CRF1 Receptor Signaling via the ERK1/2-MAP and Akt Kinase Cascades: Roles of Src, EGF Receptor, and PI3-Kinase Mechanisms

In the present study, we determined the cellular regulators of ERK1/2 and Akt signaling pathways in response to human CRF1 receptor (CRF1R) activation in transfected COS-7 cells. We found that Pertussis Toxin (PTX) treatment or sequestering Gβγ reduced CRF1R-mediated activation of ERK1/2, suggesting the involvement of a Gi-linked cascade. Neither Gs/PKA nor Gq/PKC were associated with ERK1/2 activation. Besides, CRF induced EGF receptor (EGFR) phosphorylation at Tyr1068, and selective inhibition of EGFR kinase activity by AG1478 strongly inhibited the CRF1R-mediated phosphorylation of ERK1/2, indicating the participation of EGFR transactivation. Furthermore, CRF-induced ERK1/2 phosphorylation was not altered by pretreatment with batimastat, GM6001, or an HB-EGF antibody indicating that metalloproteinase processing of HB-EGF ligands is not required for the CRF-mediated EGFR transactivation. We also observed that CRF induced Src and PYK2 phosphorylation in a Gβγ-dependent manner. Additionally, using the specific Src kinase inhibitor PP2 and the dominant-negative-SrcYF-KM, it was revealed that CRF-stimulated ERK1/2 phosphorylation depends on Src activation. PP2 also blocked the effect of CRF on Src and EGFR (Tyr845) phosphorylation, further demonstrating the centrality of Src. We identified the formation of a protein complex consisting of CRF1R, Src, and EGFR facilitates EGFR transactivation and CRF1R-mediated signaling. CRF stimulated Akt phosphorylation, which was dependent on Gi/βγ subunits, and Src activation, however, was only slightly dependent on EGFR transactivation. Moreover, PI3K inhibitors were able to inhibit not only the CRF-induced phosphorylation of Akt, as expected, but also ERK1/2 activation by CRF suggesting a PI3K dependency in the CRF1R ERK signaling. Finally, CRF-stimulated ERK1/2 activation was similar in the wild-type CRF1R and the phosphorylation-deficient CRF1R-Δ386 mutant, which has impaired agonist-dependent β-arrestin-2 recruitment; however, this situation may have resulted from the low β-arrestin expression in the COS-7 cells. When β-arrestin-2 was overexpressed in COS-7 cells, CRF-stimulated ERK1/2 phosphorylation was markedly upregulated. These findings indicate that on the base of a constitutive CRF1R/EGFR interaction, the Gi/βγ subunits upstream activation of Src, PYK2, PI3K, and transactivation of the EGFR are required for CRF1R signaling via the ERK1/2-MAP kinase pathway. In contrast, Akt activation via CRF1R is mediated by the Src/PI3K pathway with little contribution of EGFR transactivation.

In the present study, we determined the cellular regulators of ERK1/2 and Akt signaling pathways in response to human CRF 1 receptor (CRF 1 R) activation in transfected COS-7 cells. We found that Pertussis Toxin (PTX) treatment or sequestering Gβγ reduced CRF 1 R-mediated activation of ERK1/2, suggesting the involvement of a G i -linked cascade. Neither G s /PKA nor G q /PKC were associated with ERK1/2 activation. Besides, CRF induced EGF receptor (EGFR) phosphorylation at Tyr 1068 , and selective inhibition of EGFR kinase activity by AG1478 strongly inhibited the CRF 1 R-mediated phosphorylation of ERK1/2, indicating the participation of EGFR transactivation. Furthermore, CRF-induced ERK1/2 phosphorylation was not altered by pretreatment with batimastat, GM6001, or an HB-EGF antibody indicating that metalloproteinase processing of HB-EGF ligands is not required for the CRF-mediated EGFR transactivation. We also observed that CRF induced Src and PYK2 phosphorylation in a Gβγ-dependent manner. Additionally, using the specific Src kinase inhibitor PP2 and the dominant-negative-SrcYF-KM, it was revealed that CRF-stimulated ERK1/2 phosphorylation depends on Src activation. PP2 also blocked the effect of CRF on Src and EGFR (Tyr 845 ) phosphorylation, further demonstrating the centrality of Src. We identified the formation of a protein complex consisting of CRF 1 R, Src, and EGFR facilitates EGFR transactivation and CRF 1 R-mediated signaling. CRF stimulated Akt phosphorylation, which was dependent on G i /βγ subunits, and Src activation, however, was only slightly dependent on EGFR transactivation. Moreover, PI3K inhibitors were able to inhibit not only the CRF-induced phosphorylation of Akt, as expected, but also ERK1/2 activation by CRF suggesting a PI3K dependency in the CRF 1 R ERK signaling. Finally, CRF-stimulated ERK1/2 activation was similar in the wild-type CRF 1 R and the phosphorylation-deficient CRF 1 R-386 mutant, which has impaired agonist-dependent β-arrestin-2 recruitment; however, this situation INTRODUCTION Behavioral, cognitive, neuroendocrine, and autonomic responses to stress are regulated by CRF 1 and CRF 2 receptors (CRF 1 R and CRF 2 R) (1)(2)(3). The preferred mode of signal transduction by both CRF receptors was initially believed to be activation of the G s /adenylyl cyclase/PKA signaling pathway (1)(2)(3). Subsequently, CRF 1 R and CRF 2 R were also found to signal via the PLC/PKC cascade stimulating intracellular calcium mobilization and IP3 formation (1)(2)(3)(4). Besides, both CRF receptors can activate mitogen-activated protein (MAP) kinase cascades in neuronal, cardiac, and myometrial cells endogenously expressing CRF 1 R or CRF 2 R and in recombinant cell lines expressing either receptor (2,3,5,6). Several reports suggested that cellular background directed CRF 1 R to signal selectively via a specific MAP kinase pathway. For example, agonist-activated CRF 1 Rs stimulated phosphorylation of ERK1/2 and p38 MAP kinases in PC12 and fetal microglial cells (7,8) while CRF 1 Rs activated ERK1/2 but not JNK and p38 in CHO cells (9). In human mast cells and HaCaT keratinocytes, on the other hand, CRF 1 Rs induce phosphorylation of p38 but not ERK or JNK MAP kinases (10,11). Most studies suggest, however, that the ERK1/2 cascade is the MAP kinase pathway preferentially used by CRF receptors (5,9,12,13).
In addition, activation of CRF 1 R or CRF 2(b) R can stimulate phosphorylation of Akt (5,21). CRF 2(b) R Akt signaling in HEK293 cells is mediated by pertussis-sensitive G proteins and PI3K but not by cAMP-stimulated activation of PKA or EPAC, or by PKC (21). The mechanisms regulating Akt signal transduction by CRF 1 R, however, have not been investigated. Because upstream kinase pathway mediation of CRF 1 R signal transduction via the ERK and Akt cascades are not wellunderstood, the primary goal of this study was to test the hypothesis that Src tyrosine kinase and EGFR transactivation are essential regulators of these CRF 1 R signaling pathways. We also sought to determine the relative importance of G protein βγ subunits, second messenger kinases, and PI3K in the activation of the ERK1/2 and Akt cascades by the CRF 1 R. The results of our study indicate that upstream utilization of Src and PI3K are involved in ERK and Akt signal transduction by the agonistactivated CRF 1 R in COS-7 cells, without mediation by PKA and PKC, while transactivation of the EGFR is mainly required for CRF 1 R to stimulate phosphorylation of ERK but not for Akt activation.

