Inhibition of the Interaction between Tyrosine-based Motifs and the Medium Chain Subunit of the AP-2 Adaptor Complex by Specific Tyrphostins*

Several intracellular membrane trafficking events are mediated by tyrosine-containing motifs found within the cytosolic domains of certain integral membrane proteins. Many of these tyrosine motifs conform to the consensus YXXΦ (where Φ represents a bulky hydrophobic residue). This YXXΦ motif has been shown to interact with the medium chain subunits of adaptor complexes that generally link relevant integral membrane protein cytosolic domains to the clathrin coat involved in vesicle formation. The motif YXXΦ is also very similar to motifs that are targets for phosphorylation by tyrosine kinases. Tyrosine kinase inhibitors known as tyrphostins are structural analogues of tyrosine, and so it is possible that tyrphostins could also inhibit interactions between medium chains and YXXΦ motifs. TGN38 is a type I integral membrane protein containing a tyrosine motif, YQRL, within the cytosolic domain. We have previously shown that this motif interacts directly with the medium chain subunit of the plasma membrane localized AP-2 adaptor complex (μ2). We have investigated a range of tyrphostins and demonstrated a specific inhibition of the interaction between μ2 and the TGN38 cytosolic domain by tyrphostin A23 throughin vitro analysis and the yeast two-hybrid system. These data raise the exciting possibility that different membrane traffic events could be inhibited by specific tyrphostins.

Several intracellular membrane trafficking events are mediated by tyrosine-containing motifs found within the cytosolic domains of certain integral membrane proteins. Many of these tyrosine motifs conform to the consensus YXX⌽ (where ⌽ represents a bulky hydrophobic residue). This YXX⌽ motif has been shown to interact with the medium chain subunits of adaptor complexes that generally link relevant integral membrane protein cytosolic domains to the clathrin coat involved in vesicle formation. The motif YXX⌽ is also very similar to motifs that are targets for phosphorylation by tyrosine kinases. Tyrosine kinase inhibitors known as tyrphostins are structural analogues of tyrosine, and so it is possible that tyrphostins could also inhibit interactions between medium chains and YXX⌽ motifs. TGN38 is a type I integral membrane protein containing a tyrosine motif, YQRL, within the cytosolic domain. We have previously shown that this motif interacts directly with the medium chain subunit of the plasma membrane localized AP-2 adaptor complex (2). We have investigated a range of tyrphostins and demonstrated a specific inhibition of the interaction between 2 and the TGN38 cytosolic domain by tyrphostin A23 through in vitro analysis and the yeast two-hybrid system. These data raise the exciting possibility that different membrane traffic events could be inhibited by specific tyrphostins.
The trafficking and intracellular targeting of many transmembrane proteins relies upon various signal motifs found within the cytosolic domains of these proteins. One of the best characterized of these targeting motifs centers around a critical tyrosine residue often found within the sequence YXX⌽ (where ⌽ represents a bulky hydrophobic residue) (1,2). Various studies have shown that tyrosine-based motifs conforming to this consensus YXX⌽ motif can directly interact with the medium chain () subunits of clathrin-associated adaptor complexes, thus allowing the efficient incorporation of transmembrane proteins containing these motifs into clathrin-coated vesicles (3)(4)(5)(6)(7)(8). Such interactions have been shown for both 1 and 2, the medium chain subunits of AP-1 and AP-2, the trans-Golgi network (TGN), 1 and plasma membrane-associated clathrin adaptor complexes, respectively. Interactions have also been shown between tyrosine motifs and the more recently identified AP-3 complex medium chain subunits (3A and the neuronal specific 3B) (9). These interactions are critically dependent on the tyrosine residue, with mutations of this residue to alanine abolishing any in vitro interaction (3,4,7). Structural studies on peptides containing these motifs have shown that the critical tyrosine residue projects away from the peptide backbone, residing in either a tight turn or an ␣-helix structure (10 -13).
