Altered Properties of the Branched Chain Amino Acid-preferring Activity Contribute to Increased Cleavages after Branched Chain Residues by the “Immunoproteasome”*

The multicatalytic proteinase complex (MPC, protea-some) is assembled from 14 nonidentical protein subunits. It expresses five distinct proteolytic activities, including a chymotrypsin-like activity, cleaving after hydrophobic residues, and a branched chain amino ac-id-preferring component (BrAAP), cleaving preferen-tially after branched chain residues. Exposure of cells to interferons leads to replacement of the X, Y, and Z subunits by the LMP2, LMP7, and MECL1 subunits. This “immunoproteasome” is critical to processing of certain antigens. The enzymatic basis for enhanced antigen processing has not been determined. To gain insight into this question, we examined sites and relative rates of cleavage of bonds in denatured, reduced, carboxyamidomethylated lysozyme, a 129-amino acid protein, by MPC from bovine spleen, in which the X, Y, and Z subunits are replaced by LMP2, LMP7, and MECL1. We compared cleavages to those catalyzed by MPC from bovine pituitary, which contains only the X, Y, and Z subunits. We found marked increases in the rates and number of cleavages after branched chain residues in reduced, carboxyamidomethylated lysozyme by the spleen MPC. This was largely due to accelerated cleavages of bonds after a F - X -Br motif, where F is a hydrophobic residue, X is a small neutral or polar residue, and Br is a branched chain residue. Inhibitors with these structural properties were selective and potent inhibitors of the BrAAP activity of the spleen MPC. The above findings

The multicatalytic proteinase complex (MPC) 1 is a ubiqui-tous, intracellular, proteolytic particle that is essential for survival of eukaryotic cells (for reviews see Refs. [2][3][4][5]. It represents a major extralysosomal system for proteolysis that is involved in selective, targeted degradation of cytosolic, nuclear, and membrane-bound proteins. As the proteolytic core of the 26 S proteasome, it functions in degradation of multiple essential regulatory proteins including B-type cyclins and proto-oncogene products (3,6,7). Through its role in degradation of I-B (8,9), it serves to activate the NF-B transcription factor and thus participates in signal transduction.
An ancestor of the MPC has been identified in the archeabacterium Thermoplasma acidophylum (10). This more primitive MPC is assembled from 14 copies of each of two subunits designated ␣ and ␤. The MPC from eukaryotes is assembled from 14 nonidentical, constitutively expressed subunits, which can be classified as seven ␣ and seven ␤ subunits, based on their homology to the subunits of the Thermoplasma enzyme (5). These subunits are assembled with an ␣ 7 ␤ 7 ␤ 7 ␣ 7 symmetry into a cylindrical particle with a water-filled tunnel (10,11).
The eukaryotic MPC expresses five distinct proteolytic activities (12), which can be distinguished by the nature of the amino acid residue preferred in the P1 position. These include a trypsin-like activity cleaving after basic residues, a peptidylglutamyl peptide-hydrolyzing (glutamyl) activity cleaving after acidic residues, and a chymotrypsin-like activity, cleaving after hydrophobic residues including aromatic residues. A branched chain amino acid-preferring component (BrAAP), preferring cleavages after leucine, isoleucine or valine was recently described, and an activity preferring cleavages between two small neutral amino acids (SNAAP) has also been identified.
Identification of genes for two interferon-␥-inducible MPC subunits within the major histocompatibility complex (MHC) class II gene region (13)(14)(15)(16) and subsequent experiments showing that MPC inhibitors block class I antigen presentation (17,18), suggest that the MPC is a major factor involved in processing cytoplasmic and nuclear proteins into antigenic peptides. These 8-or 9-amino acid peptides are transported into the endoplasmic reticulum where they are bound to major histocompatibility complex class I molecules (19,20). This complex is transported to the cell surface and displayed to cytolytic, CD8 ϩ T-lymphocytes.