Cell Culture and Transfection
COS-7 cells (from the American Type Culture Collection) were cultured at 37 • C in a humidified atmosphere of 95% air, 5% CO 2 , in DMEM supplemented with 10% FBS, 100 µg/ml streptomycin, and 100 units/ml penicillin (COS-7 growth medium). Transient transfections were performed using LipofectAMINE (Life Technologies: Gaithersburg, MD) as described previously (27). Cells were seeded at 8 × 10 5 cells/10-cm dish in COS-7 medium and cultured for 3 days before transfection. COS-7 cells were transfected in 5 ml/dish OptiMEM containing 10 µg/ml LipofectAMINE with empty vector, pcDNA3 encoding the HA-CRF 1 R or the HA-CRF 1 R-386 mutant. In certain experiments, cells were co-transfected with plasmids containing: HA-CRF 1 R and mock (empty vector); HA-CRF 1 R and ct-βARK; HA-CRF 1 R and dn-Src; HA-CRF 1 R and dn-Akt, or HA-CRF 1 R and fulllength β-arrestin-2. After replacing the transfection medium with fresh growth medium, transfected COS-7 cells were cultured for 1 day. Subsequently, cells were re-seeded in 6-well plates and cultured for an additional day prior to the experiment.

Western Blot Methods
The protocols for measuring total and phosphorylated ERK1/2, c-Src, PYK2, Akt, and EGFR have been previously published (28,29). After cells were cultured to 60-70% confluence, they were serum-deprived for 24 h. On the day of the experiment, cells were treated with the indicated ligands and inhibitors. No significant changes in the basal level of ERK1/2 or Akt phosphorylation were observed in cells pretreated with inhibitors, except for BIM, which showed a small increase in ERK1/2 activation (Supplementary Figure S1). After treatment, cells were placed on ice, the media was aspirated, and the cells were washed twice with ice-cold PBS and lysed in 100 µl of Laemmli sample buffer 1X. The lysates were briefly sonicated, heated at 95 • C for 5 min, and centrifuged for 5 min at 14,000 rpm. Resulting supernatants were loaded in separate lanes of a 10% SDS-PAGE gels and electrophoresed. Next, Western transfer on to PDVF membranes was completed. The Western blots were then probed with specific antibodies targeting phosphorylated and non-phosphorylated forms of ERK1/2, c-Src, PYK2, Akt, and EGFR for primary immunodetection. After blots were probed with horseradish peroxidase-conjugated secondary antibody, protein bands were visualized with enhanced chemiluminescence ECL reagent (Amersham Pharmacia Biotech, Piscataway, NJ or Pierce Biotechnology, Rockford, IL) and scanned using the GS-800 Calibrated Imaging Densitometer (Bio-Rad). The labeled bands were quantified using the Quantity One 4.6.3 software program (Bio-Rad).

Co-immunoprecipitation Assay
COS-7 cells transfected with HA-CRF 1 R were grown in 10-cm dishes and serum-deprived for 24 h before treatment with 100 nM CRF for 10 min at 37 • C. Cells were washed twice with ice-cold PBS and lysed in Nonidet P-40 solubilization buffer (50 mM Tris-HCl, 150 mM NaCl, 2 mM Orthovanadate, 1 mM NaF, 1% Nonidet P-40, 10% Glycerol, 2 mM EDTA, pH 7.4, containing protease inhibitors). After immunoprecipitation of HA-CRF 1 Rs with anti-HA monoclonal antibody (HA.11; Covance, San Diego, CA) and protein A/G PLUS-Agarose (Santa Cruz Biotechnology, CA), the proteins were resolved by SDS-PAGE, Western blotted, and probed with anti-EGFR polyclonal or anti-HA monoclonal antibodies, followed by a horseradish peroxidase conjugate to identify co-immunoprecipitated proteins. Blots were also stripped with stripping buffer (100 nM Glycine-HCl, pH 2.7) and reprobed with anti-c-Src polyclonal antibody. Western blot detection of co-immunoprecipitated Src was carried out as described above. Blots were visualized and quantified, as indicated above.

