Phosphorylation of Enabled by the Drosophila Abelson Tyrosine Kinase Regulates the In Vivo Function and Protein-Protein Interactions of Enabled

Abstract

Drosophila Enabled (Ena) is a member of a family of cytoskeleton-associated proteins including mammalian vasodilator-stimulated phosphoprotein and murine Enabled that regulate actin cytoskeleton assembly. Mutations in Drosophila ena were discovered as dominant genetic suppressors of mutations in the Abelson tyrosine kinase (Abl), suggesting that Ena and Abl function in the same pathway or process. We have identified six tyrosine residues on Ena that are phosphorylated by Abl in vitro and in vivo. Mutation of these phosphorylation sites to phenylalanine partially impaired the ability of Ena to restore viability to ena mutant animals, indicating that phosphorylation is required for optimal Ena function. Phosphorylation of Ena by Abl inhibited the binding of Ena to SH3 domains in vitro, suggesting that one effect of Ena phosphorylation may be to modulate its association with other proteins.

Much of the interest in the c-Abl phosphotyrosine kinase (PTK) stems from the involvement of Bcr-Abl oncoproteins in human chronic myelogenous leukemia (CML) and acute lymphocytic leukemia (ALL) (17, 48). The kinase activity of the oncogenic Bcr-Abl fusion proteins is higher than that of the c-Abl PTK and is required for their transforming ability (37). The requirement of Bcr-Abl kinase activity for transformation suggests that phosphorylation of specific substrates by these proteins is important for the development and progression of CML and ALL. Recent work by a number of groups has identified several potential substrates of the Bcr-Abl or c-Abl PTKs. Among these are p62Dok (10, 61), Abi-1 (47), Abi-2 (13), c-Crk (18, 42), p130 CAS (36), Cbl (2), FAK (24), and paxillin (44). While phosphorylation of these proteins may be involved in Bcr-Abl-mediated oncogenesis, their importance in this process has yet to be determined.

Although the role of Bcr-Abl in CML and ALL has been extensively studied, much less is known about the normal functions of c-Abl. In mammalian cells, c-Abl is found in both the nucleus and cytoplasm (55, 60). Recent experiments suggest that its distribution between these compartments is dynamic and may be affected by integrin-mediated cell adhesion (32). In the nucleus, Abl is thought to exert cytostatic effects (23, 45) and is stimulated following DNA damage (5, 46). In contrast to the proposed roles of nuclear Abl, much less is known about the functions of Abl in the cytoplasm.

In Drosophila melanogaster, Abl is detected only in the cytoplasm and may play a role in axon outgrowth or pathfinding during development of the central nervous system (CNS) (20). To identify cytoplasmic substrates of the Abl PTK that play important roles in Abl signaling, we have undertaken a genetic analysis of Abl in Drosophila. The Drosophila Abl PTK (DAbl) is structurally and functionally related to human and murine Abl proteins (26, 27). The SH3, SH2, and kinase domains of mammalian c-Abl can substitute for these domains of DAbl, demonstrating that the functions of these domains are conserved. Both the mammalian and fly proteins have long carboxy-terminal domains that mediate binding to the actin cytoskeleton (38, 59). Flies lacking Abl die at the end of pupal development and, when this mutation is combined with mutations in other components of the Abl signaling pathway, display defects in the axonal architecture of the embryonic CNS (11, 13, 19). The lethality and CNS defects of abl mutants are suppressed by reducing the gene dosage of enabled (ena) (22). Like abl mutants, Drosophila embryos lacking Ena display defects in the axon organization of the CNS and peripheral nervous system (21). Ena was cloned and found to encode a novel Abl substrate that interacts with the Abl SH3 domain in vitro (21).

Ena is related to the human vasodilator-stimulated phosphoprotein (VASP) and to murine Enabled (mEna), both of which have been implicated in regulating F-actin assembly (19). Members of the Ena/VASP family are localized to actin stress fibers and focal adhesions, suggesting that they may play a role in cytoskeletal structure or assembly (19, 41). Expression of mEna results in actin-rich membrane projections, suggesting that Ena can promote actin polymerization or stability (19). A role for VASP in actin polymerization is suggested by its involvement in directing actin filament assembly at one pole of the intracellular pathogen Listeria monocytogenes (11). The effects of mEna and VASP on actin polymerization appear to result from interactions with profilin, which binds monomeric actin and promotes its assembly into F-actin (19, 40). Like its mammalian homologs, Drosophila Ena binds profilin and is localized to the actin cytoskeleton, suggesting that it may also affect cytoskeletal assembly (1). Thus, Ena/VASP proteins can promote actin polymerization and may normally function to regulate F-actin assembly during cell migration (35).

Ena and VASP are phosphorylated during development, which suggests that some aspects of their function may be regulated by phosphorylation. VASP is phosphorylated by cyclic nucleotide-dependent protein kinases in response to inhibitors of platelet activation, and this phosphorylation correlates with alterations in integrin adhesion (28). Multiple mEna isoforms are generated by alternative splicing, and some of these variants contain phosphotyrosine (19). While the significance of this phosphorylation is unknown, it suggests that the functions of some mEna isoforms may be regulated by phosphorylation. Drosophila Ena is phosphorylated when expressed with the Abl PTK, which suggests that it may be an Abl substrate (21). In addition, the level of Ena phosphorylation is decreased in Abl mutant animals, indicating that Abl may phosphorylate Ena during normal development (21).

We have examined the phosphorylation of Ena and found that it is directly phosphorylated by Abl. Ena is phosphorylated on multiple tyrosine residues, most of which are found near proline-rich sequences that mediate Ena’s binding to the Abl and Src SH3 domains. We show that phosphorylation is important for Ena function, as a phosphorylation-defective Ena protein is impaired in its ability to restore viability to ena mutants. Phosphorylation of Ena on these tyrosine residues reduces its interaction with the Abl and Src SH3 domains in vitro, suggesting that phosphorylation of Ena in vivo may attenuate formation of complexes with proteins that interact with the proline-rich domain of Ena.

