HLA Class I Molecules Reflect an Altered Host Proteome Following Influenza Virus Infection

Cell Lines and Transfectants

HeLa cells (ATCC CCL-2) cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin were electroporated with a pcDNA 3.1- expression vector (Invitrogen) encoding secreted HLA-A*0201. The cytoplasmic and transmembrane domains of HLA-A*0201 were removed by PCR mutagenesis. The production of sHLA-A*0201 was measured by ELISA using anti-W6/32 (20) capture and β2-microglobulin primary antibodies (21). The peptide repertoire of HLA-A*0201 was examined as it is the most common HLA-A allele in North America (22).

Virus Production

Influenza viruses A/Puerto Rico/8/34 (PR8), A/Oklahoma/7485/01 (7485), and A/Oklahoma/309/06 (309) were propagated in Madin-Darby canine kidney (MDCK) cells. Roller bottles were seeded with MDCK cells at a density of 20,000 cells/cm², cultured for 2 days, washed with PBS, and inoculated with virus in 10 ml CaMg PBS at a MOI of 0.05. Infection medium (DMEM/F12K supplemented with 6% ITS, 1% penicillin/streptomycin, 1% non-essential amino acids, and 1% sodium pyruvate) was added to the cell culture after 2 h incubation at 37°C. Virus production was monitored daily by HA titration on human red blood cells. Cell supernatant containing virus was centrifuged at 3,000 rpm for 30 min to remove cell debris and stored in 200 ml aliquots at −80°C.

Cell Pharm Infection

7.5 × 10⁹ HeLa cells were pelleted, washed 3 times with CaMg PBS, and incubated with virus at a MOI of 200 for 2 h at 37°C. Influenza-infected HeLa cells were inoculated into a cell pharm hollow fiber bioreactor (Biovest International) and cultured in DMEM supplemented with 6% ITS. Cellular glucose consumption, pH, and sHLA production were monitored daily. The percentage of HeLa cells infected with influenza was measured by intracellular staining with an anti-nucleoprotein antibody (Meridian Life Science Inc.) and anti-serum targeting the influenza core (23) followed by flow cytometric analysis.

Peptide Isolation and Purification

Approximately 25 mg of sHLA-A*0201 peptide complex was affinity purified with W6/32 antibody from the supernatant of naïve and influenza infected HeLa cells. Peptide was eluted from the HLA class I binding groove with a 10% acetic acid boil and separated from class I HLA by passage through a 3 kDa membrane (Millipore). Naïve and infected peptide pools were separated by RP-HPLC into 40 peptide containing fractions. Prior to mass spectrometric analysis, 10% of the naïve and infected peptide pools underwent 14 cycles of N-terminal Edman degradation to confirm the HLA-A*0201 origin of the eluted peptide (data not shown) (24).

Mass Spectrometric Analysis

Peptides in naïve and influenza-infected RP-HPLC fractions were sprayed 3 times via nanospray into a Qstar Elite quadrupole time-of-flight (Q-TOF) mass spectrometer to create a MS ion map for each fraction. The MS ion maps for corresponding naïve and infected fractions were aligned at 20 amu increments to identify ions (putative peptides) unique to influenza infected cells. Peptides exhibiting a fold increase of at least 1.5-fold following infection were identified by summing the intensity of each ion in 3 naïve and infected MS ion maps and calculating the normalized fold increase in the intensity of the ion in the infected versus the naïve MS ion maps (24). A threshold of 1.5-fold or more was established after mass spectrometric analysis of the HLA class I peptide pools isolated from repeat naïve cell pharm runs revealed the extent of variability in peptide presentation between cell pharm runs (unpublished). This threshold limits the impact that stochastic or random changes in protein processing have upon the data presented herein. The identity of peptides in infected MS ion maps was determined by MS/MS fragmentation followed by amino acid sequencing de novo and/or with MASCOT (25). MS/MS fragmentation of peptides in pre, post, and corresponding RP-HPLC fractions of three separate uninfected cell pharm runs and affinity purifications confirmed the absence of unique peptides in the naïve class I HLA-A*0201 peptide pool.

Predicted Binding Affinity

The binding affinity of unique and increased peptides (8-, 9-, 10-, and 11-mers) to HLA-A*0201 was predicted with the SMM binding algorithm accessed on the Immune Epitope Database (IEDB) and Analysis Resource website (26).

Pathway Analysis

Functional analysis of host proteins encoding unique or increased HLA-A*0201 peptides following influenza infection was performed with Ingenuity Pathways Analysis software (Ingenuity Systems). Analysis was limited to direct protein interactions.