Statistical Analysis
Data are presented as mean ± S.E.M. Analyses of variances (ANOVAs) across experimental groups were performed using PRISM TM , Version 8.0 for macOS (GraphPad Software, Inc., San Diego, CA). If the one-way ANOVA was statistically significant, planned post-hoc analyses were performed using Dunnet or Bonferroni's multiple comparison tests to determine individual group differences.

CRF-Induced ERK1/2 Phosphorylation Is Dependent on G i Protein and the Gβγ Subunits
CRF treatment (100 nM) of COS-7 cells transiently transfected with HA-CRF 1 R caused transient phosphorylation of ERK1/2 that reached a peak at 5-10 min and declined thereafter toward the basal level over the next 30 min ( Figure 1A). CRF (100 nM) also caused time-dependent phosphorylation of ERK1/2 in CRF 1 R-expressing HEK293 and CHO-K1 cells (data not shown), but the rate and magnitude of CRF-induced ERK1/2 activation was considerably less in these cell lines compared to COS-7 cells. In contrast, CRF 1 Rs expressed in SK-N-MC neuroblastoma cells (4) failed to signal via the ERK1/2 cascade while fibroblast growth factor induced strong ERK1/2 phosphorylation in this cell line (Supplementary Figure S2). Therefore, all subsequent experiments studying ERK1/2 signaling were performed in COS-7 cells transfected with HA-CRF 1 R cDNA. CRF-induced ERK1/2 activation was concentrationdependent, with a significant increase at 10 nM CRF (∼2.4fold increase over control) and maximal effect over the 0.1-1 µM range (∼5.2-fold increase over control, Figure 1B). The EC 50 was 25 nM and the maximum occurred at 100 nM for the CRF-induced ERK1/2 phosphorylation.
Most of the known actions of CRF 1 Rs are mediated through the G s /PKA signaling cascade, but some of the physiological actions of CRF are also known to occur through activation of G q or G i proteins (30). To determine the contributions of G s /PKA-dependent mechanisms to MAP kinase activation, COS-7 cells were pretreated with the PKA inhibitor H89 (500 nM) for 30 min prior to stimulation with CRF (100 nM). As shown in Figure 2A, the PKA inhibitor failed to inhibit CRF-stimulated ERK1/2 phosphorylation. Furthermore, the magnitude of CRF 1 R-mediated activation of ERK1/2 was similar in COS-7 cells pretreated for 30 min with the highly selective PKA inhibitor Rp-cAMP (0-100 µM) or vehicle (Supplementary Figure S3). We next explored the involvement of G q /PKC in CRF 1 R ERK signaling. A 30-min pretreatment of COS-7 cells with the PKC inhibitor BIM (1 µM), increased (∼1.6-fold increase over CRF stimulation) rather than decreased CRF-stimulated ERK1/2 activation ( Figure 2B). In contrast, BIM pretreatment inhibited ERK1/2 phosphorylation resulting from PMA-induced PKC activation ( Figure 2B). Thus, our data suggest that neither G s /PKA nor G q /PKC are required for the CRF 1 R-mediated ERK1/2 signaling in COS-7 cells.
On the other hand, the release of Gβγ subunits during GPCR coupling to G protein, particularly through G i , has an important role in downstream signaling in the ERK1/2 cascade (31,32). Thus, we examined the role of G i and Gβγ in CRF-stimulated ERK1/2 activation by two different experimental approaches: treatment with the G i protein inhibitor pertussis toxin (PTX) and by co-transfecting COS-7 cells with plasmids encoding the carboxyl terminus of βARK containing its βγ-binding domain (ct-βARK) (Supplementary Figure S4), which sequesters βγ, and the CRF 1 R. COS-7 cells expressing CRF 1 Rs pretreated with 100 ng/ml PTX showed a marked reduction in CRFinduced ERK1/2 phosphorylation (Figure 2C), suggesting the coupling of CRF 1 R to G i to mediate ERK activation. Moreover, overexpressing ct-βARK in COS-7 cells co-expressing CRF 1 Rs significantly reduced CRF-induced ERK1/2 phosphorylation by ∼40% ( Figure 2D). Thus, our data implicate Gβγ subunits from PTX-sensitive heterotrimeric G proteins in CRF 1 R-mediated activation of ERK1/2.

Transactivation of the EGFR During CRF 1 R ERK1/2 Signaling
Because transactivation of receptor tyrosine kinases (RTKs), especially the EGFR, is often an important mechanism used by GPCRs to activate ERK1/2 (33, 34), we investigated the role of EGFR transactivation in CRF 1 R-mediated ERK signaling. In COS-7 cells transiently transfected with HA-CRF 1 R, EGF stimulation of the endogenous EGFRs caused ERK1/2 phosphorylation in a time-dependent manner, reaching a maximum effect after 5 min of stimulation, which persisted for at least 30 min (∼6.0-fold increase over time 0, Figure 3A). ERK1/2 phosphorylation was also increased by EGF (0-100 ng/ml) in a concentration-dependent manner (EC 50 = 0.23 ng/ml, Figure 3B). Thus, these results are consistent with the wellestablished role of EGFR in ERK1/2 signaling (35).
When COS-7 cells expressing HA-CRF 1 R were pretreated with the EGFR tyrosine kinase inhibitor AG1478 (100 nM, 30 min), a significant inhibition (∼80%) of CRF-induced maximal ERK1/2 phosphorylation was observed ( Figure 3C). A concentration-dependent inhibition was observed with AG1478 concentrations of 0-1,000 nM with an IC 50 of 10 nM ( Figure 3D). Importantly, phosphorylation of the EGFR at Tyr 1068 was detected with Western blots in COS-7 cells beginning at 2 min and becoming maximal at 5-10 min of CRF exposure (100 nM) ( Figure 3E). Tyr 1173 of the EGFR was phosphorylated  Figure S5). Together, these results indicate that CRF-activated CRF 1 R triggers phosphorylation of two critical amino acids located within the autophosphorylation loop that are required for EGFR activation (36,37). Thus, CRF 1 R signaling rapidly transactivates the EGFR, in agreement with a study reporting that Ucn1 stimulated EGFR transactivation in CRF 1 Rexpressing HEK293 cells (5).
Altogether, these results exclude a role of MMP in CRF-induced transactivation of the EGFR and subsequent phosphorylation of ERK1/2 in COS-7 cells.

