors, ATP-induced intracellular calcium release was measured in FN-null MEFs cultured in the presence or absence of fibronectin. Treatment of cells with ATP caused a dose-dependent release of intracellular calcium, which was not affected by the presence of fibronectin ( Fig. 4A). Thus, fibronectin does not affect the ability of free IP 3 to bind to IP 3 receptors and evoke intracellular calcium release. To next determine whether fibronectin attenuates PDGF-induced PLC␥1 activation, PLC␥1 was immunoprecipitated from FN-null MEFs cultured in the presence or absence of fibronectin and stimulated with PDGF. Immunoblot analysis of the phosphotyrosine content of PLC␥1 immunoprecipitates indicated that fibronectin does not affect either PLC␥1 expression or PDGF-induced PLC␥1 phosphorylation (Fig. 4, B and C).

Fibronectin attenuates PDGF-induced PI3K activation via a substrate-independent mechanism
PLC␥1-mediated IP 3 generation also requires targeting of PLC␥1 to the plasma membrane (10), which is mediated by PIP 3 , a product of class IA PI3Ks (9, 10). To determine whether fibronectin affects PI3K activation, FN-null MEFs were incubated overnight in the absence or presence of fibronectin and then stimulated with PDGF for 2, 5, or 10 min. In the absence of fibronectin, PDGF caused a time-dependent increase in tyrosine phosphorylation of the p55 regulatory subunit of PI3K (Fig.  5, A and B). In contrast, PDGF-induced PI3K phosphorylation was significantly attenuated in cells cultured overnight in the presence of fibronectin (Fig. 5, A and B). Tyrosine phosphorylation of the alternatively spliced PI3K p85 regulatory subunit in response to PDGF was largely undetected, in either the presence or absence of fibronectin (Fig. 5A). Importantly, AKT phosphorylation was observed 2 min after PDGF treatment, in both the absence and presence of fibronectin (Fig. 5A), indicating that fibronectin selectively blocks PI3K activation.

Fibronectin attenuates intracellular calcium release
Fibronectin attenuated PDGF-induced PI3K p55 phosphorylation in a dose-dependent manner (Fig. 6, A and B). In cells pretreated with 200 nM fibronectin, the relative band density of PDGF-induced phospho-PI3K was 21 Ϯ 8% of that detected in cells lacking a fibronectin matrix (Fig. 6, A and B). Parallel studies showed that fibronectin did not reduce the expression of either the PI3K p55 regulatory subunit (Fig. 6C) or the PI3K p110 catalytic subunit (Fig. 6D). Taken together, these data indicate that fibronectin blocks activation of PI3K by PDGF.
Fibronectin fibril structure and, hence, function are influenced by the underlying adhesive substrate (24). Thus, studies were conducted to compare the ability of fibronectin to attenuate PDGF-induced calcium release in cells adherent to either collagen (an ␣1␤1 and ␣2␤1 integrin-binding substrate) or vitronectin (an ␣v␤3 integrin-binding substrate) (31). Collagenor vitronectin-adherent FN-null MEFs were incubated for 20 h in the absence or presence of fibronectin, and then PDGF-induced intracellular calcium release was measured. Importantly, collagen-and vitronectin-adherent FN-null MEFs assemble fibronectin fibrils into the ECM at a similar rate (24). PDGFinduced calcium release was significantly attenuated by fibronectin in both vitronectin-and collagen-adherent cells (Fig. 7A). As with collagen-adherent cells (Figs. 5 and 6), fibronectin attenuated PDGF-induced PI3K p55 phosphorylation of vitronectin-adherent cells (Fig. 7B, 2 min). Again, fibronectin treatment did not affect PDGF-R␤ protein levels or PDGF-induced phosphorylation of PDGF-R␤ Tyr 857 in vitronectin-adherent cells (Fig. 7B). Notably, and in contrast to collagen-adherent cells (Fig. 5A), phosphorylation of PI3K p55 was detected in vitronectin-adherent cells in the absence of PDGF (Fig. 7B). This basal PI3K activity of vitronectin-adherent cells was also attenuated by fibronectin (Fig. 7B), providing evidence that fibronectin-mediated attenuation of PI3K p55 phosphorylation is not specific to PDGF/PDGF-R signaling.

