MCP Danish Cancer Society
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/mcp.M600302-MCP200 on February 16, 2007.
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
M600302-MCP200v1
6/5/923    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Glossary
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Marcilla, M.
Right arrow Articles by López de Castro, J. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Marcilla, M.
Right arrow Articles by López de Castro, J. A.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Molecular & Cellular Proteomics 6:923-938, 2007.
© 2007 by The American Society for Biochemistry and Molecular Biology, Inc.

Research

Proteasome-independent HLA-B27 Ligands Arise Mainly from Small Basic Proteins*,S

Miguel Marcilla, Juan J. Cragnolini and José A. López de Castro{ddagger}

From the Centro de Biología Molecular Severo Ochoa (Consejo Superior de Investigaciones Científicas and Universidad Autónoma de Madrid), Facultad de Ciencias, Universidad Autónoma, 28049 Madrid, Spain


   ABSTRACT
TOP
 ABSTRACT
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
Many of the constitutive peptide ligands of HLA-B27, a molecule strongly associated with spondyloarthritis, are proteasome-independent. Stable isotope tagging, mass spectrometry, and epoxomicin-mediated inhibition were used to determine their percentage, structural features, and parental proteins. Of 104 molecular species examined, 29.8% were proteasome-independent, paralleling the level of HLA-B27 re-expression in the presence of epoxomicin after acid stripping. Proteasome-dependent and -independent ligands differed little in peptide motifs, flanking sequences, and cellular localization of the parental proteins. In contrast, whereas the former set arose from proteins whose size and isoelectric point distribution largely reflected those in the human proteome, proteasome-independent ligands, other than a few matching signal sequences, were almost totally derived from small (about 6–16.5 kDa) and basic proteins, which account for only 6.6% of the human proteome. Thus, a non-proteasomal proteolytic pathway with strong preference for small proteins is responsible for a significant fraction of the HLA-B27-bound peptide repertoire.

Major histocompatibility complex (MHC)1 class I molecules constitutively bind large peptide repertoires arising from degradation of endogenous proteins and present them at the cell surface. Most of these ligands are produced in the cytosol. They or their immediate precursors are introduced into the lumen of the endoplasmic reticulum (ER) by the transporter associated with antigen processing (TAP) where they bind to the nascent class I molecule in a process of assisted loading involving calreticulin, ERp57, protein-disulfide isomerase, and tapasin, which together with TAP and the MHC molecule form the peptide-loading complex (1–6). The properly folded MHC-peptide complexes migrate to the cell surface. The class I-bound peptide repertoires consist of several thousands of molecular species (7, 8), which arise from a broad spectrum of proteins from essentially all cell compartments (9, 10). The proteasome, a multicatalytic complex of the cytosol and nucleus, is the major protease involved in the generation of MHC class I ligands (11, 12). Although the proteasome might directly generate some of these ligands (13–16), it seems that many require further processing of proteasomal products by the cytosolic protease tripeptidyl-peptidase II (TPPII) and by amino peptidases of the cytosol and ER (12, 17–21).

Some MHC class I ligands can be generated by proteases other than the proteasome (22, 23). Peptides coming from signal sequences of proteins resident in or entering the ER to be incorporated into the exocytic route are paradigmatic. These signal sequences are cleaved by the signal peptide peptidase and can be further processed in the ER, generating TAP-independent ligands (24, 25), or in the cytosol, becoming TAP-dependent (26, 27). Generation of TAP-independent ligands other than those derived from signal sequences, involving processing in the ER, has been reported (28). Besides participating in peptide trimming (18, 20), TPPII can generate some viral epitopes (29, 30). In dendritic cells, the lysosomal enzyme cathepsin S was shown to generate a TAP-independent class I ligand for cross-presentation in vivo (31). Furine, a protease of the Golgi, can also generate class I ligands (32–34). This enzyme, or closely related proprotein convertases, may also provide suitable class I ligands upon failure of quality control mechanisms for peptide loading in the ER (35), although the actual contribution of this enzyme to the constitutive MHC class I-bound peptide repertoires in cells with an intact processing-loading pathway has not yet been confirmed. These findings do not challenge the general rule that the proteasome pathway is the major source of MHC class I ligands. This is despite the fact that a very substantial degradation of cellular proteins, especially those with long half-lives, takes place in lysosomes (36). Thus, it is assumed that peptide transfer from the lysosomal compartment to the MHC class I presentation pathway must be very inefficient. Furthermore a major source of MHC class I ligands consists of newly synthesized polypeptides that fail to reach the native state and are targeted to the proteasome for degradation (37–39) rather than proteins that are degraded at the end of their lives.

Despite the global significance of the proteasome pathway, MHC class I allotypes differ widely in the amount of proteasome-dependent ligands that they present (40). A previous study addressed this issue by means of acid stripping of the class I molecules expressed at the cell surface and quantification of their re-expression in the presence of proteasome inhibitors (23). Some class I molecules were poorly re-expressed upon proteasome inhibition, whereas others, in particular B*2705, were expressed to a substantial extent. Sequencing of peptides re-expressed in the presence of proteasome inhibitors from HLA-B27 and other class I molecules failed to reveal any obvious bias in the C-terminal peptide motifs or in the parental proteins.

The high surface re-expression of HLA-B27 in the presence of proteasome inhibitors after acid stripping (23) was convincing evidence for a significant contribution of proteasome-independent pathways to shaping the HLA-B27-bound peptide repertoire. However, a limitation of this approach is that quantitative removal of HLA class I-bound peptides by acid washing can only be indirectly assessed by flow cytometry, and it is difficult to rule out that a small amount of peptides may resist acid removal, especially in HLA-B27 whose association with ß2-microglobulin is particularly strong (41, 42). This might complicate the assignment of proteasome inhibitor-resistant ligands among those sequenced from acid-washed cells.

To circumvent this problem and to re-examine the nature of proteasome-independent HLA-B27 ligands, we envisaged a different approach that was independent of the previous removal of surface-expressed peptides. This approach, which used techniques of quantitative expression proteomics (43) that were recently applied to identifying MHC ligands (44, 45), was based on metabolic labeling of cellular proteins with [15N]Arg (46). This method allows the labeling of virtually all HLA-B27 ligands because Arg at P2 is a nearly universal motif of B27-bound peptides (10, 47). Upon labeling with [15N]Arg the mass spectrum of a peptide will show a selective increase in the intensity of the peak corresponding to the monoisotopic mass of the peptide plus 2 Da per Arg residue, according to the two 15N atoms of the labeled Arg side chain. Labeling HLA-B27-positive cells with [15N]Arg in the presence or absence of epoxomicin, a specific and potent proteasome inhibitor, provided a sensitive and reliable method to unambiguously distinguish proteasome-dependent and -independent HLA-B27 ligands. Using this approach in conjunction with MS-based peptide sequencing, it was possible to reveal a clear-cut differential distribution of proteasome-dependent and -independent ligands, depending on features of their parental proteins.


   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Cell Lines, Monoclonal Antibodies (mAbs), and Inhibitors—
C1R is a human lymphoid cell line with low expression of its endogenous HLA class I molecules (48). C1R-B*2705 transfectants were described elsewhere (49). Cells were cultured in RPMI 1640 medium supplemented with 10% FCS (both from Invitrogen). The mAb ME1 (IgG1; specific for HLA-B27, -B7, and -B22) (50) and W6/32 (IgG2a; specific for a monomorphic HLA class I determinant) (51) were used. E-poxomicin, an irreversible and specific inhibitor of the poteasome (52), and carbobenzoxy-L-leucyl-L-leucyl-L-leucinal (MG132), a potent reversible inhibitor of the proteasome and calpains (11), were from Calbiochem. Brefeldin A (BFA), which blocks egress of MHC-peptide complexes from the ER (53), was from Sigma-Aldrich.