Exposure of cells to interferon-␥ results in expression of LMP2 and LMP7, the two subunits encoded by genes in the MHC class II gene region, and also of a third interferon-␥inducible gene product, MECL1 (13-16, 21, 22), also called LMP10. Upon expression, these subunits replace the constitutively expressed Y, X, and Z subunits, respectively. Studies of mice lacking a functional LMP2 or LMP7 gene (23,24) and of antigen presentation in cells lacking these genes (25) provide evidence of an important role for these subunits in processing MHC class I-restricted antigens. The form of MPC containing the interferon-␥-inducible subunits has therefore been referred to as the "immunoproteasome." However, our understanding of how these changes in structure facilitate enzymatic formation of antigenic peptides is limited. Because antigenic peptides presented on class I molecules most often have hydrophobic or basic amino acid residues at their C terminus, it was proposed that incorporation of LMP2 and LMP7 facilitates cleavage of peptide bonds after such residues. Consistent with this hypothesis, cleavage after the leucine residue at the C terminus of an antigenic epitope contained within a model 25-mer peptide was accelerated upon incorporation of LMP2 and LMP7 (26). However, efforts to explain the changes in cleavages in the model 25-mer substrate associated with incorporation of the interferon-␥-inducible subunits, through measurement using synthetic substrates of corresponding changes in chymotrypsin-like, trypsin-like, and glutamyl activities revealed little apparent relationship. Although cleavages after hydrophobic residues in the model peptide were increased after incorporation of these subunits, activity of the chymotrypsin-like component, measured with the fluorogenic substrate N-succinyl-Leu-Leu-Val-Tyr-MCA, 2 was decreased (26,27). The authors concluded that studies with chromogenic substrates are poor predictors of interactions of MPC with extended peptides, leaving in question the enzymatic basis for the improved efficiency with which MPC containing interferon-␥-inducible subunits generates antigenic peptides. Alternatively, incorporation of interferon-␥inducible subunits may cause changes in activity of other proteolytic activities capable of hydrolyzing bonds after hydrophobic residues, such as the BrAAP activity.
The later hypothesis draws support from comparisons of structure and activities of MPCs from bovine pituitary and bovine spleen (28). Pituitary MPC contains the X, Y, and Z subunits and virtually none of the LMP2, LMP7, or MECL1 subunits, while in spleen MPC over 90% of the X, Y, and Z subunits are replaced by LMP2, LMP7, and MECL1. Spleen MPC thus provides an attractive model for further study of the enzymatic properties resulting from incorporation of the three interferon-␥-inducible subunits. Studies using chromogenic substrates (28) revealed that the trypsin-like and SNAAP activities were similar for spleen and pituitary MPCs. Activity of the glutamyl activity of spleen MPC was reduced by over 80%; a reduction in chymotrypsin-like activity of greater than 50% was also observed. By contrast, activity of the branched chain amino acid-preferring component was increased by approximately 20-fold, with an increase in catalytic efficiency of approximately 100-fold and a significantly broader substrate specificity. The goal of the present study was to explore the possible importance of activation of the BrAAP component of spleen MPC in facilitating cleavages after hydrophobic residues in peptides and proteins and, potentially, in processing antigenic peptides for presentation on class I molecules. Toward this aim, we compared the sites and relative rates of degradation of denatured lysozyme by the MPC from pituitary and spleen. Lysozyme contains 129 amino acids and thus offers a diverse group of possible cleavages after branched chain residues, providing a good model for detecting possible differences in ability to catalyze such cleavages by the two forms of MPC. The findings reveal marked differences in cleavages after branched chain residues in lysozyme by the two forms of MPC. This was largely due to accelerated cleavages by spleen MPC of bonds having a ⌽-X-Br motif, where ⌽ is a hydrophobic residue, X is a small neutral or polar residue, and Br is a branched chain residue. Inhibitors with these structural properties were selec-tive and potent inhibitors of the BrAAP activity of the spleen MPC, suggesting an important role for this activity in cleaving such bonds. The frequent occurrence of this structural motif in antigenic peptides eluted from MHC class I molecules suggests an important functional role for these changes in enzymatic properties.

