Specificity of LIM domain interactions with receptor tyrosine kinases.

LIM domains, Cys-rich motifs containing approximately 50 amino acids found in a variety of proteins, are proposed to direct protein·protein interactions. To identify structural targets recognized by LIM domains, we have utilized random peptide library selection, the yeast two-hybrid system, and glutathione S-transferase fusions. Enigma contains three LIM domains within its carboxyl terminus and LIM3 of Enigma specifically recognizes active but not mutant endocytic codes of the insulin receptor (InsR) (Wu, R. Y., and Gill, G. N. (1994)J. Biol. Chem. 269, 25085-25090). Interaction of two random peptide libraries with glutathione S-transferase-LIM3 of Enigma indicated specific binding to Gly-Pro-Hyd-Gly-Pro-Hyd-Tyr-Ala corresponding to the major endocytic code of InsR. Peptide competition demonstrated that both Pro and Tyr residues were required for specific interaction of InsR with Enigma. In contrast to LIM3 of Enigma binding to InsR, LIM2 of Enigma associated specifically with the receptor tyrosine kinase, Ret. Ret was specific for LIM2 of Enigma and did not bind other LIM domains tested. Mutational analysis indicated that the residues responsible for binding to Enigma were localized to the carboxyl-terminal 61 amino acids of Ret. A peptide corresponding to the carboxyl-terminal 20 amino acids of Ret dissociated Enigma and Ret complexes, while a mutant that changed Asn-Lys-Leu-Tyr in the peptide to Ala-Lys-Leu-Ala or a peptide corresponding to exon16 of InsR failed to disrupt the complexes, indicating the Asn-Lys-Leu-Tyr sequence of Ret is essential to the recognition motif for LIM2 of Enigma. We conclude that LIM domains of Enigma recognize tyrosine-containing motifs with specificity residing in both the LIM domains and in the target structures.

Most LIM proteins contain more than one LIM domain. The sequence of an individual LIM domain is, in general, more closely related to the same LIM domain in analogous proteins from other species than to other LIM domains within the same protein (8). Although the NMR structure of LIM2 of CRP resembles the DNA binding domain of the GATA-1 transcription factor (3), most available evidence indicates that LIM domains function in protein⅐protein rather than protein⅐DNA interactions.
Two structural targets for LIM domains have been identified. Using gel overlay techniques, Schmeichel and Beckerle (15) found that the LIM domains of zyxin interacted with the LIM-only protein CRP. Specificity was evident from the observation that LIM1 but not LIM2 or LIM3 of zyxin-bound CRP. Feurstein, et al. (16) also found evidence for LIM⅐LIM interactions involving CRP but did not observe specificity for the LIM domain. The carboxyl-terminal LIM domain of the cytoplasmic protein Enigma was found to specifically interact with exon 16 of the insulin receptor (InsR) (17). Mutations in exon 16 that disrupted the major endocytic code and ligand-induced endocytosis of InsR (18) also disrupted interaction with Enigma. The endocytic code of InsR, like that of many receptors (19), consists of 4 -6 amino acids that form a tyrosine-containing tight turn (20). A generalized tight-turn motif, which functioned in endocytosis of mutant EGFR (21), and which contained two copies of an Asn-Asn-Ala-Tyr-Phe motif interacted with a wider range of LIM domains, suggesting that specific Tyr-based tight turns would provide interaction targets for specific LIM domains. There is functional evidence for LIM domain interactions with a variety of transcription factors (22)(23)(24), suggesting that additional target specificities exist.