Acid Stripping and Flow Cytometry—
About 106 C1R-B*2705 transfectant cells were either untreated, preincubated for 2 h with 1 µM epoxomicin, or preincubated for 30 min with BFA (10 µg/ml). Cells were centrifuged, and pellets were resuspended in 500 µl of stripping buffer (0.5 M glycine, 1% bovine serum albumin, pH 2.5) and incubated for 2 min. This suspension was neutralized by adding Dulbecco’s modified Eagle’s medium (Invitrogen) to a final volume of 15 ml. Cells were centrifuged and resuspended in 2 ml of RPMI 1640 medium supplemented with 10% FCS in the presence or absence of BFA (10 µg/ml) or 1 µM epoxomicin. Flow cytometry was performed in a FACSCalibur instrument (BD Biosciences) as described previously (54).

Isotopic Labeling of C1R-B*2705 Cells—
The strategy used is summarized in Fig. 1. C1R-B*2705 transfectants were distributed in three culture flasks (about 1.5 x 108 cells/flask) and incubated for 4 h in Dulbecco’s modified Eagle’s medium without Arg and supplemented with 10% FCS. One of the flasks was then supplemented with standard (14N) Arg (100 µg/ml), a second flask was supplemented with 100 µg/ml L-[guanido-15N2]arginine·HCl (Cambridge Isotope Laboratories, Andover, MA), in which two nitrogen atoms of the guanidinium group have been replaced with 15N, and the third flask was treated with 1 µM or, in other experiments, with 0.2 and 2.5 µM epoxomicin for 30 min prior to the addition of 100 µg/ml 15N-tagged Arg to ensure that the proteasome was inhibited from the start of labeling, and the inhibitor was left for the entire labeling period. After 5 h, the cells were washed twice in 20 mM Tris, 150 mM NaCl, pH 7.5. Pellets were stored at –70 °C for further processing. In some experiments 20 µM MG132 was used instead of epoxomicin in the same conditions except that starving of cells in the absence of Arg before labeling was carried out for 2 h. All incubations were done at 37 °C. Peptide labeling was quantified by the labeling ratio, which was defined as follows. Ratio = ((15N + inh.) – 14N)/(15N – 14N) where 14N, 15N, and (15N + inh.) are the percent intensities of the relevant isotopic peak, relative to the monoisotopic peak, in the MALDI-TOF MS spectrum of the peptide in the absence of labeling or upon labeling with [15N]Arg in the absence of inhibitor or in its presence, respectively.


Figure 1
View larger version (25K):
[in this window]
[in a new window]

  		FIG. 1. Stable isotope tagging-based strategy to distinguish between proteasome-independent and -dependent HLA-B27 ligands. A, three equal aliquots of C1R-B*2705 cells were subjected to Arg starving for 4 h and subsequently incubated for 5 h either in the presence of standard (14N) Arg, 15N2-tagged Arg, or 15N2-tagged Arg in the presence of various concentrations of epoxomicin. The proteasome inhibitor was added 30 min before the addition of the isotope, to ensure inhibition of the proteasome upon labeling, and left for the entire labeling period. HLA-B27-bound peptide pools were isolated in parallel from the three cell lysates by immunopurification of HLA-B27 and acid extraction and subjected to HPLC in consecutive runs under identical conditions. Labeling of HLA-B27 ligands was detected by high resolution MALDI-TOF MS as an increased intensity of the corresponding isotopic peak in the absence of epoxomicin (in the example A + 4, corresponding to a peptide with 2 Arg residues, where A is the monoisotopic peak). In the presence of the inhibitor, labeling of proteasome-dependent peptides is inhibited, and the MS spectrum of the peptide is indistinguishable from that of the unlabeled control. Labeling of a proteasome-independent peptide will be unaffected by the inhibitor so that the MS spectrum will be similar upon labeling either in the presence or absence of epoxomicin. B, for peptide sequencing, the HLA-B27-bound peptide pool was isolated from 1010 cells and subjected, in parallel to the labeled samples and with the same conditions, to HPLC and MALDI-TOF MS analysis of the individual chromatographic fractions. The relevant peptides were identified from highly matched MALDI-TOF spectra of correlative HPLC fractions on the basis of identity in molecular mass and retention time. Sequencing was done by nanoelectrospray MS/MS.

 
Isolation of HLA-B27-bound Peptides—
B*2705-bound peptides were isolated from about 1.5 x 108 C1R-B*2705 transfectant cells as described previously (55). Briefly cells were lysed in 1% Igepal CA-630 (Sigma-Aldrich) in the presence of a mixture of protease inhibitors. After ultracentrifugation, the soluble fraction was subjected to affinity chromatography using the W6/32 mAb. HLA-B27-bound peptides were eluted with 0.1% aqueous TFA at room temperature, filtered through Centricon 3 devices (Amicon, Beverly, MA), concentrated, and subjected to HPLC fractionation in a Waters Alliance system (Waters, Milford, MA) using a Vydac 218TP52 column (Vydac, Hesperia, CA) at a flow rate of 100 µl/min as described previously (15). Fractions of 50 µl were collected. For peptide sequencing, which required higher amounts of material, the same procedure was used but starting from about 1010 C1R-B*2705 cells. The relevant peptides to be sequenced were identified as those with the same monoisotopic mass and retention time as the labeled peptides by comparing the MALDI-TOF MS spectra of correlative and highly matched HPLC fractions from this peptide pool and those from the labeling experiments obtained from consecutive chromatographic runs under identical conditions (Fig. 1).

Mass Spectrometry—
HPLC fractions were analyzed by MALDI-TOF MS using a Bruker Reflex IVTM or an Autoflex mass spectrometer (both from Bruker Daltonics, Bremen, Germany) equipped with the SCOUTTM source operating in positive ion reflector mode. Dried HPLC fractions were resuspended in 0.5 µl of TA (33% aqueous acetonitrile, 0.1% TFA), loaded onto the MALDI plate, and allowed to dry at room temperature. Then 0.5 µl of matrix solution ({alpha}-cyano-4-hydroxycinnamic acid in TA) at 2 mg/ml were added and allowed to dry again.

Peptide sequencing was carried out by quadrupole ion trap nanoelectrospray MS/MS in an LCQ instrument (Finnigan Thermoquest, San Jose, CA) using the Xcalibur 2.0 software (Thermo Scientific) or in an Esquire 3000Plus ion trap mass spectrometer (Bruker Daltonics) using the Bio Tools 2.2 software (Bruker Daltonics) after on-line chromatographic separation of samples as described previously (56). Interpretation of mass spectra was done manually but assisted by various software tools as follows. Manual inspection of the spectrum usually allowed us to determine a partial sequence. This information together with the m/z of the parent ion (in most cases charge 2 parent ions were used for sequencing in this study) was used as input data for a MASCOT (version 2.1) search (Matrix Science) in the human protein entries of the Mass Spectrometry Protein Sequence Database (MSDB) (release August 31, 2006) (Imperial College, London, UK) using a window of 0.8 m/z units. Of the 20 output sequences showing the highest scores in this preliminary search, those few showing the canonical Arg at P2 motif of HLA-B27 ligands (10, 47) and absence of "prohibited" residues for HLA-B27 binding, such as N-terminal or C-terminal Pro, were selected. From each of these sequences, a list of theoretical fragment ions was generated using the MS-Product tool (67) as an assistance to match the putative candidate sequences to our experimental MS/MS spectrum (see the supplemental data). At this stage, usually one single proper match was obtained. If ambiguity existed for more than one sequence or if the sequence determined had not been reported previously as an HLA-B27 ligand, the corresponding synthetic peptides were obtained, and their MS/MS spectra were matched for identity with our experimental spectra.

