Walker A Lysine Mutations of TAP1 and TAP2 Interfere with Peptide Translocation but Not Peptide Binding*

We generated mutants of the transporter associated with antigen-processing subunits TAP1 and TAP2 that were altered at the conserved lysine residue in the Walker A motifs of the nucleotide binding domains (NBD). In other ATP binding cassette transporters, mutations of the lysine have been shown to reduce or abrogate the ATP hydrolysis activity and in some cases impair nucleotide binding. Mutants TAP1(K544M) and TAP2(K509M) were expressed in insect cells, and the effects of the mutations on nucleotide binding, peptide binding, and peptide translocation were assessed. The mutant TAP1 subunit is significantly impaired for nucleotide binding relative to wild type TAP1. The identical mutation in TAP2 does not significantly impair nucleotide binding relative to wild type TAP2. Using fluorescence quenching assays to measure the binding of fluorescent peptides, we show that both mutants, in combination with their wild type partners, can bind peptides. Since the mutant TAP1 is significantly impaired for nucleotide binding, these results indicate that nucleotide binding to TAP1 is not a requirement for peptide binding to TAP complexes. Peptide translocation is undetectable for TAP1·TAP2(K509M) complexes, but low levels of translocation are detectable with TAP1(K544M)·TAP2 complexes. These results suggest an impairment in nucleotide hydrolysis by TAP complexes containing either mutant TAP subunit and indicate that the presence of one intact TAP NBD is insufficient for efficient catalysis of peptide translocation. Taken together, these results also suggest the possibility of distinct functions for TAP1 and TAP2 NBD during a single translocation cycle.

The transporter associated with antigen processing (TAP) 1 is a critical component of the major histocompatibility complex (MHC) class I antigen presentation (1)(2)(3). TAP functions to translocate peptides from the cytosol to the ER. Binding of peptides to newly synthesized MHC class I molecules in the ER stabilizes the MHC class I heterodimer and allows transit of MHC class I-peptide complexes to the cell surface for immune surveillance by T cells (4). Two structurally related subunits of the TAP transporter, TAP1 and TAP2, form a complex on the ER membrane that is necessary and sufficient for peptide translocation from the cytosol into the ER. The cytosolic face of TAP1⅐TAP2 complexes contains a binding site for peptides (5), which can function in the sequestration of peptides derived from proteasomal proteolysis. A recently discovered protein called tapasin is associated with the TAP1⅐TAP2 complex (6,7). Tapasin has been shown to enhance the expression level of TAP1 and increase peptide transport by TAP complexes (8). However, tapasin is not required for peptide binding by TAP1⅐TAP2 complexes or for translocation per se, since TAP1⅐TAP2 complexes expressed heterologously in insect cells can bind and transport peptides (9).
TAP is a member of the ATP-binding cassette (ABC) family of transmembrane transport proteins (10). Members of this family transport various substrates across cellular membranes in an ATP-dependent manner. Known substrates of ABC transporters include amino acids, peptides, proteins, sugars, and lipids. The typical ABC transporter has two hydrophobic membrane-spanning regions with multiple membrane-spanning segments and two cytosolic nucleotide binding domains (NBD). The NBD contains several conserved sequence motifs. These include the Walker A and Walker B sequence motifs, which are characteristic of nucleotide binding folds (11). The membranespanning segments and the NBD can occur in a single polypeptide as in the mammalian cystic fibrosis transmembrane conductance regulator and P-glycoprotein, as two separate polypeptides as in the TAP transporters, or as four separate polypeptides as in many bacterial ABC transporters (10). TAP1 and TAP2 each comprise one membrane-spanning region with several membrane-spanning segments and one cytosolic NBD. In vitro experiments with intact TAP domains as well as with each truncated NBD have shown that both TAP1 and TAP2 bind ATP and ADP (12)(13)(14)(15). Indirect evidence for the occurrence of ATP hydrolysis during translocation comes from observations that the addition of exogenous ATP is required for peptide translocation by TAP and that translocation is not supported by nonhydrolyzable ATP analogs or ADP (16). However, the catalysis of ATP hydrolysis by TAP complexes or either NBD remains to be directly demonstrated. It is not known whether ATP hydrolysis by TAP1, TAP2, or the complex is required for a completion of a catalytic cycle. Furthermore, the requirement for nucleotide binding to each TAP subunit upon substrate (peptide) interactions with the TAP complex also remains to be unambiguously defined. Peptide binding to TAP was initially suggested to be nucleotide binding-independent, based upon studies of peptide binding to microsomal membranes expressing wild type TAP1 and TAP2 in the presence or absence of apyrase, an enzyme that depletes ATP and ADP (5). However, based upon more recent studies using TAP1 and TAP2 mutants that both lacked nucleotide binding activity, it was suggested that peptide binding by TAP is dependent upon nucleotide interactions with one or both subunits of the TAP1⅐TAP2 complex (17). Further experiments are required to resolve this discrepancy and to establish whether nucleotide binding to TAP1, TAP2, or both subunits alters the affinity of TAP1⅐TAP2 complexes for peptides.