The sequence YXX⌽, which interacts with medium chain adaptor subunits, bears a remarkable similarity to sequences that can become phosphorylated on the tyrosine and, once phosphorylated, can interact with Src homology 2 (SH2) domains (14). This observation could indicate possible similarity in the binding pockets of medium chain adaptor subunits and tyrosine kinases that phosphorylate YXX⌽ type motifs. However, at present there are very few reports of tyrosine-based motifs that can interact with chains and can also become phosphorylated and interact with SH2 domains (15)(16)(17).
Tyrphostins are chemical compounds structurally analogous to the side chain of tyrosine that were originally developed as substrate-competitive inhibitors of the epidermal growth factor receptor tyrosine kinase (18,19). These chemicals have been widely used as tyrosine kinase inhibitors for the last 10 years in studies on the physiological roles of many different tyrosine kinases. Some tyrphostins have also been reported to inhibit endocytosis, autophagy, and vesicle formation from the TGN, which has been taken to indicate a possible role of tyrosine kinases in the control of these processes (20,21).
TGN38 is a type I integral membrane protein predominantly localized to the TGN that has been shown to reach the plasma membrane and become rapidly internalized and targeted back to the TGN (22)(23)(24)(25)(26). The cytosolic domain of TGN38 contains a consensus tyrosine-based motif, SDYQRL, which has been shown to be essential for the internalization of this protein via clathrin-mediated endocytosis (27). We and others have previously shown that this sequence and the complete cytosolic domain of TGN38 can specifically interact with 2, the medium chain subunit of the plasma membrane-associated clathrin adaptor complex, and that this interaction is critically dependent on the tyrosine residue (3,7). Due to the similarity between the sequences recognized by adaptor complex medium chains and at least some tyrosine kinases, we were interested in determining whether any tyrphostin tyrosine kinase inhibitors could inhibit the interaction of TGN38 and 2. The cytosolic domain of TGN38 is a particularly suitable sequence for such a study because we have already shown that the tyrosine residue in the SDYQRL motif in the cytosolic domain of TGN38 can be phosphorylated in vitro by the insulin receptor tyrosine kinase, whereupon it can bind to the SH2 domains of Syk (15). We have utilized an in vitro assay to test a range of tyrphostins and have shown a strong inhibition of the interaction between TGN38 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and 2 by tyrphostins A23 and A46. We have also demonstrated a strong inhibition of the interaction between TGN38 and 2 by tyrphostin A23 in yeast two-hybrid growth assays.

EXPERIMENTAL PROCEDURES
All reagents were purchased from Sigma (Poole, UK) unless otherwise stated Far Western Blot Interaction Assay-Soluble fusion proteins of the C-terminal two-thirds of 2 (fused to thioredoxin; TRX-2⌬N) and the wild type or Y333A mutant cytosolic domain of TGN38 (fused to glutathione S-transferase; GST-TGN38wt and GST-Y333A) were expressed and purified as described previously (7). The C-terminal two-thirds of 2 was used because this portion of 2 has previously been shown to be the domain responsible for interaction with tyrosine-based motifs, and the solubility of the protein is improved by deletion of the N terminus (6,28). 5 g of GST-TGN38wt, GST-Y333A, or GST alone were immobilized on nitrocellulose membrane strips by slot blotting, and the membranes were incubated in three changes of blocking buffer (Tris-buffered saline, 0.02% v/v Tween 20, 10% w/v dried milk powder) for 1 h at room temperature. Membranes were incubated with 10 g of TRX-2⌬N in the presence or absence of different tyrphostins (at concentrations of 1 order of magnitude above and 1 order of magnitude below the commercially reported IC 50 concentration for tyrosine kinase inhibition) in 2 ml of blocking buffer for 90 min at room temperature. Membranes were washed in six changes of washing buffer (Tris-buffered saline, 0.02% v/v Tween 20) for 1 h at room temperature, incubated with monoclonal anti-thioredoxin antibodies (Invitrogen), 1:2500 in blocking buffer for 30 min at room temperature, washed in six changes of washing buffer for 1 h at room temperature, incubated with horseradish peroxidase-conjugated anti-mouse antibodies, 1:10000 in blocking buffer for 30 min at room temperature, washed in six changes Trisbuffered saline for 1 h at room temperature, and developed using the chemiluminescence detection system (Boehringer Mannheim).
