The β2-adrenergic receptor activates pro-migratory and pro-proliferative pathways in dermal fibroblasts via divergent mechanisms

Dermal fibroblasts are required for skin wound repair; they migrate into the wound bed, proliferate, synthesize extracellular matrix components and contract the wound. Although fibroblasts express β2-adrenergic receptors (β2-AR) and cutaneous keratinocytes can synthesize β-AR agonists (catecholamines), the functional significance of this hormonal mediator network in the skin has not been addressed. Emerging studies from our laboratory demonstrate that β2-AR activation modulates keratinocyte migration, essential for wound re-epithelialization. Here we describe an investigation of the effects of β2-AR activation on the dermal component of wound healing. We examined β2-AR-mediated regulation of biological processes in dermal fibroblasts that are critical for wound repair: migration, proliferation, contractile ability and cytoskeletal conformation. We provide evidence for the activation of at least two divergent β2-AR-mediated signaling pathways in dermal fibroblasts, a Src-dependent pro-migratory pathway, transduced through the epidermal growth factor receptor and extracellular signal-regulated kinase, and a PKA-dependent pro-proliferative pathway. β2-AR activation attenuates collagen gel contraction and alters the actin cytoskeleton and focal adhesion distribution through PKA-dependent mechanisms. Our work uncovers a previously unrecognized role for the adrenergic hormonal mediator network in the cutaneous wound repair process. Exploiting these divergent β2-AR agonist responses in cutaneous cells may generate novel therapeutic approaches for the control of wound healing.


Introduction
The healing of cutaneous wounds is a dynamic, well-organized and complex process requiring the orchestration of many different cell types and cellular processes (Martin, 1997). Dermal fibroblasts are actively involved in this process. They migrate to the wound site, proliferate, synthesize extra-cellular matrix components, form granulation tissue (Grinnell, 1994), and generate mechanical forces within the wound to initiate wound contraction (Gabbiani et al., 1972). Wound contraction can be beneficial to overall wound healing by decreasing the wound area and forming a mechanically strong reparative scar.
In certain situations, however, fibroblast presence and activity can be deleterious to wound healing. Undesirable wound contracture can occur, particularly as a consequence of burn and trauma wounds, and can result in both cosmetic and functional problems (Fang and Alexander, 1990;Rudolph, 1992;Skalli, 1988;Vande Berg and Rudolph, 1985). Additionally, the accumulation of abnormally large numbers of fibroblasts within a healing wound can also result in a fibrotic, contracted scar (Redden and Doolin, 2003). Understanding the mechanisms that regulate dermal fibroblast migration, proliferation and wound contraction could, therefore, be beneficial for devising novel therapies to regulate fibrosis and wound contraction to ultimately improve the wound healing process.
␤2-ARs are the only class of ␤-AR expressed on the three major cell types of human skin: keratinocytes (Steinkraus et al., 1996) dermal fibroblasts (McSwigan et al., 1981) and melanocytes (Gillbro et al., 2004). Emerging studies from our laboratory point to a role of the ␤2-adrenergic signaling pathway in wound healing. We recently showed that the ERK signaling pathway in keratinocytes is remarkably attenuated by ␤2-adrenergic receptor (␤2-AR) activation, resulting in marked diminution of keratinocyte migration by a cAMP-independent (Chen et al., 2002), phosphatase PP2A-dependent mechanism (Pullar et al., 2003). These findings imply that ␤2-adrenergic signaling could impair wound re-epithelialization, essential for wound healing (Martin, 1997). Indeed, we observe a ␤-AR agonist-mediated delay in both human and murine skin wound healing (Pullar et al., 2006). However, as dermal fibroblast migration, proliferation and wound contraction are also required for wound repair, we sought to determine how ␤2-AR activation might affect these cells, which are so crucial to the repair process.