Toward a definition of the requirement for nucleotide binding and hydrolysis by each TAP subunit for peptide binding and translocation, we generated mutants of TAP1 (K544M) and TAP2 (K509M) that were altered at a conserved lysine residue of the Walker A motif (GXXGXGK(S/T)) of each protein. Based upon the crystal structure of the ATP binding subunit of histidine permease, a bacterial ABC transporter, the Walker A lysines of TAP1 and TAP2 are predicted to be located on the phosphate binding loop (P-loop) and form contacts with the ␤-phosphate of ATP (18). Mutations of this lysine in other ABC transporters have been shown to impair nucleotide binding and/or hydrolysis and impair substrate translocation (19 -23). We observe that the Walker A lysine mutations in TAP1 and TAP2 have distinct effects upon nucleotide binding to each subunit, with nucleotide binding being significantly impaired in the TAP1(K544M) mutant but not in the TAP2(K509M) mutant. The observation that the TAP1(K544M) mutation is significantly impaired for nucleotide binding by TAP1 also allowed us to determine the correlation between nucleotide binding by TAP1 and peptide binding to TAP1⅐TAP2 complexes. Additionally, assays of peptide translocation by each mutant allowed insights into the requirement for nucleotide hydrolysis by TAP2 during a translocation cycle. Taken together, these results allow for the refinement of a model for the mechanistic steps involved during a single peptide translocation cycle.

Construction of Baculoviruses Encoding Mutant TAP Subunits-
The cDNAs for TAP1 and TAP2 were obtained from Dr. John Trowsdale. Polymerase chain reaction was used to introduce a sequence encoding a His 6 tag to the 3Ј-end of TAP1 for the TAP1-His construct, to introduce BamHI sites at both ends of the TAP1 cDNA, and to introduce BglII sites at both ends of the TAP2 cDNA. The modified cDNAs were subcloned into pPCR2.1. Site-directed mutagenesis was performed by the Kunkel method (24). The mutated cDNAs were sequenced, excised from pPCR 2.1, and ligated into pAcUW51 transfer vector. The transfer vectors and BaculoGold DNA (PharMingen) were cotransfected into insect cells as described in Ref. 43. Plaque assays were used to isolate pure viruses, which were used to reinfect cells for virus amplification.
Insect Cell Infection, Lysis, Immunoprecipitation, and Binding to ATP Beads-2 ϫ 10 7 Sf21 cells were infected with the appropriate baculovirus at a multiplicity of infection of 5-20, depending on the expression level of the individual virus. After ϳ72 h, the cells were harvested and lysed in 1 ml of lysis buffer (10 mM Tris, 10 mM phosphate, 130 mM NaCl, 1% Triton X-100, pH 7.5) supplemented with protease inhibitors (Complete mini-EDTA free tablets; Roche Molecular Biochemicals). The cells were lysed on ice for 45 min, and the lysates were cleared by centrifugation at 40,000 ϫ g for 45 min at 4°C. 0.5 ml of the lysate was then immunoprecipitated with 0.75 l of anti-His ascites, 200 l of anti-TAP1 (148.3) hybridoma supernatant (9), or 200 l of anti-TAP2 (435.3) (25) hybridoma supernatant. The samples were centrifuged again at 14,000 ϫ g for 5 min at 4°C, and the supernatants were transferred to an appropriate volume of protein A-Sepharose beads prewashed in wash buffer (10 mM Tris, 10 mM phosphate, 130 mM NaCl, 0.5% Triton X-100, pH 7.5) and incubated for 1 h at 4°C with shaking. The beads were then washed three times in wash buffer before adding SDS-PAGE sample buffer. Samples were separated on 8% SDS-PAGE gels, transferred to nitrocellulose membranes, and immunoblotted with 5 l of anti-His ascites, 150 l of anti-TAP1 hybridoma supernatant, or 250 l of anti-TAP2 hybridoma supernatant overnight. The membranes were then incubated with 5 l of goat anti-mouse alkaline phosphatase-conjugated antibody (American Qualex) for 2 h and developed with the Bio-Rad alkaline phosphatase conjugate substrate kit.
Cells were infected and lysed in an identical manner for the ATP beads binding experiment. Cells were lysed using the lysis buffer described above, containing 5 mM MgCl 2 . 0.5 ml of each cleared lysate was incubated for 2 h with an appropriate amount of preswollen ATP-, ADP-, or AMP-conjugated agarose beads (Sigma) or immunoprecipitated with antibodies as described above. The beads were then washed three times with lysis buffer, run on gels, transferred to membranes, and immunoblotted in the same manner as above.
Microsome Preparations-Microsomes were prepared as described previously (9). Briefly, 1 ϫ 10 8 SF21 cells were infected with the appropriate baculovirus at an multiplicity of infection of 5-20, depending on the expression level of the individual virus. After 72 h, the cells were harvested, washed once in ice-cold PBS, and then resuspended in 800 l of ice-cold cavitation buffer (250 mM sucrose, 25 mM potassium acetate, 5 mM magnesium acetate, 0.5 mM calcium acetate, 50 mM Tris, pH 7.4) supplemented with a protease inhibitor mixture. Cells were lysed by repeatedly drawing the suspension through a 26-gauge needle. After centrifugation at ϳ500 ϫ g for 5 min, the supernatant was added to 5 ml of 2.5 M sucrose in gradient buffer (2.5 M sucrose, 150 mM potassium acetate, 5 mM magnesium acetate, 50 mM Tris, pH 7.4). This was then overlaid with 3 ml each of 2.0 M sucrose in gradient buffer, then 1.3 M sucrose in gradient buffer, and finally with 800 l of cavitation buffer. This was centrifuged overnight at 80,000 ϫ g, and the microsomal fraction at the interface of the 2.0 and 1.3 M sucrose layers was collected. This fraction was diluted into 5 ml of PBS, 1 mM dithiothreitol, and centrifuged for 1 h at 100,000 ϫ g, and the pellet was resuspended in 1 ml of PBS, 1 mM dithiothreitol and frozen in aliquots at Ϫ70°C. The total protein content was determined by a BCA assay (Pierce).