We recently found that in addition to InsR, Enigma specifically interacted with the receptor tyrosine kinase Ret (25). Given that Enigma, which contains three LIM domains at its carboxyl terminus, 2 was found to interact with two receptor tyrosine kinases, determining the molecular basis of this recognition became important. We found that LIM2 of Enigma specifically recognized Ret whereas LIM3 of Enigma specifically recognized InsR. Detailed analysis of the target sites indicated both LIM2 and LIM3 recognize Tyr-containing motifs located outside the tyrosine kinase cores of Ret and InsR. Individual LIM domains thus have the ability to distinguish between Tyr-based motifs providing a mechanism for specificity in both the LIM domain and in the target. This is of special interest given the requirement of the target of LIM2 in Ret for mitogenic signaling and of the target of LIM3 in InsR for endocytosis.  (26). Fluorescence-activated cell sorting employing the 528 monoclonal anti-EGFR antibody was used to isolate a subline expressing high levels of this chimeric receptor. The 12CA5 anti-HA antibody was purchased from BAbCO, Berkeley, CA; the anti-InsR antibody (27) was a gift of Dr. Lynn Seely, University of California San Diego, La Jolla, CA, and the anti-phosphotyrosine antibody PY-20 was from Transduction Laboratories (Lexington, KY). A rabbit polycloncal anti-Ret antibody was raised against the 20-amino acid peptide (Lys-Arg-Arg-Asp-Tyr-Leu-Asp-Leu-Ala-Ala-Ser-Thr-Pro-Ser-Asp-Ser-Ilu-Tyr-Asp) of the carboxyl terminus of the shorter splice isoform of Ret (28). The residue numbering in the present report follows that of Ret/ ptc2. For comparison with residue numbering in c-Ret, see Durick et al. (29).

Materials-The Caenorhabditis elegans
Synthesis and Purification of Peptides-Peptides were synthesized on an Advanced Chemtech MPS 350 at the Center for Molecular Genetics Peptide Synthesis Facility (University of California, San Diego) using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. Peptides were dissolved in water and purified on a Sephadex G-15 column equilibrated in 0.12 M triethylammonium bicarbonate. Peak fractions (monitored at A 274 or A 280 ) were dried in a speed vac (Savant), resuspended in water, and dried until the pH became neutral (total of four times). Peptides containing Trp were quantitated using A 280 and an extinction coefficient of 5600 M Ϫ1 cm Ϫ1 . Tyr-containing peptides were quantitated using A 274 and an extinction coefficient of 1400 M Ϫ1 cm Ϫ1 . The peptide lacking any absorptive amino acid was quantitated using amino acid analysis.
Mammalian Expression Vectors-The HA expression plasmid was constructed by subcloning a DNA fragment that encoded a Met initiation codon and an HA epitope tag into the mammalian expression vector pcDNA3 (Invitrogen) at the BamHI and EcoRI sites of the polylinker region. cDNA clones corresponding to various regions of Enigma were excised from the pJG4-5 yeast expression vector using the restriction enzymes EcoRI and XhoI and fused in-frame to the HA epitope-tagged vector. The Ret/ptc2 and mutant Ret/ptc2 constructs in the RC/CMV expression plasmid (Invitrogen) were prepared as described elsewhere (29). All constructs were confirmed by sequencing (U. S. Biochemical Corp.) Production of GST Fusion Proteins-Construction of plasmids expressing GST fusion proteins with the LIM domains of Enigma has been described (17). Two polymerase chain reaction primers were synthesized and used to isolate a DNA fragment encoding the carboxyl-terminal 61 amino acids (residues 536 -596) of Ret. The polymerase chain reaction product bearing 5Ј EcoRI and 3Ј Sa1I sites was cloned into the polylinker region pGEX-KG (Pharmacia Biotech Inc.). GST fusion proteins were prepared using standard procedures and the fusion proteins were immobilized on glutathione agarose beads (Sigma).
