|Home | About | Journals | Submit | Contact Us | Français|
All three members of the oxytocinase sub-family of M1 aminopeptidases, ERAP1 (ERAAP), ERAP2 and PLAP (IRAP), have been implicated in the generation of MHC class I-presented peptides. ERAP1 and 2 trim peptides in the endoplasmic reticulum for direct presentation whereas PLAP has been recently implicated in cross presentation. The best characterized member of the family, ERAP1, has unique enzymatic properties that fit well with its role in antigen processing. ERAP1 can trim a large variety of long peptide sequences and efficiently accumulate mature antigenic epitopes of 8–9 amino acids long. In this study we evaluate the ability of PLAP to process antigenic peptide precursors in vitro and compare it to ERAP1. We find that, similarly to ERAP1, PLAP can trim a variety of long peptide sequences efficiently and, in most cases, accumulates appreciable amounts of correct length mature antigenic epitope. Again, similarly to ERAP1, PLAP continued trimming some of the epitopes tested and accumulated smaller products effectively destroying the epitope. However, the intermediate accumulation properties of ERAP1 and PLAP are distinct and epitope dependent, suggesting that these two enzymes may impose different selective pressures on epitope generation. Overall, although PLAP has the necessary enzymatic properties to participate in generating or destroying MHC class I presented peptides, its trimming behavior is distinct from that of ERAP1, something that supports a separate role for these two enzymes in antigen processing.
The ability of cytotoxic T lymphocytes to recognize and eradicate diseased cells relies on the surface presentation of small peptide fragments bound onto specialized receptors of the class I Major Histocompatibility Complex (MHC class I) (1). These antigenic peptides are the product of complex intracellular proteolytic pathways. MHC class I molecules present peptides 8–10 residues long that are either derived from the degradation of intracellular cytosolic proteins (classical pathway) or endocytosed extracellular proteins (cross-presentation pathway) (2, 3). During the last few years an important role for aminopeptidases, enzymes that trim the N-terminus of peptides or proteins, has been established in the final stage of the peptide processing pathway. Peptides generated by the proteasome and translocated into the ER by the Transporter Associated with Antigen Processing (TAP) often carry N-terminal extensions compared to the mature epitopes (4). These epitope precursors have to be trimmed by ER resident aminopeptidases so that they can bind onto nascent MHC class I molecules (5). ER aminopeptidase 1 (ERAP1, also named ERAAP and A-LAP) has been demonstrated to play a crucial role in trimming peptide precursors leading to the generation or destruction of mature antigenic peptides (6–13). ERAP1 has also been demonstrated to play important roles in the cross-presentation pathway (14). In the absence of ERAP1, many peptide precursors are not appropriately processed and are therefore not presented on the cell surface (10). However, ERAP1 can also destroy some epitopes by further trimming them to smaller products that cannot bind onto MHC class I molecules (7, 8, 11, 15). ERAP1 can therefore either generate or destroy mature antigenic peptides. ERAP1’s epitope-destructive properties appear to be physiologically relevant, by restricting the variety of epitopes generated and influencing the peptide pool available for presentation. It is currently well established that ERAP1 affects the generation of the antigenic peptide repertoire and in that context affects immunodominance in vivo (9, 10, 12, 13, 16). Although absent from mice, many other species also have a highly homologous ER aminopeptidase 2 (ERAP2, also named L-RAP) that can directly interact with ERAP1 and has been demonstrated to participate in antigenic peptide processing, albeit with a role that is secondary or accessory to ERAP1 (17).
ERAP1 and ERAP2 are zinc metallopeptidases that belong to the oxytocinase sub-family of M1 aminopeptidases (18). ERAP1 has unique biochemical properties that fit with its role in antigen presentation. In contrast to most aminopeptidases, it can trim large peptides efficiently and process a large number of peptide sequences (7, 15). Furthermore, it trims peptides in a sequence dependent manner and accumulates peptides 8–9 residues long, the optimal length for MHCI binding. For this property ERAP1 has been called a “molecular ruler” since it was considered to be the only aminopeptidase with a length-dependent trimming behavior (11, 15). The structural basis of the unusual enzymatic properties of ERAP1 is not clear. However, it has been hypothesized that ERAP1’s peptide trimming properties are a result of specific structural features of its substrate binding site (11, 15).