Src Mediation of CRF 1 R ERK1/2 Signaling
We then investigated the role of Src kinase, which can serve as an important upstream regulator of GPCR signaling via the ERK1/2 cascade (42,43). Importantly, 100 nM CRF caused marked phosphorylation of Src at Tyr 416 , which is a requirement for Src activation (44), reaching a maximum at 10 min (∼3.5-fold increase over time 0), and persisting for more than 30 min ( Figure 5A). This activation was dependent on Gβγ release since ct-βARK expression reduced the CRF-induced Src phosphorylation (Figure 5B), and as expected, CRF-induced Src phosphorylation was prevented by pretreatment with the selective Src family kinase inhibitor PP2 ( Figure 9B). To further evaluate the role of Src in CRF 1 R ERK1/2 signaling, COS-7 cells were co-transfected with the CRF 1 R and a dn-Src. Overexpression of inactive Src prevented ERK1/2 activation by CRF ( Figure 5C). Other experiments demonstrated that PP2 pretreatment abolished CRF-stimulated ERK1/2 phosphorylation (Figure 5D), in a concentrationdependent manner (0-20 µM, IC 50 = 2 µM) ( Figure 5E). These findings support our hypothesis that Src plays a central role in CRF 1 R ERK1/2 signaling.
We next determined if CRF-stimulated Src activation is required for CRF 1 R-induced transactivation of EGFRs. In this context, previous research has established that Src can activate EGFR signaling by phosphorylating Tyr 845 of the EGFR protein (45,46). As shown in Figure 6A, we found that 100 nM CRF stimulated in a time-dependent manner marked phosphorylation of EGFR at Tyr 845 beginning at 2 min and becoming maximal at 10 min. This effect was blocked by pretreatment of the cells with PP2 (Figure 6B). In a recent study by Perkovska et al. (47), it was shown that V 1b vasopressin receptor interacts with Src at basal state, suggesting the formation of a GPCR/Src complex that facilitates MAP kinase activation. To evaluate if a CRF 1 R/Src complex exists under basal conditions, we analyzed CRF 1 R immunoprecipitates for the presence of coprecipitated Src under basal and CRF-stimulated conditions. As shown in Figure 6C, 100 nM CRF induced a robust interaction between the CRF 1 R and Src after 10 min stimulation (∼8.0fold increase over control). Interestingly, it was also observed that under the same immunoprecipitation conditions, the EGFR is also present in the CRF 1 R/co-precipitated complex, even in the absence of CRF stimulation ( Figure 6D). After stimulation with 100 nM CRF for 10 min, we observed a significant increase in the CRF 1 R/EGFR interaction (∼2.5-fold increase over control). These observations suggest that CRF promotes the formation of a multiprotein complex that would allow rapid EGFR phosphorylation at Tyr 845 by Src, present in this complex.
It has been shown that, in parallel to Src activation by many GPCRs, the proline-rich tyrosine kinase 2, PYK2, is also phosphorylated and activated, and in association with  Src is required for the subsequent transactivation of EGFR (44,48). Therefore, we decided to assess whether activation of CRF 1 R leads to PYK2 phosphorylation in COS-7 cells. As shown in Figure 6E, 100 nM CRF caused rapid phosphorylation of PYK2 in a time-dependent manner (0-30 min), reaching a maximum effect at 5 min and persisting for at least 30 min of stimulation. Interestingly, and as expected, CRF-mediated PYK2 phosphorylation was also dependent on Gβγ release ( Figure 5B).