Materials
The MPC was isolated from bovine pituitaries and spleen as an apparently homogeneous preparation essentially as described (28). Aliquots of the enzyme were stored at Ϫ70°C. S-Carboxyamidomethylated lysozyme (RCM-lysozyme) was prepared as described (29). Z-GGF-pAB, Z-GPALG-pAB, Z-GPAGG-pAB, and Z-GPAAG-pAB were synthesized as described previously (12). Aminopeptidase N was isolated from hog kidneys as described (30). Peptide aldehydes were made by oxidation of the corresponding peptide alcohols by a modification (31)(32)(33) of the dimethyl sulfoxide-carbodiimide reaction of Pfitzner and Moffatt (34). Inhibitors gave essentially a single peak by HPLC. All other reagents were of the highest quality available and were obtained from Sigma or from Fischer.

Methods
Degradation of RCM-lysozyme by MPC-Reactions contained 0.5 ml of RCM-lysozyme (0.135 mg/ml) in 50 mM Tris acetate, pH 8.0, and 5 l of MPC (5 g). Reactions were carried out at 37°C and were stopped at various times by adding 2 l of 85% phosphoric acid.
Reversed Phase HPLC and Microsequencing-Reactions were carried out as above and terminated after 60 min. Equal volumes of reactions with pituitary or spleen MPC were injected into an HPLC fitted with a Hypersil ODS 2 ϫ 150-mm column equilibrated with 0.15% trifluoroacetic acid, 1% methanol, and 2% acetonitrile in water (solvent A). Peptides were eluted with a gradient of solvent B (0.1% trifluoroacetic acid, 1% methanol, and 2% water in acetonitrile) from 0 to 2% B over 5 min, holding at 2% B until 10 min, and then increasing to 60% B over the next 110 min in a linear manner. Eluting peptides were detected by absorbance at 220 and 295 nm using a Linear Instruments 206PHD multiwavelength detector connected to a computerized data recording system. Effluent was collected in 1-min fractions by an automated fraction collector. For sequencing, material from three runs was pooled and concentrated. NH 2 -terminal amino acid sequencing by Edman degradation was performed using an automated sequencer. Where multiple sequences were present, they were distinguished based upon their distinctly different yields.
Analysis of the Degradation of Z-GPAAG-pAB-Activity toward the substrate was measured in a two-step coupled assay performed in the presence of excess aminopeptidase N as described previously (12). All assays were performed at a substrate concentration of 1 mM. Determination of rate constants for inactivation of MPC activities responsible for cleaving the substrate was as described (35). Sites of cleavage of the substrate were determined by separation of degradation products by HPLC and subsequent amino acid analysis of the products as described (36).
Determination of K i Values-Activity of the MPC toward Z-GGF-pAB (a substrate of the chymotrypsin-like activity) and Z-GPALG-pAB (a substrate of the branched chain-preferring activity) was determined as described (12,28). The rate of degradation of Z-GPALG-pAB and Z-GGF-pAB followed Michealis-Menten kinetics. For the chymotrypsinlike and SNAAP activities, K i values were assumed to be equal to the IC 50 value, since the substrate concentration in the assays (1 mM) was significantly less than K m (Ͼ5 mM in all cases). For the branched chain-preferring activity, K i values were calculated as IC 50 /(1 ϩ [S]/K m ), where [S] represents concentration of substrate. Published values for K m for pituitary (5.2 mM) and spleen (0.14 mM) MPCs were used; assays were conducted at substrate concentrations of 1 mM.