The third member of this sub-family is named PLAP (Placental Leucine Aminopeptidase, also known as Insulin Regulated Aminopeptidase, IRAP) (18, 19). PLAP shares with ERAP1 and ERAP2 a high degree of homology in both the Zn(II) catalytic centre and the characteristic helical C-terminal domain (20). However, PLAP contains an N-terminal extension that presumably constitutes a transmembrane tail and is hypothesized to contribute to intracellular vesicle localization (18). PLAP has been found to co-localize with the glucose transporter GLUT4 in intracellular vesicles and degrade vasopressin as well as several other peptide hormones (21–23). PLAP has also been detected in intracellular vesicular trafficking pathways that are stimulated after antigenic stimuli, suggesting that PLAP may somehow be involved in the immune response (24). Interestingly, PLAP protein levels have been shown to be stimulated by interleukin-1β (25). All these observations have hinted towards a role for PLAP in antigen presentation. This hypothesis was confirmed recently by the demonstration that PLAP participates in cross-presentation, especially in inflammatory monocyte-derived dendritic cells (26, 27).
The high degree of sequence conservation between ERAP1 and PLAP in combination with fact that the two enzymes perform similar biological functions, raise the question as to whether the unique enzymatic properties of ERAP1 that make it suitable for antigen processing, are also shared by PLAP. On the other hand, the involvement of PLAP in a separate location and pathway of antigen presentation than ERAP1 raises the question as to whether these two aminopeptidases have different enzymatic properties.
In an effort to evaluate the enzymatic properties of PLAP in the context of antigenic peptide processing, we expressed a soluble active form of PLAP in a mammalian cell line and used the purified protein in in vitro assays to test the enzyme’s ability to trim antigenic peptide precursors and generate the correct length mature epitopes in direct comparison to ERAP1. We focused our analysis on the accumulation of trimming intermediates, a feature that links the molecular and enzymatic properties of an aminopeptidase to its suitability to perform appropriate biological functions. Accumulation of correct length trimming intermediates that corresponded to antigenic epitopes has been previously thought to be a property unique to ERAP1. We tested the trimming of 14 different precursor peptides of different lengths and sequences that correspond to 10 different antigenic epitopes. PLAP was able to efficiently generate the antigenic epitope from 10 out of the 14 precursors tested, whereas ERAP1 generated the epitope from 13 out of the 14 precursors. However, in many cases (9/14) PLAP readily trimmed the mature epitope and generated smaller products whereas ERAP1 was less effective in accumulating smaller intermediates (6/14 cases). Interestingly, the in vitro epitope generation properties of ERAP1 and PLAP varied depending on the epitope and precursor tested as did the overall trimming patterns and rates. Our findings suggest that PLAP shares key critical properties with ERAP1 that make it appropriate for intracellular antigen processing. However, PLAP’s epitope-dependent trimming behavior however does not mirror that of ERAP1 suggesting that the two related enzymes can impose different selective pressures in antigen presentation.
293F cells, 293Fectin™, Freestyle™ 293 expression medium, opti-MEM I transfection medium were bought from Invitrogen. All peptides were purchased from Genscript, New Jersey USA. Peptides were purified by reverse-phase HPLC and were >95% pure. MonoQ™ column was from Pharmacia. Ni-NTA agarose was purchased from Qiagen. L-leucine-7-amido-4-methyl coumarin (L-AMC) was from Sigma-Aldrich. Trifluoroacetic acid (TFA) was purchased from Applichem and acetonitrile from Merck.