PI3K Mediation of CRF 1 R ERK1/2 and Akt Signaling
PI3Ks can mediate important biological actions of GPCRs, including cell proliferation or survival, by serving as an upstream regulator of Akt and ERK cascades (49,50). As shown in Figure 7A, 100 nM CRF caused rapid phosphorylation of Akt, an effect that was decreased by PTX pretreatment (Figure 7B) or ct-βARK overexpression (Figure 7C), similar to the previously observed effect on the CRF-induced ERK1/2 phosphorylation, suggesting the participation of G i protein and Gβγ subunits in this process. It is important to note that none of the observed effects of PTX and ct-βARK on CRF actions were present on EGF-stimulated ERK1/2 and Akt phosphorylation (Supplementary Figure S7). Pretreatment with selective PI3K inhibitors, wortmannin (100 nM) (Figure 8A), or LY294002 (10 µM) (Figure 9C) abolished CRF 1 R-mediated Akt signaling activation. Similarly, inhibition of PI3K by 100 nM wortmannin abolished the stimulatory action of EGF on Akt (Figure 8A), thereby demonstrating that the PI3K pathway is required for both CRF-and EGF-induced Akt phosphorylation. Considering that an upstream PI3K mechanism can also regulate CRF 1 R and CRF 2 R signaling via the ERK1/2 cascade in A7r5, CATH.a, and transfected CHO cells (9, 12), we investigated the potential role of PI3K in the activation of ERK1/2 by HA-CRF 1 Rs expressed in COS-7 cells. In this context, activation of RTKs, such as the EGFRs, has been shown to recruit PI3K and activate ERK1/2 (50-53). However, contradictory data on PI3K involvement in EGFR-induced ERK1/2 phosphorylation have been reported (54)(55)(56). In this regard, to find out if EGFmediated ERK1/2 phosphorylation observed in COS-7 cells is depending on PI3K activation, we analyze the effect of 100 nM wortmannin on the EGF ERK1/2 activation. As shown in Figure 8A, pretreatment with wortmannin was unable to inhibit the effect of EGF, suggesting that PI3K does not participate in this mechanism. In contrast, pretreatment with wortmannin abolished CRF-stimulated ERK1/2 phosphorylation ( Figure 8A) in a concentration-dependent manner (0-100 nM), confirming an intermediary role for PI3K in CRF 1 R ERK signaling ( Figure 8B). To examine the contribution of CRFmediated activation of Akt to the phosphorylation of ERK1/2, we evaluated the effect of the dn-Akt mutant. As shown in Figure 8C, overexpression of dn-Akt had no significant effect on ERK1/2 activation after stimulation with CRF ( Figure 8C), suggesting that Akt does not participate in the activation of ERK1/2 by CRF. Because in the present work we do not show evidence about impairment of kinase activity of the dn-Akt, it will be necessary the use of other approaches, such as genetic tools or inhibitors, to provide more evidence regarding the possible no effect of Akt on the ERK 1/2 pathway. Because in the present work we do not show evidence about impairment of kinase activity of the dn-Akt, it will be necessary the use of other approaches, such as genetic tools or inhibitors, to provide more evidence regarding the possible Akt lack of effect on ERK 1/2 pathway. Consequently, our results suggest that PI3K can regulate the transduction of CRF 1 R signals through the ERK cascade, possibly independently of Akt.

Src Acts Upstream and PI3K Downstream of the EGFR During CRF-Induced ERK1/2 Activation
Since we found that CRF-induced EGFR transactivation mediates ERK1/2 phosphorylation through Src-and PI3Kdependent mechanisms, we next determined if CRF-induced PI3K activation occurs upstream or downstream of Src and EGFR. Wortmannin inhibition of PI3K had no effect on CRF-induced phosphorylation of EGFR at Tyr 845 (Figure 9A), suggesting that PI3K acts downstream of Src and EGFR. Consistent with these results, we also observed that wortmannin pretreatment did not alter CRF-induced phosphorylation of Src at Tyr 416 ( Figure 9B). Therefore, CRF 1 R-stimulated transactivation of EGFR and phosphorylation of ERK1/2 mediated by Src was independent of PI3K. It has been reported that the PI3K/Akt signaling pathway can be activated at least by two independent mechanisms: (i) EGFR transactivation (57), and (ii) upstream Src activation (58,59). We observed that CRF-induced Akt activation was completely inhibited by the Src inhibitor, PP2 (Figure 9C), suggesting that Src is an upstream regulator of PI3K and Akt. We next measured the effect of the specific EGFR tyrosine kinase inhibitor, tyrphostin AG1478, on CRF-stimulated Akt phosphorylation. As observed in Figure 9D, while CRF-induced ERK1/2 activation was totally dependent on EGFR transactivation (Figure 9D), Akt phosphorylation was only partially dependent. Thus, we hypothesize that PI3K/Akt pathway signaling by CRF 1 R may involve two mechanisms: (i) a strong dependence on upstream Src activating PI3K and then Akt ( Figure 9C); (ii) a weak dependence on EGFR transactivation ( Figure 9D).

Role of β-Arrestin-2 in the CRF-Mediated ERK1/2 Activation
In recent years, it has been identified that β-arrestin proteins play an important role in mediating the actions of GPCRs, particularly those related to activation and regulation of Src and mitogenic pathways, in particular, the ERK1/2 signaling cascade (60). To determine the role of β-arrestins in the CRF-mediated ERK1/2 activation observed above, we used a phosphorylation-deficient mutant CRF 1 R, which also shows a diminished agonist-dependent β-arrestin-2 recruitment (24). As shown in Figure 10A, COS-7 cells transiently transfected with HA-CRF 1 R-386 mutant showed a similar response in ERK1/2 phosphorylation compared to that observed with CRF 1 R. The apparent independence of CRF-mediated activation of ERK1/2 from β-arrestin-2 could be explained by the low β-arrestin expression level  previously detected in COS-7 cells (61,62). To assess this possibility, we evaluated the effect of β-arrestin-2 overexpression in COS-7 cells, since CRF 1 R activation has been shown to lead to selective recruitment of βarrestin-2 in both HEK293 cells and neurons (24,63). As observed in Figure 10B, cells co-expressing HA-CRF 1 R and β-arrestin-2 showed a significant increase in the CRFmediated ERK1/2 phosphorylation, suggesting that β-arrestin involvement in CRF 1 R ERK1/2 signaling depends on its cellular expression levels.