We demonstrate that ␤2-AR activation is both promotogenic and pro-mitogenic in dermal fibroblasts. Dermal fibroblast-mediated collagen gel contraction was attenuated upon ␤2-AR activation and we observed changes in the ␤2-AR activation transactivated the EGFR in dermal fibroblasts The activation of the epidermal growth factor receptor (EGFR) is required for cell motility (Glading et al., 2000) and ␤-AR activation can transactivate the EGFR in COS-7 cells Pierce et al., 2000). We, therefore, reasoned that ␤2-AR activation may be stimulating fibroblast migration by transactivating the EGFR. The EGFR was immunoprecipitated from unstimulated and ␤2-AR-activated cell lysates and probed with either an anti-EGFR antibody or an anti-phosphotyrosine antibody. Roughly equal quantities of EGFR were immunoprecipitated from each lysate (Fig. 1D). Whereas tyrosine phosphorylation of the EGFR was undetectable in the absence of ␤-AR agonist, ␤2-AR activation phosphorylated the EGFR on tyrosine residues (Fig. 1D,E). Thus, ␤2-AR activation does transactivate the EGFR in human dermal fibroblasts. Dermal fibroblasts were plated onto collagen-coated glass coverslips at a concentration of 125 cells/mm 2 in FM for 3-6 hours at 37°C. Migration experiments were performed in FM in the presence or absence of ␤-AR agonist (10 nM-100 M). The migration of each single cell was monitored over a 1-hour period. The speed and distance traveled, at a concentration of 1 M ␤-AR agonist, are represented graphically in A and B, respectively. The ␤-AR-mediated dose-dependent increase in distance traveled is represented graphically in C (␤-AR agonist 10 nM-100 M). The data are representative of three independent experiments with three different fibroblast strains (n=50). Values plotted are mean ± s.e.m. *P<0.01 between ␤-AR agonist and controls. Cells were starved of growth factors in DMEM for 16 hours. 1-2ϫ10 7 cells were either left un-treated or treated with 1 M ␤-AR agonist in DMEM for 10 minutes at 37°C. After treatment, cell lysates were prepared and the EGFR was immunoprecipitated. EGFR antibodyassociated proteins were electrophoresed on two separate 10% polyacrylamide gels at the same time and transferred to membranes. Membranes were immunoblotted with either an EGFR antibody (EGFR WB) or an anti-phosphotyrosine antibody (PY WB; D). Gels were aligned to allow correct identification of the EGFR protein. The data shown are representative of three independent experiments from three separate cell strains. Three blots from three separate experiments were scanned for EGFR tyrosine phosphorylation and densitometry was performed using a gel plotting macro in NIH Image 1.62. Data was averaged, statistically analyzed and represented graphically (Fig. 1E). Values plotted are mean ± s.e.m. (n=3). *P<0.01 between ␤-AR agonist and control.
␤2-AR activation increased the phosphorylation of ERK in dermal fibroblasts ERK is known to regulate cell motility (Klemke et al., 1997), therefore, we examined whether ␤2-AR activation increased ERK phosphorylation in human dermal fibroblasts. Equal protein loading was demonstrated by immunoblotting with an anti-ERK antibody ( Fig. 2A). ERK was rapidly phosphorylated upon ␤2-AR activation, achieving a maximal increase of twoto threefold compared to unstimulated cells, 5 minutes after ␤-AR agonist addition. The phosphorylation of ERK remained significantly elevated for at least 30 minutes, returning to within basal levels after 1 hour (Fig. 2B). A time course of ERK phosphorylation in the absence of ␤-AR agonist confirmed that there was no change in phosphorylation during the time course of our experiment (results not shown).
The mechanism for the ␤2-AR-mediated increase in ERK phosphorylation was dependent on the Srcmediated transactivation of the EGFR ␤2-AR-mediated ERK activation is dependent on the transactivation of the EGFR in COS-7 cells Pierce et al., 2000). To determine whether a similar mechanism was responsible for the ␤2-AR-mediated ERK activation observed in dermal fibroblasts we pre-treated cells with the EGFR kinase inhibitor, AG1478 (10 M) (Kim et al., 2003;Kim et al., 2002) for 90 minutes prior to ␤2-AR activation. The ERK immunoblot demonstrates equal protein loading in all lanes (Fig. 3A). Although the level of basal ERK phosphorylation appeared higher in cells pre-treated with AG1478 in the absence of ␤-AR agonist, we could no longer detect an EGF-mediated increase in ERK phosphorylation confirming complete inhibition of the EGFR kinase. AG1478 pre-treatment completely prevented the ␤2-AR-mediated increase in ERK phosphorylation demonstrating that EGFR transactivation is essential for ␤2-AR-mediated ERK phosphorylation in dermal fibroblasts (Fig. 3A,B).
Src has been suggested to play a role in ␤3-AR-mediated ERK activation in brown adipocytes (Lindquist et al., 2000) and a human salivary gland cell line (Yeh et al., 2005). To determine if Src played a role in ␤2-AR-mediated ERK activation in dermal fibroblasts we pre-treated cells with the Src inhibitor PP2 (10 M) (Nam et al., 2002) for 6 hours prior to ␤-AR agonist addition. The ERK immunoblot demonstrates equal protein loading in all lanes (Fig. 3C). PP2 pre-treatment completely prevented any ␤-AR agonist mediated increase in ERK phosphorylation (Fig. 3C,D). It therefore appears that ␤2-AR-mediated ERK phosphorylation was Src and EGFR transactivation-dependent in dermal fibroblasts.