Assignment of the Parental Proteins of HLA-B27 Ligands—
This was done on the basis of unambiguous matching with a single human protein in the UniProtKB database release 9.5 (January 23, 2007) using the Fasta 3 software (www.ebi.ac.uk/fasta) after taking into account the database redundancy due to multiple entries for the same protein. In some cases, a peptide ligand matched several closely related members of a protein family (i.e. histone families). In these cases a single entry for a representative member was chosen with the understanding that the same ligand can arise from more than one member of such families.

Databases and Statistical Analysis—
The molecular mass and theoretical pI of the assigned proteins was obtained from the UniProt KB database release 9.5 (January 23, 2007; www.expasy.org/sprot). Subcellular localization of the proteins was obtained from the UniProtKB release 9.5 (January 23, 2007) and DAVID2006 (57) databases. Proteome analysis was performed with 15,495 entries from the human annotated protein database in the UniProtKB/Swiss-Prot database and assisted by the JvirGel 2.0 software (75). Statistical analyses were carried out using the {chi}2 or, for smaller samples, the Fisher’s exact test. p values <0.05 were considered as statistically significant.


   RESULTS
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
Proteasome-dependent and -independent Ligands Can Be Distinguished by Isotopic Labeling of the B27-bound Peptide Repertoire—
To determine the role of the proteasome in generating the B*2705-bound peptide repertoire an approach based on stable isotope tagging of the HLA-B27 ligands was developed (Fig. 1). 15N-Tagged Arg was used to achieve the labeling of virtually every single peptide. C1R-B*2705 cells were subjected to Arg starving. Then equal aliquots were either supplemented with standard (14N) Arg, with the same amount of 15N-tagged Arg, or treated with epoxomicin prior to the addition of the 15N-tagged Arg and incubated in the presence of the inhibitor. The B*2705-bound peptide pool was isolated from each aliquot and fractionated by HPLC, and each fraction was analyzed by MALDI-TOF MS. The peptides isolated from cells treated with 15N-tagged Arg showed a different isotopic distribution compared with that of the same peptides derived from cells supplemented with standard Arg. As the 15N-tagged Arg is 2 Da heavier than its non-tagged counterpart, labeling was detected as an increase of the intensity of the A + 2, A + 4, or A + 6 peaks (where A is the monoisotopic peak) and, proportionally, the subsequent peaks, depending on the number of Arg residues of the peptide (Fig. 2). The extent of this increase was variable among peptides as it depends on multiple factors, such as the synthesis rate of the parental protein, the efficiency in the generation of the peptide, its cytosolic stability, the transport and HLA-B27 binding efficiencies, etc., but it was highly reproducible for individual peptides (Table I).


Figure 2
View larger version (30K):
[in this window]
[in a new window]

  		FIG. 2. Stable isotope tagging of HLA-B27 ligands with [15N]Arg. Three examples corresponding to peptides with 1, 2, or 3 Arg residues (highlighted in boldface) are shown. The figure shows the expanded MALDI-TOF spectra of the corresponding peptides isolated from unlabeled (14N) or [15N]Arg-labeled (15N) cells. The spectra from unlabeled cells show the normal distribution of isotopic species whose relative intensities closely parallel the theoretical intensities. Upon labeling with 15N-tagged Arg, containing two 15N atoms, the MS spectrum is altered by a selective increase of the corresponding isotopic species: A + 2, A + 4, or A + 6 for peptides with 1, 2, or 3 Arg residues, respectively, where A is the monoisotopic peak.

 

View this table:
[in this window]
[in a new window]

  		TABLE I Reproducibility of [15N]Arg labeling and inhibition of B*2705 ligands

Six examples from two independent experiments are shown. Values under the 14N, 15N, and 15N + epoxomicin (Ep) columns indicate the percent intensity of the relevant isotopic peak (A + 2, A + 4, or A + 6 for peptides with 1, 2, or 3 Arg residues, respectively) relative to the corresponding monoisotopic (A) ion peak. The concentration of epoxomicin in these experiments was 1 µM.

 
We reasoned that the isotopic labeling of a proteasome-independent ligand isolated from cells treated with 15N-tagged Arg would be significant regardless of the presence or absence of epoxomicin in the culture medium because its generation would not be abrogated by inhibition of the proteasome. In contrast, no labeling should be detected in proteasome-dependent ligands when isolated from cells treated with epoxomicin plus 15N-tagged Arg because these ligands would not be generated in the presence of the inhibitor. Thus, the isotopic distribution of proteasome-dependent ligands isolated from [15N]Arg-labeled cells in the presence of epoxomicin or from unlabeled cells should be essentially the same (Fig. 1).

To test this assumption we compared the isotopic distributions of RRFFPYYVY, a B27 ligand known to be generated in a proteasome-dependent fashion (15), isolated from control, 15N-tagged Arg-, and 15N-tagged Arg plus 1 µM epoxomicin-treated cells (Fig. 3A). As expected, 15N labeling was detected in the ligand from untreated but not from epoxomicin-treated cells. On the other hand, the peptides IRAPPPLF and ARLQTALLV are derived from the signal sequences of cathepsin A and cytokine A22, respectively, and are presented by HLA-B27 in the TAP-deficient T2 cells,2 suggesting that they are processed in the ER. In both cases we detected a significant, but about 19–24% lower, 15N labeling of the ligands when isolated from epoxomicin-treated cells relative to cells supplemented with 15N-tagged Arg in the absence of the inhibitor (Fig. 3, B and C). Because these two ligands are probably proteasome-independent their lower labeling in the presence of epoxomicin cannot be explained by partial inhibition of the proteasome (see "Discussion").


Figure 3
View larger version (38K):
[in this window]
[in a new window]

  		FIG. 3. Metabolic labeling allows proteasome-dependent HLA-B27 ligands to be distinguished from proteasome-independent HLA-B27 ligands. A, MALDI-TOF MS spectra of the proteasome-dependent peptide RRFFPYYVY (15) isolated from unlabeled cells (14N), [15N]Arg-labeled cells in the absence of epoxomicin (15N), and labeled cells in the presence of a 1 µM concentration of the inhibitor epoxomicin (Ep). The expanded MS spectrum of the peptide from unlabeled cells shows the normal distribution of isotopic species. Upon labeling with 15N-tagged Arg, containing two 15N atoms, the MS spectrum is altered by a selective increase of the corresponding isotopic species, A + 4 in this example where A is the monoisotopic peak. In the presence of epoxomicin this alteration is not observed. The percent intensity of the A + 4 peaks relative to the corresponding monoisotopic peaks in each situation is indicated. B, MALDI-TOF MS spectra of the TAP-independent IRAPPPLF ligand, matching the signal sequence of cathepsin A, isolated in the same three situations. C, MALDI-TOF MS spectra of the TAP-independent ARLQTALLV ligand, matching the signal sequence of cytokine A22, in the same conditions.