Different Products Are Formed by the Two Forms of MPC-
To determine if rates or sites of cleavage of RCM-lysozyme differed for pituitary and spleen MPCs, we first compared products formed during degradation of RCM-lysozyme by separation of products formed after a 60-min incubation by reversed phase HPLC (Fig. 1). Clear differences were apparent when chromatograms of reactions with pituitary MPC (Fig. 1A) were compared with those for reactions with spleen MPC (Fig.  1B). The most notable was the presence in reactions with spleen MPC of a tall cluster of peaks eluting at 53-57 min. Comparison of chromatograms of peptides produced after 60 or 120 min or 24 h revealed that the retention times and relative peak areas were preserved for all time points, except for the disappearance of several minor peaks with long retention times that were present at 20 or 60 min but had diminished after 120 min and disappeared by 24 h (not shown). These peaks appear to represent intermediates formed during the initial reaction that are further degraded. No new peaks appeared over this period. These findings indicate that the elution profiles observed after a 60-min incubation of RCM-lysozyme with either MPC reflect the final products of degradation.
Cleavage Sites Differ for the Two Forms of MPC-The above findings suggested that the two MPCs cleaved different bonds in RCM-lysozyme. To further examine this possibility, cleavage sites were identified by partial NH 2 -terminal sequencing of peptides after separation by HPLC. NH 2 -terminal sequences thus obtained are shown in Table I, along with a partial, deduced sequence NH 2 -terminal to the scissile bond. In some cases, for example sequences 1, 2, 3, 7, and 10 (spleen MPC), peptides having the same NH 2 -terminal sequence were identified in more than one HPLC fraction. This apparently is due to liberation of the final products of degradation in addition to several intermediates. This interpretation is supported by the finding that most of these peptides were found in late eluting peaks.
The relative abundance of peptides can be taken as a measure of the efficiency of cleavages leading to their formation. Several arguments indicate that the efficiency with which peptides were carried through HPLC separation and amino acid sequencing was similar for reactions with pituitary and spleen MPC. Detection of eluting peptides by their absorbance at 295 nm yielded the chromatograms shown in Fig. 1, C and D. Integration of the total area for peaks derived from reactions with pituitary MPC (2.2 arbitrary units) yielded the same value as that obtained for chromatograms of reactions with spleen MPC, indicating that recovery of peptides containing tryptophan was essentially the same for the two chromatographies. Because six tryptophan residues are dispersed throughout lysozyme and each chromatogram at 295 nm closely resembles its counterpart at 220 nm from the same sample, this finding also argues that recovery of peptides was similar for the two reactions. In addition, because there were approximately twice as many cleavages observed by spleen MPC as compared with pituitary MPC, one would expect that the sum of the initial yields for sequences from reactions with spleen MPC would be approximately 2 times greater than that for sequences from reactions with pituitary MPC (24 versus 13 cleavages). Indeed, the total for spleen (1223 pmol) was 2-fold greater than that for pituitary MPC (600 pmol). These data indicate that overall recovery of peptides after HPLC separation and by amino acid sequencing was similar for fragments generated in reactions with pituitary and spleen MPC, although the initial yields may not be exactly comparable between two different peptide sequences. The initial yields therefore serve as an approximate quantitative basis for identification of major differences in the preferences for cleavage sites, as described below.
Correlation of Changes in Catalytic Activities with Changes in Cleavage Sites-Changes in catalytic activities of the spleen MPC might be reflected as a difference in number, location, or relative rates of cleavages. The number of cleavages after basic, acidic branched chain, aromatic, and other amino acids by spleen and pituitary MPC, deduced from the data in Table I, is shown in Fig. 3A. Based on consideration of the initial yields for the peptides (Table I), the relative abundance of products resulting from these types of cleavages could be estimated (Fig.  3B). These reflect the overall efficiency of cleavages after these types of amino acids. Cleavages by spleen MPC are more numerous (24 as compared with 13 for pituitary MPC), due primarily to an increased number of cleavages after branched chain amino acids (peptides 10 -16) and alanine (peptides 17-21, included in the "other" group in Fig. 3, A and B). Both the number of cleavages after branched chain residues and the abundance of peptides resulting from such cleavages were increased in reactions with spleen MPC. There was an overall decrease in the abundance of peptides derived from cleavage after acidic and aromatic residues by this form of MPC. The number of cleavages by spleen MPC after residues in the "other" category, which includes alanine, asparagine, and serine, was markedly increased, but it is noteworthy that peptides derived from such cleavages were present in uniformly small amounts. The number and location of cleavages after basic residues were similar for the two MPCs (Table I, Fig. 2, Fig. 3). Also, the relative amounts of peptides derived from such cleavages were similar for reactions with the two types of MPC. Overall, the above changes correlate well with the depressed chymotrypsin-like and glutamyl activities and markedly activated branched chain-preferring activity of spleen MPC.