The Yeast Two-hybrid System-Reagents and procedures for the Lex-A based-yeast two-hybrid system were utilized (30). cDNAs coding wild type and mutant Ret/ptc2 were used to create LexA-Ret/ptc2 fusion proteins. cDNA fragments coding all the tested LIM domains were isolated by using Pfu polymerase and oligonucleotides that contained EcoRI and XhoI sites and fused in-frame to the pJG4-5 vector. Plasmids that direct the synthesis of LIM domains and full-length Enigma were introduced into EGY48/1840 yeast that contained different LexA fusion Ret/ptc2 constructs using the lithium acetate procedure. All constructs were confirmed by dideoxynucleotide sequencing, and the expression of fusion proteins of the appropriate size was confirmed by Western blotting with anti-HA antibody (BAbCO). ␤-Galactosidase activity was visualized on 5-bromo-4-chloro-3-indoyl ␤-D-galactoside galactose-containing plates or measured in solution according to Current Protocols of Molecular Biology (31). Yeast cells were grown in the presence of glucose or galactose and resuspended in an equal volume of buffer Z (100 mM NaPO 4 , 10 mM KC1, 1 mM MgSO 4 , 50 mM ␤-mercaptoethanol). Cells were diluted 1:10 or 1:20 with buffer Z and permeabilized by SDS and chloroform. o-Nitrophenyl ␤-p-galactopyranoside was used as a substrate and the reaction was stopped when a medium-yellow color had developed. Activity was calculated according to: units ϭ 1000 (A 420 /A 600 tv) when t ϭ time and v ϭ volume (17).
Affinity Precipitation and Peptide Competition Assays-NIH3T3 cells overexpressing the EGFR/Ret chimera or EGFR were treated with 100 nM EGF for 10 min at 37°C. Treated and untreated cells were lysed with a solution containing 50 mM Hepes (pH 7.4), 1% Triton X-100, 10% glycerol, 150 mM NaCl, 5 mM KCl, 1 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 10 mM benzamidine, 10 g/ml of aprotinin and leupeptin. Rat1 cells (HIRC) overexpressing InsR (18) and 293 cells expressing Enigma or Ret/ptc2 were lysed in a similar fashion. The 293 cells were transfected with the HA-Enigma or deletion mutants using the calcium phosphate precipitation procedure (32). Expression of Ret/ptc2 and mutant Ret/ptc2 was accomplished similarly. Cells were harvested 48 h after transfection. The lysates were incubated with GST fusion proteins that were immobilized on glutathione agarose beads for 1 h at 4°C with continuous agitation. For assay of peptide competition, the indicated concentrations of peptides were mixed with lysates prior to incubation with GST fusion proteins. Beads were then washed four times with lysis buffer. Material bound to beads was resolved by electrophoresis and probed with antibodies to Ret, InsR, Enigma, phosphotyrosine, or HA.

Identification of the Recognition Motif for LIM3 of Enigma-
To determine the recognition motif for the LIM3 domain of Enigma, a random peptide library selection technique was used to study the consensus binding site. Random peptide library selection has been successfully used to determine the sequence specificity of the peptide-binding sites of SH2 and SH3 domains as well as the optimal substrates of protein kinases (33,34). We constructed a fixed tyrosine peptide library comprising peptides of the sequence: Met-Ala-X-X-X-X-Tyr-X-X-X-X-Ala-Lys-Lys-Lys, where X indicates all amino acid except Trp, Cys, Ser, Thr, or Tyr. Trp and Cys were omitted to avoid problems with sequencing and oxidation. The total theoretical degeneracy of this library is 15 8 . The Met-Ala sequence at the amino terminus provides two amino acids to verify that peptides from this mixture are being sequenced. Sequencing of these two residues also provides quantitation of the peptides present. Ala-12 provides a second quantitation and an estimate of how much peptide loss occurred during sequencing. The poly-Lys tail prevented wash-out during sequencing and improved the solubility of the mixtures.
LIM3 of Enigma was expressed as a GST fusion protein in Escherichia coli. The fusion protein was immobilized on glutathione agarose and incubated with the tyrosine peptide library. Unbound peptides were washed away and bound peptides were released by acid and subjected to micro-sequencing. The amino acids preferentially selected by LIM3 of Enigma at positions Ϫ4, Ϫ3, Ϫ2, Ϫ1 amino-terminal to the Tyr residue and ϩ1, ϩ2, ϩ3, ϩ4 carboxyl-terminal to the Tyr residue are shown in Fig.  1A. The greatest selectivity was observed at the Ϫ1 and ϩ2 positions where Pro was preferred. At the Ϫ2 position, glycine was the preferred amino acid and at the ϩ3 position, both Val and Ile were highly selected. Phe was preferred at position ϩ4.
