T cell mediated immunity against influenza H5N1 nucleoprotein, matrix and hemagglutinin derived epitopes in H5N1 survivors and non-H5N1 subjects

Background Protection against the influenza virus by a specific antibody is relatively strain specific; meanwhile broader immunity may be conferred by cell-mediated immune response to the conserved epitopes across influenza virus subtypes. A universal broad-spectrum influenza vaccine which confronts not only seasonal influenza virus, but also avian influenza H5N1 virus is promising. Methods This study determined the specific and cross-reactive T cell responses against the highly pathogenic avian influenza A (H5N1) virus in four survivors and 33 non-H5N1 subjects including 10 H3N2 patients and 23 healthy individuals. Ex vivo IFN-γ ELISpot assay using overlapping peptides spanning the entire nucleoprotein (NP), matrix (M) and hemagglutinin (HA) derived from A/Thailand/1(KAN-1)/2004 (H5N1) virus was employed in adjunct with flow cytometry for determining T cell functions. Microneutralization (microNT) assay was performed to determine the status of previous H5N1 virus infection. Results IFN-γ ELISpot assay demonstrated that survivors nos. 1 and 2 had markedly higher T cell responses against H5N1 NP, M and HA epitopes than survivors nos. 3 and 4; and the magnitude of T cell responses against NP were higher than that of M and HA. Durability of the immunoreactivity persisted for as long as four years after disease onset. Upon stimulation by NP in IFN-γ ELISpot assay, 60% of H3N2 patients and 39% of healthy subjects exhibited a cross-reactive T cell response. The higher frequency and magnitude of responses in H3N2 patients may be due to blood collection at the convalescent phase of the patients. In H5N1 survivors, the effector peptide-specific T cells generated from bulk culture PBMCs by in vitro stimulation displayed a polyfunction by simultaneously producing IFN-γ and TNF-α, together with upregulation of CD107a in recognition of the target cells pulsed with peptide or infected with rVac-NP virus as investigated by flow cytometry. Conclusions This study provides an insight into the better understanding on the homosubtypic and heterosubtypic T cell-mediated immune responses in H5N1 survivors and non-H5N1 subjects. NP is an immunodominant target of cross-recognition owing to its high conservancy. Therefore, the development of vaccine targeting the conserved NP may be a novel strategy for influenza vaccine design.


INTRODUCTION
Influenza A virus infection is common in a wide range of avian and mammalian hosts (Webster et al., 1992;Horimoto & Kawaoka, 2005;Taubenberger & Kash, 2010;Long et al., 2019). Some avian influenza A virus subtypes, i.e., H5N1, H5N6, H7N7, H7N9 and H9N2, are able to cross the species barrier and infect humans (Wong & Yuen, 2006 Yu et al., 2013;Puzelli et al., 2014;Pan et al., 2016;Peacock et al., 2019). Among these avian influenza subtypes, the H5N1 highly pathogenic avian influenza (HPAI) virus is the most virulent and the most fatal in humans and animals (Peiris, Jong & Guan, 2007;Korteweg & Gu, 2008). The first outbreak of the H5N1 HPAI virus in humans occurred in Hong Kong in 1997. Its re-emergence in 2003 affected several countries in various continents, in particular, the Asia-Pacific region (Chan, 2002;Horimoto & Kawaoka, 2005;Peiris, Jong & Guan, 2007). From 2003 to 10 July 2020, the World Health Organization (WHO) reported a cumulative number of 861 human cases with 455 deaths leading to the fatality rate of approximately 53% (World Health Organization, 2020). A person infected with the H5N1 HPAI virus mostly exhibits severe pneumonia, while asymptomatic infection or mild illness was scant (Hinjoy et al., 2008;Le et al., 2013). Almost of the cases contracted the H5N1 virus infection through contact with sick or dead poultry. Rare cases of person-to-person transmission have been documented (Ungchusak et al., 2005). Interestingly, nearly all villagers living in the H5N1 outbreak areas are naïve to this virus infection (Hinjoy et al., 2008;Dejpichai et al., 2009), which anticipated that their immune system is naïve to this virus subtype.
In this study, we assessed the T cell response to H5N1 NP, M, and HA epitopes using archival peripheral blood mononuclear cells (PBMCs) from individuals who recovered from H5N1 HPAI viral infections from 2004 to 2005 using flow cytometry and IFN-γ ELISpot assay. Moreover, we investigated the cross-reactive T cell response to H5N1 proteins in individuals who recovered from the H3N2 influenza virus infection. Our study may aid the design of a candidate T cell-based universal vaccine for broad-viral subtype protection.