 
A Significant Portion of the HLA-B27-bound Peptide Repertoire Is Generated in the Presence of Epoxomicin—
In an initial experiment proteasome inhibition was carried out with 1 µM epoxomicin. Upon MALDI-TOF analysis of the B*2705-bound peptide pool 91 ion peaks were amenable to further analysis on the following basis. 1) They showed sufficiently high intensity for good detection of multiple isotopic peaks, and 2) the relevant isotopic peak increased at least 20% upon 15N labeling in the absence of inhibitor relative to its intensity in unlabeled cells (Fig. 4). A given ligand was considered to be proteasome-independent if the intensity of the corresponding isotopic peak upon labeling in the presence of epoxomicin increased by at least 40% of the increase of that peak in the absence of the inhibitor (labeling ratio, ≥0.4). In contrast, the proteasome-dependent ligands were those that were labeled in the absence but not in the presence of epoxomicin (labeling ratio, ≤0.2). The threshold of 0.4 for the labeling ratio was adopted because proteasome-independent peptides may be less labeled in the presence of epoxomicin than in its absence due to indirect effects following proteasome inhibition, such as down-regulation of protein synthesis (see below and "Discussion"). The labeling of 62 of the 91 peptides (68.1%) was abolished with 1 µM epoxomicin (Fig. 4A), suggesting a key role of the proteasome for their generation. In 29 ligands (31.9%) labeling was significant in the presence of epoxomicin, suggesting that they were proteasome-independent. Very similar results were obtained when using MG132 instead of epoxomicin as proteasome inhibitor. Although fewer peptides were analyzed, the pattern of inhibition with MG132 for each of the peptides analyzed was the same as with epoxomicin (Fig. 4C).


Figure 4
View larger version (113K):
[in this window]
[in a new window]

  		FIG. 4. Isotopic labeling of HLA-B27 ligands in the presence of proteasome inhibitors. A total of 91 and 17 ion peaks showing sufficient intensity in their MALDI-TOF MS spectra were analyzed with 1 µM epoxomicin (A and B) and 20 µM MG132 (C), respectively. Their elution position in HPLC (Fraction N), monoisotopic mass (A; M + H+), the percent relative intensity of the corresponding A + 2, A + 4, or A + 6 peak in the absence of labeling (14N), upon [15N]Arg labeling (15N), or upon labeling in the presence of epoxomicin (15N + Ep) or MG132 (15N + MG132), the number of Arg residues (R), and the labeling ratio (Ratio; see "Experimental Procedures") are indicated. Peptides were classified as proteasome-dependent (inhibitor-sensitive) or -independent (inhibitor-insensitive) when the labeling ratio was ≤0.2 or ≥0.4, respectively. The peptide FRYNGLIHR (A; Fraction N, 155; M + H: 1175.3) was assigned as proteasome-dependent due to its total inhibition with 2.5 µM epoxomicin (Fig. 5). Within each set, the peptides are ordered by their number of Arg residues, which was inferred from the isotopic peak that increased upon [15N]Arg labeling and by molecular mass. Ion peaks that were sequenced are indicated. Ion peaks analyzed with additional concentrations of epoxomicin (Fig. 5) are indicated with (*) by their HPLC Fraction N. Ion peaks assigned with both epoxomicin and MG132 are labeled with one (*) or two (**) asterisks in the Sequence column if they were inhibitor-sensitive or -insensitive, respectively. Ion peaks without asterisks in this column were assigned only with epoxomicin.

 
Generation of HLA-B27 Ligands in the Presence of Variable Concentrations of Epoxomicin—
In a recent study (58), incubation of HeLa cells with 0.15 µM epoxomicin abrogated the chymotrypsin-like activity of the proteasome by about 85%, but its trypsin-like activity was only inhibited by about 28%. In contrast, 2 µM epoxomicin was required to achieve extensive inhibition of both activities (about 98 and 87%, respectively). The caspase-like activity of the proteasome was virtually not affected by this inhibitor.

Thus, in a second set of experiments, stable isotope tagging of HLA-B27 ligands was carried out in the presence of 0.2 and 2.5 µM epoxomicin, respectively. This was done to analyze the effect of the progressive inhibition of the trypsin-like activity of the proteasome on the generation of HLA-B27 ligands and to assess whether the proteasome dependence assignments made on the basis of inhibition with 1 µM E-poxomicin were reliable or might be altered in the presence of higher concentration of this inhibitor.

A total of 56 ion peaks fulfilling the same criteria concerning high intensity and labeling as in the previous paragraph were amenable to further analysis (Fig. 5). Peptides were assigned as proteasome-dependent when their labeling was totally inhibited with 2.5 µM epoxomicin as explained in the previous paragraph (see also Fig. 5 legend). A total of 35 (62.5%) and 21 (37.5%) ion peaks were assigned as proteasome-dependent (Fig. 5A) and -independent (Fig. 5B) ligands, respectively. These percentages are similar to those obtained with a 1 µM concentration of the inhibitor.


Figure 5
View larger version (82K):
[in this window]
[in a new window]

  		FIG. 5. Isotopic labeling of HLA-B27 ligands in the presence of various concentrations of epoxomicin. A total of 56 ion peaks were analyzed with 0.2 and 2.5 µM epoxomicin. Labeling ratios with the low (L) and high (H) concentrations of inhibitor are indicated. Proteasome-dependent (A) an proteasome-independent (B) ligands were assigned based on the inhibition of labeling at the highest concentration of the inhibitor. Conventions are as in Fig. 4 except that the threshold for assigning proteasome-independent ligands was a labeling ratio of 0.3. This threshold value, which is somewhat lower than the one used with a 1 µM concentration of the inhibitor, was adopted because it was observed with a TAP-independent ligand (ARLQTALLV; M + H: 984.6) that is presumably generated in the ER and therefore does not reflect partial inhibition of the proteasome. Ion peaks also analyzed at 1 µM epoxomicin (Fig. 4) are indicated with (*) by their HPLC Fraction N. TAP-independent ligands are labeled with (**) in the Sequence column.

 
The ion peak series analyzed with either 1 µM (Fig. 4) or 0.2/2.5 µM (Fig. 5) epoxomicin included a total of 24 proteasome-dependent and 19 proteasome-independent peptides analyzed in both set of experiments (Fig. 6). As many as 20 (83.3%) of the 24 proteasome-dependent peptides in this series showed a labeling ratio ≤0.2 already with 0.2 µM epoxomicin, indicating that most of the proteasome-dependent ligands are not generated under conditions in which the chymotrypsin-like, but not the trypsin-like activity, is inhibited. Only one proteasome-dependent peptide (FRYNGLIHR; M + H: 1175.3) showed some labeling with 1 µM epoxomicin but not with a 2.5 µM concentration of the inhibitor. The dose-dependent inhibition of this peptide by epoxomicin suggests that it can be generated if the trypsin-like activity of the proteasome is not fully inhibited.


Figure 6
View larger version (57K):
[in this window]
[in a new window]

  		FIG. 6. Comparison of the labeling ratios of 43 HLA-B27 ligands at various concentrations of epoxomicin. Conventions are as in Figs. 3 and 4. This figure summarizes data from both figures to facilitate comparisons. TAP-independent ligands are marked with (**).

 
Among the 19 proteasome-independent peptides in this series, 13 (68.4%) showed similar label incorporation at all epoxomicin concentrations so that the decreased labeling of some of them in the presence of the inhibitor was dose-independent (Fig. 6). Moreover two TAP-independent ligands, which are very likely generated in the ER in a proteasome-independent way, also showed decreased labeling in the presence of the inhibitor (Fig. 6). These results strongly suggest that partially decreased labeling in the presence of epoxomicin is due to indirect effects, such as down-regulation of protein synthesis, rather than to partial inhibition of the proteasome.

Four peptides (M + H: 970.4, 1291.4, 1341.3, and 1419.5), two of which were derived from the same protein, showed decreasing labeling as a function of epoxomicin concentration (Fig. 6). However, they were assigned as proteasome-independent because significant labeling was still obtained with the highest concentration of the inhibitor where the proteasomal contribution to the generation of these ligands is very unlikely. This result might be explained by a dose-dependent effect of the inhibitor on the synthesis of the parental proteins. One of the proteasome-independent peptides (VRLLLPGELAK; M + H: 1208.5) showed a dose-dependent increase of labeling with epoxomicin. This result suggests that either the parental protein of this ligand is up-regulated upon proteasome inhibition or that the ligand is actually destroyed by the proteasome.

