Medium Chains of Adaptor Complexes AP-1 and AP-2 Recognize Leucine-based Sorting Signals from the Invariant Chain*

Interactions between tyrosine- and leucine-based sorting signals in the cytoplasmic tails of transmembrane proteins and adaptor complexes AP-1 and AP-2 are believed to be the first step in the formation of clathrin-coated vesicles that deliver these proteins to their destination. Medium chains of AP-1 and AP-2 have been reported to interact with tyrosine-based sorting signals in a number of in vitro assays. In the present study we found that recombinant medium chains could interact with leucine-based sorting signals from the cytoplasmic tail of the invariant chain. Medium chains may therefore be responsible for the proper recognition of both tyrosine and leucine sorting signals by AP-1 and AP-2 complexes.

ponents of AP-1 or AP-2 and any of the leucine signals has been reported so far.
The invariant chain (Ii) contains two independent leucine signals in its cytoplasmic tail that are responsible for directing the major histocompatibility class II-Ii complexes to the endocytic compartments (7)(8)(9). The cytoplasmic tail of Ii has been shown to be important for AP-1 recruitment to the major histocompatibility class II-Ii complexes at the trans-Golgi network (10). Furthermore, a recent phage display-based study from our laboratory 2 has identified a short conserved sequence in the medium chains of the adaptor complexes that recognized a variety of tyrosine and leucine sorting signals, including both Ii signals. We therefore decided to test the ability of recombinant 1 and 2 to interact with the endosomal sorting signals from Ii.

MATERIALS AND METHODS
Reagents-The QIAexpress Type IV kit, Qiaex II DNA purification kit, and monoclonal anti-His antibody were from Qiagen. Materials for polyacrylamide gel electrophoresis and horseradish peroxidaseconjugated to goat anti-mouse IgG were from Bio-Rad. The PVDF membrane and ECL reagents were from Amersham Pharmacia Biotech. Oligonucleotides were synthesized by Medprobe (Norway). Dynabeads coated with streptavidin were from Dynal. Other reagents were from Sigma.
Peptides-Peptides containing the first leucine-based (LI) signal from Ii and its alanine mutant were synthesized at the Biotechnology Center of Oslo (Norway). Their sequences were M 1 DDQRDLI-SNNEQL 14 K and M 1 DAARDAASNNEQL 14 K, respectively. Note that Asp 3 and Gln 4 residues shown to be a part of the sorting motif (11,12) were changed to alanine in the mutant peptide in addition to Leu 7 and Ile 8 . Aliquots of these peptides were biotinylated (1:1 molar ratio of peptide and biotin) for use in binding assays. Biotinylated peptides containing the second (ML) signal and its mutant were kindly provided by Dr. G. Banting (University of Bristol, United Kingdom). Their sequences were biotin-N 10 NEQLPMLGRR 20 and biotin-N 10 NEQLPAAG-RR 20 , respectively (numbers are given according to the position of the residue in the p33 form of Ii).
Expression and Purification of Histidine-tagged Proteins-DNA coding for mouse 1 and rat 2 was kindly provided by Dr. T. Kirchhausen (Harvard Medical School). The full-length medium chains were cloned in frame into the type IV pQE30 vector (Qiagen) to express constructs containing a histidine (His) tag at their N termini. The oligonucleotides used for PCR amplification were 5Ј-TCCCGGGGATCCATGTCCGCCA-GCGCCGTCTAC-3Ј and 5Ј-TCTAGGCAAAGCTTTCACTGGGTCCGG-ACCTGATA-3Ј for 1 and 5Ј-GAGCTCGGTACCATGATCGGAGGCTT-ATTCATC-3Ј and 5Ј-TCTAGGCAAAGCTTCTAGCAGCGGGTTTCGT-AAAT-3Ј for 2 . Amplified constructs were purified with Qiaex II kit (Qiagen) and cloned into BamHI and HinDIII (for 1 ) or KpnI and HindIII (for 2 ) sites of pQE30. Proteins were expressed in the bacterial strain M15[pREP4] (Qiagen) according to the manufacturer's protocol. Both proteins formed inclusion bodies, which were solubilized in 6 M guanidinium hydrochloride containing 10 mM ␤-mercaptoethanol. Proteins were purified in one step under denaturing conditions on nickelnitrilotriacetic acid resin (Qiagen) according to specifications of the manufacturer. Purified proteins were diluted to a concentration of 10 -20 g/ml and refolded in the Binding buffer (0.1 M Tris, 5 mM EDTA, 0.1% Triton X-100, pH ϭ 7.5). Prior to the binding assay, proteins were centrifuged for 1 h at 100,000 ϫ g (Airfuge) to remove the insoluble matter. His-tagged dihydrofolate reductase was expressed from the control plasmid pQE16 supplied with the kit and purified according to manufacturer's recommendations. Protein concentration was determined from Coomassie-stained gels by comparison with protein standards.