Ethical issue
This study was approved by Institutional Review Boards from the Faculty of Medicine Siriraj Hospital, Mahidol University, under approval number Si213/2005. Written informed consent was obtained from all non-H5N1 individuals and H5N1 survivors or their parents for participation in this study (Kitphati et al., 2009;Noisumdaeng et al., 2014).

Human subjects and blood specimen
Sera, plasma and peripheral blood mononuclear cells (PBMCs) were obtained from 37 participants including four H5N1 Thai survivors, 10 H3N2 patients and 23 healthy individuals (Kitphati et al., 2009;Noisumdaeng et al., 2013;Noisumdaeng et al., 2014). Survivor nos. 1 and 2 were adults, while survivor nos. 3 and 4 were young children. All of them were infected with H5N1 clade 1 virus. A total of 20 sequential blood samples were collected from these survivors at intervals for up to four years after disease onset (Kitphati et al., 2009). The demographic data of the H5N1 survivors and time at blood specimen collection are shown in supplementary Table S1A. The H3N2 patients were diagnosed by real time reverse transcription-polymerase chain reaction (real time RT-PCR), and virus isolation together with serodiagnosis when possible. Demographic data of non-H5N1 infected subjects is presented in supplementary Table S1B. Sera and plasma samples were kept frozen at −20 • C until used. PBMCs were separated from anti-coagulated blood using Ficoll-Hypaque, (IsoPrep, Robbins Scientific Corporation, Sunnyvale, CA) density gradient centrifugation and stored in a freezing medium containing 10% DMSO (Sigma, MO) in fetal bovine serum (FBS) (Gibco R , NY) and cryopreserved in liquid nitrogen.

Designing the overlapping peptides and peptide-pool matrix
Overlapping peptides spanning the entire NP, M1, M2 and HA proteins of H5N1 KAN-1 virus (GenBank accession no., AAV35112, AAV35110, AAV35111 and AAS65615, respectively) were synthesized in the PEPscreen R custom peptide libraries format (Sigma-Genosys, Singapore). According to supplementary Table S2, the amino acid sequence identities of each protein were conserved among various H5N1 viruses circulating in Thailand during the study period. According to the manufacturer, all peptides were analyzed by MALDI-TOF mass spectrometry and their average crude purity is greater than 70%. Each peptide was 20 amino acids long with 10 amino acid residues overlapping, except the last peptide of the protein which may be shorter. There were a total of 49 NP peptides, 25 M1 peptides, 10 M2 peptides and 56 HA peptides as shown in Tables S3A-S3C. Peptide powder was dissolved in dimethyl sulfoxide (DMSO, Sigma) to the concentration of 50 mg/ml, then aliquot and kept at −80 • C . The concentrate peptide solution was further diluted in serum-free Roswell Park Memorial Institute 1640 medium (RPMI, Gibco) to the concentration of 2.5 mg/ml and stored at −80 • C in aliquots. These peptides were mixed into individual pools according to the two-dimensional matrix system as shown in Tables S3D-S3F. Each peptide pool contained an individual peptide at working concentration of 40 µg/ml. There were a total of 14 pools (A1-A7 and B1-B7) for NP, 12 pools (A1-A6 and B1-B6) for M and 15 pools (A1-A8 and B1-B7) for HA. The DMSO concentration in each assay was less than 0.1% (v/v).