When the two peptide sets analyzed with either 1 or 0.2/2.5 µM epoxomicin were jointly considered (Figs. 4 and 5), the proteasome dependence of 104 different HLA-B27 ligands could be established. Of these, 73 (70.2%) and 31 (29.8%) were assigned as proteasome-dependent and -independent, respectively.

Surface Expression of HLA-B*2705 in the Presence of Epoxomicin Parallels the Percentage of Proteasome-independent Ligands—
It was shown previously that the surface re-expression of HLA-B*2705 after acid stripping in the presence of the conventional proteasome inhibitors lactacystin and N-acetyl-L-leucinyl-L-leucinyl-L-norleucinal was very substantial and higher than for other MHC class I molecules (23). We repeated these experiments, using similar experimental conditions, with the more specific inhibitor epoxomicin, which, to our knowledge, does not inhibit any known protease other than the proteasome, and using the ME1 mAb, which does not react with untransfected C1R cells. The results (Fig. 7) indicated that B*2705 re-expression in the presence of 1 µM epoxomicin was {cong}34% of the re-expression in the absence of the inhibitor after subtracting the expression levels in the presence of BFA. This result is similar to that reported by Luckey et al. (23) and to the percentage of proteasome-independent ligands described in the previous paragraph, suggesting that the percentage of these ligands determined by stable isotope tagging is representative of the whole B*2705-bound peptide pool.


Figure 7
View larger version (25K):
[in this window]
[in a new window]

  		FIG. 7. Surface re-expression of HLA-B*2705 after acid stripping (STP) in the presence of epoxomicin. A, C1R-B*2705 cells were either untreated or preincubated for 2 h with 1 µM epoxomicin or for 30 min with 10 µg/ml BFA. Then they were acid-washed, allowed to re-express HLA-B27 for 4 h in the presence or absence of these inhibitors, and subjected to flow cytometry with ME1. This mAb does not stain untransfected C1R cells (not shown). A representative experiment, of a total of three independent experiments, is shown. B, mean ± S.D. of three experiments showing the percentage of HLA-B27 re-expression in the presence of epoxomicin or BFA relative to its re-expression in the absence of inhibitors. Ab, antibody; Epox, epoxomicin.

 
Proteasome-independent and -dependent Ligands Differ Little in Their Peptide Motifs, Flanking Sequences, and Subcellular Localization of Their Parental Proteins—
A total of 19 proteasome-independent and 31 proteasome-dependent ligands were sequenced by MS/MS (Fig. 8). In the former group, only three peptides were derived from signal sequences, whereas the others corresponded to internal protein sequences. A comparison of the structural features of proteasome-dependent and -independent peptides failed to reveal any statistically significant differences in residue usage at the N- and C-terminal positions (P1, PC) or in the adjacent residues within the peptides (P2-P3, PC – 2, PC – 1) or in the parental proteins (N – 2, N – 1, C + 1, C + 2) (Fig. 8 and data not shown). The only exceptions were a marginally increased frequency of Tyr at P3 and Leu at PC (p = 0.046) among proteasome-dependent ligands and Arg at PN – 2 among proteasome-independent ones (p = 0.043). No obvious differences in charge or overall chemical character were observed between the two peptide sets. For instance, the average pI of the proteasome-independent and -dependent ligands sequenced in this study was 10.45 ± 2.07 and 10.42 ± 1.29, respectively. Moreover both the proteasome-dependent and -independent peptides were detected along the whole HPLC chromatogram, indicating no obvious bias in the retention times of peptides from both sets. These results suggest that the proteasome dependence of B*2705 ligands is largely unrelated to the structure of the peptides and to the flanking sequences of their parental proteins.


Figure 8
View larger version (87K):
[in this window]
[in a new window]

  		FIG. 8. Amino acid sequences of proteasome-independent and proteasome-dependent HLA-B27 ligands. Within each set peptides are ordered by size and alphabetically. Peptides that were reported previously as HLA-B*2705 ligands (10, 74) are indicated with P. Peptides whose sequence was directly confirmed with synthetic peptides in this study are indicated with S. The corresponding parental proteins, their accession numbers (AN) in the Swiss-Prot database, cellular localization, molecular mass (MW), and theoretical isoelectric point (pI) are given. Proteasome-independent ligands arising from signal sequences are labeled with asterisks (*). Polypeptides giving rise to more than one ligand are indicated in boldface. The total number of parental polypeptides of proteasome-independent and -dependent ligands was 16 and 28, respectively.

 
Analysis of the subcellular localization of the parental proteins (Fig. 8) showed no statistical differences between proteasome-dependent and -independent ligands. Although proteins of the exocytic route were similarly represented in the two groups, the three proteasome-independent ligands from these proteins came from their signal sequences, whereas the proteasome-dependent peptides from proteins of the exocytic compartment corresponded to internal sequences. The polypeptides giving rise to more than one of the sequenced ligands were counted only once in this and all analyses concerning parental proteins.

Proteasome-independent Ligands Are Derived Mainly from Basic Proteins of Low Molecular Weight—
A striking difference with regard to protein size was observed among the parental proteins of proteasome-dependent and -independent ligands. With only one exception, proteasome-independent ligands from regions other than signal sequences were derived from low molecular mass (approximate range, 6–16.5 kDa) and basic (pI > 7.0) proteins, whereas proteasome-dependent ligands were derived mainly from proteins ranging from about 12 kDa to more than 200 kDa and showed little bias in the pI of the parental proteins (Fig. 9, A and B).


Figure 9
View larger version (27K):
[in this window]
[in a new window]

  		FIG. 9. Proteasome-independent B*2705 ligands are derived mainly from basic proteins of low molecular mass. A, molecular mass of the parental proteins from proteasome-independent (white bars) and -dependent (gray bars) B*2705 ligands. Bars represent the number of B*2705 ligands arising from proteins within the specified molecular mass ranges. The three peptides derived from signal sequences were not included. B, distribution of size and isoelectric point of the parental proteins of proteasome-independent (white bars) and -dependent (gray bars) HLA-B*2705 ligands. This distribution is compared with the corresponding distribution of these parameters in the human proteome (black bars). Proteins were classified as small (molecular mass, ≤16.5 kDa) and big (molecular mass, >16.5 kDa). Basic and acidic refer to the theoretical pI of the proteins. The differences between the percentage of small basic, big acidic, and big basic parental proteins of proteasome-independent ligands and those of proteasome-dependent or proteins from the human proteome were statistically significant. No statistical differences were found between the two latter sets except for small basic proteins (p = 0.006). C, the molecular mass of the parental proteins from proteasome-dependent ({circ}) and -independent (•) B*2705 ligands is plotted versus their theoretical isoelectric points. The value of 16.5 kDa corresponds to the second highest molecular mass (16.567 kDa) observed among parental proteins of proteasome-independent ligands derived from internal sequences. The three parental proteins of ligands matching signal sequences were not included. D, the same plot for 15,495 annotated human protein entries in the Swiss-Prot database (UniProtKB/Swiss-Prot). E, the same plot for 145 parental proteins from a registry of constitutive B*2705 ligands (10).

 
The size and pI distribution of the parental proteins of proteasome-dependent ligands reflected approximately that of the human proteome except for some over-representation of small basic proteins (21.4 versus 6.6%, p = 0.006), which might be due to the preference of HLA-B27 for Arg-containing peptides (10). In contrast, the parental proteins of proteasome-independent ligands deviated much more significantly from the human proteome in both parameters (p = 5.1 x 10–32 for small basic proteins) because small basic proteins account for only 6.6% of the human proteins (Fig. 9, C and D).