Because Pro was preferred at positions Ϫ1 and ϩ2, a second library with the sequence Met-Ala-X-X-X-X-X-Pro-X-X-Pro-X-X-X-Ala-Lys-Lys-Lys in which Pro was fixed with two intervening amino acids was designed to further test selectivity. This library lacks only Cys and Trp and has a degeneracy of 18 10 .
This library also included Tyr at the 10 degenerate positions. The general motif determined by this library was similar to those found with the Tyr-fixed peptide library (Fig. 1B). In addition, a Tyr residue was highly selected at position ϩ4. By comparison of these two motifs from two peptide library selections, the peptide sequence of Gly-Pro-Hyd-Gly-Pro-Hyd-Tyr/ Phe-Ala was determined to be the recognition motif for LIM3 of Enigma (Fig. 1C). This peptide sequence is highly homologous to the sequence of exon 16 of InsR.
To confirm the binding motif for LIM3/Enigma, peptides were tested for their capacity to disrupt the complex of LIM3/ Enigma with holo InsR. HIRC cell lysates containing InsR were incubated with GST-LIM3/Enigma without or with competitor peptides. As shown in Fig. 2, binding of InsR to GST-LIM3/ Enigma was inhibited by a 12-amino acid peptide (Asp-Gly-Pro-Leu-Gly-Pro-Leu-Tyr-Ala-Ser-Ser-Asn) corresponding to exon 16 of the InsR but not by mutant peptides (Asp-Gly-Pro-Leu-Ala-Pro-Leu-Ala-Ala-Ser-Ser-Asn and Asp-Gly-Ala-Leu-Gly-Ala-Leu-Tyr-Ala-Ser-Ser-Asn). The single substitution of the Leu immediately preceding the Tyr for Ile did not affect its ability to compete for InsR binding, confirming the random peptide library selection of a hydrophobic residue at position ϩ3. These peptide competition results demonstrate that both the Pro and Tyr residues are required to mediate interaction of LIM3 of Enigma with exon 16 of InsR.
Differential Recognition of Ret/ptc2 and InsR by LIM Domains of Enigma-When an oncogenic form of Ret, Ret/ptc2, was used in a yeast two hybrid screen to identify interacting proteins, several SH2 domain containing proteins and Enigma were isolated (25). To determine the domains of Enigma responsible for this interaction with Ret/ptc2, regions of Enigma were cloned into pJG4 -5 and tested for their ability to bind Ret/ptc2 expressed as a LexA fusion protein in pEG202 in yeast. As shown in Fig. 3A, full-length Enigma bound similarly to Ret/ptc2 and exon 16/InsR. The amino-terminal 279 amino acids of Enigma did not interact with either Ret/ptc2 or exon 16/InsR. The carboxyl-terminal 275 amino acids containing the three LIM domains were thus responsible for protein⅐protein interactions with both Ret and InsR. When the three LIM domains of Enigma were divided into individual LIM domains, LIM2 of Enigma bound Ret/ptc2 but not exon 16/InsR. Conversely, LIM3 of Enigma bound exon 16/InsR but not Ret/ptc2. These results demonstrate that LIM2 of Enigma was responsible for the association of Enigma with Ret/ptc2 and could be physically separated from LIM3 of Enigma which was responsible for Enigma association with InsR. LIM1 of Enigma, which bound two atoms of zinc characteristic of LIM domains 3 did not associate with either RET/ptc2 or InsR.
The specificity of interaction of Ret/ptc2 and InsR with other LIM domains was further examined. As shown in Fig. 3B, LIM domains of Mec-3, Isl-1, Lmx-1, zyxin, CRP, and paxillin did not recognize Ret/ptc2 or exon 16/InsR. The specificity for recognition thus resides in the LIM domains of Enigma.