Peptide Translocation Assay-For the translocation assays, a peptide of the sequence RRYNASTEL was synthesized, and iodinated using the chloramine-T labeling chemistry to a specific activity of 100 -150 Ci/ g. Microsome-based translocation assays were carried out as previously described (9). 30 g of insect cell microsomes containing wild type or mutant TAP1⅐TAP2 complexes or single subunits were added to 150 l of assay buffer (PBS, 0.1% bovine serum albumin, 1 mM dithiothreitol, pH 7.3) containing 10 mM MgCl 2 and 5 mM ATP (ϩATP samples) or 0.03 units/l of apyrase (ϪATP) (Sigma). Radioiodinated RRYNASTEL peptide was then added, and the samples were incubated at 37°C for 15 min. The samples were then centrifuged at 4°C at 8,800 ϫ g and washed once with 250 l of the appropriate assay buffer. The pellets were then resuspended in 250 l of lysis buffer (50 mM Tris, 150 mM NaCl, 1% Nonidet P-40, pH 7.4) and incubated on ice for 1 h. After centrifugation at 8,800 ϫ g for 5 min at 4°C, the supernatants were transferred to ConA-Sepharose (Amersham Pharmacia Biotech) beads and incubated for 2 h at 4°C with agitation. Beads were washed twice with the appropriate assay buffer, and the radioactivity was determined using a Beckman 5500 ␥-counter. Results from microsome-based translocation assays were verified using Sf21 cell-based assays, as per previously described protocols (26).
Peptide Binding Assay-Peptide binding assays were carried out using a recently described procedure (27). All peptides were synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. For the binding experiments, a peptide was synthesized with the sequence RRYQKCTEL. For fluorochrome labeling, the peptide was dissolved in PBS, 33% N,N-dimethylformamide. 5-Iodoacetamidofluorescein (Molecular Probes, Inc., Eugene, OR) was then added at a 1.2:1 molar ratio to peptide. The reaction was allowed to incubate in the dark for 2 h at room temperature. The modified peptide was purified by reverse-phase HPLC. Peptide concentrations were normalized between experiments by measuring the absorbance at 495 nm (A 495 ϭ 0.28 for 0.01 mg/ml peptide).
For the peptide binding assays, the fluorescence emission signal ( ex/em ϭ 470/515 nm) was recorded using a PTI fluorimeter equipped with a microstirring device. A 5-mm round quartz cuvette was filled with 180 l of assay buffer (PBS, 1 mM dithiothreitol, 5 mM MgCl 2 , pH 7.4) at room temperature, and the fluorescent peptide RRYQKCTEL was added with vigorous stirring (2.5-80 nM final concentration). The fluorescence signal was then allowed to stabilize. 20 l of insect cell microsomes treated either with 0.03 units/l of apyrase or with 1 mM ADP or ATP (for at least 30 min on ice) were added to the cuvette. The decrease in fluorescence due to TAP-dependent quenching of peptide fluorescence was monitored until equilibrium was achieved (280 s). The time dependence of quenching was determined by fitting the data to an exponential function, y ϭ F f ϩ ⌬F*e Ϫkt (where F f is the fluorescence value at steady state, ⌬F is the net change in fluorescence, and k is the rate constant) using nonlinear least squares analysis in the Prism software package (Graph Pad software). The total fluorescence quenching signal (⌬F) was plotted as a function of peptide concentration. The curve was fitted to the equation y ϭ ⌬F * [P]/(K d ϩ [P]), and K d values were determined by nonlinear least squares analysis.
After stabilization of fluorescence signals, unlabeled peptide was added at a concentration of ϳ30 M, and the fluorescence recovery was monitored over several seconds as bound fluorescent peptide dissociated. The dissociation rate constants, k d , were estimated by fitting the fluorescence recovery signal to a monoexponential function, y ϭ F e ϩ ⌬F * e Ϫkdt (where F e is the end point fluorescence value, ⌬F is the net change in fluorescence, and k d is the dissociation rate constant).
Each set of binding experiments included a control with microsomal membranes prepared from uninfected Sf21 cells. Fluorescence quenching signals were also sometimes observed with microsomes derived from uninfected cells, but recovery was typically not observed (Figs. 4 and 5). At peptide concentrations below 80 nM, the quenching signals observed for the control microsome preparations (if any) were typically significantly lower than those observed with microsome preparations containing wild type or mutant TAP complexes (Fig. 5, A-C). Variations in the quenching signal observed for microsome preparations from uninfected cells (ranging from negative amplitudes to low positive amplitudes as shown in Fig. 5, A-C) might arise due to peptide degradation during storage. In general, the lowest background signals were observed for experiments conducted within 1 week of HPLC purification of peptide, when the purified peptide was stored at Ϫ20°C under dessication.