Because there was a close correlation between the expression level of HLA-B27 in the presence of epoxomicin after acid stripping and the percentage of proteasome-independent ligands and because these came mostly from small basic proteins, we wondered whether the parental proteins of known HLA-B27 ligands reflected the size distribution of the human proteome or that observed in our study. When the molecular mass of 145 parental proteins of a large set of HLA-B27 ligands from a published registry (10) was plotted versus the pI of the proteins, the observed distribution (Fig. 9E) was more similar to that obtained with the parental proteins of the proteasome-dependent and -independent ligands in this study (Fig. 9C) than to the human proteome (Fig. 9D). Yet the percentage of small and basic parental proteins from the ligands sequenced in this study was 2.5-fold higher than among the parental proteins of known HLA-B27 ligands (43.2 and 17.2%, respectively; p = 0.0008). Thus, in the peptide set analyzed in our study, characterized by high abundance and good labeling of the peptides, the percentage of proteasome-independent ligands might be overestimated relative to the whole HLA-B27-bound repertoire.

Two of the three proteasome-independent ligands that came from signal sequences of proteins of the exocytic route arose from big acidic proteins: IRAAPPPLF (cathepsin A; molecular mass, {cong}54 kDa; pI of the protein, 6.16) and RRLALFPGVA (ERp57; molecular mass, {cong}56.8 kDa; pI of the protein, 5.98). These ligands are probably TAP-independent and processed in the ER because at least the first ligand is presented by B*2705 in the TAP-deficient T2 cells.2


   DISCUSSION
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
In this study we have 1) established a sensitive and reliable method to distinguish proteasome-dependent from -independent HLA-B27 ligands, 2) applied this method, together with MS-based peptide sequencing, to identify multiple ligands from both sets, and 3) demonstrated that a large majority of the proteasome-independent HLA-B27 ligands analyzed arise from basic proteins of low molecular mass, a subset that accounts for a small percentage of the human proteome.

A first and critical aspect of our study concerns the method used to assess proteasome dependence. Its reliability requires that, under our experimental conditions, the inhibition of the proteasome activity is achieved in a reproducible and essentially quantitative way. This is particularly important because epoxomicin is more effective in inhibiting the chymotrypsin- than the trypsin-like activity of the proteasome (58). To account for this, experiments were performed at various epoxomicin concentrations, the highest of which should almost totally inhibit both proteasomal activities (58). In our experiments all but one of the peptides assigned as proteasome-dependent presented a similar pattern of virtually total inhibition of labeling with 1 µM epoxomicin. At this concentration, the chymotrypsin-like activity of the proteasome should be totally inhibited, but some trypsin-like activity might remain. The single exception found in this set required 2.5 µM epoxomicin to reach total inhibition.

It might be argued that the decreased labeling of many of the peptides assigned as proteasome-independent in the presence of epoxomicin could be due to partial inhibition of the proteasome. There are several reasons why we consider that this possibility is very unlikely. First, a majority (68.4%) of the proteasome-independent peptides analyzed at various concentrations of the inhibitor showed a similar decrease of labeling at all the concentrations tested or at least at the two higher concentrations, which would not be expected if the decrease were due to partial inhibition of the proteasome. Second, as mentioned above, at the highest concentration of epoxomicin used in our experiments both the chymotrypsin- and the trypsin-like activities are almost completely inhibited (58). Although the caspase-like activity still remains, it is very unlikely that, in the absence of the other activities, this may play any significant role in the generation of HLA-B27 ligands. The caspase-like activity of the proteasome would preferentially generate peptides with C-terminal acidic residues, which are not allowed for HLA-B27 binding (10). Third, TAP-independent peptides derived from signal sequences of proteins of the exocytic route also showed decreased labeling when isolated from epoxomicin-treated cells, although these ligands are generated in a proteasome-independent way. Fourth, similar results were obtained with MG132, which inhibits the three proteasomal activities (59). It is unlikely that two chemically different inhibitors fail to fully inhibit the generation of the same proteasome-dependent ligands while completely blocking the rest. Fifth, there is an alternative explanation to the decreased labeling of peptides in the presence of epoxomicin: a variable reduction of labeling is expected for many proteins when exposing cells to epoxomicin because proteasome inhibition alters the rate of protein synthesis (60). Thus, ligands derived from proteins whose synthesis rate is decreased in these conditions will be worse labeled when isolated from epoxomicin-treated cells even if these ligands are generated in a proteasome-independent way. Indeed both the reduced labeling of TAP-independent ligands and the similar effect observed with both epoxomicin and MG132 would be expected if decreased labeling of the peptides is a consequence of the inhibition of the proteasome at the expression level or synthesis rate of the parental proteins.

The increased labeling of three proteasome-independent ligands might also be explained if they are derived from proteins whose expression is increased in response to epoxomicin. For instance, increased expression of stress response proteins upon proteasome inhibition has been reported (61, 62). Increased labeling of peptide ligands could also result directly from the inhibition of the proteasome because this protease is known to destroy some peptide epitopes (62–64).

The similar structural features of the peptides from both subsets suggest that the nature of the HLA-B27 ligands is mainly determined by the requirements of stable interaction with the HLA molecule and not by the origin of the peptide. This result was not necessarily expected a priori because most of the peptide motifs of HLA-B27 at individual positions, except Arg at P2, consist of multiple residues rather than a single one. For instance, the C-terminal peptide motif of B*2705 consists of basic, aliphatic, and aromatic residues (10). Because the C-terminal residue of MHC class I ligands is directly generated by the proteasome (65), alternative proteases, or incomplete inhibition of the trypsin-like activity of the proteasome, could alter the relative frequencies of the C-terminal residues; this was not observed except for a marginal increase of Leu among proteasome-dependent ligands, which should be reassessed with a higher number of peptide sequences. Differences in the flanking sequences of HLA-B27 ligands could arise from the fact that different proteases may be affected in distinct ways by residues flanking their cleavage sites. However, at the N-terminal end of the peptide, putative specificity differences among degrading enzymes may be obscured by amino peptidase-mediated trimming that adjusts many MHC class I ligands to the appropriate size and N-terminal motif (12, 19, 21, 66). Our results are also consistent with the previous suggestion that proteasome-independent ligands are generated by a protease of relaxed specificity or by multiple proteases (23). Some but no dramatic differences in the cellular origin of the parental proteins were apparent. It was remarkable that a number of proteasome-independent ligands arose from mitochondrial proteins of the respiratory chain, raising the interesting possibility of a putative mitochondrial processing of these ligands, in line with previous speculations (68) and with the finding that the same protein expressed in the cytosol or in the mitochondria produced different MHC class I-restricted peptide epitopes (69). In addition, proteasome-independent ligands derived from internal sequences of secreted proteins were not found. However, a putative correlation between the subcellular origin of the parental proteins and the proteasome dependence of the MHC class I ligands should be substantiated with significantly higher peptide numbers.

The main finding of our study was that, with few exceptions, proteasome-independent ligands arose from basic proteins of low molecular mass, whereas proteasome-dependent ligands arose from a protein set whose size and pI distribution was much more similar to those of the human proteome. The few exceptions found among proteasome-independent ligands were of two kinds. The first exception was peptides arising from signal sequences of large and/or non-basic proteins. Although some such ligands are TAP-dependent and may require proteasomal processing (26, 27), a majority is TAP-independent and is presumably generated in the lumen of the ER (24, 25). Indeed two of the proteasome-independent ligands from signal sequences found in this study are expressed on HLA-B27 transfectants of the TAP-deficient T2 cells.2 The second exception concerns a lactate dehydrogenase (LDH)-B-derived peptide. A very similar ligand, matching a sequence of the LDH-A subunit, was also reported as proteasome-independent (23). LDH is normally degraded by the lysosomal proteolytic pathway (70). In addition, LDH-A is monoubiquitylated and targeted to lysosomal degradation under conditions of oxidative stress (71), raising the possibility of a putative lysosomal origin of HLA-B27 ligands derived from LDH; this deserves further analysis. The differences in pI found among the parental proteins of proteasome-independent and -dependent ligands did not apply to the ligands themselves because the pI of both peptide sets was very similar.