DISCUSSION
In the present study, we investigated the molecular mechanisms associated with the activation of ERK1/2 and Akt signaling cascades by the human CRF 1 R in COS-7 cells. Our data suggest that agonist-stimulated CRF 1 R promotes G i activation and Gβγ release which, in turn, stimulate phosphorylation and activation of Src kinase. Once Src is active, it mediates ERK1/2 phosphorylation by at least two independent signaling mechanisms: (i) phosphorylation and transactivation of the EGFR, (ii) activation of PI3K. Interestingly, CRF 1 R-induced Akt phosphorylation also requires Src-mediated activation of PI3K as the main mechanism, but it is mostly independent of EGFR transactivation. Defining the molecular mechanisms for ERK1/2 signaling by a GPCR has become a significant focus of signal transduction research due to the multifaceted pathways mediating signaling via the ERK1/2-MAP kinase cascade. A significant role of the ERK1/2-MAP kinase pathway has been recognized in the biological action of both CRF 1 R and CRF 2 R. ERK1/2 is widely distributed in the brain and is considered an essential regulator of the molecular processes involved in response to stress (6,64). It is well-established that most GPCRs signal via ERK1/2-MAP kinase cascades through distinct G i -, G s -, and G q -dependent signaling pathways. In the case of the CRF 1 R, it has been identified that the G s /PKA pathway is importantly involved in the activation of MAP kinase cascades (12,15,18). In contrast, we found that pretreating CRF 1 R-expressing COS-7 cells with PKA inhibitors H89 or Rp-cAMP did not alter the ability of CRF to stimulate ERK1/2 phosphorylation. Although earlier research proposed that high cellular expression of the serine-threonine kinase B-Raf molecularly switches "upstream" ERK1/2 activation by G s -coupled GPCRs to a PKA mechanism (14), pretreating fetal hippocampal cells with the PKA inhibitor H89 only produced a small reduction in CRF 1 R-mediated ERK phosphorylation despite very high hippocampal levels of B-Raf (18). Furthermore, H89 failed to inhibit CRF 1 Rmediated ERK signaling in brain-derived CATH.a, rat fetal microglial, locus coeruleus, and transfected CHO cells (8,9,12,17). In fact, ERK activation by CRF 1 R in HEK293 cells was markedly decreased after the third intracellular loop's Ser 301 was phosphorylated by PKA (65). Thus, a cAMP-dependent PKA → Rap1 → B-Raf mechanism does not always mediate ERK1/2 signaling by G s -coupled receptors. EPAC, a guanine nucleotide exchange factor that is activated by intracellular cAMP, has been shown to regulate activation of Rap1 and ERK1/2 without the involvement of PKA (66). G s -coupled CRF 1 R signaling can stimulate ERK1/2 phosphorylation by activating upstream EPAC2 independent of PKA in certain cell lines (17,67). Interestingly, neither Epac nor PKA was found to mediate Akt cascade signaling by CRF 2(b) R in HEK293 (21).
The versatility of the CRF 1 R to activate different signaling pathways has allowed its coupling to G q proteins to be identified (4). G q conveys a signal to activate PKC which then triggers MAP kinase cascades. Thus, it has been shown that G q /PLC/PKC cascade signaling by CRF 1 R activated by Ucn1 contributes to phosphorylation of ERK1/2 in CRF 1 R-expressing myometrial, CHO, HEK293, and rat hippocampal cells (12,13,18). However, in pituitary AtT20 cells and CATH.a cells, PKC is not involved in Ucn1-stimulated ERK1/2 phosphorylation (12). In our study, pretreatment with the PKC inhibitor, BIM, increased rather than reduced CRF-stimulated ERK1/2 phosphorylation, suggesting that PKC may negatively regulate CRF 1 R ERK1/2 signaling in COS-7 cells, although the specific mechanism for this effect remains to be determined.
The use of PTX in our study suggests the participation of G i protein in the CRF-dependent activation of ERK and Akt pathways. Interestingly, it is now well-established that during GPCR/G i signaling, Gβγ release can activate a myriad of effectors to modulate diverse signaling pathways downstream of GPCRs, including Src, which in turn activate EGFR to promote ERK1/2 activation (43,68,69). Gβγ-activated Src can also associate PYK2. When we blocked that action of Gβγ subunits in COS-7 cells by overexpressing the ct-βARK peptide, which is a Gβγ subunit scavenger (70,71), CRF-stimulated ERK phosphorylation was decreased by ∼40%. Moreover, ct-βARK overexpression markedly reduced phosphorylation of Src and Akt. In agreement, another group has also found that CRF 1 R ERK1/2 signaling is only partially dependent on Gβγ, although their study did not assess the role of Gβγ subunits in the activation of upstream ERK1/2 pathways. Differences in CRF 1 R-mediated activation of the ERK1/2-MAP kinase cascade are probably attributable to variations  in the signaling properties of transfected CRF 1 Rs expressed in different cell lines utilized in these studies. We are presently investigating other upstream factors including βarrestins that regulate Src and EGFR mediation of CRF 1 R ERK1/2 signaling.
Our experiments did demonstrate that CRF-stimulated phosphorylation of ERK1/2 and EGFR occurred in parallel, while pretreatment with the EGFR kinase inhibitor, AG1478, caused a concentration-dependent inhibition of CRF-stimulated ERK1/2 phosphorylation. In agreement, it has been shown that EGFR transactivation is required for Ucn1-stimulated ERK1/2 phosphorylation in transfected HEK293 cells (5). In contrast to our data indicating that a MMP/HB-EGF ligand mechanism was not involved, however, this group reported that MMP generation of an HB-EGF ligand transactivated the EGFR during CRF 1 R ERK1/2 signaling (5). Therefore, EGFR transactivation can play a critical role in CRF 1 R signaling via the ERK1/2-MAP kinase cascade. Earlier studies have implicated a PI3K-dependent mechanism in CRF 1 R ERK1/2 signaling based on the observation that pretreatment with PI3K inhibitors attenuated sauvagineand Ucn1-stimulated ERK1/2 phosphorylation in CRF 1 Rexpressing CHO and HEK293 cells (5,9,12). PI3K is also involved in CRF 2(b) R-stimulated ERK1/2 activation in CHO, A7r5, and mouse neonatal cardiomyocyte cells (12,19). However, the activation sequence of PI3K, EGFR, and ERK1/2 during CRF 1 R signaling has not been fully elucidated. Here we observed that pretreating CRF 1 R-expressing COS-7 cells with the PI3K inhibitors wortmannin and LY294002 inhibited CRF-stimulated phosphorylation of ERK1/2 and Akt. Previous studies suggest that PI3K activity is required for Gβγ-mediated MAP kinase signaling pathway at a point upstream of Sos and Ras activation (50,72). Because we also found that AG1478 abolished phosphorylation of ERK1/2 while only decreasing Akt phosphorylation 25% in transfected COS-7 cells stimulated with CRF, upstream activation of the PI3K/Akt pathway by CRF 1 R is not strongly dependent on EGFR transactivation. In this context, our study suggests that Src acts as a critical mediator of PI3K activation, independent of EGFR transactivation, which, in turn, stimulates Akt and ERK1/2 phosphorylation. Previous studies have shown that activated Src directly associates with PI3K through interaction between the SH3 domain of Src and the proline-rich motif in the p85 regulatory subunit of PI3K, thereby increasing the specific activity of PI3K (59). Furthermore, intermediary proteins have also been identified to mediate Src-induced PI3K activation, such as p66Shc, Rap1, and FAK. Thus, our study raises the possibility that Src activates PI3K, although the specific mechanism for this effect remains to be determined.
For certain GPCRs, Src has been shown to induce EGFR transactivation, stimulate the PI3K-Akt pathway, and activate the ERK1/2 cascade (43,48,70). A novel finding in our study is the rapid and parallel phosphorylation of Src, PYK2, the EGFR, Akt, and ERK1/2 in CRF 1 R-expressing COS-7 cells stimulated with CRF. Importantly, we demonstrated that inhibiting Src function with PP2 markedly reduced or abolished the CRFstimulated activation of Src, PYK2, the EGFR, and ERK1/2, suggesting that Src has a central role in regulating CRF 1 R ERK1/2 signaling. Thus, our results clearly show that Src triggers signal transduction by two important pathways culminating in ERK activation by CRF 1 R: (i) EGFR activation of the classical Ras/Raf/MEK/ERK pathway, and (ii) PI3K regulation and subsequent activation of ERK1/2 (Figure 11).