Expression and ATP Binding by TAP1 and TAP2
Mutants-We modified the TAP1 cDNA by polymerase chain reaction, to introduce a sequence encoding a C-terminal hexahistidine tag (TAP1-His), to facilitate the biochemical analyses of TAP1 expression and function. The TAP1-His sequence was further modified by site-directed mutagenesis (24), in order to alter a conserved lysine residue in the Walker A motif (Lys 544 ) to a methionine. A similar modification was introduced into the cDNA sequence encoding TAP2, in order to alter the Walker A lysine (Lys 509 ) to a methionine. Mutant clones were sequenced and ligated into the baculovirus transfer vector pAcUW51. Recombinant baculoviruses were generated encoding histidinetagged TAP1 (TAP1-His), the corresponding Walker A lysine mutant (TAP1(K544M)-His), and the TAP2 Walker A lysine mutant (TAP2(K509M)). The anti-His antibody (Covance Scientific) was used to screen for TAP1 expression, and the TAP2specific antibody 435.3 (anti-TAP2) (25) was used to screen for TAP2 expression. Baculoviruses encoding wild type TAP1 (TAP1) and TAP2 (TAP2) were obtained from the laboratory of Dr. Robert Tampé (9). For comparisons of nucleotide binding by each mutant or wild type TAP subunit, insect cells were infected with baculoviruses encoding TAP1-His, TAP1(K544M)-His, TAP2, or TAP2(K509M) for ϳ72 h. Detergent lysates of cells from each infection were incubated with ATP, ADP, or AMP-Sepharose beads for 2 h. Subsequently, the beads were washed, and proteins were eluted from the beads by boiling in the presence of SDS-polyacrylamide gel electrophoresis buffer. Proteins associated with the beads were analyzed by SDSpolyacrylamide gel electrophoresis and immunoblotting with anti-His or anti-TAP2 antibodies (Fig. 1). Lysates were also immunoprecipitated with anti-His or anti-TAP2 antibodies to directly visualize the relative expression levels of mutant versus wild type proteins. At comparable expression levels of TAP1-His and TAP1(K544M)-His (Fig. 1A, lanes 4 and 8), strong binding of TAP1-His to ATP-Sepharose and ADP-Sepharose beads is visualized (Fig. 1A, lanes 1 and 2), whereas TAP1(K544M)-His binding to ATP and ADP beads is barely detectable (Fig. 1A, lanes 5 and 6). These observations indicate that the K544M mutation impairs nucleotide binding to TAP1. By contrast, TAP2(K509M) binding to ATP and ADP beads does not appear to be significantly impaired relative to wild type TAP2 (Fig. 1B, lanes 1 and 2 compared with lanes 5 and 6).
Subunit Association and Peptide Translocation by Wild Type and Mutant TAP Complexes-It has previously been shown TAP1 and TAP2 associate into stable complexes and that both subunits are required for peptide binding and translocation (28,29). Prior to analysis of the peptide binding and translocation properties of the mutants, it was therefore necessary to demonstrate that each mutant could form TAP1⅐TAP2 complexes. To assay the formation of TAP1(K544M)-His⅐TAP2 complexes, cells were infected with viruses encoding TAP1(K544M)-His and TAP2, or with TAP1-His and TAP2 as the positive controls. Detergent lysates were immunoprecipitated with either anti-His or anti-TAP2 antibodies. The samples were subsequently separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes, and the membranes were stained with anti-His or anti-TAP2. We observed that TAP1(K544M)-His associates with TAP2 as does TAP1-His ( Fig. 2A). Likewise, TAP2(K509M) associates with TAP1 as does wild type TAP2, as measured by coimmunoprecipitation analyses with the TAP1-specific antibody 148.3 (anti-TAP1) (9) and anti-TAP2 (Fig. 2B).
Peptide translocation experiments were carried out as previously described (9,26). The model substrate RRYNASTEL was used, which was 125 I-labeled. Microsomal membrane preparations containing wild type or mutant TAP complexes were incubated with the radiolabeled peptide in the presence or absence of ATP for 15 min at 37°C. Since the peptide sequence contains an NAS glycosylation motif, translocated peptide was retained in the ER rather than being exported (30). Free peptide was separated by centrifugation, the microsomal membranes were lysed in detergent, and glycosylated peptide was quantified by determining their binding to ConA-Sepharose beads. The experiments were carried out with microsome preparations expressing wild type and mutant TAP1⅐TAP2 complexes. Microsome preparations expressing single TAP subunits were used as negative controls, since it has previously been shown that microsomal preparations expressing wild type TAP1 alone or wild type TAP2 alone do not translocate peptides (9,25). The results of a representative experiment are indicated in Fig. 3. Within an experiment, we defined a translocation signal as positive when the average cpm ϩATP /cpm ϪATP ratio was at least 2-fold above the ratio observed for the single subunit control. By this criterion, results from three independent translocation experiments indicated that TAP1⅐TAP2- FIG. 1. Binding of TAP1-His, TAP1(K544M)-His, TAP2, and TAP2(K509M) to nucleotide-agarose beads. Insect cells were infected with the indicated baculoviruses for ϳ72 h. The binding of each protein to ATP-, ADP-, or AMP-agarose beads was assessed, and the total expression levels of wild type and mutant TAP1 and TAP2 were assessed using immunoprecipitations (IP) and immunoblotting with anti-His and anti-TAP2 antibodies, respectively. A, lanes 1 and 2 show that TAP1-His binds to ATP and ADP. Lanes 5 and 6 show that TAP1(K544M)-His is severely impaired in its ability to bind ATP and ADP relative to wild type TAP1-His. B, lanes 1 and 2 show that TAP2 binds to ATP and ADP. Lanes 5 and 6 show that the binding pattern for TAP2(K509M) is similar to wild type TAP2 and indicate that the mutant is not deficient in nucleotide binding.