As we have demonstrated that ␤2-AR-mediated ERK activation is both EGFR and Src dependent, we wondered if the ␤2-AR-mediated transactivation of the EGFR was also Src dependent. Dermal fibroblasts were pre-treated with the Src inhibitor, PP2 (10 M) for 6 hours prior to ␤2-AR activation. The EGFR immunoblot demonstrates equal protein loading in all lanes (Fig. 3E). While the ␤2-AR-mediated transactivation of the EGFR receptor was observed within 15 minutes, as described above (in Fig. 1C) PP2 completely prevented its tyrosine phosphorylation, demonstrating that the ␤2-AR-Journal of Cell Science 119 (3) mediated transactivation of the EGFR was also Src dependent ( Fig. 3E,F).
The ␤2-AR-mediated increase in dermal fibroblast migration was Src dependent As ERK plays a pivotal role in fibroblast motility (Glading et al., 2000) and we have demonstrated here that ␤2-ARmediated ERK phosphorylation was dependent on Src activity, we hypothesized that the ␤2-AR-mediated increase in dermal fibroblast migration might also be Src dependent. Dermal fibroblasts were pre-treated with the Src inhibitor PP2 for 6 hours prior to observing single cell migration, in the presence or the absence of ␤-AR agonists. PP2 completely prevented the ␤2-AR-mediated augmentation of dermal fibroblast migration ( Fig. 4) demonstrating that the ␤2-AR-mediated increase in dermal fibroblast migration was also Src-dependent.
␤2-AR activation increased the proliferation of dermal fibroblasts via a cAMP-dependent mechanism ␤-AR activation can augment (Colombo et al., 2001) or Dermal fibroblasts were starved of growth factors in DMEM for 16 hours as described. 1-2ϫ10 6 cells were either left un-treated or treated with 1 M ␤-AR agonist in DMEM for 5-60 minutes at 37°C. After treatment, cell lysates were prepared, electrophoresed on 10% polyacrylamide gels and transferred to membranes. Membranes were immunoblotted with either an anti-phospho ERK antibody or an anti-ERK antibody (A). Three blots from three separate experiments were scanned for P-ERK and densitometry performed using a gel plotting macro in NIH Image 1.62. Data was averaged, statistically analyzed and represented graphically (B). Values plotted are means ± s.e.m. (n=3). *P<0.01 between ␤-AR agonist and controls (0). The data shown are representative of three independent experiments from three separate cell strains.

Fig. 3.
Dermal fibroblasts were starved of growth factors in DMEM for 16 hours as described. 1-2ϫ10 6 cells were preincubated with either DMEM alone (0, 5-60 minutes ISO) or DMEM containing either 10 M AG1478 (A,B) for 90 minutes or 10 M PP2 for 6 hours at 37°C (C,D). Cells were either untreated (control, 0, 5-60 minutes ISO) or stimulated with DMEM containing inhibitor and 1 M ␤-AR agonist for 5-60 minutes at 37°C, unless otherwise noted. After treatment, cell lysates from each experiment were prepared and electrophoresed on the same 10% polyacrylamide gels and transferred to membranes. Membranes were immunoblotted with either an anti-ERK antibody, a anti-phospho ERK antibody (P-ERK) an anti-EGFR antibody (EGFR) or an antiphosphotyrosine antibody (PY). Three blots from separate AG1478 or PP2 experiments were scanned for p-ERK or PY and densitometry performed using a gel plotting macro in NIH Image 1.62. Data was averaged, statistically analyzed and represented graphically (B,D,F). Values plotted are means ± s.e.m. (n=3). *P<0.01 between conditions and controls. # no significant difference between AG1478/control and AG1478/␤-AR agonist or PP2/control and PP2/␤-AR agonist. The data shown are representative of three independent experiments from three separate cell strains.
conversely, decrease (Liu et al., 2004) cell proliferation, depending on the cell type studied. Thus it was important to determine what effect ␤2-AR activation would have on human dermal fibroblast proliferation. Therefore, human dermal fibroblasts were grown in the presence or absence of ␤-AR agonist (1 M). ␤2-AR activation significantly increased fibroblast proliferation, with a maximum augmentation of 55% at day 6 ( Fig. 5A). To ensure that the ␤-AR-mediated increase in fibroblast proliferation rate was not limited to a specific concentration of ␤-AR agonist we performed proliferation experiments at a range of ␤-AR agonist concentrations from 10 nM to 10 M. We observed a dosedependent increase in proliferation rate. ␤-AR activation was pro-mitogenic at all concentrations of ␤-AR agonist tested (results not shown).
␤2-AR can couple to G s (Xiao et al., 1999) increasing intracellular cAMP levels and activating downstream cAMPdependent kinases such as PKA and EPAC (exchange proteins directly activated by cAMP) (Hanoune and Defer, 2001). To determine if the ␤2-AR-mediated increase in dermal fibroblast proliferation was cAMP-dependent we initially incubated cells in the presence of sp-cAMP, an active cAMP analog (Van Haastert et al., 1984), to increase the concentration of intracellular cAMP. The growth rate of fibroblasts maintained in the presence of ␤-AR agonist, sp-cAMP or both, were practically indistinguishable from each other, hinting that the ␤2-AR-mediated pro-mitogenic effects were cAMP-dependent (Fig. 5B).