Yeast Two-hybrid Growth Assays-Yeast cells (strain L40 (29)) were co-transformed with vectors encoding the wild type TGN38 cytosolic domain fused to the LexA DNA binding domain (pBTM116-TGN38wt) and 2 fused to the VP16 transactivating domain (pVP16-2) or pVP16 alone, or transformed with a positive control vector encoding the LexA DNA binding domain fused to the VP16 transactivating domain (pLexA-VP16) as described previously (7). Quantitative growth assays were performed as described previously (28). Briefly, overnight cultures of each yeast strain were grown at 30°C in synthetic yeast medium containing histidine to allow all strains to grow to saturation. 10-ml aliquots of fresh synthetic medium lacking histidine were inoculated with 5 ϫ 10 Ϫ3 A 600 units of each saturated overnight culture in triplicate, in the presence or absence of tyrphostins (at a concentration of 1 order of magnitude above the reported IC 50 for tyrosine kinase inhibition). Cultures were incubated at 30°C for 72 h, and A 600 measurements were taken at the intervals shown in the figure legends. 3-Amino-1,2,4-triazole (3-AT) inhibition curves were generated by inoculating 2.5 ϫ 10 Ϫ3 A 600 units of the appropriate saturated overnight yeast cultures into triplicate 5 ml of synthetic medium lacking histidine, supplemented with 0, 0.1, 1, 10, or 100 mM 3-AT, and incubated at 30°C in the presence or absence of the appropriate tyrphostins (at a concentration 1 order of magnitude above the reported IC 50 for tyrosine kinase inhibition), and A 600 measurements were taken after 48 h. Repeat additions of appropriate tyrphostins (the same amount as originally added to each culture) were added to the yeast cultures every 24 h to counteract degradation of the tyrphostins.

RESULTS
Previous work has shown that the medium chain () subunits of clathrin-associated adaptor complexes can interact with tyrosine residues found within the motif YXX⌽ (where X ϭ any amino acid and ⌽ ϭ bulky hydrophobic residue) (3)(4)(5)(6)(7)(8)(9). Tyrosine residues that can be phosphorylated and bind to SH2 domains can also be found in remarkably similar sequence motifs, i.e. YXX⌽ (14). It has been reported that a single tyrosine motif can act as both a chain binding region and as a target for a tyrosine kinase and subsequent SH2 domain binding (16,17), but this is the only evidence for such an occurrence in vivo to date. However, peptides corresponding to this motif where the tyrosine is in the phosphorylated form no longer show binding to chains (4, 6). Due to the similarity of the motifs that interact with chains and some tyrosine kinases, we were interested in whether a family of structural analogues of the tyrosine side chain (tyrphostins, tyrosine kinase inhibitors) can disrupt the interaction between chains and their target tyrosine motifs.
Far Western Blot Assays-We utilized a far Western blot procedure to analyze the effect of a range of tyrphostin chemicals on the interaction of the cytosolic domain of TGN38 (containing the motif SDYQRL) and the AP-2 adaptor complex medium chain (2). Of all tyrphostins tested, only A23 and A46 showed a complete inhibition of the TGN38-2 interaction at a concentration 10 times above the reported IC 50 for inhibition of the epidermal growth factor receptor kinase (Fig. 1). Virtually all other tyrphostins at the same relative concentrations (with respect to the reported IC 50 for tyrosine kinase inhibition) demonstrated no inhibition of the TGN38-2 interaction, with signals the same as control levels (data for tyrphostins A1 and A47 shown in Fig. 1). Tyrphostins A8 and A25 demonstrated a slight inhibition of TGN38-2 interaction at concentrations 10 times above their IC 50 , with the interaction signal being weaker than control levels but not abolished (data not shown). Table I shows qualitative results of all tyrphostins tested, with complete inhibition of TGN38-2 interaction shown as ϩϩϩ, partial inhibition shown as ϩ, and no inhibition shown as Ϫ. All experiments were repeated 3-5 times.
