The Viral Transcription Group Determines the HLA Class I Cellular Immune Response Against Human Respiratory Syncytial Virus*

The cytotoxic T-lymphocyte-mediated killing of virus-infected cells requires previous recognition of short viral antigenic peptides bound to human leukocyte antigen class I molecules that are exposed on the surface of infected cells. The cytotoxic T-lymphocyte response is critical for the clearance of human respiratory syncytial virus infection. In this study, naturally processed viral human leukocyte antigen class I ligands were identified with mass spectrometry analysis of complex human leukocyte antigen-bound peptide pools isolated from large amounts of human respiratory syncytial virus-infected cells. Acute antiviral T-cell response characterization showed that viral transcription determines both the immunoprevalence and immunodominance of the human leukocyte antigen class I response to human respiratory syncytial virus. These findings have clear implications for antiviral vaccine design.

adults, HRSV poses a severe health risk for immunocompromised (5,6) or elderly individuals (7,8). Despite the immune mechanisms involved in HRSV disease and protection are not completely understood, it is known that the cytotoxic T lymphocytes (CTLs) are required to clear virus-infected cells (9). Like for all paramyxoviruses, the single-stranded, negativesense RNA genome of this enveloped virus is sequentially transcribed by viral RNA polymerase (vRNAP) into separate mRNAs, which are involved in transcription initiation at a single 3Ј promoter. Additionally, this process involves a sequential start-stop-restart mechanism (1). The vRNAP occasionally fails to reinitiate the downstream mRNA at each stoprestart junction, which leads to the loss of transcription of further downstream genes (1,10); hence, there is an mRNA synthesis gradient that is inversely proportional to the distance of the gene from the 3Ј end of the genome. Thus, the promoter-distal genes are expressed less efficiently (10,11).
Translated viral mRNA yields proteins that can be further degraded by proteasomes (12), and in some cases, by other cytosolic proteases (13), which generate an extremely diverse pool of peptides both in sequence and length that can be translocated to the endoplasmic reticulum (ER) lumen by transporters associated with antigen processing. Among them, only a small fraction with a correct size or NH 2 -terminally extended precursors can be used for antigen presentation by direct ligand binding to human leukocyte antigen (HLA) class I molecules or by precursor editing and customization by ER-resident aminopeptidase activity (14), respectively, to yield the final viral ligand. Finally the stable trimolecular peptide-HLA-␤ 2 -microglobulin complexes are transported to the infected cell membrane where they can be recognized by antiviral T lymphocytes, an event that will end in the killing of the infected cell (15).
The general antigen processing and presentation rules that are applicable to individual viruses or families of viruses are largely unknown, despite the importance of the different elements of the HLA class I antigen processing and presentation pathway (e.g. ubiquitination, proteasome, cytosolic, and ERresident proteases, TAP, chaperones, and HLA peptide binding). Additionally, the TCR repertoire or regulatory T cells, among others, have been described in multiple studies that contribute to the antigen processing and presentation of individual epitopes (e.g. influenza A and HIV, which are summarized in (16)). We are interested in the identification of viral ligands that are presented by several frequent HLA class I molecules in HRSV-infected cells to analyze how the immune system selects natural HLA class I ligands and epitopes. Immunoproteomics analysis of peptide pools from HRSVinfected cells has led to the identification of several new naturally processed ligands from different viral proteins that together with ligands identified in previous studies (17,18), define both the nature and hierarchy of the T-cell class I specific response against HRSV. of 0.1% TFA, and the peptides were eluted with 80% acetonitrile in 0.1% TFA. The peptide samples were then concentrated to ϳ20 l using vacuum centrifugation (17,28).