MATERIALS AND METHODS

Molecular biology.

DNA was purified by standard protocols (3). Site-directed mutagenesis of ena and abl cDNAs was done by the method of Deng and Nickoloff (14). Oligonucleotides containing single-base substitutions that changed individual tyrosine codons to phenylalanine were incorporated into the His-tagged ena cDNA. The EnaYF⁶ mutant cDNA was generated by subcloning the Y129F, Y311F, Y329F, Y354F, Y370F, and Y530F mutations into the same construct. Mutagenesis of the Abl SH3 domain was performed to change tryptophan 118 to alanine. The entire mutagenized DNA fragments were sequenced to confirm the presence of the desired mutations and to make sure that no other mutations were introduced during the mutagenesis.

Purification of recombinant Ena and Abl.

DNA encoding six histidine residues was added to the COOH terminus of Ena by PCR, and this His-tagged Ena construct was subcloned into the baculovirus vector pVL1393 (Invitrogen). Spodoptera frugiperda SF9 cells were cotransfected with 2.5 μg of pVL1393-Ena plus 200 ng of Baculogold viral DNA (Pharmingen) to recover recombinant virus. High-titer virus stocks were generated and used to infect 2 × 10⁸ SF9 cells. At 48 h after infection, cells were harvested and lysed in 20 ml of 0.5% Triton X-100–20 mM NaPO₄ (pH 7.8)–500 mM NaCl–1 mM Pefabloc–1 μg each of pepstatin, leupeptin, and aprotinin per ml. After lysis, cell debris was pelleted at 12,000 × g for 20 min, and the lysate was incubated for 1 h at 4°C with 2 ml of Ni-nitrilotriacetic acid (NTA) agarose (Qiagen). The resin was washed twice with lysis buffer and then twice with 20 mM NaPO₄ (pH 6.3)–500 mM NaCl. Nonspecifically bound proteins were eluted with wash buffer plus 100 mM imidazole, and Ena protein was eluted in 20 mM PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid); pH 6.9]–500 mM NaCl–300 mM imidazole and stored at 4°C.

Virus stocks expressing DAbl were established as described above and used to infect SF9 cells. Lysates from Abl-infected cells were prepared as described above and applied to Ni-NTA resin. Abl protein bound weakly to Ni-NTA agarose and was eluted with 50 mM imidazole. Peak fractions were pooled and bound to S-Sepharose in 10 mM HEPES (pH 6.9)–100 mM NaCl. Abl protein was eluted with 200 mM NaCl, concentrated in a Centricon-30 filtration unit, and frozen at −80°C.

In vitro phosphorylation and peptide mapping.

For mapping studies, wild-type or mutant Ena proteins were purified from transfected S2 cells with Ni-NTA agarose and concentrated prior to phosphorylation. Purified Ena proteins were incubated with approximately 100 ng of Abl in 40 μl of kinase buffer (20 mM PIPES [pH 6.9], 10 mM MgCl₂, 10 mM MnCl₂, 5 mM dithiothreitol, 1 mM Na₃ VO₄) containing 10 μM unlabeled ATP and 20 μCi of [γ-³²P]ATP. Phosphorylated Ena proteins were immunoprecipitated from the kinase reactions, electrophoresed through sodium dodecyl sulfate (SDS)–7.5% acrylamide gels, and transferred to nitrocellulose membranes. Filter pieces containing labeled Ena proteins were rinsed in water and incubated for 1.5 h in 70% formic acid containing 100 mg of CNBr per ml. Eluted peptides were dried to remove CNBr and dissolved in 4 μl of pH 1.9 buffer (7). Samples were spotted to 0.1-mm-thick cellulose plates and resolved by electrophoresis for 30 min at 1,000 V in pH 1.9 buffer. After electrophoresis, plates were dried and peptides were resolved in the second dimension by ascending chromatography in phospho-chromo buffer (7).

SH3 binding experiments.

Phosphorylated Ena protein was prepared by incubating purified Ena (2.5 μg) with 100 ng of purified Abl in kinase buffer supplemented with 100 μM ATP for 1 h. Unphosphorylated Ena samples were prepared at the same time in kinase buffer lacking ATP. Kinase reactions were diluted to 0.5 ml with IP (immunoprecipitation) buffer (0.5% Triton X-100, 50 mM Tris [pH 8.0], 150 mM NaCl, 5 mM EDTA, 1 mM Na₃VO₄) and incubated with glutathione S-transferase (GST)–SH3 fusion proteins immobilized on glutathione-Sepharose for 2 h at 4°C. Beads were washed with IP buffer, and bound proteins were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) followed by Western blotting with anti-Ena antibodies.

MALDI-TOF mass spectrometry.

Phosphorylated or unphosphorylated His-tagged Ena proteins (5 to 50 pmol) were purified by SDS-PAGE and detected by using the Bio-Rad copper stain. Gel slices containing Ena were macerated and incubated for 12 h with one crystal of CNBr in 100 μl of 0.1 M HCl at room temperature. Peptide digestion products were recovered from the supernatant and by additional extraction of the gel with 150 μl of 50% acetonitrile–0.05% trifluoroacetic acid (TFA). The eluted peptides were lyophilized and dissolved in 20 μl of 50% acetonitrile–0.05% TFA. A small amount (0.5 μl) of this peptide solution was mixed with 0.5 μl of a saturated solution of α-cyano-4-hydroxycinnamic acid in 50% acetonitrile–0.05% TFA and allowed to air dry on a stainless steel sample plate. MALDI-time-of-flight (TOF) mass spectra (29, 30) of this mixture were obtained in an MP1 mass spectrometer (Ciphergen Biosystems, Palo Alto, Calif.). Samples were irradiated with a UV nitrogen laser to create peptide ions whose masses were determined by timing their flight down the 0.6-m flight tube of the mass spectrometer. Masses were determined by comparing the TOF of specific peptides to that of known internal reference standards. Data were acquired and analyzed with the Seldi molecular recognition system software (Ciphergen Biosystems).