Yeast Two-hybrid Growth Assays-The yeast two-hybrid system was utilized to further test the inhibition of the TGN38-2 interaction. Yeast expressing two proteins that interact, one fused to a DNA binding domain and one fused to a transactivating domain, can drive the expression of various reporter genes. One such reporter gene is the HIS3 gene that allows the growth of the yeast strain in the absence of histidine. The affinity of the interaction between two proteins is proportional to the rate of growth of the expressing yeast in the absence of histidine, and thus growth curves can be used to analyze the strength of an interaction between two proteins. Yeast expressing the TGN38 cytosolic domain fused to the LexA DNA binding domain and 2 fused to the VP16 transactivating domain were grown in the presence or absence of various tyrphostins, at a concentration 10 times above the reported IC 50 of tyrosine kinase inhibition. The rate of growth was followed by measuring the optical density at 600 nm of the cultures at various time points. Yeast cells expressing the positive control construct, LexA fused to VP16, thus acting as a permanent interaction, were also grown under the same conditions, and the growth rate was determined. TGN38-2 yeast cultures demonstrated a significantly slower rate of growth in the presence of tyrphostin A23 ( Fig. 2A), whereas positive control LexA-VP16 yeast cells were largely unaffected by the same treatment (Fig.  2B). This shows that tyrphostin A23 can inhibit the interaction of TGN38 with 2 in living yeast cells. No other tyrphostin tested demonstrated inhibition of the TGN38-2 interaction (data not shown). This included A46, which showed a strong inhibition in the far-Western blot assay. This was surprising considering the inhibition demonstrated by A23 in the growth assays but may be due to yeast cells being able to either exclude A46 from their cytosol or rapidly metabolize A46 to an inactive form.
Yeast Two-hybrid Growth Inhibition Assays-The affinity of an interaction between two proteins in the yeast two-hybrid system can also be demonstrated by using growth inhibition assays. 3-AT is an inhibitor of the yeast HIS3 gene product and therefore inhibits growth of histidine prototrophs that arise due to the TGN38-2 interaction. A higher affinity interaction between the TGN38 cytosolic domain and 2 in the two-hybrid system leads to a higher expression level of the HIS3 gene, and so a greater concentration of 3-AT is required to inhibit the growth of these yeast cultures. Yeast expressing the LexA-TGN38 cytosolic domain and the 2-VP16 fusion proteins were grown in the absence of histidine, in the presence of various 3-AT concentrations, and in the presence or absence of various tyrphostins, and the relative growth was determined by optical density readings at 600 nm after 48 h. Positive control LexA-VP16 yeast cells were also grown under the same conditions, and growth levels were determined. Growth of TGN38-2 yeast was inhibited at much lower 3-AT concentrations in the presence of tyrphostin A23 compared with the same yeast in the absence of A23 (Fig. 3A). Growth was almost completely inhib- interaction assay ϩϩϩ indicates a strong inhibition of the interaction between TGN38 cytosolic domain and 2, with no detectable signal observed. ϩ indicates a weak inhibition of the interaction between TGN38 cytosolic domain and 2, with a slightly weaker signal than control levels observed. Ϫ indicates no inhibition of the interaction between TGN38 cytosolic domain and 2, with no difference in the signal compared with control levels. The chemical structures for each tyrphostin are shown.