Influenza Strain Specificity Among Class I HLA-A*0201 Self Peptides

In this study we assessed the impact of influenza A virus infection upon class I HLA presentation of host encoded peptides. We initially gathered HLA/peptide complexes from naïve and influenza infected cells. A catalog of self peptides eluted from naïve cells was utilized for comparison to the class I repertoire of cells infected with multiple strains of influenza: PR8, a well characterized H1N1 influenza A virus laboratory strain, and two recent influenza A (H1N1 7485 and H3N2 309) circulating strains. Comparative mass spectrometry of HLA-A*0201 peptides gathered from naïve and infected cells identified 20 host-encoded peptides unique to infected MS ion maps (Fig. 1A–C) and 347 different self peptides with presentation increased 1.5 fold or greater following infection (Fig. 1D). Upregulated host peptides were increased by 1.5 to 33.09 fold following influenza infection with an average increase of 4-fold.

Many host peptides demonstrated increased presentation following infection, yet only a handful were increased by multiple strains of the influenza A virus (Fig. 2). With 347 upregulated host ligands characterized following influenza infection, 39 were increased by both strains of H1N1 influenza A (PR8 and 7485) and 6 were shared by recent H1N1 and H3N2 influenza A virus isolates 7485 and 309 (Fig. 2). Only five host-derived peptides (GVYDGEEHSV-MAGEB2, KLDQDLNEV-SMC3, TLYEHNNEL-AAAS, LLDVVHPA-CCT7, and SLFEGTWYL-HMGCS1) exhibited increased presentation following infection with all three influenza strains studied (Fig. 1D and 2, Table 2). These data show that each viral strain has a unique signature of altered self ligands, that influenza A H1N1 strains overlap by approximately 10% of their upregulated ligands (39/347), and that a neglible number of upregulated ligands (6/347) overlap between the H3N2 and H1N1 isolates circulating in the human population.

Although overlap in the presentation of unique and increased host ligands between virus strains was minimal, it is possible that the class I HLA of influenza infected cells sample different peptides of the same host protein. Therefore, we determined the overlap in source proteins revealed by HLA-A*0201 following influenza infection. The 20 unique and 347 increased host HLA-A*0201 peptides are encoded by 297 different proteins. Figure 3 confirms the minimal overlap between the three influenza strains in the source proteins sampled by class I HLA, reflecting the strain specificity of unique/increased self HLA-A*0201 peptides following PR8, 7485, and 309 influenza infection. Approximately 3% of the host proteins (9/297) encoding unique/increased ligands were revealed by class I HLA following infection with all three influenza A virus strains (Fig. 3). The two H1N1 influenza virus strains PR8 and 7485 share 15% (44/297) of host proteins, the recent H1N1 and H3N2 influenza isolates (7485 and 309) overlap in 2% (6/297) of the host proteins exposed by HLA-A*0201, and class I HLA sampled only one protein exclusive to PR8 and 309 influenza strains (Fig. 3). As influenza viruses diverge, so does the signature of class I HLA presented self ligands associated with each viral strain: Unique patterns of self-ligand presentation and host proteins sampled distinguish the various strains of influenza.

Physical Characteristics Distinguish Upregulated and Unique Ligands

Nineteen of 20 unique self peptides resulting from influenza infection are not nonamers and vary in length from 7 to 13 amino acids (Table 1). In contrast, 78% of host peptides with increased presentation in multiple influenza strains are nine amino acids in length (Table 1 and 2). Unique and upregulated ligands were further distinguished by their HLA-A*0201 binding affinity (Table 1 and 2). The average HLA-A*0201 peptide unique to infected cells has a lower (IC50 7570 nM) binding affinity compared to self peptides with increased presentation (IC50 971 nM) (Table 1 and 2). These data indicate that class I HLA presented self peptides unique to influenza infected cells vary in size and have a substantially lower HLA-A*0201 binding affinity when compared to ligands with increased presentation which are predominantly nonamers and exhibit a much higher predicted HLA-A*0201 binding affinity.

Different Influenza Strains are Related in Their Host Interaction

The biologic function of the host proteins most frequently sampled by HLA class I molecules following influenza infection is poised to reveal host cell functions disrupted during viral replication. The functions of source proteins encoding multiple unique and/or increased HLA-A*0201 self peptides were compared to identify common cellular functions altered by viral replication. Table 3 illustrates that infection with H1N1 influenza virus strains PR8 and 7485 triggered the class I presentation of multiple peptides encoded by proteins involved in cellular metabolism. In particular, proteins involved in aerobic and anaerobic glycolysis (i.e. GAPDH, LDHA, LDB, and PKM2) were frequently sampled by class I HLA. Numerous peptides derived from several proteins that function in protein translation elongation such as RPL10A and RPLP0 were presented by HLA-A*0201 following 7485 H1N1 and 309 H3N2 influenza infection respectively indicating a disturbance in protein synthesis (Table 3). In addition, PR8 infection displayed several unique/increased class I peptides encoded by proteins that process RNA (CPSF1, IMP3, and SNRNPG) and chaperone proteins (HSPA2 and CCT7) in the host cell nucleus (Table 3).