Transfections and Western blot analysis.

Drosophila S2 cells were transiently transfected with ena and abl cDNAs in the pPac-PL expression vector. Cells were harvested after 60 h and lysed in IP buffer. Ena protein was isolated from the lysates by immunoprecipitation, resolved by SDS-PAGE, and analyzed by Western blotting with antiphosphotyrosine monoclonal antibody 4G10 (Upstate Biotechnology) or anti-Ena antibodies.

For in vivo phosphorylation experiments, cells were metabolically labeled with ³²P_i as described previously (21), and Ena was immunoprecipitated from cell lysates with anti-Ena antibodies.

Analysis of Ena mutant transgenes.

His-tagged Ena and EnaYF⁶ cDNAs were cloned into the pUAST vector, which contains a minimal promoter preceded by five GAL4 binding sites (8). Germ line transformations were performed as described previously (43), and stocks which contained both the Ena transgenes and heterozygous ena^GC5 mutations were generated. Ubiquitous expression of the Ena transgenes was driven by the GAL4-e22c enhancer trap (39). For rescue experiments, ena^GC5/Cyo; UAS-Ena virgin females were crossed to ena^GC1, GAL4 e22c/Cyo males. Progeny were scored for the presence of rescued Cy⁺ ena^GC1/ena^GC5 flies. Three wild-type and three EnaYF⁶ transgenes were tested for rescue, and Table 2 presents the results of three independent experiments. The number of expected ena mutant progeny was one-half of the number of the observed ena heterozygous progeny (Cy). Percent rescue for each cross was calculated by dividing the number of observed ena mutant progeny (Cy⁺) by the expected number of mutant offspring.

TABLE 2.

Effects of expression of wild-type and phosphorylation-defective Ena

Transgene	Expt no.	No. of ena mutant flies		% Rescueb	Avg % rescue/line	Overall % rescuec
		Expecteda	Observed
wt1A	1	65	63	97
	2	59	45	76	83
	3	48	37	77
wt1B	1	82	70	85
	2	74	59	80	83	86
	3	52	44	85
wt10A	1	87	83	95
	2	76	59	78	90
	3	52	51	98
YF14A	1	108	57	53
	2	97	65	67	57
	3	67	35	52
YF31A	1	102	45	44
	2	109	68	62	55	53
	3	47	28	60
YF31C	1	102	30	29
	2	81	36	44	45
	3	138	85	62

^aCalculated as one-half of the number of the observed ena heterozygotes (Cy). 

^bDetermined by dividing the observed number of Cy⁺ ena homozygotes by the expected number of ena mutants. 

^cAveraged over all experiments. The mutant transgenes were less able to rescue ena mutants than the wild-type transgenes, as determined by χ² analysis (P = 3.3e⁻¹¹). 

RESULTS

Abl phosphorylates Ena on multiple tyrosine residues.

Ena becomes tyrosine phosphorylated when expressed with the Abl PTK in Drosophila S2 cells, which suggests that Ena may be a substrate of the Abl kinase (21). To determine whether Ena is a direct substrate of Abl, we used purified Ena and Abl in in vitro kinase assays. Ena was phosphorylated in the presence of purified Abl (Fig. 1A, lane 3), demonstrating that Ena is a direct substrate of the Abl PTK. To map the Abl phosphorylation sites in Ena, in vitro-labeled Ena was cleaved with CNBr and the phosphorylated peptides were resolved by electrophoresis and thin-layer chromatography (TLC) (7). Four distinct phosphorylated species were observed in the two-dimensional (2-D) maps (Fig. 1B), indicating that Abl phosphorylates Ena on multiple tyrosine residues. Spot A was a mixture of phosphorylated species which could be resolved under different electrophoretic conditions into inorganic phosphate and two phosphorylated peptides (Fig. 1C). A nearly identical pattern of phosphorylated peptides was obtained when Ena was isolated from metabolically labeled cells expressing both Ena and Abl (Fig. 1D). Aside from insoluble material remaining at the origin, the only difference was the presence of peptide E in the in vivo map. Phosphoamino acid analysis of these peptides indicated that peptide E contained phosphoserine, while all of the Abl-dependent phosphopeptides contained phosphotyrosine (not shown). The similarity of the maps indicated that Abl phosphorylated the same residues both in vitro and in vivo.

FIG. 1.

Ena is phosphorylated by Abl on multiple tyrosine residues. (A) Ena is an Abl substrate. In vitro phosphorylation reactions were performed with 1 μg of purified Ena and 100 ng of purified Abl proteins in the presence of 20 μCi of [γ-³²P]ATP. Ena was phosphorylated in vitro when incubated with Abl (lane 3). No signal was seen in reactions containing only Abl (lane 1) or Ena (lane 2). Sizes are indicated in kilodaltons. (B) Multiple phosphorylated species were observed in a 2-D peptide map of in vitro-phosphorylated Ena. Ena was cleaved with CNBr, and the peptides were resolved by thin-layer electrophoresis in pH 1.9 buffer (horizontal axis) followed by ascending chromatography (vertical axis). Samples were applied to the TLC plate at the origin (+). (C) Spot A from the initial map shown in panel B contains a mixture of phosphopeptides which was recovered from the TLC plate and resolved by electrophoresis at pH 1.9 (horizontal axis) followed by electrophoresis at pH 3.5 (vertical axis). Under these conditions, the mixture was resolved into three phosphopeptides and inorganic phosphate (P_i). (D) A similar pattern of phosphopeptides was observed when Ena was metabolically labeled in cells expressing the Abl kinase. Peptide E in this map contained phosphoserine, while all of the other peptides were phosphorylated on tyrosine (not shown).