FIG. 2. Tyrphostin A23 inhibits the growth of yeast expressing
LexA-TGN38 and 2-VP16 but not yeast expressing LexA-VP16. Yeast transformants were grown in ϪHis liquid medium in the presence or absence of tyrphostin A23 and growth levels determined by A 600 readings at the time points shown. Growth rates depend on the expression level of the HIS3 gene product that is proportional to the affinity of interaction between the separately expressed LexA and VP16 fusion proteins. A, LexA-TGN38 ϩ 2-VP16 expressing yeast, ϩA23 (E) and ϪA23 (q). B, LexA-VP16 expressing yeast, ϩA23 (E) and ϪA23 (q). All A 600 readings are the means Ϯ S.D. of triplicate results. Where not shown, error bars lie within the data points.
ited at 1 mM 3-AT in the presence of A23, but in the absence of A23 growth was only completely inhibited at 100 mM. The growth of positive control LexA-VP16 yeast was unaffected by the presence of A23 (Fig. 3B). This further demonstrates a specific inhibition of the TGN38 cytosolic domain 2 interaction in the yeast two-hybrid system by tyrphostin A23. All other tyrphostins tested showed no difference in the 3-AT inhibition of growth in the presence or absence of each tyrphostin (data not shown).

DISCUSSION
Medium chain () subunits of the clathrin-associated adaptor complexes have previously been shown to interact with motifs conforming to the consensus YXX⌽ (where X ϭ any amino acid and ⌽ ϭ bulky hydrophobic residue) found within the cytosolic domains of many transmembrane proteins (3)(4)(5)(6)(7)(8). Furthermore, various tyrosine kinases can phosphorylate tyrosine residues found within the same consensus motifs, leading to the interaction of the phosphorylated motif with SH2 domains (14). However as yet only two individual YXX⌽ inter-nalization motifs have been reported to also act as substrates for tyrosine kinases (15)(16)(17). The observation that chains and at least some tyrosine kinases can recognize very similar motifs leads to the possibility that both chains and tyrosine kinases may have similar binding pockets for the target tyrosine. Structural studies on peptides containing the YXX⌽ motif from proteins internalized from the cell surface via AP-2-associated clathrin-mediated endocytosis have shown that the side chain of the critical tyrosine residue is exposed either in a tight turn or as part of an ␣-helix (10 -13).
Structural analogues of tyrosine, known as tyrphostins, have been widely used over the last 10 years as potent tyrosine kinase inhibitors in numerous studies (18 -21). Because tyrphostins presumably inhibit tyrosine kinases by competitive interaction with the tyrosine binding pocket, it is possible that these same chemicals could inhibit the interaction of chains with their target YXX⌽ sequences in the same manner. However, as yet there are no published reports of the effect of these tyrosine analogues on the interaction of YXX⌽ motifs with adaptor complex chains.
We have demonstrated a potent specific inhibition of the in vitro interaction between the cytosolic domain of TGN38 (containing the motif SDYQRL) and the medium chain subunit of the AP-2 complex, 2, by two tyrphostins, A23 and A46. All other tyrphostins tested showed little or no effect on the level of in vitro interaction between TGN38 and 2. Comparison of the structures of the tyrphostin chemicals shows very specific structural features of the tyrphostins required for the inhibition of the TGN38-2 interaction (Table I). Both tyrphostins A23 and A46 have hydroxyl groups in positions 3 and 4 of the phenyl ring with A46 containing an amide group in place of one of the nitrile groups found in A23 (A23 ϭ ␣-cyano-(3, 4-dihydroxy)-cinnamonitrile, A46 ϭ ␣-cyano-(3,4-dihydroxy)-cinnamide). However, tyrphostins A8 and A25 have almost identical structures to A23 but with 4-monohydroxylated or 3,4,5trihydroxylated phenyl rings, respectively (A8 ϭ ␣-cyano-(4hydroxy)-cinnamonitrile, A25 ϭ ␣-cyano-(3, 4, 5-trihydroxy)cinnamonitrile) but show only slight or no inhibition of the TGN38-2 interaction. Also tyrphostin A47 has a very similar structure to A46 with the oxygen atom of the cinnamide substituted with a sulfur atom to give a thiocinnamide (A47 ϭ ␣-cyano-(3,4-dihydroxy)thiocinnamide), but A47 also shows no inhibition of the TGN38-2 interaction.