Binding Assay-Biotinylated peptides containing leucine signals * This work was supported by the grants from the Norwegian Cancer Society and the Norwegian Research Council (to O. B. and D. G. R.). 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.
were immobilized on Dynabeads coated with streptavidin (Dynal) overnight at 4°C on a rotating wheel. The amount of Dynabeads used for a single assay was 20 l, which was sufficient to immobilize about 0.05 g of a biotinylated peptide. The unbound peptide was washed out with copious amounts of Binding buffer, and Dynabeads were further blocked with 5% BSA in Binding buffer for 60 min at room temperature. His-tagged medium chains were added to the immobilized peptides typically at 1-2 g/assay (unless specified otherwise) in a total volume of 450 l of Binding buffer containing 1% BSA and incubated on a rotating wheel for 60 min at room temperature. The peptides immobilized on Dynabeads were gently washed three times on a magnet with 1 ml of Binding buffer. Bound medium chains were rescued with SDS-polyacrylamide gel electrophoresis loading buffer and resolved by SDS-polyacrylamide gel electrophoresis on 10% polyacrylamide gels. Proteins were transferred to PVDF membranes (Amersham Pharmacia Biotech) in a mini-trans-blot electrophoretic transfer cell (Bio-Rad) in 25 mM Tris, 192 mM glycine, 10% methanol. Membranes were blocked overnight with 5% skim milk in phosphatebuffered saline at 4°C. Membranes were then incubated with anti-His IgG (1:5,000 dilution) for 60 min at room temperature, washed with 0.2% Tween 20 in phosphate-buffered saline, and further incubated with a horseradish peroxidase-conjugated anti-mouse IgG (1: 3,000 dilution) for 1 h. Bands were detected with ECL reagents (Amersham Pharmacia Biotech) using Kodak X-Omat AR film. Images were scanned using Adobe Photoshop software, and band intensities were quantified with the Gel-Pro Analyzer program. Different exposures of films were quantified to ensure the linearity of the signal. High contrast prints of the images are shown.
Inhibition Experiments-His-tagged 1 (1 g) was incubated with different amounts of competing peptides in 400 l of Binding buffer overnight at 4°C. The mixtures were added to LI peptide immobilized on Dynabeads and incubated for 60 min at room temperature in the presence of 1% BSA. Subsequent steps were performed as described for the binding assay.

RESULTS
We first examined the interactions between the two leucine signals from Ii and 1 . Equal amounts of peptides encoding the wild type and the mutant signal immobilized on Dynabeads were incubated with 1 . As shown in Fig. 1A, 1 preferentially bound to the peptide containing the first signal (LI signal hereafter). Interactions between 1 and the peptide-containing alanine mutant of LI signal were over 5-fold weaker as compared with the wild type peptide (Fig. 1B). Similarly, we found that 1 interacted with the peptide containing the second leucine signal (ML signal hereafter) over 4-fold stronger than with the peptide containing mutated ML signal (Fig. 1, C and  D). To rule out the possibility that the observed interactions were dependent on the short histidine tag and not on 1 sequence itself, we investigated interactions between the peptides containing LI and ML signals and His-tagged dihydrofolate reductase. No binding to either LI or ML signal was observed (data not shown).
We then investigated the dependence of binding of 1 to LI signal on the concentration of the medium chain. Equal amounts of the LI peptide immobilized on Dynabeads were incubated with the increasing amounts of 1 . As shown in Fig. 2, binding reached saturation at about 2 g of 1 . To further demonstrate the specificity of interactions between 1 and LI signal, we studied them in the presence of competing peptides. A fixed amount of 1 (1 g/assay) was incubated overnight with various concentrations of either wild type or mutant LI peptide (0 -1,200 molar excess of a peptide over the medium chain), and the ability of 1 to bind to the wild type LI signal was then assayed. As shown in Fig. 3 (A and  B), binding of 1 to LI signal was strongly inhibited by the wild type LI peptide, whereas the inhibition by the mutant peptide was much less pronounced.