The HLA class I peptides that were immunoprecipitated with each HLA-specific mAb were analyzed by LC-MS/MS using an Orbitrap XL mass spectrometer (Thermo Scientific, San Jose, CA) that was fitted with a capillary HPLC column (Eksigent, Dublin, CA) (17,28). The peptides were resolved on homemade Reprosil C18-Aqua capillary columns (75 micron ID) (29) with a 7-40% acetonitrile gradient for 2 h in the presence of 0.1% formic acid. The 7 most intense masses that exhibited single-, double-, and triple-charge states were selected for fragmentation from each full mass spectrum with CID.
Database Searches-Sequest 3.31 (Thermo-Fisher) (30) was used for peak-list generation from the LC-MS/MS data. The peaks were identified using Proteome Discoverer 1.0 SP1 (Thermo-Fisher) and Bioworks Browser 3.3.1 SP1 (Thermo-Fisher) software programs (30) as well as the human and virus parts of the NCBI database (Jan 2013), which included 656,486 proteins. The search was not limited by enzymatic specificity; the peptide tolerance was set to 0.005 Da, and the fragment ion tolerance was set to 0.5 Da (17,31). This search was not limited by any methodological bias (e.g. individual protein selection or HLA consensus scoring algorithm use). The identified peptides were selected if the following criteria were met: Sequest Xcorr Ͼ1.4 for singly, Ͼ2.2 for doubly, and Ͼ2.9 for triply charged peptides; P(pep) less than 1 ϫ 10 Ϫ3 ; and a mass accuracy of 0.005 Da (17,31). When the MS/MS spectra fit more than one peptide, only the highest scoring peptide was analyzed. No peptides were found in a search of a reversed database. The purpose of the filtering criteria was to identify candidate HRSV peptide from the MS/MS scans for further manual inspection to determine whether the MS/MS fragment ion fingerprints matched the identified peptide sequences. Additionally, the corresponding synthetic peptide was made and its MS/MS spectrum was used to confirm the assigned sequence.
HLA/Peptide Stability Assays-The following synthetic peptides were used as controls in the HLA/complex stability assays: VACV A10L 688 -696 (ILDRIITNA, HLA-A*02-restricted) (28), A34R 82-90 (LPRP-DTRHL, HLA-B*07-restricted) (32), and CMV pp65 7-15 (RCPEMISVL, HLA-Cw1-restricted) (33). The different RMA-S transfectant cells were incubated at 26°C for 16 h. This allowed for empty HLA class I molecule expression (without antigenic peptide) at the cell membrane that was stable at 26°C but not at 37°C. The cells were washed and incubated for 2 h at 26°C with various peptide concentrations in medium without FBS. The cells were maintained at 37°C for an additional 2 h and then collected for flow cytometry. This method allowed for the empty HLA class I molecules to become internalized, and thus we were able to discriminate between bound or unbound peptides. HLA expression levels were measured using monoclonal PA2.1 (anti-HLA-A*02) and ME1 (anti-HLA-B*07) Abs, as previously described (34). Data were acquired on a FACSCanto flow cytometer (BD Biosciences, San Jose, CA) and analyzed using BD FACSDiva software, version 6 (BD Bioscience). The cells that were incubated without peptides exhibited peak fluorescence intensities close to the background staining that were observed with the secondary Ab alone. The fluorescence index was calculated for each time point as the ratio of the mean peak channel fluorescence of the sample to that of the control incubated without peptide. Peptide binding was also expressed as EC 50 , which is the molar concentration of the peptide at 50% of the maximum fluorescence obtained in a concentration range of 0.01-200 M.