One change that must be considered is the marked increase in the number of cleavages after alanines seen in reactions with spleen MPC. To gain insight into the possible enzymatic basis for this difference, we studied degradation by spleen MPC of a model substrate predicted to be cleaved after alanine, Z-GPAAG-pAB; amino acid analysis of the products formed, after separation by reversed phase HPLC, indicated that the Ala-Gly bond was the sole cleavage site for both spleen and pituitary MPC. Spleen MPC degraded the substrate approximately 6-fold faster than pituitary MPC at a substrate concentration of 1 mM (1.14 versus 7.07 mol/mg/h for pituitary and spleen MPCs, respectively). There are two possible explanations for the increased rate of hydrolysis of the substrate. The decreased ability of the chymotrypsin-like activity of spleen MPC to cleave bonds after hydrophobic residues could be explained by a shift in specificity of the chymotrypsin-like component favoring smaller residues. Alternatively, the broadened specificity and increased catalytic efficiency of the BrAAP-like component of spleen MPC could facilitate such cleavages. To examine the roles of the chymotrypsin-like and BrAAP activities of spleen MPC in degrading this substrate, we utilized the marked difference in sensitivity of the two activities to inactivation by 3,4-dichloroisocoumarin (DCI) (12,28,37). The chymotrypsin-like activity is easily distinguished from the BrAAP activity by the fact that the former is rapidly inactivated by low concentrations of DCI, whereas the BrAAP component of spleen MPC is inactivated slowly (12,28,37). Pseudo-first order rate constants of inactivation were therefore determined for spleen MPC using Z-GPAAG-pAB or Z-GGF-pAB (a substrate for the chymotrypsin-like activity) as substrates. In these experiments, DCI was used at a concentration of 50 M. Inactivation of the catalytic component responsible for degradation of Z-GGF-pAB by spleen MPC proceeded rapidly with a value for k obs /[I] of 62.9 M-1S-1 (three determinations), where [I] represents concentration of inhibitor. By contrast, inactivation of degradation of Z-GPAAG-pAB by DCI proceeded slowly, with a value for k obs /[I] of 13.1 M-1S-1 (three determinations), indicating that the chymotrypsin-like activity does not contribute to degradation of the substrate. This reaction was somewhat faster than that observed for inactivation of the degradation of Z-GPALG-pAB (a substrate of the BrAAP component), which proceeded with a k obs /[I] of 6.9 M-1s-1 . One likely explanation for this finding is that the substrate is degraded by both the SNAAP and branched chain-preferring activities, as was described previously for degradation of the substrate by the pituitary MPC (12).
Additional Features of Substrate Specificity Revealed by  (Fig. 1, A and B) and subjected to NH 2 -terminal amino acid sequencing as described under "Reversed Phase HPLC and Microsequencing." Dashes in sequences indicate the scissile bond. Sequences to the right of the dash are those obtained by Edman degradation. Relative abundance of the specific cleavage is indicated as the initial yield during Edman degradation. Where an amino-terminal sequence was present in more than one peak, the relative abundance was taken as the sum of the initial yields for all occurrences of the sequence.  Amino acids are designated using conventional single-letter codes, except that the letter C indicates S-carboxyamidomethylcysteine. Cleavage sites were deduced from NH 2 -terminal sequences of peptides separated by HPLC ( Fig. 1 and Table I). The relative efficiency of cleavages at each site was classified based on the net pmol of that cleavage as a percentage of the total pmol of NH 2 termini liberated (1223 pmol of total NH 2 termini for reactions with spleen MPC and 600 pmol of total NH 2 termini for those with pituitary MPC). Cleavages with an abundance greater than 10% of the total were classified as major (large arrows), those liberating peptides constituting less than 5% of the total as minor (small gray arrows), and those between 5 and 10% as intermediate (small black arrows).
