Tyrosine Phosphorylation of the orphan receptor ESDN/DCBLD2 serves as a scaffold for the signaling adaptor CrkL

Abstract

A quantitative proteomics screen to identify substrates of the Src family of tyrosine kinases (SFKs) whose phosphorylation promotes CrkL-SH2 binding identified the known Crk-associated substrate (Cas) of Src as well as the orphan receptor ESDN. Mutagenesis analysis of ESDN’s seven intracellular tyrosines in YxxP motifs found several contribute to the binding of ESDN to the SH2 domains of both CrkL and a representative SFK Fyn. Quantitative mass spectrometry showed that at least three of these (Y565, Y621 and Y750), as well as non-YxxP Y715, are reversibly phosphorylated. SFK activity was shown to be sufficient, but not required for the interaction between ESDN and the CrkL-SH2 domain. Finally, antibody-mediated ESDN clustering induces ESDN tyrosine phosphorylation and CrkL-SH2 binding.

1. Introduction

Virtually all eukaryotic signaling pathways employ reversible phosphorylation to effectuate cellular responses. Phosphorylation not only alters local protein structure via simple electrostatic interactions, but it also coordinates more sophisticated electrostatic interactions that include the recruitment of proteins which have phosphorylation-dependent binding domains. These domains recognize phosphotyrosine, phosphoserine or phosphothreonine residues typically within a preferred linear sequence of amino acids. Large-scale phosphoproteomics has made enormous strides toward characterizing the preferred phosphorylation motifs for specific kinases and the preferred binding motifs for individual phosphorylation-dependent binding domains [1–4]. In addition, the individual substrates of specific kinases and phosphorylation-dependent binding domains are being delineated with impressive speed and scale. As the methods and instrumentation for these types of analyses have matured, our capacity to conduct more functional proteomic analyses has also increased [1–4].

One type of functional proteomics analysis attempts to capture a functional assemblage, drawing simultaneous connections between a kinase, a substrate and a downstream effector. Here we describe the results of a screen to identify a novel substrate of the Src family of tyrosine kinases (SFKs) that when phosphorylated facilitates the binding of the Crk-Like (CrkL) adaptor protein via its SH2 domain. SFKs are perhaps the best studied tyrosine kinases owing to the early discovery of Src’s oncogenic potential [5]. Src is now known to play numerous roles in normal and aberrant biological processes [6–8], with its roles in regulating cytoskeletal changes in diverse systems ranking as some of its most important [9]. The adaptor protein Crk and its homolog CrkL have similarly been well-studied given that Crk, like Src, was initially identified as the cellular homolog of a transforming retroviral oncogene [10]. Furthermore, SFKs and Crk/CrkL have been shown to coordinate their efforts in a variety of signaling pathways with SFK substrates subsequently becoming scaffolds for Crk/CrkL. Several examples of such substrates exist and include Cas, IRS-1, Paxillin, Gab1 and p62dok [10,11]. Another important example of this coordinated activity in neuronal systems involves the scaffolding adaptor Disabled-1 (Dab1). The phosphorylation of Dab1 by SFKs induces Crk/CrkL-binding and this is critical for the proper positioning of neurons downstream of Reelin signaling during the development of the vertebrate central nervous system [12–15].

Given that SFKs and Crk/CrkL have emerged as critical collaborators in several important systems, we conducted a functional and quantitative proteomics screen toward the identification of a novel SFK substrate that also acted as a scaffold for the Crk/CrkL family of adaptors. This yielded a relatively uncharacterized scaffolding receptor, Endothelial and smooth muscle cell-derived neuropilin-like protein (ESDN), also known as DCBLD2 (discoidin, CUB and LCCL domain containing 2).