IFN-␥-secreting CD8 ϩ Cell Detection by ELISPOT-ELISPOT assays were performed as previously described (35) to detect antigenspecific CD8 ϩ T-cell activation. Briefly, purified rat anti-mouse IFN-␥ antibody (clone R4 -6A2, BD Pharmingen, San Diego, CA) was coated on 96-well MultiScreen HTS HA plates (Millipore, Billerica, MA). The plates were incubated overnight at room temperature and were blocked with medium that was supplemented with 10% fetal bovine serum for 2 h at 37°C. Duplicate cultures of erythrocyte-depleted spleen cells were prepared from HLA class I-transgenic mice at 7 days (acute response) post intranasal infection with 1 ϫ 10 6 pfu of Long strain HRSV at different dilutions with 10 Ϫ5 M peptide. The plates were incubated overnight at 37°C in a 5% CO 2 atmosphere and were then washed with PBS-T (PBS 0.05% Tween-20). The plate wells were incubated for 2 h at room temperature with biotinylated anti-mouse IFN-␥ mAb clone XMG1.2 (BD Pharmingen, San Diego, CA), washed with PBS-T, and incubated for 1 h at room temperature with horseradish peroxidase-labeled streptavidin. The plates were additionally washed before adding 3,3Ј-diaminobenzidine substrate (Sigma, St. Louis, MO) in 50 mM Tris buffer pH 7.4 that contained 0.015% hydrogen peroxide. To enumerate the IFN␥ responses, spots were counted and wells were photographed using a Leica EZ4 HD stereo microscope and LAS EZ software (Leica Microsystems, Germany). Additionally, the percentage of CD8 ϩ cells was determined after staining spleen cells with FITC-conjugated anti-mouse CD8 antibody (clone KT15, Proimmune, Oxford, United Kingdom). Events were acquired on a FACSCanto flow cytometer (BD Biosciences, San Jose, CA) and analyzed using BD FACSDiva software, version 6 (BD Bioscience).
Statistical Analysis-To analyze statistical significance of HLA/peptide stability and ELISPOT assays unpaired Student's t-tests were used. In addition, Chi-square test was used to analyze the distribution of HLA class I ligands. p values Ͻ 0.05 were considered to be statistically significant.

RESULTS AND DISCUSSION
Physiological Processing Generated Three Different Viral HLA-A*02:01 Ligands in Human HRSV-infected Cells-The HLA-A*02:01-bound peptide pool was isolated from large numbers of either healthy or HRSV-infected human cells. These peptide mixtures were subsequently separated by reversed-phase HPLC and analyzed using mass spectrometry. Using bioinformatic tools, three fragmentation spectra present in the HRSV-infected HLA-bound peptide pool, but absent in the control uninfected pool (data not shown), were resolved with high confidence parameters as HRSV protein peptides (Table I). Additionally, a human proteome database search failed to identify any of these spectra as human protein fragments, suggesting the viral origin of these peptides. The first ion peak, with an m/z of 511.8, was assigned to the viral amino acid sequence KLIHLTNAL, which spans residues 33-41 of the HRSV NS1 protein (Fig. 1, upper panel) that was previously identified (36). Moreover, two different ion peaks at m/z 525.3 and 681.9 were assigned to other viral peptides. These ion peaks corresponded to the TQFPHFSSV (Fig. 1, medium panel) and RLLEITREFSV (Fig. 1, lower panel) peptides, which span residues 315-323 of the N protein and 229 -239 of the F protein, respectively. Virtually all fragments with a relative abundance higher than 10% of the maximum signal of the three MS/MS spectra were assigned as daughter ions of the putative peptidic sequences (Fig. 1). This theoretical assignment was confirmed by MS/MS spectrum identification of the corresponding synthetic peptide (Fig. 1). Therefore, these results indicate that a total of three HLA-A*02 ligands were endogenously processed and presented in the HRSV-infected cells.