For mutant TAP1(K544M)-His⅐TAP2 complexes, a low but positive translocation signal was observed in three independent translocation experiments. In each experiment, the average cpm ϩATP /cpm ϪATP ratio for the mutant complex was approximately 2-fold above the single subunit background (ratios for the single subunit controls ranged from 0.6 to 1.7 in the different experiments). The low translocation efficiency is apparent from the higher cpm ϩATP /cpm ϪATP ratio observed for wild type complexes although the expression level was significantly reduced compared with the mutant (Fig. 3B, compare lanes 3  and 4).
Peptide Binding by Wild Type and Mutant TAP Complexes-A fluorescence quenching assay has recently been described for quantitation of peptide binding to TAP complexes, under nucleotide-depleting conditions (27). We used these assays to quantify peptide binding by wild type and mutant TAP complexes. For this purpose, the peptide RRYQKCTEL was fluorescein-labeled using 5-iodoacetamidofluorescein and HPLC-purified. Microsomal membrane preparations containing each mutant TAP complex, the corresponding wild type TAP complexes, or neither TAP subunit (uninfected microsomes) were added to different concentrations of the fluorescent peptide, and the fluorescence emission signal was monitored. For the wild type as well as for the mutant TAP complexes, increased fluorescence quenching was observed as the peptide concentration was increased (Fig. 4, A-C). After stabilization of fluorescence signals, unlabeled peptide was added at a concentration of ϳ30 M, and the fluorescence recovery was monitored over several seconds as bound fluorescent peptide dissociated. As for wild type TAP complexes, fluorescence recovery was also observed for both TAP mutants, and the magnitude of the recovery was in proportion to the magnitude of the quenching. The magnitude of the fluorescence quenching and recovery signals obtained for wild type and mutant TAP complexes were critically dependent upon the expression levels of both TAP subunits. In the experiments shown in Fig. 4, A and C, the same microsome preparations of TAP1⅐TAP2 and TAP1⅐TAP2(K509M) were used as for the translocation assays shown in Fig. 3 (Fig. 3B, lanes 1 and 2). At lower expression levels, reliable fluorescence quenching and

FIG. 2. Association of wild type and mutant TAP1 and TAP2.
Insect cells were infected with baculoviruses encoding the indicated constructs. Detergent lysates of these cells were immunoprecipitated (IP) with the indicated monoclonal antibody, and immunoblotting analysis was carried out with an antibody directed against the same or a partner subunit. A, TAP1-His/TAP2 or TAP1(K544M)-His/TAP2 interactions. Immunoprecipitations with anti-His indicate the presence of TAP2, and immunoprecipitations with anti-TAP2 indicate the presence of TAP1. B, TAP1⅐TAP2, or TAP1⅐TAP2(K509M) interactions. Immunoprecipitations and Western blots were done identically to those shown in Fig. 2A, but an anti-TAP1 monoclonal antibody (9) was used instead of anti-His to detect the untagged wild type TAP1 protein. Immunoprecipitations with anti-TAP1 indicate the presence of TAP2, and immunoprecipitations with anti-TAP2 indicate the presence of TAP1.
FIG. 3. Peptide translocation by wild type or mutant TAP complexes. A, microsomal vesicles expressing the indicated TAP complexes were assayed for their ability to import 125 I-labeled RRYNASTEL. Bar graphs represent the average of experiments done in triplicate, and the average cpm ϩATP /cpm ϪATP ratio is indicated. In three independent translocation experiments, the average cpm ϩATP /cpm ϪATP ratios for TAP1(K544M)-His⅐TAP2 complexes were 2-fold higher than for single subunit controls, whereas the average cpm ϩATP /cpm ϪATP ratios for TAP1⅐TAP2(K509M) complexes were at the same level as the single subunit controls. B, expression levels of TAP1 and TAP2 in the translocation experiments from A. Microsomal membranes used in the translocation assays shown in A were run on SDS-polyacrylamide gels and immunoblotted with anti-TAP1 or anti-TAP2 monoclonal antibodies. The resulting blot shows that while no translocation signal is observable for TAP1⅐TAP2(K509M) complexes, both TAP subunits are expressed at levels comparable with the wild type complex (compare lanes  1 and 2). The TAP1(K544M)-His⅐TAP2 complex shows a positive translocation signal, but the translocation efficiency is low. Despite a significantly lower expression level of TAP1-His and TAP2 in the TAP1-His⅐TAP2 microsome preparation, the translocation efficiency is higher than that observed with the TAP1(K544M)-His⅐TAP2 preparation (com- pare lanes 3 and 4). recovery was not observed. For example, with microsomal preparations of TAP1-His⅐TAP2 (Fig. 3B, lane 3), fluorescence recovery was not observed upon the addition of excess unlabeled peptide. Increasing the expression level of both TAP1-His and TAP2, however, resulted in binding profiles that resembled wild type TAP1⅐TAP2 complexes (data not shown).