The inactive cAMP analog, rp-cAMP (Van Haastert et al., 1984), a specific PKA inhibitor (de Wit et al., 1982) had no effect on proliferation alone, but when added to the dermal fibroblasts before the ␤-AR agonist it almost completely prevented the ␤2-AR-mediated augmentation of proliferation (Fig. 5C). The ␤2-AR-mediated augmentation of dermal fibroblast proliferation was, therefore, mediated by a cAMP/PKA-dependent mechanism.
␤2-AR activation attenuated the dermal fibroblastmediated contraction of collagen gels Fibroblast-seeded collagen gels have been widely used experimentally as a wound contraction model because they simulate fibroblast behavior in the early phases of wound healing (Grinnell, 2000). To determine whether ␤2-AR activation would alter the contraction of dermal fibroblastseeded collagen gels, collagen lattices populated with dermal fibroblasts were assembled in either the absence or presence of 10 M ␤-AR agonist. After 24 hours the addition of ␤-AR agonist had markedly delayed gel contraction (Fig. 6). The delay was maintained throughout the 5 days of the experiment and could be prevented with antagonist pre-treatment (results not shown). We also observed an inhibition of collagen gel contraction at lower concentrations of ␤-AR agonist (10 nM and 1 M, results not shown) with maximum inhibition observed at 10 M ␤-AR agonist.
As we had determined that the ␤2-AR-mediated augmentation of dermal fibroblast proliferation was mediated by a cAMP/PKA-dependent mechanism we hypothesized that the mechanism for the ␤2-AR-mediated delay in dermal fibroblast-mediated gel contraction could also be cAMP dependent. We added the inactive cAMP analog, rp-cAMP (Dostmann et al., 1990), to the collagen gels before casting, Journal of Cell Science 119 (3) at a concentration known to inactivate cAMP-mediated downstream signaling components (50 M) (Dostmann et al., 1990;Hirshman et al., 2001). Rp-cAMP alone had no effect on the contraction of the collagen gels but partially prevented the ␤2-AR-mediated delay in contraction (Fig. 6). After 24 hours, ␤-AR agonist-treated gels were only 47% contracted, whereas gels cast with dermal fibroblasts pre-treated with rp-cAMP prior to ␤-agonist addition were 59% contracted, and untreated gels were 71% contracted, indicating that the mechanism for the ␤2-AR-mediated delay in dermal fibroblast collagen gel contraction was partly cAMP/PKA dependent.
At day 4, the gels were digested with collagenase, cells were counted and the viability of the fibroblasts was assessed by Trypan Blue exclusion. Cells were 95% viable and the cell number was found to be comparable between control, ␤agonist-treated and rp-cAMP-treated gels and similar to seeding density (data not shown). Dermal fibroblasts were plated onto collagen-coated glass coverslips in FM as described and pre-treated with 10 M PP2 for 6 hours at 37°C. The migration of each single cell was monitored over a 1-hour period in FM in the presence or absence of 1 M ␤-AR agonist, as described. The speed and distance traveled are represented graphically in A and B, respectively. The data are representative of three independent experiments with three different fibroblast strains (n=50). Values plotted are means ± s.e.m. *P<0.01 between ␤-AR agonist and controls. ␤2-AR is motogenic and mitogenic in dermal fibroblasts ␤2-AR activation alters the dermal fibroblast cytoskeleton Actin remodeling plays an important role in cell motility (Pantaloni et al., 2001), proliferation (Blakesley et al., 1998;Cuadros et al., 2000;Ikeda et al., 2003;Joneson et al., 1996;Landriscina et al., 2000;Sastrodihardjo et al., 1987) and collagen gel contraction (Miki et al., 2000). Actin filaments terminate in focal adhesions, where several proteins, including vinculin, mediate interactions with the actin cytoskeleton (Burridge and Fath, 1989).
As we have demonstrated that ␤2-AR activation in dermal fibroblasts is pro-motogenic, pro-mitogenic and anticontractive, we were interested to see if it also altered cytoskeletal F-actin and focal adhesion number and size using vinculin as a focal adhesion marker (Beningo et al., 2001;Burridge and Fath, 1989).