Two Viral HLA-B*07:02 Ligands Were Endogenously Processed in the Human HRSV-Infected Cells-Similarly to HLA-A*02:01, the HLA-B*07:02-bound peptide pool was isolated from either healthy or HRSV-infected cells. Two fragmentation spectra present in the HLA-B*07:02-bound peptide pool, but absent in the control uninfected pool (data not shown), were also resolved as HRSV protein peptides (Table I). The human proteome database search failed to identify these spectra as human protein fragments, suggesting the viral origin of these HLA-B*07-bound peptides. The first ion peak, with an m/z of 471.8, was assigned to the viral amino acid sequence NPKASLLSL, which spans residues 306 -314 of the HRSV N protein (Fig. 2, upper panel) as previously identified (37). This indicated that antigen processing of this protein can generate several viral ligands bound to two different HLA class I molecules. The second ion peak, at m/z 665.9, was assigned to the RPLSLETTITSL peptide sequence (Fig. 2, lower panel), which spans residues 19 -30 of the NS2 protein. Fig. 3 shows that all significant fragments of these two MS/MS spectra were assigned as daughter ions of the putative peptide sequences. As for the HLA-A*02:01 ligands, these assignments were confirmed by MS/MS spectrum identification of the cor- responding synthetic peptide (Fig. 2). Collectively, these results indicate that a similar number of ligands were endogenously processed and presented by HLA-A*02:01 or -B*07:02 class I molecules in the same HRSV-infected cells.
HRSV Ligand Binding Affinity for the A*02:01 Molecule-The classical anchor motifs for HLA-A*02:01 binding, Leu or Met at position 2 (P2), and the aliphatic C-terminal residues (SYFPEITHI database: http://www.syfpeithi.de (38)), were present in two of the three detected HRSV viral ligands (Table  I). In contrast, the N 315-323 ligand presented a Gln at P2; however, it was co-immunoprecipitated with an HLA-A*02specific mAb and thus could be an unusual HLA-A*02-re- [MH] +2 b 9 * + y 8     stricted ligand. To confirm that HLA-A*02:01 was the HLA class I molecule that presented these three ligands, HLA/ peptide complex stability assays were performed using TAPdeficient RMA-S cells that were transfected with the HLA-A*02:01 molecule (Fig. 3A). The three viral ligands were bound to the HLA-A*02:01 class I molecules with EC 50 values in the range commonly found among other natural high-affinity ligands (Fig. 3B). These data confirm that all ligands detected in HRSV-infected cells were presented in association with the HLA-A*02:01 molecule.
The HRSV Virus Ligand Binding Affinity for the B*07:02 Molecule-Both peptides identified as bound to HLA-B*07:02 have known anchor motifs for binding to this HLA class I molecule, which are Pro at P2 and Leu at the C-terminal residues (SYFPEITHI database (38)) ( Table I). To confirm that HLA-B*07:02 is the HLA class I molecule that presents these ligands, HLA/peptide complex stability assays were performed using TAP-deficient RMA-S cells transfected with the HLA-B*07:02 molecule (Fig. 3C). Both viral ligands were bound to HLA-B*07:02 class I molecules with EC 50 values similar to those of other natural high-affinity ligands (Fig. 3D). These data confirm that the ligands detected in the HRSVinfected cells were presented in association with the B*07:02 molecule. In summary, HLA-A*02:01 and B*07:02 class I molecules can both bind high-affinity ligands derived from different HRSV proteins in infected cells.
HRSV-Infected Cells Endogenously Presented a Viral Ligand by HLA Class I Molecules that was Different from HLA-A*02:01 and -B*07:02.-We sequentially immunoprecipitated HLA class I molecules ( supplemental Fig. S1) to investigate the possibility of new viral HLA class I ligands that were presented by other HLA class I molecules expressed in the same HRSV-infected cells. One fragmentation spectrum present in the peptide HRSV-infected W6/32-bound pool, but absent in its control uninfected pool, was also resolved as a HRSV proteome peptide (Table I). Furthermore, the human proteome database searches also failed to identify this spectrum as a human protein fragment, suggesting the viral origin of this HLA-bound peptide. The ion peak, with an m/z of 514.3, was assigned to the viral amino acid sequence FISSGLYKL, which spans residues 25-33 of the HRSV G protein. Fig. 4 shows the experimentally obtained MS/MS spectra and the respective assignments. The putative peptide sequence was confirmed by MS/MS spectrum identification of the corresponding synthetic peptide (Fig. 4). The haplotype of the cell line used in the immunoprecipitation was HLA-A*02:01, -B*07:02, and -C*07:02, and thus the G 25-33 peptide was most likely a HLA-C*07:02 ligand.