Cleavages in Lysozyme-A fundamental property of the proteolytic activities of the MPC is the strong influence on substrate binding and hydrolysis exerted by residues in the P1 position (primary specificity) and the P2 and P3 positions (secondary specificity). Whereas lysozyme presents a large number of potential cleavage sites, only a limited number of cleavages were observed. Consideration of sequences flanking cleavage sites might therefore give insights into specificities of MPC proteolytic activities, and comparisons of these cleavages may reveal similarities and differences in specificities between pituitary and spleen MPCs. Further insights may be obtained from information on the relative efficiency with which particular bonds in lysozyme are hydrolyzed. The abundance of a peptide may be taken qualitatively as a reflection of the efficiency with which the relevant scissile bonds are cleaved. Thus, cleavages involved in liberating the peptides in Table I were ranked as major, intermediate, or minor based on their initial yields during amino acid sequencing by Edman degradation. A map of cleavage sites deduced from the NH 2 -terminal sequences of these peptides and the relative efficiency of the cleavages deduced from initial yields is shown in Fig. 2. In particular, it is notable that yields of peptide 9 in reactions with spleen MPC increased by approximately 30-fold, suggesting a marked increase in efficiency of hydrolysis of this bond by the complex from this organ. Peptides 10 and 11 were also formed as a result of cleavages that appeared to be more efficient for spleen as compared with pituitary MPC.
Consideration of sequences flanking scissile bonds by pituitary or spleen MPCs revealed that in general, small neutral or polar residues appeared to be preferred in the P1Ј position. No cleavages after tryptophan residues were observed for either form of MPC, suggesting that very bulky residues are not accommodated in P1. It is also of interest that tryptophan was observed in the P3 position only in two of the cleavages by spleen MPC (peptides 9 and 14), both of which were relatively minor cleavages, suggesting that this residue is too bulky to be easily accommodated by the S2 or S3 subsites of either the spleen or pituitary MPC. A notable feature of cleavages after basic residues by spleen or pituitary MPC was the presence in many cases of a basic residue in either the P2 or P3 position (cleavages 2-5 in Table I), suggesting a preference of the trypsin-like activity of the MPC for such structural features.
Upon comparison of major cleavages for spleen and pituitary MPCs, deduced from the abundance of the resulting peptides, several marked differences were observed. Hydrolysis of the Glu 35 -Ser 36 and Tyr 23 -Ser 24 bonds (cleavages 8 and 27, Table  I), major sites of cleavage by pituitary MPC, were not detected in reactions with spleen MPC. As compared with the pituitary MPC, the spleen MPC efficiently cleaved the Leu 25 -Gly 26 , Leu 84 -Ser 85 , and Ile 58 -Asn 59 bonds (cleavages 9 -11, Table I). The latter cleavage was not observed in reactions with pituitary MPC. Comparison of features of amino acid sequences adjacent to cleavages after branched chain residues revealed that the activity or activities in spleen MPC cleaving such bonds show a significantly broadened specificity. Thus, spleen MPC cleaved more efficiently those bonds after branched chain residues in which a hydrophobic residue was present in P3, with a small neutral or polar residue in P2. The Leu 25 -Gly 26 and Ile 8 -Asn 59 bonds provide clear examples of this difference.