2. Materials and Methods

2.1. Plasmids and Site-Directed Mutagenesis

The expression construct for full-length human ESDN with Myc and Flag (DDK) tags at the carboxyl-terminus in pCMV6-Entry was obtained from Origene (product #RC224483; Rockville, MD). The full-length human Fyn in pRK5 was acquired through Addgene (plasmid 16032) and was originally constructed in the laboratory of Filippo Giancotti [16]. The mutant ESDN constructs were generated by Bio Basic Inc. (Markham, ON) and are denoted as follows: ESDN-Y1F (Tyr750Phe), ESDN-Y3F (Tyr750Phe, Tyr732Phe and Tyr565Phe), and ESDN-Y7F (Tyr750Phe, Tyr732Phe, Tyr677Phe, Tyr666Phe, Tyr655Phe, Tyr621Phe, and Tyr565Phe). The bacterial expression plasmid for GST-CrkL-SH2 was a gift of A. Imamoto and was used as described previously [13]. The bacterial expression plasmid for GST-Fyn-SH2 was constructed first by PCR amplification of the SH2 domain using pRK5-Fyn as template and the following primers: 5′-CGCGCGAATTCGTTGACTCTATCCAGGCAGAAG -3′ (forward, sense) and 5′-GCGCGCGGCCGCGGTAAGCCTTGGCATCCCTTTG -3′ (reverse, anti-sense). The PCR reaction was digested with EcoRI and NotI and the purified fragment was cloned in-frame with GST in pGEX-4T-1 which had been cut with EcoRI and NotI. The new clone was sequence-verified using the University of Vermont/Vermont Cancer Center DNA analysis facility.

2.2. Cell Culture, Transfections, Stimulations and Lysis

E1A-transformed Human embryonic kidney (HEK 293E) cells were grown in DMEM (Mediatech, Manassas, VA) supplemented with 5% Fetal Bovine Serum (FBS), 5% Cosmic Calf Serum (sera were from Hyclone, Logan, UT), 50 units/ml of penicillin and 50 μg/ml of streptomycin. For SILAC experiments cells were grown as described [17] and as detailed in the Supplementary Methods. HEK 293E cells stably expressing ESDN-Myc-Flag were generated by selecting G418-resistant transfected cells. The line used in these studies represents a mixed pool of more than a 100 stably-transfected clones.

Transfections were performed by calcium phosphate precipitation when cells were at 60–75% of confluence. H₂O₂ stimulations (8 mM) were for 15 minutes. PP2 (Sigma, Saint Louis, MO) and Src Inhibitor-1 (EMD-Calbiochem, Billerica, MA) were dissolved in dimethyl sulfoxide and used at 2 or 10 μM as indicated for 20 minutes prior to, and throughout the duration of, H₂O₂ stimulations. Antibody stimulations were done by adding the antibody directly to the growth media for twenty minutes. Lysis is described in the Supplementary Methods.

2.3. Pulldowns, Immunoprecipitations, SDS-PAGE, Immunoblots and Antibodies

For the initial large-scale pulldown, ~20 mg of heavy protein extract was mixed with ~20 mg of light protein extract (1 mg/ml). The mixture was divided in two, half of which was incubated with 50 μg of GST on glutathione resin and half of which was incubated with 50 μg of GST-CrkL-SH2 on glutathione resin. The incubations were done on a rocker at 4 °C overnight. The resins were washed four times with lysis buffer and proteins were eluted with protein sample buffer (150 mm Tris pH 6.8, 2% SDS, 5% β-mercaptoethanol, 7.8% glycerol, 0.01% bromophenol blue) at 95 °C for five minutes. Small-scale pulldowns used 2–5 μg of GST, GST-CrkL-SH2 or GST-Fyn-SH2 on glutathione resin and 200–800 μg of protein extract. Immunoprecipitations were done using 200 μg of protein extract and 20 μl of anti-Flag (M2) affinity gel (Sigma). SDS-PAGE was conducted on 10% (37.5:1 acrylamide:bis-acrylamide) gels. Immunoblotting and mass spectrometry were conducted as described previously [18] and as detailed in the supplementary methods. The anti-Flag (M2) antibody was from Sigma. The anti-phosphotyrosine (4G10) and anti-GST (06-332) antibodies were from Upstate Biotechnology/Milipore (Billerica, MA). The anti-Fyn (4023) and anti-alpha-tubulin (DM1A) antibodies were from Cell Signaling Technology, Inc. (Danvers, MA). The anti-ESDN (E1781, Sigma) was raised against amino acids 399–416 of human ESDN which is in the middle of the factor V/VIII domain.