The Immunoprevalence of the HLA Class I Response was Limited by the HRSV Viral Transcription Group-Using the same experimental approach, we previously identified nine HLA-B*27:05 ligands (17) and one HLA-C*04:01 (18) ligand that along with the six different HLA class I ligands identified in the current report, raised the total number of HRSV HLA class I ligands to 16 and belonged to nine of the 11 viral proteins that are encoded by the HRSV genome in human virus-infected cells (Table II and  tified; thereby, the presence of Pro or Arg at P2 is almost mandatory for the HLA-B*07:02 or -B*27:05 ligands, respectively (SYFPEITHI Database (38)). Thus, the viral proteins with a high content for both of these amino acids are candidates for the HLA-B*07:02 and -B*27:05-restricted ligand source (Table III). However, no correlation between the number of the HLA ligands detected in the immunoproteomics analysis and both Pro and Arg content or viral protein size was found (Table  III). Therefore, neither the content in residues used by anchor motif amino acids nor the HRSV protein size were relevant in the HLA-B*07:02 and -B*27:05 class I viral ligand selection.
The HRSV genome contains 10 genes in the following order: 3Ј NS1-NS2-N-P-M-SH-G-F-M2/M2-2-L (Fig. 5). These genes are transcribed sequentially as a mRNA synthesis gradient that is inversely proportional to the distance of the gene from the 3Ј end of the genome and thus the promoter-proximal genes are expressed more efficiently (11). As for the other viruses that belong to the Mononegavirales order, the HRSV genome has been divided into three different mRNA expression level groups (10): 3Ј core protein genes, intermediate genes, and 5Ј large polymerase gene, which involve 26% (NS1, NS2, N, P, and M proteins), 25% (SH, G, F, M2, and M2-2 proteins) and 49% (L protein) of the viral proteome, respectively (Fig. 5). Thus, the analysis of the 16 natural HRSV ligands identified by mass spectrometry, which are bound to five different HLA class I molecules that cover ϳ70% of the human population, shows that most (75%) of the viral ligands detected were included in proteins encoded by the 3Ј group, whereas only 3 (19%) and 1 (6%) of them were integrated in proteins encoded by intermediate and 5Ј groups, respectively   (18). c HLA peptide binding is expressed as EC 50 (the molar concentration of the peptide at 50% of the maximum fluorescence obtained in a concentration range of 0.01-200 M), and its standard deviation. (Fig. 5). This 12:3:1 distribution of HLA class I ligands found in the immunoproteomics analysis is statistically different to an expected random distribution (4:4:8) along the viral proteome (p value ϭ 0.0083). Therefore, the proteins to which HLA class I antigen processing and presentation are addressed to, that is, the immunoprevalence of the HLA class I response, are limited by the HRSV transcription groups.
The Recognition of Eleven HLA-A*02:01, -B*07:02, and -B*27:05 Ligands by Specific T Cells in HRSV-infected HLA Transgenic Mice-To study in vivo the physiological relevance of the identified HLA class I viral ligands, which were identified in the current and the previous (17) studies, different HLA class I-A*02:01, -B*07:02, or -B*27:05 transgenic mice were infected with HRSV. Later, a physiological measurement of the functional ex vivo activity of T cells against the different HLA class I viral ligands identified using mass spectrometry was carried out. Spleen cells that specifically recognized cells that were pulsed with the three HLA-A*02:01 (Fig. 6A), the two -B*07:02 (Fig. 6B), and six of the nine -B*27:05 (Fig. 6C) peptide corresponding viral ligands indicated that the HLA  (17,18) together with the number of HLA class I epitopes identified in this report are shown in a , the transcription gradient, which was measured as the mRNA molar ratio percentage (11,53,54) is shown in b , the proteome percentage is shown in c , the HLA class I ligand percentage from the total is shown in d , the CTL epitope percentage from the total is shown in e , and the total IFN-␥ ϩ immune response percentage detected in the HLA class I-transgenic mice is shown in f .  ligands were either HLA-A*02:01, -B*07:02, or -B*27:05-restricted epitopes, respectively, and that they were simultaneously recognized as part of the acute response to HRSV. The N 315-323 ligand was the immunodominant epitope in the HLA-A*02:01 response (Fig. 6A), whereas either the two HLA-B*07:02 or six -B*27:05 epitopes that were detected displayed IFN-␥ ϩ responses of the same order ( Fig. 6B and 6C).