Ex vivo IFN-γ enzyme-linked immunospot (ELISpot) assay
IFN-γ ELISpot assay was performed to demonstrate the T cell responses against H5N1derived peptides among H5N1 survivors and non-H5N1 subjects. A 96-well polyvinylidene difluoride (PVDF) ELISPOT plate (Multiscreen TM IP, MAIPS4510, Millipore, USA) was coated with mouse monoclonal anti-human IFN-γ 1-D1K (Mabtech AB, Stockholm, Sweden) at a concentration of 15 µg/ml overnight at 4 • C, followed by blocking with RPMI supplemented with 10% FBS for 2 h. After the blocking solution was discarded, the PBMC suspensions were added at the concentration of 3 × 10 5 cells/100 µl/well. Thereafter, peptide pool at a final concentration of 4 µg/ml (for screening of T cell activity) or individual peptide (for peptide specific activity of T cells) at a final concentration of 10 µg/ml was added into each well. PBMCs incubated with medium only served as the negative control, whereas PBMCs treated with 1 and 10 µg of phytohemagglutitin (PHA) (Sigma) served as the positive controls. After incubation for 14-16 h, PBMCs were removed, and the plate was washed with 0.05% Tween-20 in PBS (PBST). The biotin conjugated-anti-IFN-γ (7-B6-1biotin, Mabtech) was used as the primary antibody to bind the released IFN-γ absorbed on the membrane lining the plate bottom. Streptavidin-labelling alkaline phosphatase was used as the secondary antibody and followed by BCIP/NBT plus (Mabtech) as the chromogenic substrate. The reaction plate was incubated until purple spots were visible on the membrane, and then the reaction was terminated. The spots which represent the secreted IFN-γ producing T cells were counted and analyzed using a KS ELISpot version 4.11 program of an ELISpot plate reader (KS ELISpot, Zeiss, Munich, Germany). The T cell activity was considered positive when the number of spot forming cells (SFCs) in the reaction wells was at least two times greater than those of the negative control wells.

Functional assays of H5N1-specific T cells
The T cell functional assay was demonstrated by intracellular cytokine staining (ICS) and flow cytometry of peptide-specific polyclonal T cell lines (effector T cells) which were generated as bulk culture by stimulating the PBMCs with an individual specific peptide. The study design was shown in Fig. S1, and the detailed method was as followed:

Establishment of Epstein -Barr virus (EBV) transformed B cell lines (TBCLs)
Autologous TBCLs were used as the target cells for T cell functional assay by flow cytometry. TBCLs were established by infecting PBMCs with EBV from supernatant of B95-8 culture in RPMI supplemented with 10% FBS and in the presence of cyclosprorin A. Clumps of transformed B lymphocytes generally appeared after 2-3 weeks of infection. TBCLs could multiply indefinitely, a condition known as immortalization.

Establishment and maintenance of peptide-specific T cell lines
A bulk culture of specific polyclonal T cell line was generated by stimulating PBMCs with peptide and used as a source of the effector T cells. Frozen PBMCs of an number of 5-10 × 10 6 cells were thawed and divided into two portions comprising 20% and 80% of total cells. The 20% portion were pulsed with 100 µg/ml of peptide before mixing with the 80% portion. The suspension of cell mixture was added into a 6-well culture plate at the amount of 2-3 × 10 6 cells/well in 10% NHS RPMI supplemented with 25 ng/ml IL-7 (Peprotec, USA) and 50 U/ml IL-2 (Chiron, Netherlands).
Clumps of proliferating T cells should be seen in the culture at approximately 2 weeks after cultivation, i.e., peptide-specific polyclonal T cell line was established. T cell lines were expanded by stimulation with autologous gamma irradiated peptide-pulsed TBCLs plus irradiated allogeneic PBMCs (2,000 rad using cesium source).