3. Results and Discussion

To identify proteins that when phosphorylated by SFKs become binding partners for the CrkL SH2 domain, we employed a quantitative functional proteomics screen. HEK 293 cells were grown in SILAC media and treated with H₂O₂ which serves as a general tyrosine phosphatase inhibitor [19], thereby dramatically boosting the levels of tyrosine phosphorylation on substrates of endogenous tyrosine kinases. While both light and heavy SILAC cells were treated with H₂O₂, heavy cells were pre-treated with PP2, a relatively specific inhibitor of SFKs [20]. Anti-phosphotyrosine immunoblots of whole cell extracts showed that H₂O₂ generally increased phosphotyrosine levels while PP2 reduced H₂O₂-stimulated phosphotyrosine levels. Furthermore, pulldown assays showed that a subset of the tyrosine phosphorylated proteins following H₂O₂ stimulation interacted with GST-CrkL-SH2 but not with GST alone. Additionally, the binding of several of these proteins to GST-CrkL-SH2 was sensitive to PP2 (Fig. 1A). We thus proceeded with a large-scale pulldown experiment using the same extracts. Heavy and light SILAC extracts were mixed one-to-one and were subjected to pulldowns with GST or with GST-CrkL-SH2. Both the GST-bound proteins and the GST-CrkL-SH2-bound proteins were eluted separately and subjected to SDS-PAGE. The gels were then cut into multiple regions and digested in-gel with trypsin. Extracted peptides were analyzed by LC-MS/MS using a linear ion trap-orbitrap mass spectrometry platform. SILAC ratios were determined for all peptides from proteins uniquely bound to the CrkL-SH2 domain. Figure 1B shows the two proteins that had SILAC ratios that were statistically different from those of all other identified proteins. These two proteins, Cas and ESDN, presented a profile consistent with SFK-dependent CrkL-SH2 binding.

Fig. 1.

SILAC-based quantitative functional proteomics identifies Cas and ESDN as PP2-sensitive tyrosine kinase substrates that interact with the CrkL-SH2 domain. HEK 293 cells were grown in control media (C), heavy (H) media, or light (L) media and treated as indicated. Whole cell extracts (WCE) were subjected to SDS-PAGE and immunoblotting directly (A, left panel) or after pulldown assays using either GST or GST-CrkL-SH2 (A, right panel). Blots were probed with anti-phosphotyrosine antibodies. *Indicates the position of GST-CrkL-SH2 background reactivity. (B) SILAC ratios for ESDN and Cas peptides are statistically different (p< 0.001) from all other proteins. The mean SILAC ratio of all peptides is the vertical light gray line, whereas the dark vertical line is the mean SILAC ratio of all peptides less those peptides from ESDN and Cas.

While the intracellular scaffolding protein Cas has been extensively studied, and was first identified as a substrate of SFKs that bound to Crk more than two decades ago [11], ESDN (Fig. 2A) is predicted to be a single-pass, orphan receptor and little is known about its signaling. The Crk/CrkL SH2 domain docks at phosphorylated tyrosine residues in the minimal consensus motif YxxP. Scansite [21] bioinformatically predicts 16 possible Crk-SH2 YxxP binding motifs in Cas and six in the intracellular domain of ESDN. One additional intracellular YxxP motif is also found in ESDN (Fig. 2B). We obtained a clone of ESDN with a carboxyl-terminal Flag tag and designed three ESDN-Flag mutant constructs with tyrosine to phenylalanine mutations at predicted Crk/CrkL-SH2 binding YxxP motifs. The constructs harbored mutations at either the top predicted, the top three predicted, or all seven YxxP tyrosines (Fig. 2B). As shown in Figure 2C, cells transfected with these constructs and treated with H₂O₂ generated ESDN molecules capable of binding to the CrkL-SH2 domain in all cases except when all seven of the YxxP tyrosines were mutated to phenylalanine.