To the best of our knowledge, our study demonstrates for the first time that Src regulates ERK and Akt signaling by the CRF 1 R. Yuan et al. (20) reported that Src was an upstream regulator of ERK signaling by both the CRF 1 R and CRF 2 R in the mouse atrial HL-1 cardiomyocytes cell line based on the effects of antalarmin (a CRF 1 R antagonist) and anti-sauvagine (a CRF 2 R antagonist). Although CRF 1 R was reported to be expressed in the human heart (73, 74), Ikeda et al. (75) reported that CRF 2(b) R is the major CRF receptor expressed in the HL-1 mouse atrial cardiomyocyte cell line with no measurable level of CRF 1 R mRNA. Their data detecting only FIGURE 11 | Schematic representation of the proposed CRF-mediated ERK1/2/Akt signaling mechanism in CRF 1 R transfected COS-7 cells. When COS-7 cells are stimulated with CRF, the overexpressed CRF 1 R leads to activation of G i protein and the subsequent dissociation of α i -GTP and βγ. Gβγ subunit is able to activate Src, which plays a central role in the activation of EGFR, through the formation of a protein complex that contains CRF 1 R, EGFR, and Src. PI3K, independently of Akt, is involved in ERK1/2 activation, presumably through Ras/C-Raf/MEK1/2. On the other hand, Src/PYK2 transactivates EGFR. Such EGFR transactivation leads to activation of the MAP kinase/ERK1/2 cascade and parallelly to weak activation of the PI3K/Akt pathway. Solid arrows indicate signaling mechanisms that have been identified. Dashed arrows indicate that the precise mechanism associated with the CRF-mediated regulation of Ras by PI3K and ERK1/2 by PKC remains to be determined. Plus sign (+) means positive regulation, and minus sign (-) means negative regulation of the ERK pathway. CRF 2 R expression in HL-1 cells is consistent with previous and more recent studies demonstrating only CRF 2 R expression in rat and mouse cardiomyocytes (19,76,77). Therefore, ERK1/2 signaling stimulated by Ucns in cardiomyocytes is mediated through CRF 2 R, which appears to be the main mediator of the cardiac stress response (78,79), rather than through CRF 1 R. Additionally, recent observations also indicate that CRF 2 R controls the cellular organization and colon cancer progression, specifically through the Src/ERK pathway (80,81). Thus, while all previous findings are relevant to CRF 2 Rs, our findings show for the first time that Src plays an important role in the regulation of ERK1/2 and Akt signaling by the CRF 1 R.
An important finding of our study was the detection of a signaling protein scaffold, which contains CRF 1 R, Src, and EGFR (Figures 6C,D). While the association between CRF 1 R and Src was totally dependent on CRF agonist activation, a constitutive interaction between CRF 1 R and EGFR was also detected, which was increased after CRF stimulation. In this context, it has previously reported that some GPCRs physically interact with EGFR in the absence of receptor ligands, a condition that may increase the efficiency of EGFR transactivation (29,(82)(83)(84). Thus, it is possible that the detected constitutive association between CRF 1 R and EGFR facilitates a more rapid CRF agonist-induced recruitment of Src to the EGFR and subsequent phosphorylation and activation of EGFR. With regard to this possibility, the presence of a putative proline-rich domain-binding SH3 motif (ProXXPro; X, any amino acid), located in the carboxyl terminus of the CRF 1 R (Pro 398 Thr 399 Ser 400 Pro 401 ) may provide a site for the direct interaction between Src and CRF 1 R after agonist stimulation. However, the CRF 1 R-386 mutant, which lacks the ProXXPro motif, induces a similar degree of ERK1/2 activation that is induced by the wild-type CRF 1 R, which suggests this putative region may not participate in the binding to Src (Figure 10A).
Moreover, there is evidence that Tyr phosphorylation of GPCRs plays a role in mediating GPCR-Src interactions (43). For instance, in studies conducted in A431 epidermoid carcinoma cells, stimulation of the β 2 -adrenergic receptor (β 2 -AR) with isoproterenol, results in phosphorylation of the receptor on Tyr 305 (43,85). The mutation of this residue to Phe abolishes Src/β 2 -AR association and impairs Src activation. This residue lies within a canonical Src SH2 binding domain, and it is proposed that Src directly binds the Tyr-phosphorylated β 2 -AR. Interestingly, the CRF 1 R has also a single putative SH2 binding domain (TyrXX-hyd; hyd, hydrophobic amino acid) located at the end of the third intracellular loop (Tyr 309 Arg 310 Lys 311 Ala 312 ), which may be a site where Src can directly interact with CRF 1 R. Further work is needed to establish the importance of this putative site in the agonist-induced CRF 1 R/Src interaction and Src activation.
β-arrestins are a small family of cytosolic proteins initially identified for their central role in GPCRs desensitization. Furthermore, β-arrestins act as adaptors in clathrin-mediated receptor endocytosis (86). In this sense, their role in CRF 1 R homologous desensitization and endocytosis is well-recognized, particularly for β-arrestin-2 (24,63,87). It is now wellestablished, however, that β-arrestins can also act as GPCRsignaling transducers that recruit and activate many other signaling molecules, including Src, MAP kinase, NF-κB and PI3K that modulate diverse cellular responses (64,86). β-arrestin regulation of CRF/CRF 1 R signaling is still not fully understood.
Under this experimental evidence and due to the importance of β-arrestins in the scaffolding and activation of Src and regulation of MAP kinase cascades, it was decided to evaluate their role in the CRF/CRF 1 R-mediated ERK1/2 activation observed in COS-7 cells. Using a phosphorylation-deficient mutant CRF 1 R, which has a decreased interaction with βarrestin-2 (24), no significant changes in the activation of ERK1/2 were detected after agonist stimulation (Figure 10A), suggesting that β-arrestin-2 is not involved in the CRF/CRF 1 R-mediated ERK1/2 activation observed in COS-7. This finding, however, can be explained in part by the low expression level of βarrestins in COS-7 cells (61,62). This hypothesis is supported by our data showing that overexpressing β-arrestin-2 in COS-7 notably increased the CRF/CRF 1 R-mediated ERK1/2 activation ( Figure 10B). Likewise, β-arrestin overexpression in COS-7 cells has been found to augment CRF 1 R internalization (24). Thus, our data provide evidence about the involvement of β-arrestin-2 in the CRF/CRF 1 R MAP kinase activation in cells with sufficient β-arrestin expression.
Our findings on signaling pathways activated by CRF 1 R help to elucidate the molecular mechanisms involved in response to stress mediated by this receptor. For instance, kinases in the ERK1/2-MAP kinase cascade, including Src and PYK2, are highly expressed in extended amygdala and forebrain neurons regulating anxiety defensive behavior and stress responsiveness (90)(91)(92). Acute stress or central CRF administration induces rapid phosphorylation of ERK1/2 in the basolateral amygdala and hippocampal neurons and prominent anxiety-like behavior in rats and mice (93)(94)(95). Furthermore, CRF 1 Rs can also signal through other cellular pathways that may be involved in post-traumatic stress disorder pathophysiology. As we showed here, CRF 1 R activated by CRF stimulated rapid phosphorylation of Akt at Ser 473 that is mediated by upstream Src and PI3K. Preclinical research has shown that activated Akt in the ventral tegmentum promotes resilience to anxiety-and depressive-like responses to stress (3,96), while high levels of phosphorylated Akt in the dorsal hippocampus and basolateral amygdala prolongs contextual and sensitized fear induced by inescapable stress (3,97). Therefore, the consequences of CRF 1 R Akt signaling during trauma and severe stress may differ depending on the brain region. Hence, ERK1/2-MAP kinase and Akt cascade signaling by CRF 1 R regulated by Src, PYK2, and EGFR may have critical roles in stress-induced anxiety and depression.