All cells plated in the absence of ␤-AR agonist showed pronounced transcytoplasmic actin stress fibers along the borders of the cells and multiple vinculin-containing focal adhesions (Fig. 7A). Pre-treating with ␤-AR agonist for 15 minutes (1 M) markedly decreased actin staining in 90% of the cells, suggestive of ␤2-AR-mediated actin depolymerization (Hirshman et al., 2001), and also decreased the number and size of vinculin-containing focal adhesions (Fig. 7B). ImageJ was used to quantify the reduction in actinand vinculin-associated fluorescence by measuring the mean pixel intensity of 25 cells from each condition. Control cells had a mean pixel intensity of 50.3±4.8. ␤-AR agonist treatment resulted in a 67% drop in mean pixel intensity to a level of 16.6±2.0.
Since we have determined that the ␤2-AR-mediated augmentation of proliferation and attenuation of collagen gel contraction were both cAMP/PKA dependent we hypothesized that the ␤2-AR agonist-mediated change in actin stress fibers and vinculin-associated focal adhesions could also be cAMP/PKA dependent. We pre-treated dermal fibroblasts with the active cAMP analog, sp-cAMP (50 M) for 45 minutes. 80% of sp-cAMP-treated cells had the same staining pattern as ␤-AR agonist-treated cells, with reduced actin and vinculin staining, indicating a cAMP-mediated mechanism (Fig. 7C).
Adding both ␤-AR agonists and sp-cAMP simultaneously resulted in a similar staining pattern (Fig. 7D). ImageJ was used to quantify the reduction in actin-and vinculin-associated fluorescence by measuring the mean pixel intensity of 25 cells from each condition. Treatment with the active cAMP analog, sp-cAMP (50 M), resulted in a 71% drop in pixel intensity to 14.6±1.4, similar to the drop in pixel intensity observed upon ␤-AR agonist treatment (67%). Combining sp-cAMP and ␤-AR agonist did not further decrease the pixel intensity. sp-cAMP pre-treatment followed by ␤-AR agonist addition resulted in a 70% drop in mean pixel intensity to 15.1±2.3.
To confirm the role of cAMP/PKA in the ␤-AR agonistmediated reduction in actin and vinculin staining, dermal fibroblasts were pre-treated with the inactive cAMP analog, rp-cAMP, to inhibit PKA (de Wit et al., 1982). There was no observed effect of rp-cAMP treatment alone on the actin or vinculin staining of dermal fibroblasts (Fig. 7E), all cells Journal of Cell Science 119 (3) resembled untreated cells. Pre-treatment with rp-cAMP, however, prevented the ␤-AR agonist-mediated decrease in actin and vinculin staining in 90% of the cells, confirming that the mechanism for the ␤2-AR-mediated alteration of cytoskeletal conformation was cAMP/PKA dependent (Fig.  7F). ImageJ was used to quantify the reduction in actin-and vinculin-associated fluorescence by measuring the mean pixel intensity of 25 cells from each condition. Conversely, the inactive cAMP analog, rp-cAMP did not significantly alter the mean pixel intensity observed in control cells. The mean pixel intensity measured in rp-cAMP-treated cells was 55.7±7.1. Additionally, rp-cAMP pre-treatment prevented the ␤-ARmediated decrease in actin/vinculin-associated immunofluorescent staining, the mean pixel intensity of rp-cAMP pre-treated, ␤-AR agonist-treated cells was 47.4±4.5, a level within the range of pixel intensity measured in control cells.

Discussion
Adrenergic receptors were identified in human skin over three decades ago (Tseraidis and Bavykina, 1972). Interestingly, ␤2-ARs are the only class of ␤-AR expressed on the three major cell types of human skin: keratinocytes (Steinkraus et al., 1996) dermal fibroblasts (McSwigan et al., 1981) and melanocytes (Gillbro et al., 2004). Keratinocytes also have the capacity to synthesize the catecholamines epinephrine and nor-epinephrine, both ligands for adrenergic receptors (Schallreuter, 1997;Schallreuter et al., 1995). With cells that express both receptors and ligands, it is becoming evident that the skin generates a localized hormonal mediator network, which has the potential to regulate its physiology.
Clues to the physiological role of the ␤2-AR/catecholamine network within skin have been previously uncovered by the demonstration of alterations within the components of this network in some epidermal skin diseases. In atopic eczema there is a point mutation in the ␤2-AR gene and a low ␤2-AR density on keratinocytes and peripheral blood lymphocytes (Schallreuter, 1997). In psoriasis, epidermal cells from psoriatic lesions demonstrate a low cAMP response to ␤2-AR activation (Eedy et al., 1990). Additionally, a paracrine role for the hormone mediator network in skin homeostasis has been demonstrated recently as keratinocyte catecholamine synthesis can regulate melanogenesis in melanocytes (Gillbro et al., 2004).