Fig. 2.

(A) ESDN domain structure and corresponding human amino acids including the seven intracellular tyrosines found in YxxP motifs. TM=transmembrane. (B) Weblogo of the Crk-SH2 binding motif aligned with the ESDN intracellular YxxP motif sequences in descending order (high, medium and low) of their Scansite predicted Crk-SH2 binding. Note putative phosphotyrosine residues are at position zero. (C) Multiple regulated phosphotyrosine residues participate in the binding of ESDN to the CrkL-SH2 domain. HEK 293 cells transfected with the indicated constructs were treated as indicated. Whole cell extracts were subjected to SDS-PAGE and immunoblotting directly with the indicated antibodies, or after being subjected to either GST-CrkL-SH2 pulldowns (upper panels) or anti-Flag immunoprecipitations (lower panels). The Ponceau stain of the membrane prior to immunoblotting indicates the levels of the transferred GST-CrkL-SH2 in each lane. *Indicates non-specific background bands.

To determine if ESDN underwent reversible tyrosine phosphorylation at YxxP tyrosines we immunoprecipitated ESDN-Flag from the same extracts used in the GST-CrkL-SH2 pulldowns (Fig. 2C). All constructs, including the Y7F, showed reversible tyrosine phosphorylation as measured by anti-phosphotyrosine immunblotting with the Y7F showing dramatically reduced anti-phosphotyrosine immunoreactivity (Fig. 2C). This is consistent with reversible phosphorylation at several ESDN YxxP tyrosines participating in the binding to the CrkL-SH2 domain. To identify sites of reversible tyrosine phosphorylation, we conducted another SILAC experiment where light cells were transfected with wild type ESDN-Flag and treated with H₂O₂ and heavy cells were transfected with wild type ESDN-Flag and left untreated. Anti-Flag immunoprecipitations were conducted on the two extracts separately. After denaturing the immune complexes, the two samples were combined and subjected to SDS-PAGE. In-gel tryptic digestion and LC-MS/MS analysis of the extracted peptides identified four ESDN tryptic peptides harboring phosphotyrosine (Supplemental Figures 1–5). These phosphopeptides were only identified in the stimulated, light samples and based on signal to noise we concluded that H₂O₂ induced a ~50 fold increase in the four identified phosphopeptides. Three of the identified phosphotyrosyl-peptides were in YxxP motifs, and consistent with an increase in phosphotyrosine at a non-YxxP motif in the ESDN-Flag Y7F mutant (Fig. 2C), one of the identified phosphopeptides was not in a YxxP motif (Y715). The sites that we identified have already been curated in PhosphoSitePlus [22], having been identified in many large-scale phosphoproteomic experiments (Supplementary Fig. 1).