Identification of Ena phosphorylation sites.

To identify which tyrosine residues were phosphorylated by Abl, we generated a panel of mutant Ena proteins in which each tyrosine residue was individually changed to phenylalanine. Mutant proteins were expressed with Abl in cultured cells, and phosphopeptide maps of these mutant proteins were generated to look for the loss of specific spots from the map. Elimination of individual tyrosine residues resulted in the loss of single phosphopeptides, which indicates that there is not an initial phosphorylation event that is required for subsequent phosphorylation at additional sites. Specific phosphopeptides were absent from the maps of four of the mutant proteins, identifying these four tyrosine residues as Abl phosphorylation sites. The phosphorylated residues and the corresponding phosphopeptides are listed in Table 1. Changing Tyr329 to Phe resulted in the loss of peptide B (Fig. 2A, second panel). The new spot in this map (spot A′) resulted from the partial resolution of the peptide mixture in spot A and was not due to a shift of spot B in the Y329F mutant. Y354 and Y370 are predicted to be on the same CNBr fragment, and peptide C was absent from the maps of both the Y354F and Y370F mutants (Fig. 2A, third and fourth panels). The observation that either mutation resulted in the loss of peptide C from the maps indicates that peptide C represents the doubly phosphorylated form of this peptide and that both Y354 and Y370 are sites of phosphorylation by Abl. Elimination of either of these two sites should yield a peptide that can still be phosphorylated at the remaining site. While we never observed the singly phosphorylated form of this peptide in the 2-D maps, the singly phosphorylated peptide was detected by MALDI-TOF mass spectrometry (see below). The singly phosphorylated species may be poorly soluble and could be obscured by the material that remained near the origin in the 2-D maps. Peptide A1 was absent from the Y311F mutant (Fig. 2B, second panel), indicating that Y311 is phosphorylated by Abl. No single Tyr-to-Phe substitution eliminated spot D, suggesting that this peptide is phosphorylated on multiple tyrosine residues.

TABLE 1.

Results of phosphopeptide mapping

Spot on 2-D map	Position (kDa) of band on SDS-PAGE	MALDI-TOF peptide mass(es) (Da)a	Phosphorylated residue(s)b
A1			Y311
B		2,418	Y329
C		2,690, 2,769	Y354, Y370
D			Multipleb
	6		Y530
	4		Y129

^aMasses given correspond to the phosphorylated forms of the peptides. 

^bPeptide D is phosphorylated on multiple tyrosine residues. 

FIG. 2.

Identification of Ena phosphorylation sites. (A) Two-dimensional peptide maps of wild-type or mutant Ena proteins containing single Tyr-to-Phe substitutions were compared to identify phosphorylation sites. Compared to the map of wild-type Ena (left panel), peptide B was absent from the Y329F mutant (arrow in the second panel) and peptide C was missing from both the Y354F and Y370F mutants (arrows in the third and fourth panels). Therefore, peptide C corresponds to the doubly phosphorylated form of a CNBr fragment that contains Y354 and Y370. (B) Analysis of spot A from the wild-type protein or the Y311F mutant demonstrated that peptide A1 was absent in the mutant (arrow). (C) CNBr digests of the wild-type protein and the Y129F and Y530F mutant proteins were analyzed by SDS-PAGE. A 6-kDa phosphopeptide is missing in the Y530F mutant (arrowhead), and a 4-kDa peptide is absent from the Y129F mutant (arrow). The 6-kDa peptide containing Y530 is also phosphorylated on serine, and this peptide (*) migrates at 5 kDa in the Y530F mutant.

Examination of CNBr digests of the mutant proteins by SDS-PAGE identified two additional phosphorylation sites. A prominent 6-kDa phosphopeptide was absent from the Y530F mutant, identifying Y530 as a major site of Ena phosphorylation. This peptide also contained phosphoserine and was observed as a faster-migrating species in the Y530F mutant protein (Fig. 2C). In addition, a less prominent 4-kDa phosphopeptide was missing from the Y129F mutant (Fig. 2C), indicating that Y129 is phosphorylated by Abl.

To verify the results described above, CNBr digests of phosphorylated or unphosphorylated wild-type Ena protein were analyzed by MALDI-TOF mass spectrometry. Mass spectra from unphosphorylated and phosphorylated Ena were compared to detect the 80-Da mass shift due to phosphorylation of specific peptides. Phosphorylation of the 2,338-Da peptide containing Y329 shifted its mass to 2,418 Da, confirming that Y329 was phosphorylated by Abl (Fig. 3A). Both an 80- and a 160-Da shift in the 2,612-Da peptide containing Y354 and Y370 was detected following phosphorylation by Abl (Fig. 3B), and these mass increases correspond to the singly and doubly phosphorylated forms of this peptide. Mass shifts due to phosphorylation of Y530 and Y129 were also detected by this analysis (not shown), confirming the results obtained from the 2-D mapping experiments.

FIG. 3.

Phosphorylation of specific Ena peptides is detected by mass spectrometry. Phosphorylated or unphosphorylated Ena was cleaved with CNBr, and the peptides were analyzed by MALDI-TOF mass spectrometry. Panel A shows a portion of the mass spectrum which contains the 2,338-Da peptide containing Y329 (arrowhead). After phosphorylation by Abl, the mass of this peptide increased by 80 Da to 2,418 Da (arrow). The peptide containing Y354 and Y370 is shown in panel B (arrowhead). The singly and doubly phosphorylated forms of this peptide (arrows) were detected following phosphorylation by Abl.