CONCLUSIONS
In summary, the data presented herein establish that the tyrosine kinase Src serves as a central upstream regulator of ERK1/2-MAP kinase and Akt cascade signaling by the human CRF 1 R in COS-7 cells. Although CRF 1 R coupling to G proteins strongly activates PKA and PKC pathways, neither second messenger kinases were involved in CRF 1 R-mediated ERK1/2 signaling. However, Gβγ released during activation of CRF 1 R by CRF, particularly from G i , stimulates phosphorylation of Src and PYK2, which in turn promotes transactivation of the EGFR through the formation of a heterotrimeric complex formed by the association of CRF 1 R, Src, and EGFR. EGFR transactivation, which occurred independent of MMP generation of the HB-EGF ligand, was essential for CRF-stimulated ERK1/2 phosphorylation while having only a small role in CRF 1 R-mediated Akt activation. Although PI3K activation contributes to CRF-stimulated ERK1/2 phosphorylation, CRF 1 R-mediated EGFR transactivation is independent of the PI3K/Akt pathway. In contrast, CRF 1 R Akt signaling while also being mediated by generation of Gβγ and phosphorylation of Src is weakly dependent on EGFR transactivation.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
JO-R, RH, and KC conceived the project. JO-R, FD, and RH designed the experiments. GP-M, AF-G, JH-A, and MD-C carried out the experiments. JO-R, GP-M, and AF-G analyzed and discussed the data. JO-R and RH wrote the manuscript. All authors read and approved the final version of the manuscript and took a due care to ensure the integrity of the work.

DEDICATION
This work is dedicated to Dr. Kevin J. Catt, who was an extraordinary scientist, mentor, and friend who passed away on October 1, 2017.