To verify the TAP dependence of the fluorescence quenching and recovery signals shown in Fig. 4, A-C, changes in fluorescence were also monitored upon the addition of microsomal preparations from uninfected insect cells (cells that expressed neither TAP subunit) to different concentrations of the fluorescent peptide (Fig. 4D). The amplitudes of fluorescence quenching signals observed for uninfected microsomes typically ranged from negative values to low positive values over the concentration range of 2.5-80 nM in several different experiments. Positive quenching profiles, if observed, were linear rather than exponential, and the end point amplitudes were significantly lower than observed for TAP-containing microsomes. Finally, in cases where positive quenching signals were observed with uninfected microsomes, fluorescence recovery was typically not observed upon the addition of excess unla-beled peptide (an example is indicated in Fig. 4D, 40 nM). Thus, the fluorescence quenching and recovery profiles for uninfected microsomes are distinct from those seen for wild type TAP complexes, as previously described (27), or for either mutant complex, as indicated in Fig. 4, B and C.
The steady state fluorescence quenching was determined from an exponential fit of the association data (averaged over two independent experiments for the same microsome preparation). The steady state fluorescence quenching values are plotted as a function of peptide concentration for microsomes containing wild type and mutant TAP complexes (Figs. 5, A-C). Each experiment summarized in Fig. 5, A-C, included a negative control with uninfected microsomes (also averaged over two independent experiments for the same microsome preparation). For most of the data points shown in Fig. 5, the amplitudes of the fluorescence quenching signals obtained with microsomes containing wild type or mutant TAP complexes were significantly greater than that obtained for microsomes from uninfected cells, over the indicated concentration range. Taken together with the fluorescence recovery data, we infer from these results that both sets of mutant TAP complexes are FIG. 4. Association and dissociation of fluorescent peptide⅐TAP complexes. Microsomal vesicles expressing the indicated TAP complexes or microsomes prepared from uninfected Sf21 cells were assayed for their ability to bind fluorescein-labeled RRYQKCTEL at the indicated concentrations. Microsomes containing wild type TAP complexes or either mutant showed increased fluorescence quenching and recovery profiles, as the peptide concentration was increased over the indicated range. Fluorescence quenching was sometimes also observed with uninfected microsome preparations, as indicated, but fluorescence recovery was typically not observed upon the addition of excess unlabeled peptide. In addition, if quenching was observed, the magnitude of the quenching was lower than that seen for TAP-containing microsomes, and the quenching profiles were linear rather than exponential. The quenching and recovery profiles shown indicate that both mutant TAP complexes are capable of binding peptides. The dissociation rate constants, k d , estimated by fitting the fluorescence recovery signals to a monoexponential function are indicated for wild type and mutant TAP complexes. For wild type TAP1⅐TAP2 and TAP1⅐TAP2(K509-M), the same microsomes were used as in the translocation assays indicated in capable of binding peptides. The apparent binding constants (K d ) were calculated from the steady state fluorescence quenching versus peptide concentration plots (Fig. 5, A-C). In addition, the dissociation rate constants, k d , were estimated by fitting the fluorescence recovery signal to a monoexponential function (Fig. 4, A-C). The calculated binding constants for the TAP1⅐TAP2(K509M) mutant indicated that the peptide binding affinity of this mutant was slightly weaker compared with wild type TAP complexes. The observations that this mutant TAP complex can bind peptides (Figs. 4C and 5C) and nucleotides (Fig. 1B) but does not translocate peptides (Fig. 3A) indicate that the mutant is arrested at a stage that follows peptide binding and nucleotide binding, most likely at a step that requires nucleotide hydrolysis.