It might be argued that the 50 ligands whose proteasome dependence was determined in this study are a very small portion of the whole B27-bound peptide repertoire, which, as for any MHC class I molecule, consists of several thousands of peptides (8). That the percentage of proteasome-independent peptides was similar to the percentage of surface re-expression of HLA-B27 in the presence of epoxomicin after acid stripping suggests that this data set may be approximately representative of the bulk of the B27-bound peptide pool. However, the percentage of small basic parental proteins in the data set analyzed in this study (a total of 44 polypeptides) was 2.5-fold higher than the corresponding percentage among the 145 parental proteins of known HLA-B27 ligands from a published registry (10). A possible explanation for this discordance is that our data set was limited by the requirements of high intensity and significant labeling of the corresponding ion peaks, and it is likely to consist of abundant peptides in the B27-bound pool. A majority of the MHC class I ligands are presented in very low amounts and are not easily amenable to this analysis. Thus, it is possible that our data set may represent more closely the percentage of proteasome-independent and -dependent ligands among the most abundant peptides than among the whole B27-bound pool.

The strong bias observed among proteasome-independent ligands toward small and basic parental proteins was not seen in a previous study (23). However, as already noted, that study was based on acid stripping and re-expression of HLA-B27 in the presence of proteasome inhibitors, and removal of previously synthesized peptides from the cell surface was only indirectly assessed by flow cytometry of acid-stripped HLA-B27. Whereas acid washing can remove the majority of peptides from the cell surface, flow cytometry cannot properly assess their quantitative removal prior to inhibition of the proteasome. If peptide removal were not complete, this might lead to misassignment of some peptides as proteasome-independent. Indeed one of the peptides that in our study was clearly proteasome-dependent (NRFAGFGIGL; see Fig. 4) was reported in that study (23) as generated in the presence of proteasome inhibitors.

Our observed bias of proteasome-independent HLA-B27 ligands toward arising from proteins of small size demonstrates the existence of a hitherto unnoticed non-proteasomal proteolytic pathway that makes a significant contribution to the HLA-B27-bound peptide repertoire. The basic character of these proteins might also be a specificity feature of this non-proteasomal activity or might just be a consequence of the preference of HLA-B27 for basic peptides. B27 ligands contain at least an Arg residue at P2 and frequently additional Arg or other basic residues at P1 and PC. Acidic residues are not allowed or are very disfavored at multiple positions such as P1, P2, P3, and PC (10). Thus, small basic proteins, due to their higher frequency of basic residues, are likely to generate more peptides with these features than small acidic proteins. Obviously this difference would not apply to larger proteins, which could contain sequences compatible with HLA-B27 ligands even if the protein has an overall acidic character. In addition, small basic proteins (≤16.5 kDa) are 2-fold more abundant than small acidic proteins in the human proteome (6.6 and 3.4%, respectively).

It seems unlikely that lysosomal degradation is a main source of proteasome-independent HLA-B27 ligands. Whereas some lysosomal activities may occasionally generate MHC class I ligands, as demonstrated in dendritic cells (31) and hypothesized in this study for LDH-derived ligands, it is clear that, globally, the lysosomal degradation pathway does not explain the observed bias toward small parental proteins because lysosomal targeting and degradation is not size-dependent (72). However, we cannot rule out the possibility that lysosomal or Golgi proteases might contribute to a small extent to the HLA-B27-bound peptide repertoire. The cytosolic protease TPPII (17) might seem a more likely candidate for proteasome-independent HLA-B27 ligands. This enzyme has endopeptidase activity (17, 18) and can generate class I epitopes (29, 30). Actually it was recently proposed that it is the major protease capable of processing peptides longer than 14 residues (20). TPPII cleaves after Lys residues but, at least in vitro, also after non-basic residues (17). Therefore, the absence of a bias toward C-terminal Lys residues among the proteasome-independent B*2705 ligands does not argue against a role of TPPII in their generation. Whether TPPII can directly generate MHC class I ligands from proteins of the size observed in this study (i.e. about 6–16.5 kDa) or might work on shorter protein fragments produced by other non-proteasomal endopeptidases remains to be explored. However, unlike the effect of epoxomicin, TPPII inhibition with its specific inhibitor butabindide did not impair the surface re-expression of HLA-B27 after acid stripping.3 This is in contrast to published evidence for other HLA class I molecules (18), but it is in full agreement with the recent observation that small interfering RNA-mediated TPPII inhibition has no effect on the generation of properly folded MHC class I proteins (20), implying that this enzyme is not required for most MHC class I antigen presentation. Our result suggests that TPPII is probably not responsible for the generation of the bulk of proteasome-independent HLA-B27 ligands. Thus, the enzyme(s) involved in the generation of this peptide subset remains to be identified, and efforts toward this aim are currently ongoing.

Finally we have not addressed the role of the non-proteasomal pathway described in this study for class I molecules other than HLA-B27. However, we examined the parental proteins of a reported series of HLA-B35 (73) and HLA-B14 (56) ligands. These two allotypes show a significantly higher proteasome dependence than HLA-B27 in acid stripping experiments (Ref. 23 and Footnote 4, respectively).4 For both B35 and B14 ligands, the percentage of small (≤16.5 kDa) and basic parental proteins (6.9 and 9.1%, respectively) was close to the corresponding percentage of these proteins in the human proteome (6.6%) as expected from the strong proteasome dependence of both allotypes. Analysis of proteasome-dependent and -independent ligands of non-B27 class I molecules with a proteasome-dependence comparable to HLA-B27 is currently under way.


   ACKNOWLEDGMENTS
 
We thank Anabel Marina and Juan P. Albar (Proteomics Department, Centro de Biología Molecular Severo Ochoa and Centro Nacional de Biotecnología, respectively) and the technical staff for assistance in MS and Manuel Ramos (Instituto de Salud Carlos III, Madrid, Spain) for sharing unpublished data. We especially thank Hidde Ploegh (Whitehead Institute, Massachusetts Institute of Technology, Cambridge, MA) and Peter van Endert (Université René Descartes, Hôpital Necker, Paris, France) for useful critiques.


  FOOTNOTES
 
Received, August 9, 2006, and in revised form, October 3, 2006.

Published, MCP Papers in Press, February 16, 2007, DOI 10.1074/mcp.M600302-MCP200

1 The abbreviations used are: MHC, major histocompatibility complex; ER, endoplasmic reticulum; TAP, transporter associated with antigen processing; TPPII, tripeptidyl-peptidase II; mAb, monoclonal antibody; MG132, carbobenzoxy-L-leucyl-L-leucyl-L-leucinal; BFA, brefeldin A; LDH, lactate dehydrogenase; HLA, human leukocyte antigen. Back

2 M. Ramos and J. A. López de Castro, unpublished data. Back

3 M. Marcilla and J. A. López de Castro, unpublished observations. Back

4 E. Merino and J. A. López de Castro, unpublished observations. Back

* This work was supported by Ministry of Science and Technology Grants SAF2003-02213 and SAF2005-03188, Comunidad Autónoma de Madrid Grant 08.3/0005/2001.1, and an institutional grant from the Fundación Ramón Areces (to the Centro de Biología Molecular Severo Ochoa). 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. Back

S The on-line version of this article (available at http://www.mcponline.org) contains supplemental material. Back