Based on the intensities of the phosphorylated peptides, there does not seem to be a single, major phosphorylation site. Rather, Y311, Y329, Y354, Y370, and Y530 appear to be phosphorylated at comparable levels and are all phosphorylated more efficiently than Y129. To confirm that phosphorylation of the sites identified above is responsible for the majority of Ena phosphorylation, we generated a construct in which tyrosines 129, 311, 329, 354, 370, and 530 were changed to phenylalanine. This mutant protein (EnaYF⁶) was expressed with Abl in transfected S2 cells to determine to what extent Abl-dependent phosphorylation had been eliminated. The amount of phosphotyrosine was approximately eightfold lower on the EnaYF⁶ mutant protein than on wild-type Ena (Fig. 4A, lanes 2 and 4), demonstrating that this mutant protein is defective in Abl-dependent phosphorylation. The low level of phosphorylation seen in the EnaYF⁶ protein was not eliminated by further introduction of individual Tyr-to-Phe substitutions (not shown). Thus, the residual phosphorylation of the EnaYF⁶ protein is not due to phosphorylation at a single site and likely results from a low level of phosphorylation at a number of minor sites. Phosphorylation of the EnaYF⁶ protein by purified Abl in vitro was also less than that of the wild-type Ena protein (Fig. 4B), indicating that the reduction seen in vivo is likely due to a reduction in phosphorylation by Abl. Because Ena is also phosphorylated when expressed with Dsrc64B (21), we tested whether elimination of the Abl-dependent phosphorylation sites affected phosphorylation of Ena by Dsrc64B. While wild-type Ena was phosphorylated when expressed with Src, no phosphorylation of the EnaYF⁶ protein was detected (Fig. 4C). This result indicates that there is some overlap between the sites on Ena that are recognized by Abl and Dsrc64B.

FIG. 4.

Phosphorylation of EnaYF⁶ is greatly reduced. (A) Western blots of Ena immunoprecipitates from transfected cells show that elimination of six phosphorylation sites reduces Ena phosphorylation. Wild-type Ena (Ena^wt; lanes 1 and 2) and EnaYF⁶ (Ena^YF; lanes 3 and 4) proteins were expressed either alone (lanes 1 and 3) or with the Abl PTK (lanes 2 and 4). Ena was immunoprecipitated and examined by Western blotting using antiphosphotyrosine (anti-ptyr) monoclonal antibody 4G10 (left panel). As determined by densitometry, the amount of ptyr on EnaYF⁶ was approximately eightfold lower than that on wild-type Ena. Western analysis with anti-Ena (right panel) showed that equal amounts of wild-type and mutant proteins were loaded on the gel. Sizes are indicated in kilodaltons. (B) Phosphorylation of EnaYF⁶ by Abl in vitro is less than that of wild-type Ena. Purified wild-type (wt) and EnaYF⁶ (YF) proteins were incubated in vitro with the Abl PTK, and the phosphotyrosine content of Ena proteins was determined by Western blotting with an antiphosphotyrosine antibody. (C) EnaYF⁶ is not phosphorylated by Dsrc64B. Wild-type and EnaYF⁶ proteins were immunoprecipitated from transfected cells expressing Dsrc64B and examined by Western blotting with antiphosphotyrosine antibody 4G10. Wild-type Ena protein is phosphorylated when expressed with Dsrc64B. No phosphotyrosine was detected on EnaYF6 that had been expressed with Dsrc64B.

Ena phosphorylation sites are clustered in the proline-rich domain.

The locations of the six identified phosphorylation sites in Ena are shown in Fig. 5A. Five of the six sites are clustered near a proline-rich region of Ena that mediates binding to the SH3 domain of Abl (21), suggesting that SH3 binding might enhance phosphorylation of these residues. To examine the requirement of the Abl SH3 domain in Ena phosphorylation, we generated a mutant Abl protein in which a tryptophan residue conserved in all SH3 domains was changed to alanine. Although mutation of this residue disrupted binding of SH3 domains to proline-rich ligands (54), this mutant Abl protein was still capable of phosphorylating Ena when expressed in transfected cells (Fig. 5B, lane 3). In addition, Ena proteins lacking the SH3 binding sites are phosphorylated by Abl (21). These results indicate that binding of the Abl SH3 domain to Ena is not essential for phosphorylation of Ena by Abl.

FIG. 5.

Ena phosphorylation sites are clustered in the central proline-rich domain. (A) Positions of the six identified phosphorylation sites in Ena. The EVH1 and EVH2 domains are highly conserved between Drosophila Ena, mEna, and human VASP (19). The central proline-rich domain of Ena contains binding sites for SH3 domains and for profilin. Five of the six sites are clustered near proline-rich motifs that mediate interaction with the Abl SH3 domain. (B) Binding of the Abl SH3 domain to proline-rich sequences is not required for phosphorylation of Ena. Ena was immunoprecipitated from S2 cells transfected with Ena and Abl expression constructs and analyzed by antiphosphotyrosine Western blotting. Comparable levels of Ena phosphorylation was observed when Ena was expressed with wild-type Abl (lane 2) or AblW118A (lane 3). No phosphorylation of Ena was observed in the absence of Abl expression (lane 1). Western blotting of the same gel with Ena antibodies confirmed that equal amounts of Ena protein were loaded in all lanes (not shown). Sizes are indicated in kilodaltons. (C) An alignment of the phosphorylation sites and flanking sequences. Shaded residues indicate amino acids found at identical positions in three or more phosphorylation sites. There is a preference for Gly at the −3 and +1 positions and for Asn at the −4 position. The consensus sequence for c-Abl phosphorylation, determined by Songyang et al. (50), is shown below the consensus sequence of the Ena phosphorylation sites. Shaded residues indicate amino acids that were highly enriched in peptides phosphorylated by c-Abl in vitro. The SH2 binding consensus sequence of c-Abl (YENP [51]) is also shown.