Mapping the Interaction Site of of Ret with Enigma-Deletion of the carboxyl terminus of Ret/ptc2 distal to the conserved tyrosine kinase core abolished both mitogenic activity and Enigma binding (25). To determine whether the carboxyl-terminal region of Ret is sufficient to support the interaction, the carboxyl-terminal 61 amino acids of Ret/ptc2 (residues 536 to 596) were expressed as a GST fusion protein (GST-CЈ/Ret) and tested for their ability to bind Enigma. GST and GST-CЈ/Ret were immobilized on glutathione agarose and mixed with fulllength, amino-terminal or carboxyl-terminal domains of Enigma that were expressed as HA epitope-tagged fusion proteins in 293 cells (Fig. 4A). Equal amounts of GST and GST-CЈ/Ret were assessed for their ability to bind these Enigma proteins (Fig. 4B). GST-CЈ/Ret bound full-length Enigma and 3 D. Winge, personal communication.

FIG. 1. Substrate specificity of LIM3 of Enigma determined by binding degenerate peptide libraries.
Values in parentheses indicate the relative selectivity for the amino acids and X indicates no selectivity. The one-letter amino-acid code is used. LIM3 of Enigma was expressed as a GST fusion protein and immobilized on glutathione agarose. A Tyr-fixed random peptide library (A) or a Pro-fixed random peptide library (B) was presented to the immobilized GST-LIM3/Enigma. The unbound peptides were washed away and the retained peptide mixture was sequenced. The consensus peptide sequence is compared to the endocytic code of exon 16 of InsR (C). the carboxyl terminus containing the three LIM domains of Enigma but failed to bind the amino terminus of Enigma. There was no binding of any of the three forms of Enigma protein to GST. These results indicate that the carboxyl-terminal 61 amino acids of Ret contain the motif that is required for the association of Ret with Enigma.
To map the binding site within the carboxyl terminus of Ret, peptides were used to specifically disrupt the complexes of Ret with Enigma. Four 20-amino acid peptides (Fig. 5A) were assayed for their ability to disrupt the interaction of Ret with Enigma. Full-length Enigma was mixed with the indicated peptides and interacted with immobilized GST-CЈ/Ret. As shown in Fig. 5B, incubation with wild type peptide, Asn-Lys-Leu-Tyr, and mutant peptide, Asn-Lys-Leu-Phe, were sufficient to displace Enigma binding to the carboxyl terminus of Ret, while the Ret mutant peptide Ala-Lys-Leu-Ala and the peptide corresponding to exon 16/InsR were without effect. These results indicate that the sequence Asn-Lys-Leu-Tyr in the carboxyl terminus of Ret is necessary for interaction with the LIM2 domain of Enigma. Mutation of the Asn and Tyr residues abolished interaction but a Phe substitution for Tyr was tolerated.
Specificity was further assessed by using these peptides to disrupt interaction of an EGFR/Ret chimera (26) with immobilized GST-LIM2 of Enigma. NIH3T3 cell lysates expressing EGFR/Ret were incubated with GST-LIM2/Enigma in the presence of the indicated peptides, and bound receptors were detected by an anti-Ret antibody. Similar to the results in Fig. 5B, peptides Asn-Lys-Leu-Tyr and Asn-Lys-Leu-Phe displaced EGFR/Ret binding to LIM2/Enigma (Fig. 5C). However, the peptide with the two amino acid mutation to Ala-Lys-Leu-Ala and the exon 16/InsR peptide failed to compete for the binding, confirming the Asn-Lys-Leu-Tyr sequence at the carboxyl terminus of Ret is the core recognition site for LIM2/Enigma. EGFR alone did not interact with Enigma (data not shown).
Although the mutant peptide Asn-Lys-Leu-Phe blocked the interaction of Enigma and Ret using GST-fusion protein assays, mutation of Tyr 586 to Phe in Ret/ptc2 decreased this interaction as assayed in a yeast two hybrid system and decreased Ret/ptc2-stimulated DNA synthesis in microinjection experiments in mouse fibroblasts (25). To clarify these differing results, the effects of replacement of Tyr 586 with Phe in Ret/ ptc2 on Enigma binding were quantitated. Wild-type Ret/ptc2 (Tyr 586 ) and mutant Y586F Ret/ptc2 (Phe 586 ) were expressed in 293 cells and the relative affinities of these proteins for the LIM domains of Enigma were measured. As shown in Fig. 5D the affinity of Tyr 586 exceeded that of Y586F Ret/ptc2 for Enigma by approximately 5-fold. Deletion of the carboxyl terminus containing this region, i.e. Ret/ptc2 truncated at residue 574 completely abolished the interaction (data not shown). LIM2 of Enigma thus recognized the Phe substituted carboxyl terminus of Ret but with lower affinity compared to wild type Ret/ptc2 with the Tyr-containing sequence. The differing results using the yeast two hybrid system and in vitro peptide competition are explained by the decreased affinity of Y586F compared to wild type Ret/ptc2 for LIM 2 of Enigma.