To determine associations among influenza mediated changes in the infected cell, host proteins encoding HLA-A*0201 unique/increased self peptides were mapped by Ingenuity Pathway Analysis software. Pathway analysis revealed that 45% (134/297) of the host proteins encoding unique/increased HLA-A*0201 ligands directly interact in the host cell (Fig. 4). Five prominent hubs of influenza/host protein interaction were revealed by these direct interactions: HLA-B, beta-actin (ACTB), heat shock protein 90 (HSP90AB1), cyclin-dependent kinase 2 (CDK2), and annexin A2 (ANXA2). All three influenza A virus isolates influenced related source proteins in these five hubs as detected by HLA-A*0201 peptide sampling (Fig. 4).

Almost half of the proteins in the network directly interact with HLA-B (31/134) and/or beta-actin (31/134), hinting at a major reorganization in the HLA-B class I presentation pathway and cytoskeleton during influenza virus infection (Fig. 4). Approximately 30% of network source proteins directly interact with HSP90AB1 (18/134), CDK2 (11/134), and/or ANXA2 (10/134) molecules which function in protein folding, cell cycle, and calcium-ion dependent phospholipid binding respectively, indicating additional host pathways likely modified by influenza replication (Fig. 4). Moreover, host changes that did not directly interact with one another in these pathways were nonetheless clustered in the same regions. For example, several ribosomal proteins that do not directly interact with one another in the host cell circle around ACTB (RPS3A, RPS8, RPS12, RPS18, RPS19, RPLP0 and RPL7A) and ANXA2 (RPS3A, RPS18, RPS19, RPL19, and RPL35). Collectively, these data reveal that despite minimal overlap in peptide presentation, the host proteins sampled by class I HLA reveal common cellular functions altered by influenza virus infection. Source proteins encoding host HLA class I peptides with unique or increased presentation following influenza infection form a highly integrated web of functional and physical interaction.

The Influenza A Virus Directly Interacts with Unique and Increased Self Peptide Source Proteins

Literature reveals that seven HLA-A*0201 host peptides with increased or unique presentation following influenza A virus infection are derived from five proteins that are directly involved in viral propagation (Table 4). The presentation of a nine amino acid peptide derived from actin (FAGDDAPRA) was upregulated following infection with 7485 and 309 influenza virus strains (Table 1). The binding of influenza nucleoprotein (NP) molecules to the actin cytoskeleton helps maintain cytoplasmic retention of vRNPs following nuclear export (27). Infection with the two H1N1 strains resulted in increased presentation of two nonamers (IINEPTAAA and LLDVTPLSL) derived from heat shock protein 70 (HSP70) (Table 1). During influenza infection, HSP70 facilitates nuclear export of vRNPs by binding influenza matrix 1 (M1) molecules attached to vRNPs (28, 29). Infection with the H1N1 strains also resulted in the increased presentation of two nonamers (ILDKKVEKV-PR8 and ALLSSGFSL-7485) derived from heat shock protein 90 (HSP90) (Table 4). During influenza infection, the inducible (HSP90AA1) and constitutively (HSP90AB1) expressed forms of HSP90 promote viral replication by facilitating the nuclear import and assembly of influenza polymerase proteins (30–32). PR8 infection revealed the unique presentation of an undecamer peptide derived from nucleolin (TLQEVFEKATF) (Table 2). During influenza infection the non-structural protein 1 (NS1) protein colocalizes with nucleolin in the nucleoli of the host cell, although the significance of this interaction is yet unknown (33). Influenza 309 infection identified a 12 amino acid peptide encoded by annexin A2 (SVIDYELIDQDA) unique to HLA-A*0201 of infected cells (Fig. 1, Table 2), and annexin A2 is positioned to impact viral maturation by activating plasmin which cleaves the HA0 molecule into the HA1 and HA2 subunits (34–36). Thus, within the host cell the influenza virus hijacks actin, HSP70, HSP90, and annexin A2 to carry various stages of viral replication. Collectively, these data indicate that five proteins encoding HLA-A*0201 ligands unique (nucleolin and annexin A2) or increased (actin, HSP70, and HSP90) following influenza infection directly interact with the influenza A virus during various stages of viral replication.