A soluble version of human PLAP (shPLAP) was cloned between the EcoR1 and XhoI sites of the multi-cloning site of the mammalian expression vector pcDNA6/myc-His. The final expression construct contains the gene for PLAP starting at amino acid position 155, lacking the extended N-terminal domain of the gene that contains the putative transmembrane region of the protein. The N-terminus of the construct contained the signal sequence from the adenoviral E3/19K protein, so as to target the protein to the secretory pathway during expression. The final construct contained the signal sequence from the adenoviral protein E3/29K fused upstream of the PLAP gene, to allow for targeting of the expressed protein to the secretory pathway of the cell.
The pcDNA6/myc-his construct containing the gene for shPLAP was used to transfect 293F cells grown in suspension culture at 8% CO2 and 37°C. Transfection was performed using the 293Fectin transfection reagent according to the freestyle293™ expression system (Invitrogen) instructions. Three days post tranfection the cell supernatant was harvested and expression confirmed by L-AMC assay and by western blotting using an anti-5xHis-HRP conjugate antibody (Qiagen). The cell supernatant was extensively dialyzed against 10mM sodium phosphate buffer at pH 8.0 and the final composition adjusted to 50mM sodium phosphate pH 8.0, 300mM NaCl, 5mM imidazole. 1ml of NiNTA agarose slurry was added to the dialysate and the mixture was gently stirred at 4°C for 1hr to allow for protein binding to the column. Elution was performed using the same buffer containing increasing concentrations of imidazole. Elution was followed by L-AMC aminopeptidase activity assay. The majority of the activity was eluted at 150mM imidazole and the corresponding fractions were pooled and dialyzed in 10mM Tris buffer, pH8.0. The protein was further loaded onto a monoQ™ anion-exchange column attached onto a FPLC system and eluted using a linear NaCl gradient. One major peak was evident from the chromatogram that also corresponded to the majority of L-AMC activity. The corresponding fractions were pooled, concentrated and stored at −20°C in 20mM Tris pH 8.0, 100mM NaCl, 50% glycerol. Protein was quantified based on its absorbance at 280nm using an extinction coefficient of 1.67 mg/ml OD280, calculated based on the amino acid sequence of the expressed construct. Protein integrity was assessed by SDS-PAGE and western-blotting using the anti-5xHis-HRP conjugate antibody. The protein was >98% pure based on SDS-PAGE.
Human recombinant ERAP1 was expressed in insect cells after infection with baculovirus as previously described (11). The recombinant protein was harvested from the cell medium post infection and was purified by a combination of Ni-NTA chromatography, anion-exchange chromatography and size-exclusion chromatography again as previously described (11).
The aminopeptidase activity of human recombinant ERAP1 and PLAP during purification and storage was monitored by following the change of solution fluorescence upon incubation of the recombinant enzyme with the fluorigenic substrate L-leucine-7-amido-4-methyl coumarin (L-AMC). 140 μL of a 100μM L-AMC solution containing 20mM Tris pH 8.0 and 100mM NaCl was incubated with either ERAP1 or PLAP and the fluorescence of the solution was monitored using a QuantaMaster4 (Photon Technologies International) fluorimeter using an excitation wavelength of 380nm and following the emission at 460nm. The generated time-dependent change of the signal was calibrated using an AMC standard curve. The slope of the time-course is directly proportional to the enzymatic activity.
The N-terminal trimming of model antigenic peptide precursors by either ERAP1 or PLAP was followed separating the trimming products on a reverse-phase column attached to an HPLC system. For the trimming reactions, 100μM peptide in 20μL of 50mM Tris pH 8.0 and 100mM NaCl was incubated with 1–2000nM of either ERAP1 or PLAP at 37°C. In all cases the peptide was in excess over the enzyme. At the end of the incubation, the reaction was stopped by adding TFA to a final concentration of 0.5% and storing the reaction at −20°C. The trimming products were analyzed on a reverse-phase column (Chromolith™, Merck) equilibrated with 0.05% TFA. Elution was performed using a linear concentration gradient of acetonitrile. Eluted peptides were typically detected by absorbance at 220nm. Product peaks were identified either based on their elution time as defined by running control peptides or by spiking part of the reaction with a control peptide. Novel product peaks were collected during elution from the column and lyophilized for further analysis by ESI-MS to assist with peptide identification.