We have also shown a significant inhibition of TGN38-2 interaction by tyrphostin A23 in the yeast two-hybrid system. These results support the observed in vitro data, demonstrating the same reproducible inhibition in a different environment. The environment in which the protein-protein interaction occurs in the yeast two-hybrid system is more reminiscent of in vivo conditions, suggesting that the observed inhibition in the far Western blot analysis is not an artifact of a purely in vitro assay. The inhibition of the interaction in the yeast twohybrid system is signified by an inhibition of the growth of the yeast cultures, and so it is conceivable that the growth inhibition was due to inhibition of some essential tyrosine kinase within the yeast cells. However, this seems unlikely due to the fact that the positive control yeast cultures showed a very similar growth curve irrespective of the presence of tyrphostin A23. Also other tyrphostins tested showed no inhibition of yeast growth even though they were at the same relative concentrations with respect to the reported IC 50 values for tyrosine kinase inhibition. The fact that tyrphostin A46 showed no inhibition of the TGN38-2 interaction in the yeast two-hybrid assays was surprising considering the potent inhibition observed in the in vitro analysis. However, this could be due to several factors. The yeast cells could be able to exclude A46 but FIG. 3. Tyrphostin A23 increases the susceptibility to 3-AT of yeast expressing LexA-TGN38 and 2-VP16 but has no effect on the 3-AT susceptibility of yeast expressing LexA-VP16. Yeast transformants were grown in ϪHis liquid medium, in the presence or absence of tyrphostin A23, and in the presence of varying concentrations of the histidine biosynthesis inhibitor 3-AT. Growth levels were determined by A 600 readings after 48 h of growth. A, LexA-TGN38 ϩ 2-VP16 expressing yeast, ϩA23 (E) and ϪA23 (q). B, LexA-VP16 expressing yeast, ϩA23 (E) and ϪA23 (q). All A 600 readings are the means Ϯ S.D. of triplicate results. not A23 from their cytosol or efficiently metabolize A46 but not A23. It should be noted that even A47, which differs from A46 only in the fact that the oxygen atom of the cinnamide group has been substituted by a sulfur atom, fails to inhibit the interaction between 2 and the cytosolic domain of TGN38 in vitro, thus demonstrating that a minor modification to the structure of A46 by the yeast cells may well be sufficient to lead to its inactivation. Another alternative would be that some component of the yeast cytoplasm/nucleoplasm could prevent A46 from interacting with 2 and so inhibit the interaction with TGN38 cytosolic domain, but this same component had no effect on A23.
In summary our data show that at least one tyrphostin appears to be able to specifically inhibit the interaction of a tyrosine-based motif with the AP-2 medium chain subunit, 2. As far as we are aware this has never been demonstrated before. Furthermore it is conceivable that these tyrphostins or other structural analogues of tyrosine could inhibit the interaction between tyrosine-based motifs and the other adaptor complex medium chains and possibly as yet unidentified medium chain homologues. This observation could lead to the possibility of inhibitor chemicals being designed based around the tyrphostin backbone that could inhibit specific membrane traffic pathways. Previous studies with tyrphostins have often shown potent inhibition of membrane traffic events at concentrations lower than those used in this study, and thus the previous data still indicate that tyrosine kinases play a role in the membrane traffic events studied. However, the results presented here suggest that caution should be used in future interpretation of the effects of tyrphostins on membrane traffic events due to the possible inhibition of chain interactions. Hopefully these data will lead to further research into the effect of these chemicals on membrane traffic in eukaryotic cells.