Tyrosine Phosphorylation of Ret Is Not Required for Binding to LIM2 of Enigma-To investigate whether tyrosine kinase activation was required for the association between LIM2/ Enigma and Ret/ptc2, the EGFR/Ret chimeric protein was used. Because the ligand for the Ret tyrosine kinase receptor is unknown, the chimera generated by fusing the extracellular and transmembrane domains of EGFR and the intracellular domain of Ret was used (Fig. 6A). EGF activated the Ret tyrosine kinase activity and mitogenic responses of this chimera (26). NIH3T3 cells overexpressing EGFR/Ret were treated without or with EGF and cell lysates were mixed with GST or GST-LIM domains of Enigma. As shown in Fig. 6B, only GST-LIM2/Enigma interacted with EGFR/Ret. GST, GST-LIM1, or GST-LIM3 of Enigma did not bind. Ligand treatment did not effect the binding of EGFR/Ret to LIM2 of Enigma (left panel). Phosphorylation of the EGFR/Ret chimera was also examined by Western blotting using an anti-phosphotyrosine antibody. GST-LIM2 of Enigma interacted with phosphorylated as well as unphosphorylated EGFR/Ret receptors. The interaction of LIM2 of Enigma with Ret is thus independent of ligand activation and Ret autophosphorylation.

DISCUSSION
The growing number of proteins that contain one or more LIM domains function in a variety of pathways and locations within the cell, implicating LIM domains as versatile protein modules that are capable of acting in diverse cellular contexts. Although the NMR structure of LIM2 of CRP resembles the DNA binding domain of the GATA-1 transcription factor, no direct evidence that a LIM domain binds to nucleic acids has been presented. Indeed, a lack of affinity for target DNA sequences has been reported for the LIM domains of Mec-3 (35).

FIG. 2. Peptide competition with InsR binding to LIM3 of Enigma (A) lysates of Rat1 cells that overexpressed InsR were incubated with GST-LIM3/Enigma in the presence of varying concentrations of the indicated peptides.
GST beads were washed and the associated InsR eluted and detected by Western blotting using an antibody specific to the ␤ subunit of InsR. B, bound receptor was quantitated by scanning densitometry and plotted against the concentration of competitor peptide. The amount bound in the absence of competitor peptide was set at 100%.
Most available evidence indicates that LIM domains function in specific protein⅐protein interactions (1,8).
The present studies demonstrate that LIM2 of Enigma specifically interacts with Ret while LIM3 of Enigma specifically interacts with InsR. The Asn-Lys-Leu-Tyr sequence at the carboxyl terminus of Ret was essential for the formation of the Ret⅐Engima complex. For the interaction of InsR with Enigma, the Gly-Pro-Leu-Gly-Pro-Leu-Tyr sequence of the juxtamembrane region of InsR was required. Both LIM2 and LIM3 recognized Tyr-containing motifs located outside of the tyrosinekinase cores of Ret and InsR. Although the recognition motifs for LIM2 and LIM3 of Enigma share sequence similarity, they were not exchangeable, demonstrating that the two LIM domains have the ability to distinguish between two Tyr-based motifs. These results also indicate that individual LIM domains within a single protein have distinct partner preferences. Because the structural features of LIM domains are highly conserved, sequences other than the conserved residues that are involved in metal coordination must be important for defining the selectivity of individual LIM domains for their particular partner. LIM1 of Enigma failed to recognize either Ret or InsR and is likely to have a yet unidentified target protein in cells.