The rate of hydrolysis of the fluorigenic substrate L-AMC by either ERAP1 or PLAP was measured as described above by substituting the appropriate buffer system to achieve the required pH value. The buffers used were the following: pH range 4–5.9, citrate; pH range 6.0–6.8 phosphates; pH range 6.9–7.9 HEPES; pH range 8.0–8.9 Tris and pH range 9.0–9.8 Glycine. All reaction solutions contained 50mM buffer, 100mM NaCl and 100μM L-AMC.
We used a mammalian cell line expression system to express a soluble version of human PLAP for in vitro assays, as described in the Materials and Methods section. The pcDNA6/myc-his plasmid vector contained the human PLAP gene starting at amino acid 155, lacking the PLAP N-terminal domain that is postulated to contain a transmembrane region that would attach the protein on the cell’s vesicular or plasma membrane. The designed construct more readily resembles the soluble version of PLAP that can be created after proteolytic digestion of the full length protein (23). The expressed amino acid sequence encompasses the section of the enzyme that shares up to 50% identity with full length ERAP1 or ERAP2. Recombinant PLAP was purified to homogeneity by affinity purification (Ni-NTA column) and anion-exchange chromatography and was found to be active as an aminopeptidase (Figure 1, panel A). Compared to human recombinant ERAP1, PLAP was significantly more active towards the hydrolysis of the model fluorigenic substrate L-AMC, having a specific activity of 233±14 × 103 mol AMC/mol enzyme × min versus the specific activity of ERAP1 that was found to be 3.1±0.6 × 103 mol AMC/mol enzyme × min.
Recently, Saveanu et al (26) demonstrated that PLAP affects cross-presentation in endosomal compartments. Since the pH of endosomal compartments becomes more acidic than the pH of the ER where ERAP1 primarily operates, we asked whether the optimal pH dependence of PLAP activity is different than the pH dependence for ERAP1. To test this, we measured L-AMC hydrolysis by ERAP1 or PLAP in varying pH solutions within the range of 4–10. Both aminopeptidases displayed maximum hydrolytic efficiency at pH 7.0 (Figure 1, panel B). However, PLAP retained significant hydrolytic activity over a wider range of pH compared to ERAP1. Specifically, PLAP was found to be very active at mildly acidic conditions and retained almost 70% of its activity at pH 6.0, compared to ERAP1 that retained about 30% of its maximum activity at that pH. Additionally, PLAP retained 50% of its activity at pH 8.0. This property appears consistent with the proposed role of PLAP in endosomes that are gradually acidified after endocytosis or ones that are initially alkalinized (28, 29). It should be noted however, that both aminopeptidases displayed no detectable activity under stronger acidic conditions (i.e. pH 5.0).
To evaluate PLAP’s in vitro ability to trim long peptides and accumulate intermediates that correspond to actual antigenic peptides we used 14 model antigenic peptide precursors to 10 known antigenic epitopes. The peptides used were of varying size and sequence designed to test different aspects of PLAP’s trimming capabilities. All experiments were done in direct comparison to trimming by ERAP1, an enzyme whose peptide processing properties are well established. We focused our analysis on the accumulation of trimming intermediates and more specifically the 8mer or 9mer mature antigenic epitope since up to now this has been considered to be a characteristic and unique property of ERAP1. Several tests were performed for each peptide, with varying amounts of enzyme in the range of 1 nM to 2 μ, in order to identify all product peaks that accumulate during trimming. Accumulated product peaks were characterized by the use of peptide controls and mass spectrometry. In almost all trimming reactions examined, use of similar concentrations of ERAP1 and PLAP was sufficient to trim the initial substrate and generate product peaks, indicating that PLAP is equally effective as ERAP1 in trimming long peptides. In each case, characteristic chromatograms are depicted that highlight product accumulation.