The calculated binding and rate constants for peptide interactions with TAP1⅐TAP2 complexes and TAP1-His⅐TAP2 are very similar to each other and to TAP1(K544M)-His⅐TAP2 complexes (Fig. 4, A and B, and Fig. 5, A and B; data not shown for TAP-His⅐TAP2 complexes). As mentioned above, the peptide binding experiments shown in Figs. 4 and 5 were carried out in the presence of apyrase (ATP and ADP-depleting conditions). It is a formal possibility that TAP-associated nucleotides cannot be accessed by apyrase and that one or both TAP subunits in wild type TAP complexes remain nucleotide-occupied in the presence of apyrase. However, since TAP1(K544M)-His is significantly impaired for nucleotide binding (Fig. 1), we expect that in the presence of apyrase, the nucleotide occupancy will be low for the TAP1(K544M)-His mutant in TAP1(K544M)-His⅐TAP2 complexes. The observation that the calculated binding constants and dissociation rate constants for TAP1(K544M)-His⅐complexes do not differ significantly from the values derived for wild type TAP1⅐TAP2 complexes when apyrase is present or when ADP is present (Table I) indicates that nucleotide interaction with TAP1 is not a requirement for peptide interactions with TAP1⅐TAP2 complexes. DISCUSSION It has previously been observed that the isolated nucleotide binding domains of TAP1 and TAP2 are capable of binding nucleotide (9,13,15). In addition, it has been shown that wild type TAP1⅐TAP2 complexes bind to ␣-32 P-labeled 8-azido-ATP and to ATP and ADP-Sepharose beads (14,17). Here we show that detergent extracts of isolated TAP subunits are also capable of binding nucleotide. We generated mutants of TAP1 and TAP2 that were altered at structurally analogous residues in the Walker A motif of TAP1. We observed that the TAP1(K544M) mutation significantly reduced nucleotide binding by TAP1 but that the TAP2(K509M) mutation did not significantly alter nucleotide binding by TAP2. These observations point to structural differences in the nucleotide binding pockets of TAP1 and TAP2. Indeed, the NBD of TAP1 and TAP2 show only about 51% sequence identity; thus, structural differences might be expected at the atomic level. These differences are likely to contribute to differences in the respective nucleotide binding pockets, which might further lead to differing affinities for nucleotide and differences in catalytic properties for nucleotide hydrolysis. It remains to be addressed whether such differences exist and, if so, whether there are any functional consequences.
Here we report that the K509M mutation in TAP2 abrogates peptide transport by TAP1⅐TAP2(K509M) complexes, although ATP binding by this mutant is not significantly different from wild type. Impairment in peptide translocation does not arise from structural disruptions induced by the mutation, since TAP1⅐TAP2(K509M) complexes are capable of binding both peptides and nucleotides (Figs. 1 and 3). Furthermore, based upon limited proteolytic digestion analysis, the proteolysis profiles observed for TAP1⅐TAP2(K509M) complexes closely parallel the profiles seen for TAP1⅐TAP2 complexes. 2 Thus, nucleotide hydrolysis by TAP complexes containing mutant TAP2 appears to be impaired. Likewise, the TAP1(K544M) mutation reduces peptide translocation efficiency of TAP complexes even 2 S. Arora and M. Raghavan, unpublished observations. FIG. 5. Equilibrium binding curves for fluorescent peptide/ TAP interactions. Microsomal vesicles expressing the indicated TAP complexes or microsomes prepared from uninfected Sf21 cells were assayed for their ability to bind fluorescein-labeled RRYQKCTEL over the indicated concentration range. At each peptide concentration, the steady state fluorescence quenching was determined and plotted as a function of peptide concentration, and the K d values were estimated. The averages of data points from two separate experiments with the same microsome preparation are indicated on each plot. Each set of binding experiments included a control with microsomal membranes prepared from uninfected cells, which are also indicated on each plot. For all of the TAP-containing microsomes indicated, the same microsome preparations were used as in the translocation assays indicated in Fig. 3. The total protein concentration on each microsome preparation was 0.7 mg/ml TAP1⅐TAP2 (A), 1.1 mg/ml TAP1(K544M)-His⅐TAP2 (B), and 1.3 mg/ml TAP1⅐TAP2(K509M) (C). For the uninfected microsomes, the total protein concentration was 1.3 mg/ml. Four independent sets of binding experiments comparing peptide interactions with TAP1(K544M)-His⅐TAP2 microsomes and uninfected microsomes verify that these mutant complexes are capable of binding peptides by the following criteria: (i) the overall fluorescence quenching and recovery profiles resemble those obtained with wild type TAP complexes rather than uninfected microsomes, and (ii) the amplitudes of the quenching signals are greater than those obtained for uninfected microsomes. By similar criteria, four separate sets of experiments comparing peptide binding by TAP1⅐TAP2(K509M) microsomes and uninfected microsomes verify that this mutant complex is also capable of binding peptides.
when no residue modifications are introduced into the TAP2 subunit. Thus, nucleotide hydrolysis by TAP complexes containing mutant TAP1 also appears to be impaired. We are presently unable to directly demonstrate impaired ATPase activity of TAP complexes containing either mutant TAP subunit, since we do not have a working assay to measure the ATPase activities of TAP1 or TAP2 NBD or of TAP1⅐TAP2 complexes. ATPase activities have been previously demonstrated for purified NBD of other ABC transporters (for example, Refs. 31-33). Using purified TAP1 and TAP2 NBD and TAP NBD complexes (15), we were unable to unambiguously demonstrate ATPase activities. The lack of measurable ATPase activity for TAP NBD suggested that intact membrane-associated TAP complexes might be required for observation of the enzymatic activity. Expression of the human MDR1 gene in Sf9 cells has previously been shown to generate a membrane ATPase activity that is significantly stimulated by drugs known to interact with P-glycoprotein (34). We are attempting to develop an analogous assay using microsomal membrane preparations expressing wild type TAP1⅐TAP2 complexes but thus far have been unable to demonstrate a TAP-specific ATPase activity.