An alignment of the Ena phosphorylation sites and flanking sequences is shown in Fig. 5C. None of the phosphorylation sites identified in Ena are found in mEna or VASP. The sites that are phosphorylated in Ena do not closely resemble the consensus derived from analysis of peptides phosphorylated by the mammalian c-Abl PTK. c-Abl preferentially phosphorylates tyrosine residues immediately preceded by Ile and followed by Ala and Pro residues at the +1 and +3 positions (50). The phosphorylation sites in Ena show a preference for Gly at the +1 and −3 positions and Asn at the −4 position (Fig. 5C). The differences between c-Abl and DAbl phosphorylation sites could be a result of differences in the methods used to identify phosphorylation sites. Songyang et al. (50) used a degenerate peptide library to identify peptides that were preferentially phosphorylated by c-Abl in vitro. The phosphorylation sites identified here were recognized in the context of an intact substrate and may reflect effects of secondary and tertiary protein structure on substrate specificity.

Many kinases with SH2 domains phosphorylate tyrosine residues in contexts favorable to binding to their own SH2 domains (50). In some cases, phosphorylation of a substrate leads to stable association of the kinase to the substrate via its SH2 domain and subsequent phosphorylation of additional tyrosine residues (15, 36). The Ena phosphorylation sites do not resemble the consensus sequence for Abl SH2 binding (51) (Fig. 5C). Consistent with this finding, we have not detected a stable association between Ena and Abl following phosphorylation, suggesting that the Abl SH2 domain does not bind to phosphorylated Ena. These data suggest that the multiple phosphorylation sites in Ena do not result from processive phosphorylation following binding of the Abl SH2 domain to an initial phosphorylated residue.

Phosphorylation is required for full Ena function.

The genetic and biochemical interactions between Abl and Ena suggest that phosphorylation of Ena by Abl may affect its function during development. To examine the importance of phosphorylation for Ena function, the phosphorylation-defective EnaYF⁶ mutant was introduced into the Drosophila germ line, and three independently derived lines were tested for the ability to rescue the lethality of ena null mutants. When expressed ubiquitously via the binary UAS/GAL4 system (8), wild-type Ena rescued 86% of the expected ena mutant progeny to adulthood with no obvious developmental defects (Table 2). In contrast, phosphorylation-defective Ena restored viability to only 53% of the level for the expected ena mutants. The reduced rescue by EnaYF⁶ demonstrates that phosphorylation is necessary for optimal Ena function.

The difference between the ability of wild-type and EnaYF⁶ to rescue ena mutants was not due to dominant effects of expressing the phosphorylation-defective protein, as no detectable phenotype was observed in wild-type flies when either the wild-type or mutant protein was expressed in a variety of tissues and developmental stages. The partial ability of EnaYF⁶ to rescue ena mutants indicates that the overall conformation of this protein was not dramatically affected by the multiple Tyr-to-Phe substitutions. Indeed, this protein accumulated to levels comparable to those of wild-type Ena (Fig. 6) and maintained all of the in vitro binding properties of the wild-type protein (Fig. 7 and data not shown). While it is possible that the presence of multiple Tyr-to-Phe substitutions had an effect on the function of the EnaYF protein in addition to eliminating phosphorylation, this is an intrinsic problem with site-directed mutagenesis that is not easily addressed experimentally.

FIG. 6.

Wild-type and EnaYF⁶ mutant proteins are expressed at comparable levels. The amounts of Ena protein expressed from three wild-type transgenes and two EnaYF⁶ mutant transgenes was determined by Western blot analysis. Lysates were prepared from ena mutant pupae expressing wild-type (lanes 1 to 3) or EnaYF⁶ mutant (lanes 4 and 5) protein, and protein from an equal number of pupae was loaded in each lane. The pupae used in this experiment (ena^GC1/ena^GC5) make no Ena protein; therefore, all of the Ena protein detected in this experiment was expressed from the transgenes. Sizes are indicated in kilodaltons.

FIG. 7.

Phosphorylation reduces Ena’s interaction with the Abl SH3 domain. One microgram of purified wild-type or EnaYF⁶ protein was phosphorylated in vitro by the Abl PTK. Reactions in which ATP was omitted were performed as negative controls. Equal amounts of the samples were incubated with glutathione beads containing GST, or a GST-Abl SH3 fusion protein, and the amount of bound protein was determined by Western blotting with anti-Ena. Aliquots of the input protein show that equal amounts of wild-type (lanes 1 and 2) and mutant (lanes 3 and 4) proteins were added to all reactions (A). No binding was seen with GST alone (B). Binding of wild-type Ena was reduced after phosphorylation (compare lanes 1 and 2 in panel C); only a slight reduction in binding was seen after phosphorylation of the EnaYF⁶ mutant protein (lane 4). Sizes are indicated in kilodaltons.

Phosphorylation affects association with SH3 domains.

The proximity of the Ena phosphorylation sites to proline-rich motifs that mediate binding to SH3 domains (Fig. 5A) suggested that phosphorylation of Ena might affect its binding to SH3 domains. To examine this, we compared the binding of GST-Abl SH3 and GST-Src SH3 fusion proteins to unphosphorylated Ena or Ena which had been phosphorylated by Abl in vitro. Equal amounts of purified wild-type and EnaYF⁶ mutant proteins were incubated with the Abl kinase in the presence or absence of 10 μM ATP. Western blots of the phosphorylation reactions analyzed with anti-Ena antibodies demonstrated that equal amounts of Ena protein were added to the binding reactions (Fig. 7A). The majority of wild-type Ena was shifted upward due to hyperphosphorylation at multiple sites (Fig. 7A, lane 2). A less dramatic shift in mobility of the EnaYF⁶ protein (Fig. 7A, lane 4) is consistent with the lower level of phosphorylation on this protein. The phosphorylated and unphosphorylated samples were then incubated with GST-SH3 proteins bound to glutathione-Sepharose beads, and bound Ena protein was detected by Western blotting. No binding to beads containing GST alone was seen (Fig. 7B). The Abl SH3 domain consistently bound less well to Ena that had been phosphorylated by Abl than to unphosphorylated Ena (Fig. 7C; compare lanes 1 and 2). This effect on SH3 binding is primarily due to phosphorylation of the sites identified above, as binding of the EnaYF⁶ mutant protein to the Abl SH3 domain was only slightly reduced by Abl phosphorylation (Fig. 7C, lanes 3 and 4). The same effect was seen with a GST-Src SH3 fusion protein (not shown). These results demonstrate a specific effect of phosphorylation on the biochemical properties of Ena and suggest that phosphorylation of Ena in vivo may modulate its association with SH3 domain-containing proteins or other proteins that bind to Ena’s central proline-rich domain.