Fibronectin attenuates intracellular calcium release
Tyr 751 were similar in the absence and presence of fibronectin (Fig.  8, A-C). These data indicate that the attenuation of PDGF-induced PI3K activation by fibronectin is not due to loss of docking sites on PDGF-R␤. Rather, these data provide evidence that fibronectin inhibits PI3K activation directly.
Thus, to determine whether ECM fibronectin acts as a general inhibitor of agonist-induced PI3K activation, effects of fibronectin on PDGF-, epidermal growth factor (EGF)-, and fibroblast growth factor-2 (FGF-2)-induced PI3K activation were determined. Each of these growth factors triggers intracellular calcium release via the PI3K signaling axis (29,32). FNnull MEFs were incubated overnight in the presence or absence of fibronectin, and PI3K p55 phosphorylation in response to PDGF, EGF, and FGF-2 was examined. Fibronectin inhibited EGF-induced PI3K p55 phosphorylation (Fig. 9, A and B), pro-viding further evidence that inhibition of PI3K activity by fibronectin is not mediated at the level of the PDGF-R␤. Of note, FGF-2 did not activate either PI3K p55 or AKT but did stimulate ERK1/2 activation (Fig. 9A). These results are expected, as FGF-2 activates class II PI3Ks, which do not contain the PI3K p55 regulatory subunit (9).

Discussion
In this study, we report that cell-assembled, ECM fibronectin fibrils selectively attenuate intracellular calcium release in response to PDGF. The inhibitory effects of fibronectin were striking, as intracellular calcium release in response to a saturating concentration of PDGF was blocked completely by an overnight exposure to 50 g/ml fibronectin. Remarkably, the inhibitory activity of fibronectin was confined to the PI3K-  (25 nM) or an equal volume of PBS and then exposed to 30 ng/ml PDGF (ϩ) or the vehicle control (Ϫ) for 2 min. Proteins were extracted with SDS-RIPA buffer, immunoprecipitated using an anti-PLC␥1 mAb, and analyzed by immunoblotting using anti-PLC␥1 and phosphotyrosine monoclonal antibodies. Immunoblots shown represent one of three independent experiments performed in duplicate. C, the ratio of the average net intensity of phosphotyrosine bands to the average net intensity of PLC␥1 bands was determined. Values were normalized to the ϩPBS; ϩPDGF condition, which was set to 1. Data are presented as the average ratio Ϯ S.E. of three experiments performed in duplicate.

Fibronectin attenuates intracellular calcium release
calcium signaling axis, as the presence of a fibronectin matrix did not inhibit or alter the kinetics of several other PDGF-induced intracellular pathways, including ERK1/2, AKT, and PLC␥. It is important to note that the inhibitory effects of ECM FN were not due to binding and sequestration of PDGF within the ECM, as PDGF-induced phosphorylation of PDGF-R␤ Tyr 857 , Tyr 740 , and Tyr 751 showed similar kinetics in the absence and presence of fibronectin. A schematic of the PDGFintracellular calcium release cascade is shown in Fig. S2.
Several lines of evidence indicate that the inhibitory effect on calcium release is a specific property of ECM fibronectin fibrils. First, two different inhibitors of fibronectin matrix assembly (FUD peptides and 9D2 mAbs) ablated the ability of fibronectin to attenuate PDGF-induced intracellular calcium release. Similarly, intracellular calcium release in response to PDGF was enhanced in human dermal fibroblasts as well as mouse embryonic fibroblasts when fibronectin matrix assembly was inhibited. Second, incubation of FN-null MEFs with fibronectin for 20 h, but not for 4 h, attenuated PDGF-induced intracellular calcium release. During this 16-h time period, there is a 3-fold increase in the amount of fibronectin matrix deposited by FNnull MEFs (24), suggesting that fibronectin-mediated attenuation of PDGF-induced intracellular calcium release requires extensive fibronectin matrix assembly and maturation of fibronectin fibrils. Third, treatment of FN-null MEFs with the recombinant fibronectin matrix mimetic, FNIII1H,8 -10, for 4 h was sufficient to attenuate PDGF-induced intracellular cal-cium release. Soluble fibronectin must be assembled into an insoluble fibronectin matrix in order to adopt its bioactive ECM conformation. In contrast, FNIII1H,8 -10 was engineered to constitutively mimic the mature, ECM form of fibronectin, without the need for cell-mediated conformational changes to expose bioactive sequences (28). The ability of FNIII1H,8 -10 to mimic the signaling capacity of ECM fibronectin suggests that conformational changes to fibronectin during fibril formation are necessary for attenuation of PDGF-induced intracellular calcium release and, further, that ECM-specific conformations of the FNIII1 and/or the integrin-binding domains of fibronectin mediate these effects.
There is considerable evidence that fibronectin fragments and protomeric fibronectin can promote PDGF signaling. Both protomeric fibronectin and fibronectin fragments can enhance PDGF-induced fibroblast and smooth muscle cell proliferation, migration, and PDGF-R␤ activation (33). Fibronectin-derived fragments and peptides have also been used to enhance growth factor-mediated wound closure (33) and PDGF-mediated cell survival under stress (34). Finally, cell adhesion to protomeric fibronectin can cause transient ligand-independent activation of PDGF-R␤ (35). In contrast to these reports, the data presented herein are, to our knowledge, the first example of attenuation of PDGF signaling by fibronectin. Thus, reduced fibronectin matrix assembly or increased fibronectin matrix proteolysis may not only release the inhibitory effects of ECM fibronectin fibrils on PDGF-mediated calcium signals; proteo-