Well-studied examples exist where SFKs and Crk/CrkL not only coordinate their activities around a scaffold, but the molecular complex they form is perpetuated by a positive feedback loop. It is well understood that in unstimulated cells a SFK typically exists in a folded, low activity conformation with its SH2 domain intramolecularly bound to its CSK-phosphorylated C-terminal tail [23]. Following a variety of stimuli the folded structure of an SFK can be opened, primarily by its SH2 domain binding to a newly phosphorylated tyrosine on a neighboring molecule for which it has a higher affinity than for its own C-terminal tail [24]. This is observed for both Cas and Dab1 as they serve as SFK substrates as well as binding partners for both the SH2 domains of SFKs and Crk/CrkL [11,13,14,18]. However, Cas and Dab1 are phosphorylated at multiple residues by SFKs with some residues capable of docking and unfolding SFKs and some capable of docking Cas/CrkL [11,13,14]. We therefore examined whether the SH2 domain of a representative SFK, Fyn, showed regulated binding to ESDN and whether the same ESDN YxxP tyrosines that were required for CrkL-SH2 binding might participate in Fyn-SH2 binding. Pulldown assays with both GST-CrkL-SH2 and GST-Fyn-SH2 showed H₂O₂-induced binding to wild type ESDN but not to ESDN Y7F (Fig. 3A). Importantly, similar results were obtained when ESDN was co-expressed with Fyn in the absence of H₂O₂ Fig. 3B). These results show that active Fyn is sufficient to directly or indirectly induce ESDN tyrosine phosphorylation and CrkL-SH2 binding. However, while pharmacological inhibition of SFKs with the highly selective Src Inhibitor-1 (Src-1) [19] led to increased ESDN electrophoretic mobility, significant binding of ESDN to the CrkL-SH2 domain was retained (Fig. 3C), suggesting that additional kinases are important in promoting the ESDN-CrkL interaction. Given that we identified hydrogen peroxide induced phosphorylation of several ESDN tyrosine residues including 715 and732 (non-YxxP and YxxP sites respectively, Sup. Fig. 1), whose phosphorylation was recently shown to be dependent on SFKs [25], and given significant ESDN binding can be achieved to the CrkL-SH2 domain even when the three best-predicted CrkL-binding sites are mutated (Fig. 2C), suggest that SFKs can cooperate with non-SFKs to phosphorylate ESDN and promote a high avidity binding to the CrkL-SH2 domain. Furthermore, while SFKs physiologically phosphorylate YxxP motifs including Dab1 [13], YxxP is not the preferred motif for their kinase and SH2 domains. However, several other kinase families and SH2 domains could easily contribute to ESDN phosphorylation and ESDN binding [2–4] are these are currently being pursued.

Fig. 3.

(A) Wild type but not ESDN Y7F binds to the SH2 domain of Fyn following H₂O₂ stimulation. HEK 293 cells were transfected with ESDN-Flag wild type or ESDN-Flag Y7F and either left untreated or stimulated with H₂O₂. Whole cell extracts were subjected to SDS-PAGE and immunoblotting with the indicated antibodies directly or following pulldown assays with GST-CrkL-SH2 or GST-Fyn-SH2 fusions. The Ponceau stain of the membranes prior to immunoblotting indicates the levels of the transferred GST fusion proteins. (B) Fyn induces the binding of wild type ESDN, but not Y7F ESDN, to the SH2 domains of both CrkL and Fyn. Cells were treated as in A except that where indicated, Fyn was co-expressed with ESDN. (C) Pharmacological disruption of SFKs reduces but does not eliminate H₂O₂-induced ESDN binding to the CrkL-SH2 domain. Cells stably-expressing ESDN-Flag were pre-treated with the indicated inhibitors prior to stimulation with H₂O₂. Extracts were subjected to immunoblotting before or after GST-CrkL-SH2 pulldowns as indicated.

ESDN appears to be unique to, and highly conserved among, vertebrates, with all seven of the intracellular YxxP motifs conserved from human to fish and frog (Supplementary Fig. 6). ESDN is best understood and was originally identified as a marker of vascular endothelial cells [26,27]. However, it has also been described as a marker of platelets and some cancers [28–31]. Additionally, the extracellular domain structure of ESDN is reminiscent of neuropilins which are receptors/co-receptors involved in neuronal migration [26]. It will be important in future studies to describe the individual cell types that express ESDN in a variety of systems and to elucidate both its biological functions, and its mechanisms of signal transduction. We propose an introductory signaling model for ESDN involving SFKs and other tyrosine kinases (Fig. 4A). In this model several possible extracellular ligands could function in autocrine or paracrine fashion to cluster ESDN molecules. The clustering concentrates ESDN molecules sufficiently that the relatively inactive SFK, which is linked to the membrane via a lipid moiety, begins to phosphorylate a few ESDN molecules. This could lead to Fyn’s SH2 domain binding to phosphorylated ESDN and thereby relieving its autoinhibition [24]. Simple thermodynamic processes [32] and a positive feedback loop could then lead to a highly phosphorylated pool of ESDN and the direct binding of several types of proteins with phosphotyrosine-binding domains, including CrkL, whose N-terminal SH3 domain binds many proteins that could govern changes in cell adhesion or migration proteins [33]. Consistent with this model, we find that a presumed clustering of ESDN by treating ESDN-expressing cells with a bivalent (IgG) antibody against the Factor V/VIII extracellular domain induces ESDN tyrosine phosphorylation and its binding to the CrkL-SH2 domain (Fig. 4C).