Inhibitors with the ⌽-X-Br Motif Are Selective for the Spleen BrAAP Activity-The above shift in cleavages after branched chain amino acids could reflect a shift in the substrate specificity of one or more proteolytic activity or activities of the spleen MPC toward substrates with a general ⌽-X-Br motif, where ⌽ is a hydrophobic residue, X is a comparatively small neutral or charged residue, and Br is the branched chain residue after which cleavage occurs. In this case, inhibitors containing these structural features would be expected to be good inhibitors of one or more activities of the spleen MPC as compared with the pituitary MPC. Inhibitory properties of peptide aldehydes containing the ⌽-X-Br motif were therefore synthesized. Inhibitory constants of two such peptide aldehydes for the chymotrypsin-like, small neutral amino acid-preferring, and branched chain amino acid-preferring activities are shown in Table II Table I. Open bars, pituitary MPC; closed bars, spleen MPC. B, relative abundance of peptides resulting from cleavages after each of several types of amino acids. Open bars, pituitary MPC; closed bars, spleen MPC. Taking as an example the relative abundance of peptides formed by spleen MPC by cleavage after basic residues, the total pmol of NH 2 termini resulting from such cleavages (374 pmol) was divided by the total NH 2 termini generated by proteolysis by spleen MPC (1223 pmol) and multiplied by 100.
ponents of this form of MPC. Thus, the inhibitor shows significant selectivity for the branched chain amino acid-preferring activity. Similar, although less pronounced, findings were observed in studies with Z-GLEL-CHO, in which the alanine residue was replaced with glutamate. As compared with pituitary MPC, the compound demonstrated a significantly higher (30-fold) affinity for the branched chain-preferring component of spleen MPC, while smaller changes in K i were observed for the chymotrypsin-like and small neutral amino acid-preferring components of spleen MPC. The inhibitor showed a more than 10-fold lower K i for the branched chain-preferring activity of spleen MPC as compared with values for the chymotrypsin-like or small neutral preferring activities. Taken together, the above findings suggest a selective shift in substrate preference of this proteolytic component of spleen MPC toward substrates with the general ⌽-X-Br motif. DISCUSSION The MPC is involved in processing most peptide antigens presented on MHC class I molecules (17,38). Peptide aldehyde inhibitors of MPC block presentation of many antigens as does lactacystin, a relatively selective inhibitor of MPC (17,18,39,40). A survey of the sequences of peptides eluted from MHC class I molecules reveals that approximately 80% of such peptides end in a hydrophobic amino acid residue (20). These peptides are 8 -10 amino acids in length. Two precise cleavages are required for their production. This reaction is facilitated by ␥-interferon-inducible proteins including the LMP subunits and a protein activator designated PA28 or REG. The latter is assembled from two closely related 28-kDa subunits to form an 11 S particle (41,42). Binding of REG to the MPC causes activation of the five peptidase activities, with the BrAAP activity being most affected (41)(42)(43). Activation of the MPC by REG improves the efficiency of antigen processing in both biochemical and cellular systems (27,44,45). One consequence of activation by REG is an increased ability to catalyze the dual cleavages that liberate the 8 -10 amino acid class I-restricted peptides from larger precursors (44). A central role for the expression of the LMP subunits in antigen processing is shown by findings that LMP2 and LMP7 are required for processing of certain virus-derived antigenic peptides with C-terminal branched chain amino acids (25,46). Moreover, studies in mice indicate that loss of LMP2 or LMP7 results in a marked decrease in overall expression of MHC class I molecules on the cell surface (23,24) and decreased efficiency of expression of antigenic peptides derived from influenza virus. Similar findings came from studies of antigen presentation by cells lacking the LMP2 or LMP2 and LMP7 subunits (25,46). This study was motivated by the question of how changes in proteolytic activities of the MPC associated with incorporation of the three interferon-inducible subunits facilitate cleavages after hydrophobic amino acid residues and hence processing of antigenic peptides for presentation on MHC class I molecules. These peptides most commonly have hydrophobic residues (80%) at the carboxyl terminus (20). In this regard, it is of interest to consider the number of possible peptides of this size generated by the pituitary and spleen MPC. Eight such peptides could be predicted from the observed cleavages by pituitary MPC, whereas the observed cleavages by spleen MPC would be predicted to result in 19 peptides of this size. This observation may account for part of the improved ability of the LMP subunit-containing MPC to generate class I-restricted antigenic peptides and may complement the improved efficiency of formation of these peptides induced by activation of the MPC by REG.