Fibronectin attenuates intracellular calcium release
lyzed fibronectin fragments may serve to further amplify the effects of PDGF on cell behaviors.
PDGF-induced intracellular calcium release is initiated by PDGF-induced PDGF-R␤ trans-autophosphorylation of Tyr 857 (6). Our studies indicate that ECM fibronectin did not affect PDGF-induced PDGF-R␤ Tyr 857 phosphorylation and that PDGF did not induce phosphorylation of PDGF-R␣ (195 kDa) in FN-null MEFs. Furthermore, ECM fibronectin did not block PDGF-induced phosphorylation of tyrosine residues that bind and activate PI3K, such as PDGF-R␤ Tyr 751 and Tyr 740 , and did not inhibit PDGF-induced activation of other intracellular signaling molecules, such as AKT or ERK1/2. Rather, fibronectin specifically attenuated PDGF-induced activation of PI3K, which is required for PDGF-induced intracellular calcium release (6). In addition, fibronectin attenuated EGF-induced PI3K activation in collagen-adherent cells and reduced basal PI3K activity in vitronectin-adherent cells, indicating that fibronectin specifically attenuates PI3K activity, without affecting other PDGF-generated signals. In support of this inhibitory mechanism, others have shown that PI3K inhibition can attenuate growth factor-induced calcium signaling without affecting PLC␥1 activation (36).
An important example of the specificity of ECM fibronectin in attenuating the PDGF-PI3K-calcium signaling cascade is illustrated by studies comparing the effect of fibronectin on PDGFversus ATP-induced intracellular calcium release. The PDGF-and ATP-induced calcium release pathways are very similar but are mediated by different isoforms of PI3K and PLC (6,9,10,30). Therefore, the different isoforms of PI3K that mediate PDGFversus ATP-induced intracellular calcium release could be important for understanding the differential regulation of the calcium response to these two agonists. ATP activates PLC␤ and class IB PI3Ks (9,30), whereas PDGF activates PLC␥1 and class IA PI3K (6,9). Class IA PI3Ks have either 55-or 85-kDa regulatory subunits, whereas class IB PI3Ks utilize 101-or 85-kDa regulatory subunits (9). Thus, the ability of fibronectin to attenuate PDGF-but not ATP-induced calcium release is consistent with the inhibition of the PI3K p55 regulatory subunit that was observed, as PI3K p55 is not involved in the ATP-induced calcium release pathway.
Yet another example of the specificity of ECM fibronectin in attenuating the PDGF-PI3K-calcium signaling axis is the absence of an effect of fibronectin on PDGF-induced AKT (protein kinase B) activity. Activation of AKT is classically achieved via translocation of AKT to the PI3K product, PIP 3 , and subsequent phosphorylation of AKT on Thr 308 (by PDK1) and on Ser 473 (by mTORC2) (37). Interestingly, our results indicate that although fibronectin clearly attenuated PDGF-induced PI3K p55 phosphorylation, fibronectin did not attenuate PDGF-induced AKT activation. AKT is an important mediator of cell survival (38), and as such, cells have developed multiple mechanisms of AKT activation, which allow PI3K and AKT to operate independently (39). One such mechanism is activation of the ubiquitously expressed nonreceptor tyrosine kinase Ack1, which phosphorylates AKT on Tyr 176 , causing AKT to be activated by phosphatidic acid, rather than PIP 3 (40). Activation of PDGF-R␤ has been shown to activate Ack1-AKT signaling in cancer cells (41), suggesting that PDGF-induced AKT activation can occur without PI3K activation (40,41). Alternatively, our data indicate that fibronectin attenuates, but never completely ablates, PDGF-induced PI3K p55 phosphorylation. Agonist-induced PIP 3 production is not the rate-limiting step in AKT activation (23), and relatively small amounts of PIP 3 can facilitate normal levels of AKT activation (39). Therefore, ECM fibronectin-mediated attenuation of PDGF-induced PI3K activation may be sufficient to inhibit PDGF-induced intracellular calcium release but insufficient to block PDGF-induced AKT activation.
The mechanism by which ECM fibronectin inhibits activation of the PI3K p55 regulatory subunit is not yet known. However, we can identify a number of mechanisms that are not responsible for attenuation of PI3K activation by fibronectin. One of these mechanisms involves Src homology region 2 domain-containing phosphatase-2 (SHP2), which dephosphorylates the phosphotyrosine residues on PDGF-R␤ that are associated with activation of PI3K (Tyr 740 and Tyr 751 ) (42). PDGF-induced phosphorylation of PDGF-R␤ Tyr 740 and Tyr 751 was not affected by ECM fibronectin, strongly suggesting that fibronectin-mediated SHP2 activity is not the mechanism by which fibronectin attenuates PI3K activity.