Our laboratory has discovered a novel role for the adrenergic hormonal mediator network in modulating skin wound repair. We reported that ␤2-AR activation decreased keratinocyte migration and ERK phosphorylation in a cAMPindependent (Chen et al., 2002) and phosphatase PP2Adependent manner (Pullar et al., 2003). ␤-AR agonists decrease the re-epithelialization of both human and murine skin wounds (Pullar et al., 2006). As multiple cell types contribute to cutaneous wound healing (Martin, 1997) and ␤-AR activation can result in diametrically opposing responses in different cell types (Masur et al., 2001;Murphy et al., 1998;Salathe, 2002;Spurzem et al., 2002), it was important to examine the response to ␤2-AR activation in other cutaneous cells.
Here we demonstrate the unique effects of ␤2-AR activation on the physiological processes that contribute to the fibroblasts reparative role in the skin: migration, proliferation and contractile ability. Further, we elucidate the divergent signaling pathways by which these ␤2-AR-driven responses are generated.
We discovered that in contrast to the anti-motogenic effects of ␤2-AR activation in keratinocytes (Pullar et al., 2003;Pullar and Isseroff, 2005), the activation of ␤2-AR in dermal fibroblasts was both pro-motogenic and pro-mitogenic. The diametrically opposite response to ␤2-AR activation in fibroblasts as compared to keratinocytes underscores the importance of evaluating the ␤2-AR-mediated responses in a specific cell type. For example: ERK phosphorylation was increased by ␤2-AR activation in dermal fibroblasts yet decreased in keratinocytes (Pullar et al., 2003).
We provide evidence for the activation of divergent promotogenic and pro-mitogenic ␤2-AR-mediated signaling pathways in dermal fibroblasts. ␤2-ARs are classical GPCRs, capable of coupling to G s and increasing intracellular cAMP (Hurley, 1999;Xiao et al., 1999). Indeed, we discovered that a cAMP/PKA-dependent pathway mediated the ␤-AR agonistinduced increase in dermal fibroblast proliferation and decrease in contraction of collagen gels. On the other hand, the ␤2-AR-mediated transactivation of the EGFR, and increase in ERK phosphorylation and migration were Src dependent. The mechanism for ␤2-AR-mediated Src-dependent EGFR transactivation could be dependent on the matrix metalloprotease-mediated release of heparin-binding EGF (Kim et al., 2002;Pierce et al., 2000), clathrin-mediated endocytosis  or both. Actin cytoskeletal remodeling (Pantaloni et al., 2001) and focal adhesion turn over (Burridge and Fath, 1989) play an important role in migration. The ␤2-AR-mediated changes in cytoskeletal conformation were cAMP/PKA-dependent, however Src inhibition attenuated the ␤2-AR-mediated increase in migration, suggesting that Src could be upstream of cAMP/PKA in dermal fibroblasts. Indeed, murine embryonic fibroblasts overexpressing c-Src show enhanced ␤-ARmediated cAMP accumulation (Bushman et al., 1990).
Thus in dermal fibroblasts, divergent signaling pathways control cellular responses to ␤2-AR activation: a PKAdependent pathway controls proliferation, contractile ability and cytoskeletal conformation while a Src-dependent pathway regulates migration. These pathways are summarized in Fig. 8.
␤2-AR activation can, therefore, modulate the behavior of both keratinocytes and dermal fibroblasts in vitro. Our previous work on keratinocytes and human and murine skin wound healing demonstrates that ␤2-AR activation is detrimental to skin re-epithelialization (Pullar et al., 2003;Pullar and Isseroff, 2005;Pullar et al., 2006) and the results presented here suggest that ␤2-AR activation on dermal fibroblasts may also contribute to a ␤2-AR-mediated delay in skin wound repair. The ability of ␤2-AR activation to delay fibroblast-mediated collagen gel contraction may translate into decreased wound contraction in vivo, which could be detrimental to wound healing as dermal fibroblasts form the granulation tissue in the wound generating the mechanical forces that initiate contraction to decrease wound size (Gabbiani et al., 1972). Additionally, the ␤-AR agonist-mediated increase in fibroblast migration rate and proliferation could result in the accumulation of abnormally large numbers of fibroblasts in the wound, which may also be deleterious to the wound repair process, resulting in unwanted fibrosis and scarring (Redden and Doolin, 2003). It would, however, be premature to conclude that the ␤-AR-mediated effects we observe in dermal fibroblasts in vitro would contribute to the attenuation of wound healing in vivo and experiments are presently underway in our lab to address these questions.