DISCUSSION

Ena protein was initially identified based on the observation that reducing the gene dosage of ena restored viability to Drosophila mutants lacking the Abl PTK (21, 22). In abl mutant animals, the level of Ena protein is likely detrimental because a 50% reduction in the level of Ena rescues the lethality of the abl mutant. Although Ena protein in abl mutants is hypophosphorylated (21), it was not known whether the hypophosphorylation is responsible for the detrimental effects of Ena in abl mutants. The work presented here indicates that hypophosphorylated Ena protein, due to mutation of the tyrosine phosphorylation sites, retains substantial biological function but is not as effective in vivo as wild-type Ena, as judged by adult survival. The biochemical analysis of Ena protein presented here reveals a possible mechanism for the tyrosine phosphorylation of Ena, i.e., regulation of protein-protein interactions through the proline-rich central domain of Ena. Therefore, EnaYF⁶ may not be as effective in vivo as wild-type Ena protein because its protein-protein interactions through the proline-rich domain are not properly regulated. Complexes containing phosphorylation-defective EnaYF⁶ may be abnormally stable, thereby altering the overall kinetics of a dynamic process such as cytoskeletal remodeling during development. Hypophosphorylated Ena in abl mutant animals may be detrimental because its interactions through the proline-rich central domain are also improperly regulated, resulting in altered stoichiometry and/or kinetics of protein-protein complexes involving Ena. Heterozygous ena mutations may reestablish a more normal stoichiometry and/or kinetics to the process essentially by reducing the level of hypophosphorylated Ena protein that is too sticky.

If one consequence of abl mutations is to reduce the phosphorylation of Ena, why doesn’t expression of the phosphorylation-defective EnaYF⁶ produce a phenocopy of abl mutants? Although expression of phosphorylation-defective EnaYF⁶ in a wild-type Abl background might be expected to mimic some of the effects of abl mutations, the likely existence of other Abl substrates that would be properly phosphorylated in this background suggests that expression of EnaYF⁶ would not reproduce all of the defects observed in abl mutants. In addition, tyrosine phosphorylation of Ena in abl mutants is reduced only three- to fourfold (21), suggesting that Ena is phosphorylated by other kinases. In contrast, the elimination of multiple phosphorylation sites in EnaYF⁶ leads to a more dramatic reduction in Ena phosphorylation. This difference in Ena phosphorylation could contribute to the different phenotypic consequences in the two situations.

A possible explanation for the partial rescue observed with EnaYF⁶ is based on the observation that Ena is phosphorylated by other kinases during development. We have previously shown that phosphorylation of Ena is reduced, but not eliminated, in flies lacking the Abl PTK (21). This finding indicates that other kinases phosphorylate Ena during development and might also be able to affect its function. If these kinases recognize a different set of tyrosine residues than Abl, they could still phosphorylate EnaYF⁶ and might affect its function similarly to Abl. Ena is phosphorylated when expressed with Dsrc64B in cultured cells, suggesting that Src may also phosphorylate Ena during development (21). Elimination of the Abl-dependent phosphorylation sites in Ena also eliminated phosphorylation of EnaYF⁶ by Dsrc64B, indicating that phosphorylation of EnaYF⁶ by Dsrc64B is not likely to affect its function in vivo. Although we observed no reduction in Ena phosphorylation in flies lacking Dsrc64B (not shown), the existence of multiple Src-related kinases in Drosophila could compensate for the absence of a single family member (25, 49, 53). This would be similar to the functional overlap observed between members of the Src family kinases in mice, where the effect of removing individual kinases is partially masked by the presence of other family members (34, 52). Identification of other kinases that phosphorylate Ena should provide insight into the effect of phosphorylation on Ena function.

Our observation that phosphorylation of Ena reduces its association with SH3 domains illustrates a novel mechanism for regulating protein-protein interactions involving SH3 domains. Broome and Hunter (9) have shown that phosphorylation of Src on Y138 in the SH3 domain reduces binding to peptide ligands. This tyrosine residue is in the ligand binding groove, and its phosphorylation is predicted to decrease peptide binding by increasing the negative electrostatic potential of the ligand binding groove (9). Our data suggest a complementary model in which phosphorylation of an SH3 ligand can reduce its binding to SH3 domains. Phosphorylation of multiple residues in the proline-rich domain of Ena may increase the net negative charge of this region and interfere with binding to the hydrophobic groove of SH3 domains. Alternatively, the reduction in Ena binding to SH3 domains following phosphorylation may reflect an alteration of Ena’s secondary or tertiary protein structure. Structural analysis of proline-rich ligands bound to SH3 domains have demonstrated that the ligand assumes a left-handed type II polyproline helical conformation that interacts with aromatic residues in the SH3 domain (33, 62). Phosphorylation of multiple residues in the proline-rich domain of Ena could alter its conformation such that the SH3 binding sites would be inaccessible.