We first used a model antigenic precursor of the well characterized ovalbumin antigenic epitope SIINFEKL that is presented by the MHC class I molecule H2-Kb (30). It has been previously demonstrated both in vitro and in cell-based experiments, that ERAP1 generates the SIINFEKL peptide efficiently from N-terminally extended precursors (7, 8). In this study we used a single amino acid extension, carrying a leucine residue, an optimal N-terminal residue for both ERAP1 and PLAP (31, 32). Consistent with previously reported results, ERAP1 trims LSIINFEKL efficiently, accumulating a single product that corresponds to the SIINFEKL peptide (Figure 2, panel A). PLAP is also able to trim the LSIINFEKL peptide efficiently and similarly to ERAP1, PLAP initially generates appreciable amounts of the SIINFEKL epitope. In contrast to ERAP1, however, PLAP continues trimming, generating smaller peptides and accumulating the 7mer peptide with the sequence IINFEKL (Figure 2, panel B). Interestingly, both ERAP1 and PLAP were equally efficient in producing the mature epitope, having very similar specific rates of production (2.0 pmol/sec/μg for ERAP1 and 2.1 pmol/sec/μg for PLAP, see table I).
Inspection of the naturally occurring epitopes deposited in the SYFPEITHI database reveals that the majority of antigenic peptides are 9mers (33). Furthermore, analysis of antigenic peptide precursors generated by the immunoproteasome has suggested that the one of the most common N-terminal extensions is a single amino acid(4). As a result, the ability of an aminopeptidase to excise a single amino acid from a 10mer antigenic peptide precursor and accumulate the 9mer product is probably highly relevant to its suitability to efficiently generate antigenic epitopes in the cell. To evaluate the ability of PLAP to generate 9mer model antigenic peptides from 10mer precursors we analyzed the trimming of 7 model 10mer peptides, 5 of which correspond to naturally occurring antigenic peptide precursors. Three of the peptides we tested (sequences IRLRPGGKKK, KSLYNTVATL and KGYKDGNEYI) correspond to natural precursors of epitopes found to be cross-presented (34, 35). Figure 3 demonstrates characteristic chromatograms from trimming reactions highlighting accumulated products. ERAP1 was able to efficiently generate the HLA-A*03 9mer ligand RVYEKMALY and its variant RVYEKMALR from a 10mer precursor whereas PLAP did not accumulate the 9mer but rather trimmed it further and accumulated 8mer and 7mer products (Figure 3, panels A and B). Both ERAP1 and PLAP efficiently accumulated the HLA-A*03 and HLA-B*27 epitopes RLRPGGKKK and SRHHAFSFR respectively (Figure 3, panels C and D). Similarly both ERAP1 and PLAP accumulated significant amounts of the HLA-A*02 epitope SLYNTVATL (Figure 3, panel E). This epitope was resistant to further trimming by PLAP but was degraded to accumulate the 7mer product by ERAP1 (Figure 3, panel E). The HLA-B*27 epitope precursor DIRSSVQNKL was highly resistant to both enzymes and no detectable epitope was produced under reasonably achievable experimental conditions (Figure 3, panel F). However, incubation with larger amounts of ERAP1, led to accumulation of the 7mer and 8mer products, indicating that ERAP1 can destroy this epitope (Figure 3, panel F). Finally, both enzymes were able to accumulate the H-2-Kd epitope GYKDGNEYI as an intermediate, but continued to further trim it to accumulate the 6mer product DGNEYI (Figure 3, panel G). The results for all these peptides are summarized in Table I, including the calculated rates of production of the mature antigenic epitopes.