Fig. 4.

(A) Model depicting possible mechanism by which ESDN acts as a scaffold for the adaptor CrkL. ESDN’s ectodomains may respond to factors (X, Y, Z) that induce clustering of ESDN in the proximity of low activity SFKs. ESDN serves as a substrate for SFKs as well as unidentified tyrosine kinases at multiple tyrosine residues, some of which can bind to the SH2 domain of SFK’s and thereby relieve autoinhibition. Phosphotyrosine sites on ESDN serve as docking sites for CrkL and other phosphotyrosine-binding proteins. By virtue of primarily its N-terminal SH3 domain, CrkL may recruit additional effector proteins. (B–C) Antibody-induced clustering induces ESDN tyrosine phosphorylation and its interaction with the CrkL-SH2 domain. HEK 293 cells expressing wildtype or Y7F ESDN-Flag were treated as indicated with two μg/ml of an anti-ESDN antibody recognizing the extracellular Factor V/VIII domain. Cells were lysed and extracts were subjected to anti-Flag immunoprecipitations. Immune complexes and whole cells extracts were blotted with either anti-phosphotyrosine or anti-Flag antibodies as indicated. *Indicates non-specific band. (C) Cells were treated as in (B) except extracts were subjected to GST-CrkL-SH2 pulldowns prior to immunoblotting.

In this study we used a functional proteomics screen to identify SFK substrates that serve as scaffolds for the CrkL adaptor protein. We identified the orphan receptor ESDN. Our data describe a nascent signaling pathway and define avenues of related future research. These avenues include the identification of the extracellular ligands that activate or perhaps cluster ESDN, the identification of additional ESDN kinases, the identification of intracellular phospho-dependent ESDN binding proteins that effectuate ESDN signaling, and the identification of downstream ESDN signaling events, and ultimately functional studies of ESDN using vertebrate systems.

Highlights.

• Functional Proteomics Identifies ESDN as a novel CrkL-SH2 binding protein.

• Fyn and other tyrosine kinases induce the binding of ESDN to both the CrkL and Fyn SH2 domains.

• Mass spectrometry-based identification of regulated tyrosine phosphorylation sites on ESDN.

Abbreviations

ESDN

Endothelial and smooth muscle cell-derived neuropilin-like protein

DCBLD2

discoidin, CUB and LCCL domain containing 2

CUB

complement C1r/C1s, Uegf, Bmp1

Crk

CT10 regulator of kinase

CrkL

Crk-Like

SFK

Src family tyrosine kinase

Cas

Crk-associated substrate

WCE

whole cell extract

SILAC

stable-isotope labeling by amino acids in cell culture

PMSF

phenylmethylsulfonyl fluoride

PCR

polymerase chain reaction

MS

mass spectrometry

LC

liquid chromatography

CSK

C-terminal Src kinase

IRS-1

Insulin Receptor Substrate-1

Gab1

GRB2-associated-binding protein 1

GST

Glutathione S-transferase

RIP2

Receptor-Interacting Protein 2

Appendix A. Supplementary data

Supplementary data associated with this article include six figures, one table and a supplementary methods section.