{ddagger} To whom correspondence should be addressed: Centro de Biología Molecular Severo Ochoa, Facultad de Ciencias, Universidad Autónoma, 28049 Madrid, Spain. Tel.: 34-91-497-8050; Fax: 34-91-497-8087; E-mail: aldecastro{at}cbm.uam.es


   REFERENCES
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Sadasivan, B., Lehner, P. J., Ortmann, B., Spies, T., and Cresswell, P. (1996) Roles for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class I molecules with TAP. Immunity 5, 103 –114[CrossRef][Medline]

  2. Hughes, E. A., and Cresswell, P. (1998) The thiol oxidoreductase ERp57 is a component of the MHC class I peptide-loading complex. Curr. Biol. 8, 709 –712[CrossRef][Medline]

  3. Lindquist, J. A., Jensen, O. N., Mann, M., and Hammerling, G. J. (1998) ER-60, a chaperone with thiol-dependent reductase activity involved in MHC class I assembly. EMBO J. 17, 2186 –2195[CrossRef][Medline]

  4. Morrice, N. A., and Powis, S. J. (1998) A role for the thiol-dependent reductase ERp57 in the assembly of MHC class I molecules. Curr. Biol. 8, 713 –716[CrossRef][Medline]

  5. Ortmann, B., Copeman, J., Lehner, P. J., Sadasivan, B., Herberg, J. A., Grandea, A. G., Riddell, S. R., Tampe, R., Spies, T., Trowsdale, J., and Cresswell, P. (1997) A critical role for tapasin in the assembly and function of multimeric MHC class I-TAP complexes. Science 277, 1306 –1309[Abstract/Free Full Text]

  6. Park, B., Lee, S., Kim, E., Cho, K., Riddell, S. R., Cho, S., and Ahn, K. (2006) Redox regulation facilitates optimal peptide selection by MHC class I during antigen processing. Cell 127, 369 –382[CrossRef][Medline]

  7. Huczko, E. L., Bodnar, W. M., Benjamin, D., Sakaguchi, K., Zhu, N. Z., Shabanowitz, J., Henderson, R. A., Appella, E., Hunt, D. F., and Engelhard, V. H. (1993) Characteristics of endogenous peptides eluted from the class I MHC molecule HLA-B7 determined by mass spectrometry and computer modeling. J. Immunol. 151, 2572 –2587[Abstract]

  8. Engelhard, V. H. (1994) Structure of peptides associated with class I and class II MHC molecules. Annu. Rev. Immunol. 12, 181 –207[CrossRef][Medline]

  9. Hickman, H. D., Luis, A. D., Buchli, R., Few, S. R., Sathiamurthy, M., VanGundy, R. S., Giberson, C. F., and Hildebrand, W. H. (2004) Toward a definition of self: proteomic evaluation of the class I peptide repertoire. J. Immunol. 172, 2944 –2952[Abstract/Free Full Text]

  10. Lopez de Castro, J. A., Alvarez, I., Marcilla, M., Paradela, A., Ramos, M., Sesma, L., and Vazquez, M. (2004) HLA-B27: a registry of constitutive peptide ligands. Tissue Antigens 63, 424 –445[CrossRef][Medline]

  11. Rock, K. L., Gramm, C., Rothstein, L., Clark, K., Stein, R., Dick, L., Hwang, D., and Goldberg, A. L. (1994) Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 78, 761 –771[CrossRef][Medline]

  12. Rock, K. L., York, I. A., Saric, T., and Goldberg, A. L. (2002) Protein degradation and the generation of MHC class I-presented peptides. Adv. Immunol. 80, 1 –70[Medline]

  13. Dick, L. R., Aldrich, C., Jameson, S. C., Moomaw, C. R., Pramanik, B. C., Doyle, C. K., DeMartino, G. N., Bevan, M. J., Forman, J. M., and Slaughter, C. A. (1994) Proteolytic processing of ovalbumin and ß-galactosidase by the proteasome to a yield antigenic peptides. J. Immunol. 152, 3884 –3894[Abstract]

  14. Kessler, J. H., Beekman, N. J., Bres-Vloemans, S. A., Verdijk, P., van Veelen, P. A., Kloosterman-Joosten, A. M., Vissers, D. C., ten Bosch, G. J., Kester, M. G., Sijts, A., Drijfhout, J. W., Ossendorp, F., Offringa, R., and Melief, C. J. (2001) Efficient identification of novel HLA-A*0201-presented cytotoxic T lymphocyte epitopes in the widely expressed tumor antigen PRAME by proteasome-mediated digestion analysis. J. Exp. Med. 193, 73 –88[Abstract/Free Full Text]

  15. Paradela, A., Alvarez, I., Garcia-Peydro, M., Sesma, L., Ramos, M., Vazquez, J., and Lopez de Castro, J. A. (2000) Limited diversity of peptides related to an alloreactive T cell epitope in the HLA-B27-bound peptide repertoire results from restrictions at multiple steps along the processing-loading pathway. J. Immunol. 164, 329 –337[Abstract/Free Full Text]

  16. Yague, J., Alvarez, I., Rognan, D., Ramos, M., Vazquez, J., and Lopez de Castro, J. A. (2000) An N-acetylated natural ligand of human histocompatibility leukocyte antigen (HLA)-B39. Classical major histocompatibility complex class I proteins bind peptides with a blocked NH2 terminus in vivo. J. Exp. Med. 191, 2083 –2092[Abstract/Free Full Text]

  17. Geier, E., Pfeifer, G., Wilm, M., Lucchiari-Hartz, M., Baumeister, W., Eichmann, K., and Niedermann, G. (1999) A giant protease with potential to substitute for some functions of the proteasome. Science 283, 978 –981[Abstract/Free Full Text]

  18. Reits, E., Neijssen, J., Herberts, C., Benckhuijsen, W., Janssen, L., Drijfhout, J. W., and Neefjes, J. (2004) A major role for TPPII in trimming proteasomal degradation products for MHC class I antigen presentation. Immunity 20, 495 –506[CrossRef][Medline]

  19. York, I. A., Chang, S. C., Saric, T., Keys, J. A., Favreau, J. M., Goldberg, A. L., and Rock, K. L. (2002) The ER aminopeptidase ERAP1 enhances or limits antigen presentation by trimming epitopes to 8–9 residues. Nat. Immunol. 3, 1177 –1184[CrossRef][Medline]

  20. York, I. A., Bhutani, N., Zendzian, S., Goldberg, A. L., and Rock, K. L. (2006) Tripeptidyl peptidase II is the major peptidase needed to trim long antigenic precursors, but is not required for most MHC class I antigen presentation. J. Immunol. 177, 1434 –1443[Abstract/Free Full Text]

  21. Saveanu, L., Carroll, O., Lindo, V., Del Val, M., Lopez, D., Lepelletier, Y., Greer, F., Schomburg, L., Fruci, D., Niedermann, G., and Van Endert, P. M. (2005) Concerted peptide trimming by human ERAP1 and ERAP2 aminopeptidase complexes in the endoplasmic reticulum. Nat. Immunol. 6, 689 –697[CrossRef][Medline]

  22. Vinitsky, A., Anton, L. C., Snyder, H. L., Orlowski, M., Bennink, J. R., and Yewdell, J. W. (1997) The generation of MHC class I-associated peptides is only partially inhibited by proteasome inhibitors: involvement of nonproteasomal cytosolic proteases in antigen processing. J. Immunol. 159, 554 –564[Abstract]

  23. Luckey, C. J., Marto, J. A., Partridge, M., Hall, E., White, F. M., Lippolis, J. D., Shabanowitz, J., Hunt, D. F., and Engelhard, V. H. (2001) Differences in the expression of human class I MHC alleles and their associated peptides in the presence of proteasome inhibitors. J. Immunol.<