Overall, both enzymes were reasonably successful in accumulating appreciable amounts of the 9mer mature epitopes in vitro (6 of the 7 cases for ERAP1, panels A-E and G; 4 of the 7 cases for PLAP, panels C, D, E and G). Additionally, in 4 out of 7 cases, ERAP1 trimmed the epitope further and accumulated smaller products 6–8 amino acids long effectively destroying the epitope (panels C, E, F and G). Similarly, PLAP accumulated smaller fragments in 4 out of 7 cases (panels A, B, C and G). Epitope generation rates varied over a range of 3 orders of magnitude but were very often comparable between ERAP1 and PLAP (Table I). Interestingly, for both enzymes, the ability to generate epitopes depended on the epitope sequence examined. Overall, ERAP1 and PLAP showed significant differences in which of the tested epitopes they were able to accumulate or continue trimming to smaller products indicating that the two enzymes have significant differences in their substrate preferences.
Some antigenic peptide precursors transported into the ER are longer peptides up to 16 amino acids long (4, 36, 37). Trimming from ER resident aminopeptidases would therefore be a multi-step process that may necessitate coordinated trimming steps from more than one aminopeptidase. Recently such a mechanism has been proposed for ERAP2 (17). ERAP1 has been shown to be able to efficiently generate mature antigenic peptides from long N-terminally extended precursors (7, 8, 15). To compare PLAP’s ability to perform this function we studied the trimming of 6 antigenic peptide precursors based on 3 antigenic epitopes. The following peptides were tested: i) the KRFEGLTQR epitope, a ligand for HLA-B27 and a potential link between predisposition to the autoimmune disease ankylosing spondylitis and this MHC allele (38–40), ii) the ATFPDTLTY epitope which is a HLA-A*03 ligand and a tumor related antigen derived from human integrin alpha 6 chain (41) and iii) the cross-presented GPGRAFVTI peptide that can bind onto the H-2-Ld allele and is derived from the env protein of the Human Immunodeficiency Virus 1 IIIB (42, 43).
We first tested the ability of ERAP1 and PLAP to generate the KRFEGLTQR epitope from a 14mer N-terminally extended precursor with the sequence LNYTYKRFEGLTQR derived form the natural protein precursor sequence. Generation of the 9mer epitope requires the sequential removal of 5 N-terminal amino acids. Figure 4 shows representative chromatograms of the trimming products after incubation of the 14mer precursors with ERAP1 and PLAP respectively. ERAP1 trimmed the peptide in vitro efficiently and generated appreciable amounts of KRFEGLTQR. However, ERAP1 did not accumulate the 9mer peptide exclusively but continued trimming down to the 7mer product FEGLTQR. This is consistent with previous observations that ERAP1 can generate but also destroy some epitopes depending on their sequence (8, 11). In contrast to ERAP1, PLAP trimmed the 14mer peptide but was found to accumulate only a 13mer and the 7mer intermediate. No 9mer epitope was detectable in any of our tests indicating that in this case PLAP was unable to accumulate the 9mer and instead only destroyed the antigenic peptide by trimming it further (Figure 4).
We next evaluated the in vitro generation of the antigenic peptide ATFPDTLTY using three 13mer model precursor peptides that differ either in the preceding sequence or the C-terminal residue of the naturally occurring precursor sequence (Figure 5). In all three cases ERAP1 efficiently trimmed the peptide, accumulating as the primary intermediate the 9mer epitope. PLAP was also able to trim the peptide and accumulate some of the 9mer epitope but proceeded to trim the 9mer down to a 7mer product that was not further hydrolyzed. Interestingly, the nature of the preceding sequence or the C-terminal residue did not seem to affect the qualitative outcome of the reaction, although it did affect the kinetics of the reaction as judged from the different amounts of enzyme necessary to generate the epitope.