We also studied the interactions between the leucine signals from Ii and 2 . As shown in Fig. 3 (A and B), binding of 2 to the mutated LI signal was significantly less than to the wild type signal. However, no interactions between 2 and ML signal could be detected in our system, as both the wild type ML signal and its alanine mutant bound 2 at the background levels (data not shown). DISCUSSION In this study we demonstrated that the full-length medium chain of the AP-1 adaptor complex interacted with both leucine sorting signals from the invariant chain in an in vitro assay. We also demonstrated interactions between the LI signal from Ii and the full-length medium chain of the AP-2 adaptor complex. These interactions were specific as medium chains bound wild type signals with higher affinity than the mutated ones (Figs. 1 and 4). We chose to substitute residues important for endocytosis with alanines in the control peptides for two reasons. First, according to structure predictions based on NMR studies, the exchange of the residues in question to alanines does not distort the structure of the cytoplasmic tail of the invariant chain 3 and second, substitution of the residues in question for alanines led to inactivation of the sorting signals in vivo (7)(8)(9)11). We also showed that binding of 1 to the LI sorting signal from the invariant chain was saturable (Fig. 2). Finally, specificity of the interactions between 1 and the LI sorting signal was confirmed using competition assay. Binding of 1 to the LI signal was 75% inhibited in the presence of the 1,200 molar excess of the wild type peptide encoding LI signal over the medium chain (Fig. 3), whereas only 25% inhibition was observed with the same amount of the mutant peptide. A high excess of free peptide necessary to prevent the medium chain from binding to the immobilized peptide is probably due to the fact that only a minor fraction of the free peptide has the proper conformation for binding medium chains in the solution. Others have also found that a molar excess of 1,000 -15,000 of a free peptide was required to inhibit binding of AP-1 and AP-2 complexes to various sorting signals (e.g. Refs. 13 and 14).
Interactions between AP-1 and AP-2 complexes and both leucine and tyrosine sorting signals have been well documented (Ref. 2 and references therein and Refs. [15][16][17]. Furthermore, it has been demonstrated that interactions between tyrosine signals and AP-2 could be improved by the addition of phos-phoinositides that are phosphorylated at the D-3 position of the inositol ring or when AP-2 was in the clathrin coat (18). Medium chains of AP-1 and AP-2 were shown to recognize a variety of tyrosine-based sorting signals in yeast two-hybrid system and in a number of in vitro assays (19 -23), and this recognition is believed to be the basis for interactions between the tyrosine signals and AP complexes although interactions between the ␤-chain of AP-2 and a signal from asialoglycoprotein receptor have also been reported (24). On the other hand, the leucine signal from CD3␥ failed to interact with the medium chains in the two-hybrid system (19), leading to an opinion that leucine signals might be recognized by a different AP subunit. However, a recent phage-display study 2 has identified a sequence from the medium chains that was able to recognize a number of leucine-based signals. It is important to notice that this sequence was missing from some of constructs used in other studies (19,20,22,23) that were still able to bind tyrosine signals. One could therefore hypothesize that the medium chains contain at least two sites that are able to interact with different endosomal sorting signals.
The results presented here are mainly in line with those of Bremnes et al., 2 but in contrast to the phage-display studies our approach did not demonstrate interactions between 2 and ML signal. At present we are not able to explain this difference, but in the cell both LI and ML signals are independently involved in the internalization from the plasma membrane (8), and this might indicate that the AP-2 complexes interact with both signals in vivo. Such interactions may require additional residues around ML signal that were not present on the peptide used in our experiments. We are also aware of possible pitfalls associated with extrapolation of in vitro results to an in vivo situation as other subunits of AP complexes may be involved in the recognition of endosomal sorting signals either directly or indirectly by influencing the conformation of medium chains in the complex or the degree of their exposure to the cytoplasm. Nevertheless, studies of potential interactions between individual chains of AP complexes and sorting signals is an essential first step in understanding the detailed mechanism of how sorting signals are recognized.
Acknowledgments-We thank Dr. T. Kirchhausen for the medium chain DNA constructs and Dr. G. Banting for the peptides. We also 3 A. Motta, personal communications.