What evidence is there that endogenous cutaneous catecholamines could potentially regulate wound repair? Catecholamines form a critical component of the body's response to stress (Nankova and Sabban, 1999), that can have a deleterious effect on wound healing (Detillion et al., 2004). Surgical stress can increase post-operative plasma levels of catecholamines and cortisol, the major stress-induced hormones and cortisol has long been correlated with impaired human skin wound healing (Ebrecht et al., 2004). Normal circulating levels of epinephrine are reported to be 0.3-0.8 nmol/l in human plasma (Sedowofia et al., 1998) but increase by greater than tenfold (3-12 nmol/l) within the first 6 hours following injury (Crum et al., 1990;Matsui et al., 1991;Sedowofia et al., 1998). Since, we observed pro-motogenic, anti-contractile and pro-mitogenic effects in the nanomolar range in vitro and as this corresponds to the circulating levels seen in post trauma, our results may indeed be physiologically relevant. It is also important to note that catecholamines are rapidly metabolized by the liver (Martel, 1998), therefore, we would hypothesize that levels of hormone in the blood or plasma may not reflect the concentrations of catecholamines synthesized locally by the epidermis at sites of injury/stress, which could be higher. Additionally, topical application of ␤2-AR agonists impaired human and murine wound re-epithelialization (Pullar et al., 2006) and a ␤2-AR antagonist improved barrier recovery, as evaluated by measuring transepidermal water loss (Denda et al., 2003). The current finding that ␤2-AR activation also regulates dermal fibroblast migration, proliferation and contractile ability, processes that are all critically required for wound repair, now provides mechanistic support for the regulatory role of the catecholamine hormonal network in the repair process. Further investigation of this regulatory pathway will improve our understanding of the wound healing process.
To summarize, we have identified novel, divergent, ␤2-ARmediated pro-motogenic and pro-mitogenic mechanisms in dermal fibroblasts. The pro-motogenic pathway was Src dependent, while, the pro-mitogenic pathway, the attenuation of collagen gel contraction and alterations in cytoskeletal conformation were all cAMP/PKA dependent. Our work uncovers a previously unrecognized role for the adrenergic hormonal mediator network in cutaneous wound repair and provides tantalizing information to prompt further study of ␤2-AR modulation of the wound healing process.

Materials for cell treatments
The ␤-AR agonist, isoproterenol, inhibitors, AG 1478, PP2 and the cAMP analogs, rp-cAMP and sp-cAMP, were purchased from Calbiochem (San Diego, CA).

Dermal fibroblast growth
Human dermal fibroblasts (NHF) were isolated from neonatal foreskins obtained by routine circumcision under an approved exemption from the University of California, Davis, Institutional Review Board, as previously described (Isseroff et al., 1987). At least three fibroblast strains, between passages 3 and 7, were used in all experiments. Stock cultures were maintained as monolayers in plastic cell culture dishes from Falcon Labware (BD Biosciences, San Jose, CA) using fibroblast growth medium [FM: Dulbecco's modified Eagle's medium (DMEM, basal medium), 1% antibiotic solution from Gibco (Grand Island, NY), and 10% calf serum (Tissue Culture Biologicals, Tulare, CA)]. The cultures were incubated at 37°C in a humidified atmosphere of 5% CO 2 .

Single cell migration assay
All single cell migration assays were performed on cells plated on glass coverslips (Eppendorf, Hamburg, Germany) that had been coated for 1 hour at 37°C with 60 g/ml collagen I (Cohesion Technologies, Palo Alto, CA). Cells were plated onto the collagen-coated glass coverslips in FM at a density of 125 cells/mm 2 for 3-6 hours at 37°C. Cells were either untreated or pre-treated with PP2 (10 M) for 6 hours. ␤-AR agonist (10 nM-100 M) and PP2 were added to the FM at time 0 if required. The coverslips formed the bottom of a migration chamber to monitor individual cell migration over a 1-hour period at 37°C, as described previously (Pullar et al., 2003). The migration chamber was placed on an inverted Nikon Diaphot microscope. Time-lapse images of the cell migratory response were digitally captured every 10 minutes over a 1-hour period by Q-Imaging Retiga-EX cameras (Burnaby, BC, Canada) controlled by a custom automation written in Improvision Open Lab software (Lexington, MA) on a Macintosh G4. After the center of mass of each cell was tracked using the Open Lab software, migration speed and distance were calculated and imported to Excel (Microsoft Corporation, Redmond, WA). 'Distance' is the average total distance in m that the cells travel in a one-hour period of time. 'Speed' is the average speed in m/min that the cells travel in a 1-hour period. Statistical significance between the means of two cell populations was calculated using Student's t-test (unpaired) with P<0.01.