Target cell preparation
The peptide-pulsed autologous TBCLs or TBCLs infected with rVac-NP virus were used as the target cells or antigen presenting cells (APCs) for investigating the function of peptide-specific T cells by flow cytometry. The unpulsed TBCLs or TBCLs infected with rVac-pSC11 virus were used as the negative target cell control. Pellets of TBCLs were pulsed with 100 µg/ml of specific peptides for 1 h, followed by washing and re-suspending with 10% FBS RPMI. For preparing recombinant vaccinia virus infected TBCLs, the cells were infected with rVac-NP or rVac-pSC11 virus at a multiplicity of infection (moi) of 3. After adsorption for 1 h 30 min at 37 • C, the infected cells were washed and re-suspended with 2% FBS RPMI and incubated overnight prior to functional assay by flow cytometry.

Intracellular cytokine staining (ICS) by flow cytometry
The flow cytometry assay comprised the FACS tubes of the polyclonal peptide-specific T cell lines (effector cells) stimulated with target cells (TBCLs pulsed with specific peptide or TBCLs infected with rVac-NP virus), and the control tubes including the effector cell alone, effector cells plus unpulsed TBCLs or effector cells plus rVac-pSC11 virus infected TBCLs. Staphylococcal enterotoxin B (SEB) was used as the positive stimulation control. Briefly, 1 × 10 5 effector cells were mixed with 1 × 10 6 target cells in a one ml volume (effector: target ratio of 1:10). The cell mixtures were incubated with anti-CD107a monoclonal antibody conjugated with FITC (BD Bioscience) for 2 h before adding with the mixture of monensin and brefeldin A at the final concentration of 10 µg/ml, and further incubated for 5 h. The cells were washed with FACS washing buffer (1% FBS, 0.5% NHS, 0.5 mM EDTA, 0.1% NaN 3 in PBS) and treated with FACS Permeabilizing Solution II (BD Biosciences) for 15 min in the dark in order to fix and permeabilize the cells. Thereafter, the cells were washed and stained with a cocktail of monoclonal antibodies: anti-CD4 PerCP or anti-CD8 PerCP, anti-IFN-γ APC and anti-TNF-α PE monoclonal antibodies (BD Biosciences). The CD4 + or CD8 + T cell populations producing IFN-γ , TNF-α and upregulation of CD107a degranulation marker were analyzed by FACSCalibur TM instrument by using CellQuest TM program (BD Biosciences).

Data analysis
The data graphs on IFN-γ T cell responses in H5N1 survivors and non-H5N1 subjects were generated using GraphPad Prism software version 4.0. The difference in the magnitude of responses (SFCs/10 6 PBMCs) between the H3N2 patients and healthy subjects was analyzed by Mann-Whitney U test. The statistically significant difference was obtained when the Pvalue was less than 0.05.

Screening for T cell response against NP, M and HA in H5N1 survivors
Pool peptides derived from H5N1 NP, M or HA amino acid sequences were employed to screen for reactive T cell response in H5N1 survivors using IFN-γ ELISpot assay. The results showed that PBMCs from the four H5N1 survivors elicited T cell responses against NP, M and HA peptide pools as shown in Figs. 1A-1C, respectively. The number of IFN-γ producing T cells stimulated by NP pooled peptides was higher than those of M and HA, which suggested that NP was more immunogenic in the induction of T cell responses. We also found that the T cell response in adults (survivor nos. 1 and 2) was stronger than that in children (survivor nos. 3 and 4). The T cell reactivity against NP and M peptides persisted for as long as 4 years after onset of disease as followed that far (Figs. 1A and 1B). All four H5N1 survivors had markedly high neutralizing antibody titers against H5N1 virus as demonstrated by an ELISA-based microNT assay. Moreover, they also contained neutralizing antibodies to seasonal H1N1 and/or H3N2 viruses as shown in supplementary  Table S1.