Fibronectin attenuates intracellular calcium release
Alternatively, downstream targets of AKT can feedback to inhibit agonist-induced PI3K activation (43). However, fibronectin did not affect basal or PDGF-induced AKT phosphorylation, indicating that an AKT-mediated negative feedback loop is unlikely to play a role in fibronectin-mediated attenuation of PDGF-induced PI3K activation. Rather, we speculate that attenuation of PI3K p55 activity by fibronectin may be due to activation of Src homology region 2 domaincontaining phosphatase-1 (SHP1), which dephosphorylates and inactivates PI3K directly (44,45). SHP1 is a 68-kDa proteintyrosine phosphatase expressed predominately by hematopoi-etic cells (46), but also produced at low levels by epithelial cells (47) and mesenchymal stem cells (48). Preliminary studies show that phosphorylated SHP1 was not detected in cells cultured in the absence of fibronectin (Fig. S3A). In contrast, fibronectin treatment induced a significant increase in the phosphorylation of a 47-kDa protein that was recognized by both phospho-SHP1 and nonphospho-SHP1 antibodies (Fig. S3, A and B). FN-null MEFs expressed the nonphosphorylated 47-kDa form of SHP1 in both the absence and presence of fibronectin (Fig. S3A). SHP1 is produced in multiple isoforms (49) and can be partially activated by proteolytic cleavage, and active SHP1 fragments

Fibronectin attenuates intracellular calcium release
have been reported (50). Thus, future studies to characterize the 47-kDa form of SHP1 will advance our understanding of SHP1 expression and function and knowledge of how fibronectin and SHP1 may contribute to the integration of growth factor signaling in mesenchymal cells.
In summary, PDGF is a potent intercellular signaling molecule that can initiate a cascade of intracellular signals, leading to a variety of cell and tissue behaviors both in vitro and in vivo. The PDGF-PLC␥1-PI3Kcalcium signaling axis exerts control over cell differentiation, proliferation, migration, and changes in gene transcription (1). The ability of ECM fibronectin to selectively regulate this arm of the PDGF signaling cascade brings to light a new mechanism by which cells utilize ECM-derived signals to tune intracellular signaling pathways. A greater understanding of the precise mechanisms by which cells integrate signals from specific ECM components to control which PDGF-induced signaling pathways are activated may allow for the development of more targeted PDGF-based therapies.

Cell culture
FN-null and WT MEFs were from Dr. Jane Sottile (University of Rochester, Rochester, NY) (15). FN-null and WT MEFs were cultured under serum-and fibronectin-free conditions on collagen I-coated tissue culture flasks in a 1:1 mixture of Aim V (Invitrogen) and SF Medium (Corning). Adult NHDFs were purchased from Lonza (Walkersville, MD) (CC-2511). NHDFs were cultured in fibroblast basal medium (Lonza) supplemented with 2% fetal bovine serum, human FGF-B, and insulin (Lonza). Medium was changed every 3 days, and cells were passaged at 70 -80% confluence.