We finally evaluated the trimming of the precursor sequence KIRIQR on the N-terminus of the epitope GPGRAFVTI (Figure 6, panel A). Consistent with previous reports (17, 26), both PLAP and ERAP1 were able to efficiently generate the mature epitope. However, because in this epitope sequence the 2nd residue is a proline, this sequence is expected to be a very poor residue for many aminopeptidases of this family and as a result accumulation of the 9mer peptide may not be restricted to these specific aminopeptidases. To test whether ERAP1 and PLAP can still generate the correct 9mer even in the absence of the Pro residue, we tested trimming of the same precursor sequence attached onto the model epitope GSGRAFVTI in which the proline residue is substituted by a serine (Figure 6, panel B). ERAP1 was still able to accumulate large amounts of the 9mer product but then proceeded to trim it further to smaller products. In contrast, PLAP was able to produce the 9mer epitope almost exclusively and no smaller products were detectable (Figure 6, Panel B)
To test whether the pH of the solution can affect the ability of the enzyme to accumulate specific trimming intermediates, we performed digestions of the peptides LSIINFEKL and LRVYEKMALY (used in Figures 2 and and3)3) by PLAP in pH 6.0. We found that PLAP trimmed both peptides and accumulated intermediates in an indistinguishable manner to trimming in pH 8.0, indicating that the pH value of the solution does not affect the ability of the enzyme to generate these antigenic epitopes (data not shown).
A model aminopeptidase that shows little to no specificity for peptidic substrates should process all trimming intermediates sequentially and accumulate them to a minimal degree. Deviations from this model trimming behavior can be attributed to substrate preferences and particularly the enzyme’s difficulty to process a specific intermediate, leading to that intermediate’s accumulation. Intermediate accumulation can be indicative of the suitability of a particular aminopeptidase to perform a specific biological function by trimming specific substrates or preferentially accumulating specific products. The ability of ERAP1 to: first, trim efficiently a large variety of long peptides and second, accumulate intermediates that are 8 or 9 amino acids long has been considered to be a hallmark of the enzyme’s role in antigen processing. These properties of ERAP1 have been hypothesized to be derived from an extended peptide binding site adjacent to the catalytic site of the enzyme (11, 15). This site can accommodate long peptides, forming a large network of atomic interactions that affect the trimming rate of the substrate. The high level of sequence homology between ERAP1 and PLAP suggests the possibility that PLAP retains at least some degree of the substrate recognition properties of ERAP1. Consistent with this hypothesis, we demonstrate here that PLAP, similarly to ERAP1, can trim many different peptide sequences efficiently and accumulate intermediates that correspond to mature antigenic epitopes in vitro. In that context, PLAP has the minimum required enzymatic properties to process antigenic peptide precursors inside the cell.
Recently, the mouse homolog of PLAP, IRAP was demonstrated to play a critical role in the a cross-presentation pathway (26). We demonstrate here that PLAP is active in a slightly wider range of pH compared to the sharp pH dependence of ERAP1. This is consistent with PLAP operating in endosomal or phagosomal compartments that have a wider range of possible pH values compared to the tightly regulated ER pH. Indeed, phagosomes can initially alkalinize, reaching values of pH 8 that promotes cross-presentation, but later acidify (28, 29, 44–46). This acidification is significantly slowed down in dendritic cells compared to other cell types, with the pH of phagosomes remaining in the range of 7 to 7.5 for several hours (29, 47–50). Since we find that this is the optimal pH range for PLAP this suggests that PLAP is an appropriate aminopeptidase to perform antigen processing in cross-presentation vesicles.
According to our observations, both ERAP1 and PLAP appear suitable to generate correct length antigenic epitopes from longer precursors but to also destroy some of them by further trimming to smaller products that would not be expected to be able to bind onto MHC class I molecules. Although it is difficult to determine whether the enzyme and peptide concentrations present in the cell during antigen processing are sufficient to destroy epitopes, this in vitro behavior shows that such destruction is a possibility. Indeed, ERAP1 has been demonstrated to be able to destroy many epitopes in vivo (8, 10). Moreover, this complex behavior appears to be both epitope and enzyme dependent suggesting that sequence determinants in the precursor examined can affect how well the epitope will be produced. As seen in Table I, ERAP1 and PLAP followed different patterns in their ability to generate intermediates that correspond or do not correspond to antigenic epitopes. This was more pronounced in the case of accumulation of smaller products from further trimming of the antigenic epitope, in which PLAP was slightly more effective than ERAP1. Differences in these trimming patterns are likely the result of differences in the specificity of these related enzymes. We have previously demonstrated that ERAP1 trims peptides depending on their internal sequence and it is reasonable to assume that PLAP shares this property but displays different preferences for amino acid side chains in the peptide sequence.