Determining the T cell activity against specific epitopes on NP, M and HA
Due to strong responses to pooled NP, M and HA peptides, H5N1 survivor nos. 1 and 2 were further identified for individual T cell specific epitopes by ELISpot assay. The major epitopes recognized by survivor no. 1 were NP 1-20 , NP 111-130 ( Fig. 2A), M1 121-140 , M1 201-220 (Fig. 2B) and HA 461-480 (Fig. 2C); whereas survivor no. 2 recognized NP 411-430 (Fig. 2D), M1 1-20 , M1 91-110 , M1 241-252 (Fig. 2E), HA 41-60 , HA 251-270 and HA 291-310 (Fig. 2F). However, the magnitude of responses to NP peptides was higher than those to M and HA peptides, indicating that NP is more immunogenic and putatively a highly immunogenic protein target for T cell responses. Based on serial PBMC samples, it was demonstrated that the reactivity of the peptide-reactive T cells could persist for at least 4 years after disease onset. The summary of magnitude and longevity of T cell responses are shown in Table  S4.

Cross -reactive T cell responses against H5N1 NP peptides in non-H5N1 individuals
This study carried out the cross-reactive T cell response to H5N1 NP peptides in 33 non-H5N1 individuals, including PBMCs from convalescent blood samples of 10 H3N2 patients and 23 healthy individuals. These subjects were confirmed for their H5N1 seronegativity by microNT assay (Table S1B). PBMCs were tested against the 49 overlapping peptides spanning the entire H5N1 NP by IFN-γ ELISpot assay. The results demonstrated that the magnitude of total cross-reactive T cells varied considerably between individuals. The mean of ex vivo IFN-γ responses was 102 and 51 SFCs/10 6 PBMCs in H3N2 patients and healthy subjects, respectively; and the mean of responses was statistically significant different between both groups (Mann-Whitney U test; p < 0.05) (Fig. 3).
The ELISpot assay with individual candidate NP peptide was conducted to identify the peptide-specific reactivity. From a total of 33 non-H5N1 subjects, six (60%) of 10

H5N1 NP -specific T cells elicited cytokine production and cytotoxic function
The effector cells in bulk culures of peptide-specific T cell lines (generated by clonal expansion of T cells after in vitro stimulation [IVS] of PBMCs from H5N1 survivors no. 1 with NP 1-20 and survivors no.2 with NP 411-430 peptide) were investigated by flow cytometry for IFN-γ and TNF-α productions, and the upregulation of CD107a, the degranulating marker of cytotoxic function, and additionally their immunophenotypes. The result showed that both CD4 + and CD8 + T cells from survivor nos. 1 and 2 expressed IFN-γ + and TNF-α + in recognition to the target cells infected with rVac-NP virus or pulsed with NP 1-20 (for survivor no.1) or NP 411-430 (for survivor no.2). Nevertheless, only CD8 + T cells sufficiently expressed CD107a + , while the expression level was poor for the CD4 + T cells (Fig. 4, Fig. S2 and Table S5).

Identity of NP epitope sequences among different influenza virus subtypes
Amino acid sequences of the H5N1 NP peptides that were recognized by the H5N1 survivors and non-H5N1 subjects were aligned and analyzed for their identities with those derived from the human influenza viruses as shown in Table 2. The H5N1 NP 1-20 and NP 401-420 amino acid sequences were identical to those of the H1N1pdm virus, while the NP 111-130 and NP 221-240 were identical to the H3N2 virus. As such, the H5N1 cross-reactive T cells   found in non-H5N1 subjects  peptides) might be generated from previous exposure to the common epitopes between the H5N1 virus and the seasonal H1N1 or H3N2 viruses. Cross reactivity for the NP 111-130 and NP 221-240 was found in higher frequency than the other peptides. Interestingly, the NP 411-430 (TFSVQRNLPFERATIMAAFT) was a unique epitope recognized by H5N1 survivors only.  (Zhao et al., 2012). The total number of IFN-γ reactive T cells against NP was higher than that of M and HA as demonstrated in two H5N1 adult survivors by ELISpot assay using 20 mers-overlapping pooled peptides, which suggested that NP was more immunogenic in the induction of T cell response. NP is the most abundant viral protein synthesized in the infected cells. It contains several immunodominant epitopes that could stimulate both the humoral and cell-mediated immune responses. As such, NP was identified as the major target of subtype-specific and cross-subtypic CD4 + and cytotoxic CD8 + T lymphocytes as reported by several groups of previous investigators (Kreijtz et al., 2008;Lee et al., 2008;Rimmelzwaan & Katz, 2013;Roti et al., 2008). Our result of the 1st round ELISpot assay using pooled peptides was confirmed by the 2nd round ELISpot assay using individual peptides. T cell responses could last longer than 4 years after disease onset for as far as the PBMC samples were available. Using individual peptide, our results showed that each survivor recognized different peptides, suggesting that their HLA or their repertoire of specific T cells were different.