A lack of specific elements, such as ERAP2 (39), or differences between the human and murine antigen processing machineries (e.g. tapasin or the proteasomes) (40) could explain the absence of the T-cell recognition of the three HRSV ligands in the HLA-B*27 transgenic mouse model.

The Viral Transcription Groups also Determined the Immunodominance of the T-cell Class I-specific
Response-In each HLA transgenic model, the individual epitopes, which included the 3Ј group ligands, showed higher specific IFN-␥ ϩsecreting responses than the corresponding intermediate and 5Ј group's ligands (Fig. 6). Quantification of the overall T-cell responses specific for the 11 epitopes presented by the three different HLA class I molecules showed that most (98%, 100%, and 88% for HLA-A*02:01, -B*07:02, and -B*27:05, respectively, and 91% for the overall T-cell response) of the specific analyzed IFN-␥ ϩ responses were restricted by ligands from proteins that were encoded by the 3Ј group. Additionally, this represented only about a quarter of the viral proteome, and a minor 6 and 3% of the CD8 ϩ responses that were against epitopes encoded by proteins from the intermediate and 5Ј groups, respectively (Fig. 5). Therefore, the T-cell class I specific response hierarchy against HRSV was dependent on the viral transcription group. In contrast, the T-cell responses to the different HRSV ligands were not related to the HLA binding affinities (Fig. 7A) nor is there relationship between the HLA binding affinity of the different HRSV ligands and the viral transcription (Fig. 7B).
These results are in agreement with several observations. For instance, a ligand from the C protein included in the 3Ј transcription group of the paramyxo-measles virus was the HLA class I immunodominant epitope, and the other three ligands from the F and H proteins, which were included in the intermediate group or M protein of the 3Ј group, were subdominant (41). Interestingly, when using bioinformatic tools to analyze the distribution of the epitopes that are described in the immune databases, a different study showed that in HCV, a virus whose entire proteome is made from a single polyprotein that is translated from a single ORF, the HLA class I epitopes are not distributed along the viral proteome; however, they are concentrated in the 3Ј-terminal core protein, which is encoded by a single ORF (42). Altogether, both of these studies, along with our current report, indicate that with small RNA viruses, the transcriptional regulation or the translational control are the major characteristics that limit and determine both the nature and the hierarchy of the T-cell class I specific response: the immunoprevalence and immunodominance. Thus, further studies using immunoproteomics that identify the natural epitopes from infected cells in other different virus families are needed to determine the extension of this immune mechanism. Additionally, in two bioinformatic analyses, the cellular proteins encoded by highly abundant mRNA were found to be the much more likely sources of endogenous HLA class I ligands, but the poorly transcribed mRNA also generated a significant fraction of these ligands (43,44). In contrast, analyses of gene expression and their autologous HLA ligand densities from approximately threehundred proteins showed no correlation between mRNA expression and the abundance of renal carcinoma-associated ligands (45). Although multiple factors determine both the nature and the immunodominance of the T-cell responses of individual epitopes, some studies have shown the importance of some viral elements such as temporal virus protein expression. Thus, specific T-cell recognition of early but not late viral antigens was found both for herpesvirus (46) and some vaccinia virus strains (47,48). However, although a tight correlation between onset of protein expression and vaccinia virus epitope display has been found, no connection between immunodominance and epitope abundance was described for this poxvirus (49).