Several arguments support our conclusion that changes in enzymatic properties of the BrAAP component of MPCs containing the interferon-inducible subunits are important factors in the altered cleavage after hydrophobic residues by this form of MPC. Studies with peptide aldehyde inhibitors having the ⌽-X-Br motif revealed marked differences in the substrate specificity of the branched chain-preferring activities of spleen and pituitary MPCs. This was reflected by marked increases in affinity of the inhibitors for the BrAAP component of spleen MPC, which approached 100-fold in the case of Z-GLAL-CHO. This difference in inhibitor binding was selective for the branched chain-preferring activity, indicating that the branched chain-preferring activity of spleen MPC possesses the ability to bind in a selective manner peptides with the general motif ⌽-X-Br. This motif is quite different from that preferred by the branched chain-preferring component of pituitary MPC, consisting of a small neutral residue in P3 (including proline), a hydrophobic residue in P2, and a branched chain residue in P1. Of interest, however, inhibitors with the latter structure, such as Z-GPFL-CHO, are also good inhibitors of the branched chain-preferring component of spleen MPC, with K i values similar to that observed for this activity in pituitary MPC, indicating a general broadening of the specificity of this catalytic activity. Our findings thus extend conclusions of previous comparisons of spleen and pituitary MPCs, which found (28) that the branched chain-preferring component of spleen MPC displays a broadened primary specificity, accommodating aromatic as well as branched chain residues in P1.
While a complete understanding of the respective roles of the branched chain-preferring and chymotrypsin-like activities in cleaving bonds after hydrophobic residues is not possible from our findings, the data provide several arguments supporting a role for the enhanced properties of the branched chain-preferring component of spleen MPC in the increases in cleavages after branched chain residues in lysozyme by this form of MPC. Cleavages after aromatic residues in RCM-lysozyme, a preferred cleavage site for the chymotrypsin-like component, were reduced overall (Fig. 3B), a finding that mirrors the decrease in activity of the chymotrypsin-like activity of spleen MPC, which shows some preferences for such cleavages (47)  branched chain residues seen in the current study. Finally, the secondary specificity of the proteolytic component responsible for novel, rapid cleavages after branched chain amino acids in lysozyme parallels that of the branched chain amino acid component as determined in studies with peptide aldehyde inhibitors.
Our findings indicate an important role for changes in the activity and specificity of the BrAAP component associated with incorporation of the LMP subunits in alterations in cleavages after branched chain amino acids. The association of the changes in activity and specificity of the branched chain-preferring component of spleen MPC with the replacement of the X, Y, and Z subunits by the interferon-inducible subunits LMP2, LMP7, and LMP10 suggests an important role for these alterations in enzymatic properties of this catalytic component in facilitating cleavages required for processing antigenic peptides ending in branched chain amino acids. The central importance of these conclusions is shown by the finding that approximately 60% of peptides eluted from MHC class I molecules end in a branched chain residue (leucine, isoleucine, or valine) (20). Of interest in this regard, radiolabeling studies using [ 14 C]DCI suggest that the LMP7 subunit is involved in the active site of the BrAAP activity of spleen MPC (47), and presentation of an MHC class I-restricted peptide containing a C-terminal ⌽-X-Br motif appeared to be increased upon incorporation of LMP7 (25). When these findings are considered together with the findings of the current study, they strongly suggest that the changes in the properties of the branched chain-preferring component of spleen MPC are a major factor in the formation of many antigenic peptides having carboxylterminal branched chain amino acids.