During our analysis, even when the mature antigenic peptide was detected to be produced, the degree of its accumulation was found to vary significantly. To which degree the antigenic peptide accumulation during a trimming reaction and/or its susceptibility to further trimming are important parameters for antigen presentation in vivo, is currently unknown. On the one hand it is possible that the degree to which the mature epitope accumulates determines how much will be bound to and then presented by MHC class I molecules. On the other hand, it is possible that even very little epitope accumulation may be sufficient for antigen presentation in vivo if they are rapidly bound by MHC class I molecules. This is because MHC-bound peptides are protected from aminopeptidase degradation since their N-terminus is tightly bound by the MHC and is therefore not accessible to peptidases (51, 52). In this context, even minimal epitope accumulation in vitro may signify efficient in vivo generation of certain epitopes by either ERAP1 or PLAP because MHC class I molecules may bind and protect a fraction of the epitopes as they are generated and before they can be further hydrolyzed. Recently, trimming rates by ERAP1 has been found to correlate with antigen presentation efficiency on cultured cells, suggesting that the fundamental trimming properties of this aminopeptidase are directly relevant to its biological function and it is therefore reasonable to hypothesize that the same applies for PLAP (53).
The combination of the generating and destructive properties of ERAP1 in vivo has led to the description of the enzyme as an antigenic peptide editor, an enzyme that affects the fate of potential antigenic peptides. Consistent with this idea, in vivo experiments in mouse models have demonstrated that the presence of ERAP1 can alter immunodominance, leading to complex and hard to predict changes in the immune response(9, 10, 13). In this study we demonstrate that PLAP shares the basic properties of ERAP1 in both generating and destroying antigenic epitopes. Recent work has implicated PLAP in cross-presentation in certain cells (26, 27). The trimming patterns that affect epitope generation however appear to be distinct between ERAP1 and PLAP suggesting that if PLAP can also affect antigen presentation and immunodominance in vivo, it may do so differently compared to ERAP1. Why it would be necessary to have distinct trimming behaviors of specialized antigen processing aminopeptidases in separate antigenic processing pathways is not clear at the moment. It is possible that the two aminopeptidases encounter different pools of antigenic peptides and have adapted accordingly. Alternatively, it is conceivable that the antigenic peptide selection requirements in separate processing pathways in the cell are different and require corresponding trimming behaviors. Finally, the presence of two trimming patterns within the cell may be advantageous in terms of increasing the variety of epitopes that can be processed, therefore contributing to the efficiency of the processing machinery towards the very large number of antigenic precursor sequences that are encountered. On the other hand if CD8 T cells are generated to unique cross presented peptides, then the effector cells that would be generated would be unable to recognize target cells that are presenting peptides generated through direct presentation. While such responses would not be effective in purging abnormal parenchymal (non-cross presenting cells), they might still be useful in recognizing dendritic cells or macrophages harboring pathogens in their phagosomes. A more comprehensive investigation of the trimming properties of all three members of the oxytocinase sub-family of aminopeptidases (ERAP1, ERAP2, PLAP) may be necessary to help us understand why the cellular antigen generation machinery utilizes two distinct trimming activities to achieve what seems to be the same biological function.
In summary, our study suggests that the key enzymatic properties that make ERAP1 especially fit for intracellular antigen processing are also shared to a large degree by the homologous PLAP. However, PLAP’s epitope-dependent trimming patterns do not mirror that of ERAP1’s suggesting that the two enzymes may impose different selective pressures on antigen generation.
1Funding for this work was provided to ES by the 6th Framework Programme of the European Union (Marie Curie International Reintegration Grant 017157 to E.S) and from the National Centre for Scientific Research “Demokritos”. Funding was also provided by the NIH to KLR.