The observation that the TAP2(K509M) mutation impairs translocation by TAP1⅐TAP2(K509M) complexes although no residue alterations were introduced into TAP1 indicates that the ATPase activity at TAP1, if present, is insufficient for completion of a peptide translocation cycle. Taken together with the observation that the TAP1(K544M) mutation significantly reduces peptide translocation efficiency of TAP complexes when no residue modifications are introduced into TAP2, these results indicate a coupling between nucleotide interactions with TAP1 and TAP2. The mechanistic model that emerges from these studies differs from that recently described for histidine permease. Functional analyses of mutants of histidine permease, which contains two copies of a single NBD (HisP), have shown that presence of one intact ATP hydrolysis site is sufficient to support ATPase activity and substrate translocation (35). In chimeric histidine permease complexes containing one ATPase-active and one ATPase-inactive NBD, ligand translocation occurred at half the rate of the wild type. These observations are consistent with a mechanism for histidine permease, whereby, in wild type complexes, two molecules of ATP are hydrolyzed within a single turnover of the catalytic cycle, and hydrolysis at either NBD can result in ligand translocation. By contrast to histidine permease, in the case of the P-glycoprotein-based drug transport system, mutations of the Walker A lysine at either one NBD completely blocked drugstimulated ATPase activity of P-glycoprotein and drug transport, although no residue alterations were introduced at the second site (19,21). Other mutants, as well as single site chemical modifications of P-glycoprotein were observed to show similar effects (36 -38). Similar results were obtained for malt-ose permease (39), another bacterial transporter, which contains two copies of a single NBD (MalK). When the Walker A lysine was replaced with asparagine in both MalK subunits, maltose transport and ATPase activities were reduced to 1% of those of the wild type. When the mutation was present in only one of the two subunits, the complex had 6% of the wild-type activities. Functional coupling of ATPase activities observed for P-glycoprotein and MalK might arise if ATP hydrolysis is required at two distinct steps during a single turnover of the catalytic cycle as has been recently suggested for P-glycoprotein (40,41). Activation of the ATPase activity at each NBD might require transient conformational states that are acquired at specific steps of a translocation cycle. Modification at either ATPase site would then be expected to cause conformational trapping, resulting in an inability to complete a translocation cycle. Other models invoking alternating catalytic cycles (42) at each NBD are also consistent with the observations of functional coupling between the two NBD.
Our studies indicate that TAP may belong to a group of transporters with mechanistic similarities to P-glycoprotein and MalK. It is interesting that the TAP2(K509M) mutation abrogates peptide translocation by TAP1⅐TAP2(K509M) complexes but that the TAP1 mutant with a significant impairment in TAP1 nucleotide binding appears to, with low efficiency, mediate peptide translocation by TAP1(K544M)-His⅐TAP2 complexes. One interpretation of these observations is that nucleotide hydrolysis by TAP2 constitutes the first ATP hydrolysis step during a single peptide translocation cycle. This first hydrolytic event might be accompanied by the release of bound peptide and its translocation. Nucleotide hydrolysis by TAP1 might then be required to reset the transporter for another cycle of peptide translocation. Additional experiments are required to obtain further evidence in support of such a mechanism and to obtain greater insights into the precise role of TAP1 and TAP2 during translocation.
We analyzed the ability of each mutant TAP subunit in combination with the wild type partner to bind peptides. Using fluorescence quenching assays, the equilibrium binding constant we calculate for peptide interaction with wild type TAP1⅐TAP2 complexes (K d ϭ 19.4 Ϯ 4.8 nM) is quite similar to that previously reported (K d ϭ 12 Ϯ 1 nM) (27). However, the calculated dissociation rate constant (k d ϭ 0.013 Ϯ 0.005 s Ϫ1 ) is severalfold faster than that previously reported (k d ϭ 0.002 s Ϫ1 ) (27). The difference in the dissociation rate might arise because the binding experiments were carried out at room temperature and 10°C, respectively. Using similar sets of fluorescence quenching assays, we show here that TAP1(K544M)-His⅐TAP2 and TAP1⅐TAP2(K509M) complexes are capable of binding peptides, although the binding affinity of TAP1⅐TAP2(K509M) complexes appears weaker than wild type.

TABLE I
Comparisons of peptide binding to wild type TAP1⅐TAP2 or TAP1(K544M)-His⅐TAP2 complexes in the presence of ADP and apyrase Microsomes containing wild type TAP1⅐TAP2 complexes or TAP1(K544M)-His⅐TAP2 complexes were assayed for their ability to bind fluoresceinlabeled RRYQKCTEL using fluorescence quenching assays similar to those shown in Fig. 4. Peptide binding to both sets of microsomes was determined, and binding in the presence of ADP or apyrase was compared. The steady state fluorescence quenching values were plotted as a function of peptide concentration, and apparent K D values were determined as specified under "Experimental Procedures" and in Fig. 5. The indicated K D values represent the average from at least three independent peptide binding experiments with wild type and mutant TAP complexes. In addition, the dissociation rate constants, k d , were estimated by fitting the fluorescence recovery signals to a monoexponential function. No significant differences are discernible between peptide binding to wild type and mutant TAP complexes in the presence of ADP and apyrase. In addition, no differences are apparent for peptide binding to wild type TAP complexes in the presence of apyrase or ADP. Apparent