Intracellular calcium release
FN-null MEFs, WT MEFs (1.1 ϫ 10 5 cells/cm 2 ), or NHDFs (3.5 ϫ 10 4 cells/cm 2 ) were seeded in Dulbecco's modified Eagle's medium onto black, clear-bottom 96-well plates precoated overnight at 4°C with either 0.05 mg/ml type I collagen in 0.02 N acetic acid or 64 nM vitronectin in PBS (24). Four h post-seeding, fibronectin and/or inhibitors of fibronectin matrix assembly were added to the media. Following an overnight incubation (ϳ23 h post-seeding), cells were well spread and confluent. Following removal of media, cells were loaded with 4.6 M Fluo-4/AM (Molecular Probes, Inc., Eugene, OR) in a HEPES-based imaging buffer (127 nM NaCl, 0.56 mM MgCl 2 , 4.7 mM KCl, 0.55 mM Na 2 HPO 4 , 1.28 mM CaCl 2 , 10 mM HEPES-NaOH, 11 mM D-glucose, pH 7.4) (54). Following a 40-min incubation, the Fluo-4 -containing imaging buffer was replaced with calcium-free imaging buffer containing 200 M EGTA, and cells were incubated for 20 min to allow for intracellular de-esterification of Fluo-4. Fluorescence intensity was measured using a FlexStation multimode microplate reader (Molecular Devices). This device allows for concurrent scanning of all eight wells within a single column of the 96-well plate, taking a fluorescence reading for each of the eight wells every 1.5 s. Baseline fluorescence intensity measurements were recorded for 30 s prior to the addition of PDGF-BB or ATP. Control wells received an equal volume of the vehicle alone (PBS containing 0.1% BSA).
To quantify changes in intracellular calcium signaling in response to PDGF, raw data were filtered to reduce noise inherent in the fluorescence signal by taking a 30-s running average of the raw data (55). The PDGF-induced fluorescence increase was defined as the peak increase in fluorescence intensity (⌬F) above baseline fluorescence (F 0 ). Data are presented as the relative change in fluorescence (⌬F/F 0 ) to account for shifts in baseline fluorescence due to independent factors, including dye uptake, excitation intensity, or detector efficiency (55). ⌬F/F 0 values from individual trials were normalized to a control condition (e.g. "ϩPBS"), which was set to 1.
Cell lysates (40 g of protein/lane) or immunoprecipitates were analyzed by SDS-PAGE and immunoblotting (24). Immunoblots were blocked with either 5% milk (Carnation, Glendale, CA) or 3% BSA (heat-shock fraction V, pH 7.5; Affymetrix, Cleveland, OH) in TBS, pH 7.6, with 1% Tween 20 and 1 mM sodium orthovanadate. Immunoblots were incubated overnight with primary antibody at 4°C. Blots were then washed, incubated with goat anti-rabbit or goat anti-mouse horseradish peroxidase-linked secondary antibodies, and developed using SuperSignal West Pico chemiluminescent substrate (Thermo Scientific). After detection, blots were stripped either with 0.2 M glycine, 0.1% SDS, 1% Tween, pH 3.8, or by incubation for 1 h at 65°C with 0.7% ␤-mercaptoethanol and 2% SDS in 62.5 mM Tris-HCl, pH 6.7. Stripped blots were reprobed with secondary antibodies alone before reprobing with primary antibodies. The average net band intensity for duplicate samples was determined by densitometry using Carestream molecular imaging software (Rochester, NY). Relative band densities were averaged across multiple experiments to determine the average relative band density of each protein of interest for each treatment condition.

Statistical analyses
Unless otherwise stated, data are presented as mean -fold increase Ϯ S.E. relative to control treatment. Experiments were performed in quadruplicate (calcium signaling) or duplicate (immunoprecipitation and Western blotting) for a minimum of three independent experiments. Statistical analysis was performed using GraphPad Prism version 4 (La Jolla, CA). Unless otherwise indicated, statistical significance was determined using either an analysis of variance (ANOVA) with a Bonferroni post-test or Student's t test. Results were considered statistically significant if p values were Ͻ0.05.