Interestingly, the prevalence of HLA class I ligands (Fig. 7C) encoded by the 3Ј transcription group mRNAs correlated with the respective size of each viral protein, with the immune response being mainly focused on the N protein. Therefore, the immunoprevalence in the 3Ј transcription group was related to the size of the encoded protein. Additionally, the IFN-␥ ϩ responses were also mainly focused against the N protein (Fig. 7C). These data from HLA-transgenic mice with H-2 b backgrounds are in agreement with a previous study that used peripheral blood mononuclear cells from normal adults stimulated with HRSV in vitro, which identified that the N protein was the most strongly recognized protein by specific T cells from different donors (50). In contrast, the HRSV response in BALB/c mice, which was analyzed using an overlapping peptide library spanning the HRSV proteome, was targeted almost exclusively against CD8 ϩ T-cell epitopes restricted by H-2K d from F and M2-1 proteins included in the intermediate group (51). The use of different background strains could explain the dissimilar T-cell responses identified in mice, and these data suggest that humanized H-2 b strains resemble more the human HRSV-specific HLA class I responses than the BALB/c mouse model.
In summary, the antigen processing and presentation of HLA class I epitopes from HRSV was modulated at two levels. First, by a transcription gradient, and second, by the protein size, which resulted from highly transcribed mRNAs. A confirmation of our data could be achieved with the immunologic study of a recombinant HRSV in which the different transcription groups of wild type virus were exchanged (e.g. 5Ј-IM-3Ј or IM-5Ј-3Ј). However, a previous study proved that although gene rearrangement increases mRNA levels for 3Ј proximal genes, the replication efficiency of the recombinant virus with only two or three individual rearranged genes was also decreased in a 10-or 100-fold range, respectively (52), rendering this approach quite unfeasible.
Finally, our data have evident implications for the analysis of the CTL response as well as for vaccine development, not only for HRSV and other Paramyxoviridae family members, but also for the other Mononegavirales order viruses that share the same genomic structure, although future and extensive studies with different viruses of this order are needed. For example first, multiple HRSV studies (ϳ60%), which targeted individual viral genes, have identified MHC class I epitopes and CD8 ϩ T-cell responses both in humans and in mice against individual viral proteins encoded by poorly transcribed mRNAs (Immunoepitope Database) that would be less relevant in the natural HLA class I response. Thus, a bioinformatic analysis of the distribution of known HRSV epitopes from databases, such as the one reported for HCV (42), could mask the real contribution of the different viral proteins to the T-cell immune response. Second, HLA class I immunologic efforts must be prioritized to study the most significant larger proteins encoded by highly transcribed mRNAs not only for the Paramyxoviridae family but also with FIG. 7. Comparison between HLA binding affinities and T-cell responses for the HLA class I ligands identified by mass spectrometry. A, the HLA binding affinity versus immunogenicity for the synthetic HRSV ligands was represented as Dot plot that represents the HLA binding affinity, which is expressed as EC 50 from Table 2, versus the INF␥ ϩ T-cell response from Fig. 6. B, The HLA binding affinities, which are expressed as EC 50 from Table 2, of the different synthetic HRSV ligands that are grouped by transcriptional groups defined as indicated in Fig. 5 are shown. C, Dot plot represents the number of HLA class I ligands versus the size of the five HRSV proteins that were encoded by the 3Ј transcription group. other virus of the Mononegavirales order, such as Rhabdovirus and Filovirus. This approach would be of great relevance for virulent viral diseases, such as the diseases caused by Ebola or Marburg viruses, for which biosafety level IV is required to work with infected cells, but where the immunoproteomics study of the 3Ј core protein genes group in a nondangerous recombinant virus for example, would yield the most relevant immunological information.