In theory, phosphorylation of Ena could generate binding sites for SH2 domain-containing proteins whose association with Ena might then prevent the interaction between Ena and SH3 domains. Because these experiments were done with purified Ena and Abl proteins, the reduction in SH3 binding seen in vitro is not due to SH2-dependent association of other proteins with phosphorylated Ena. While many SH2 domain-containing kinases phosphorylate tyrosine residues in contexts favorable for binding to their SH2 domains (50), we have been unable to detect a phosphorylation-dependent association between Ena and Abl. Therefore, the effect on SH3 binding observed after Ena phosphorylation does not result from a steric hindrance due to binding of the Abl SH2 domain to Ena. In the future, it will be interesting to determine if other interactions involving SH3 domains might be regulated in a similar manner.

How might the phosphorylation of Ena affect its biological properties? Phosphorylation of Ena by Abl would be expected to generate binding sites for proteins containing SH2 or phosphotyrosine-binding (6, 31) domains. Association of phosphorylated Ena with such proteins might affect Ena’s function or facilitate the assembly of protein complexes that are important for Ena function. We have examined Ena immunoprecipitates from metabolically labeled S2 cells and have failed to detect any phosphorylation-dependent protein associations with Ena (data not shown). However, it is possible that the proteins with which Ena associates following phosphorylation are not expressed in S2 cells and thus would not have been detected in these experiments. Future experiments to identify proteins whose interactions with Ena are dependent on phosphorylation should provide insight into how Ena function is regulated by phosphorylation.

Another mechanism by which Ena’s activity could be regulated by phosphorylation involves our observation that phosphorylation inhibits Ena’s interaction with SH3 domains. Based on our finding that Ena’s interaction with SH3 domains is attenuated by phosphorylation, we propose that phosphorylation of Ena by Abl might alter its association with other proteins during axon outgrowth. In addition to SH3 domains, Ena/VASP proteins associate with the actin-binding protein profilin (1, 19, 40). Considering the observation that phosphorylation reduces Ena’s interaction with SH3 domains, it is possible that phosphorylation of Ena affects its interaction with profilin or other proteins that bind to the Ena proline-rich domain. The interaction with profilin is thought to be critical for the ability of Ena/VASP proteins to promote actin polymerization (35). Thus, modulation of this interaction by phosphorylation could provide a novel mode of regulating actin assembly mediated by Ena/VASP proteins.

Since SH3 domains can direct the cytoskeletal association of proteins (4), phosphorylation of Ena could attenuate its association with the actin cytoskeleton. Reduction in the amount of Ena at sites of cell adhesion could alter the dynamics of cytoskeletal reorganization during cell migration. Evidence that Abl transduces signals from cell adhesion molecules comes from genetic interactions between Abl and fasciclin I (16). We propose a model in which signals emanating from cell adhesion molecules are interpreted by the Abl PTK, which then phosphorylates Ena, altering its localization or function (Fig. 8). In this model, phosphorylation of Ena by Abl would provide a mechanism to regulate Ena’s association with other proteins during the dynamic process of axon pathfinding. This model is consistent with the genetic observation that reducing the amount of Ena can compensate for the absence of the Abl PTK. In the absence of phosphorylation by Abl, complexes between Ena and proteins that bind to Ena’s central proline-rich domain might be inappropriately stabilized, impairing the ability of cells to respond to extracellular cues. Mutations that reduce the amount of Ena would compensate for the absence of Abl by returning the number of Ena complexes to a level compatible with normal development.

FIG. 8.

Proposed effects of phosphorylation on Ena complexes in vivo. Ena localizes to sites of cell adhesion, perhaps through interactions with zyxin-related or SH3 domain-containing proteins. Extracellular signals are transmitted through cell adhesion molecules to Abl and phosphatases, which increase or decrease the level of Ena phosphorylation. Phosphorylation of Ena by Abl attenuates its association with proteins such as profilin or SH3 domain-containing proteins that interact with Ena’s proline-rich domain. Reducing such interactions is likely to affect Ena’s function or subcellular localization. In such a model, phosphorylation of Ena by Abl is necessary for the dynamic association of Ena with cytoskeletal proteins and a precise balance between the levels of phosphorylated and unphosphorylated Ena is required for cytoskeletal remodeling in response to extracellular cues.

The effect of phosphorylation on Ena activity is consistent with previous observations suggesting that other members of the Ena/VASP family might be regulated by phosphorylation. VASP is phosphorylated by cyclic nucleotide-dependent protein kinases in response to inhibitors of platelet activation, and this correlates with altered integrin-mediated adhesion of these cells (28, 58). In mice, specific neuronally enriched isoforms of mEna are phosphorylated on tyrosine, suggesting that the function of mEna in the nervous system might be regulated by phosphorylation (19). While the kinases responsible for mEna phosphorylation have yet to be identified, we have found that mEna is phosphorylated by Drosophila Abl in transfected cells (12). Thus, phosphorylation of Ena/VASP proteins may be a common mechanism used to regulate the function of these proteins. Further analysis of mEna phosphorylation in mice is needed to determine whether Abl or related kinases affect its function during development.

Our finding that phosphorylation by Abl regulates the function of the cytoskeletal Ena protein suggests a role for cytoplasmic Abl in regulation of the actin-based cytoskeleton. This proposed role for Abl is consistent with the observation that c-Abl transiently relocates from the nucleus to focal adhesions in response to integrin-mediated cell adhesion (32). Both c-Abl and Bcr-Abl have actin binding domains that facilitate their interactions with the cytoskeleton (37, 38, 56), suggesting that these proteins may function in signal transduction pathways that affect the structure or integrity of the cytoskeleton. Consistent with this possibility are observations that Bcr-Abl induces alterations in integrin-based cell adhesion (57). Further analysis of Ena and other Abl substrates should clarify the roles of normal and oncogenic Abl proteins in developmental processes that are disrupted in human leukemias.