DISCUSSION
Up until the present day, the H5N1-specific T cell epitopes are not well characterized due to the small number of H5N1 survivors. Powell et al. (2012) could identify three H5 HA-specific T cell epitopes (HA 160-177 , HA 344-364 and HA 439-456 ) from H5N1 asymptomatic cases; while Sun et al. (2010) could identify only one H5 HA-specific T cell epitope  ). However, our study was confined to the H5N1 NP peptides according to the limited amount of PBMCs available and the high immunogenicity of NP epitopes. Amino acid sequence identity of greater than 90% was found across NP proteins of various influenza virus subtypes (Noisumdaeng et al., 2014 (Lee et al., 2008). Moreover, the cross-reactive CD4 + T cells against H5N1 NP 401-420 andNP 113-132 were demonstrated in the healthy Caucasian descent based on A/Vietnam/1203/2004 H5N1 (VN1203) as determined by tetramer-guide epitope mapping. The NP 401-420 was restricted to HLA-DR0404; while NP 113-132 was restricted to HLA-DR1101 (Roti et al., 2008). However, the NP epitope variants which could escape from the T cell recognition have been reported (Rimmelzwaan et al., 2004;Berkhoff et al., 2007).
We further investigated the effector functions and immunophenotypes of the NP peptide-specific T cells by flow cytometry. The NP 1-20 andNP 411-430 peptides could stimulate both CD4 + and CD8 + T cell subsets in survivors, leading to clonal expansion and a high degree of polyfunction by simultaneously producing IFN-γ and TNF-α, and together with an upregulation of CD107a in recognition of the target cells pulsed with peptide or infected with rVac-NP virus. Since our peptides are 20 amino acids long; they might bind to both HLA class I (optimal epitope are 8-12 amino acids long) and class II (optimal epitope are 12-18 amino acids long) and lead to activation of CD4 + and CD8 + memory T cells. It is implied that NP protein expressed in the rVac-NP virus infected TBCLs is naturally processed and presented in association with HLA to the effector T cells (Jameson, Cruz & Ennis, 1998). Importantly, it is uncertain that the induction of such polyfunctional T cell populations might be associated with the recovery from severe diseases in our H5N1 survivors. The amino acid sequences of the avian H5N1 internal proteins were closely identical to those of the seasonal H1N1, H3N2 and H1N1pdm viruses; exposure to seasonal influenza virus infection or vaccination can generate a pool of cross-reactive memory CD4 + and CD8 + T cells which are capable of recognizing a number of conserved internal proteins form avian H5N1 virus.

CONCLUSIONS
The broader vaccines that rely on the induction of T cell-based immunity against conserved epitopes would provide broader partial protection, restrict the viral diversity in the infected host and help lower severity and mortality against overwhelming pandemic influenza virus infection. Our present study provides insight into a better understanding of the homosubtypic and heterosubtypic T cell-mediated immune responses against H5N1 virus in H5N1 survivors and non-H5N1 subjects. NP is an immunodominant target of cross-recognition owing to its high conservancy. Therefore, the development of vaccine targeting the conserved NP may be a novel strategy for influenza vaccine design.