Tyr-based motifs serve a number of functions. Tyr-containing tight turns are the essential structural feature of the endocytic codes of many proteins (19). Four to six amino acid sequences containing an essential Tyr residue also function as lysosomal and trans-Golgi to basolateral surface targeting codes (36 -38). There is no evidence for covalent modification of Tyr residues within these trafficking codes. Phosphorylated Tyr residues in specific sequence contexts serve as the recognition motif for SH2 and PTB domains (33,39,40). The present studies indicate that Tyr-containing sequences also function as recognition elements for certain LIM domains. The Gly-Pro-Leu-Tyr motif of InsR that is recognized by LIM3 of Enigma forms a Tyr-containing tight turn (20). Use of two random peptide libraries indicated that the target recognized by LIM3 of Enigma consisted of the more extended sequence Gly-Pro-Leu-Gly-Pro-Leu-Tyr-Ala. The Asn-Lys-Leu-Tyr motif of Ret that is recognized by LIM2 of Enigma resembles the endocytic sequence in the LDL receptor which forms a Tyr-containing tight turn (41).
The function of the target sequences in Ret and InsR are different. LIM3 of Enigma recognized the major endocytic code of InsR while LIM2 of Enigma interacted with the carboxyl terminus of Ret. Ret is a protein tyrosine kinase receptor implicated in several disease processes. Mutations that inactivate its tyrosine kinase result in Hirschsprung's disease characterized by defective sympathetic innervation of the large intestine (42,43). Activating mutations characterize a group of inherited multiple endocrine neoplasia type syndromes that include MEN2A, MEN2B, and familial medullary thyroid cancer (44,45). Gene rearrangements including the one fusing the type 1 regulatory subunit of cyclic AMP-dependent protein kinase to the tyrosine kinase domain of Ret (Ret/ptc2) occur as oncogenic events in papillary thyroid carcinoma (28). In a nuclear microinjection assay the mitogenic activity of Ret/ptc2 was abolished by carboxyl-terminal truncation to residue 574 or by the mutation Y586F (25,29). Deletion of sequences distal to residue 574 abolished interaction with Enigma and mutation of Tyr 586 reduced the affinity of Enigma for Ret. Additionally, co-expression of the LIM domains of Enigma blocked the mitogenic activity of Ret/ptc2, implicating involvement of Enigma in the mitogenic signaling of Ret (25). The ability of LIM3 of Enigma to recognize the active endocytic codes of InsR fulfills the first property of the endocytic mechanism, but additional functional criteria will be necessary to critically test the hypothesis that Enigma functions in endocytosis of InsR.
Interactions of LIM domains of Enigma with these two receptors did not require either tyrosine kinase activity or tyrosine autophosphorylation on their target sequences. However, for both endocytosis of InsR and mitogenic signaling by Ret, activation of tyrosine kinase activity is necessary. Interactions of Enigma with InsR and Ret are thus proposed to be necessary but not sufficient to support these biological processes. Interestingly, Tyr 1062 in holo Ret corresponding to Tyr 586 in Ret/ptc2 is phosphorylated when expressed in COS cells (46). The stoichiometry was not determined but this could provide a mechanism for the reversible association of Ret with Enigma.
Most LIM proteins contain more than one LIM domain (1,8).
Finding distinct targets for two LIM domains of a single protein not only supports the hypothesis that LIM domains function in protein⅐protein interactions but indicates a possible adaptor function to assemble multiple proteins into a complex. Alternatively, multiple LIM domains could function to differentially assemble proteins with distinct receptors. The finding that binding of InsR and of Ret to Enigma does not require the protein tyrosine kinase activity of either receptor distinguishes LIM domain interactions from those of SH2 and PTB domains which direct assembly that is dependent on tyrosine kinase activity and covalent modifications of proteins (33,39). Because the processes requiring the target sequences in InsR and Ret that are recognized by Engima do depend on the tyrosine kinase activity of these receptors, mechanisms of function of the assembled LIM domain complexes must coordinate with tyrosine kinase activity perhaps via substrate phosphorylation or protein assemblies.