Cell treatments for immunoblotting and immunoprecipitation experiments
Cells were starved in basal medium (DMEM) for 16 hours. 1-2ϫ10 6 cells were preincubated with either DMEM alone or DMEM containing either 10 M AG1478 for 90 minutes at 37°C or 10 M PP2 for 6 hours at 37°C. Cell treatments did not affect cell viability. Untreated cells were then stimulated with DMEM containing 1 M ␤-AR agonist alone or DMEM alone (control) for 5-60 minutes at 37°C, unless otherwise noted. Inhibitor-treated cells were then stimulated with DMEM containing inhibitor and 1 M ␤-AR agonist or inhibitor and DMEM alone (control) for 5-60 minutes at 37°C, unless otherwise noted. Cells were placed immediately on ice, washed twice with ice-cold phosphate-buffered saline (PBS) containing phosphatase inhibitors (50 mM NaF and 1 mM Na 3 VO 4 ) and scraped in 50 l-1 ml lysis buffer (PBS containing 0.5% Triton X-100, 50 mM NaF, 1 mM Na 3 VO 4 , leupeptin 10 g/ml, aprotinin 30 g/ml, PMSF 200 g/ml, pepstatin A 10 g/ml). The lysates were transferred into 1.5 ml tubes, incubated on ice for 20 minutes and then centrifuged at 14,000 g for 10 minutes at 4°C. The protein concentration of the samples was determined using the Bradford Assay (Bio-Rad Laboratories, Hercules, CA).

Immunoprecipitation
Anti-EGFR antibody (1005; 5 g) (Santa Cruz Biotechnology, Santa Cruz, CA) linked to 30 l of pre-washed protein A/G Sepharose beads (Amersham Pharmacia, Piscataway, NJ) was used to immunoprecipitate the desired proteins from 1 ml of lysate prepared from 1-2ϫ10 7 cells, either untreated or pre-treated with 1 M ␤-AR agonist for 10 minutes, as described in the cell treatments section. Lysates were initially pre-cleared with 150 l of pre-washed beads for 30 minutes at room temperature and then incubated with the antibody-bound beads at 4°C overnight on a rotary mixer. The beads were washed five times with lysis buffer, 1ϫ reducing sample loading buffer was added (Bio-Rad, Hercules, CA), the samples were boiled for 3 minutes and centrifuged to pellet the beads. The supernatants were loaded onto two 10% polyacrylamide Tris-HCl gels (Bio-Rad, Hercules, CA) and the proteins were separated electrophoretically followed by transfer to Immobilon membrane for immunoblotting with either an anti-phosphotyrosine antibody (Ab-4) (Oncogene, Boston, MA) or an anti-EGFR antibody (1005; Santa Cruz Biotechnology, Santa Cruz, CA), as outlined above.

Proliferation assay
Dermal fibroblasts were released from the tissue culture plate by treatment with 0.25% trypsin/EDTA (Gibco, Grand Island, NY), resuspended in FM and counted using a haemocytometer. Cells were either untreated or pre-treated with 50 M sp-cAMP or 50 M rp-cAMP for 30 minutes prior to ␤-AR agonist addition in FM. 5ϫ10 4 cells were plated per well in a 12-well plate in triplicate in FM in the presence or absence of 10 nM-10 M ␤-AR agonist, 50 M sp-cAMP or 50 M rp-cAMP. Triplicate wells were harvested and counted on days 2, 4, 6 and 8. The medium was changed every day.

Collagen gel contraction assay
A solution of bovine collagen types I (97%) and III (3 mg collagen/ml Vitrogen 100; Collagen Corp., Palo Alto, CA) was mixed with triple strength DMEM, containing 20 mM Hepes buffer (Gibco, Grand Island, NY) to maintain neutral pH, calf serum (Tissue Culture Biologicals, Tulare, CA), cells (detached by trypsin from monolayer confluent cultures) and ␤-adrenergic receptor agonists or cAMP analogs if required. The individual solutions were prepared and cooled to 4°C prior to mixing to prevent premature gelation. The final solution contained, by volume: 40% Vitrogen, 20% 3ϫ DMEM, 30% DMEM with Hepes buffer, 10% calf serum. Cells were untreated or pre-incubated with either 50 M sp-cAMP or 50 M rp-cAMP for 30 minutes at 37°C and added to the collagen gel mix, in the presence or absence of 10 nM-10 M ␤-AR agonist, 50 M sp-cAMP or 50 M rp-cAMP, just prior to gel casting at the concentration of 20,000 cells per ml. The lattices were cast, with 2 ml of the final solution per dish, in 35 mm bacteriologic dishes (Falcon Labware, BD Biosciences, San Jose, CA), which fibroblasts adhere poorly to. The mixture gelled within 30 minutes upon incubation at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . To assure even contraction, lattices were detached from the sides of the dishes after 2 hours by rimming the edges of the dishes using a sterile 100 l tip and gently shaking the dishes until the gels slid freely. Lattice retraction was measured every day by placing the dishes over a flat ruler on a black background. After maximum retraction the lattices were digested with collagenase I (1,000 U/ml; Worthington Biochemicals, Freehold, NJ) for 30 minutes at 37°C for assessment of cell number and viability by Trypan Blue exclusion. Statistical analysis was performed using the two-tailed Student's t-test packaged with