Background
The highly pathogenic (HP) avian influenza viruses (AIVs) of the H5 subtype pose a serious epidemiological problem. This applies especially to H5N1 HPAIV, which was detected for the first time among farmed geese in China in 1996 and in humans a year later [
1]. The spread of the H5N1 viruses to many regions of the world has been accompanied by frequent avian flu outbreaks in poultry that have resulted in mortality rates of up to 100% [
2]. Moreover, as of January 2017, there have been a total of 856 laboratory-confirmed human cases of H5N1 influenza, 452 of which were fatal [
3]. From 2009 onward, the emergence of the reassortant H5-subtype HPAIVs, such as H5N2, H5N5, H5N6 and H5N8, has been noted [
2]. In addition, the novel H5N8 and H5N2 HPAIVs were observed to spread rapidly and globally soon after their identification in 2014 [
4,
5]. Many populations of domestic birds have been substantially affected by these lethal viruses due to infection or mass culling [
6,
7].
In view of the threat to animal and human health and life, the H5-subtype HPAIVs are under epidemiological surveillance. Various strategies have been developed for prevention and treatment of the infections caused by these viruses. The majority of the strategies have focused on the H5 hemagglutinin (HA), which determines the high pathogenicity of AIVs and is also the main target for neutralizing antibodies. In the efforts to combat H5-subtype HPAIVs, monoclonal antibodies (mAbs) against H5 HA play a significant role. When characterized by high specificity and affinity and/or neutralizing activities, they constitute effective diagnostic and therapeutic agents. Moreover, mAbs are powerful as tools for vaccine development as well as in basic research including studies of the antigenic architecture of the HA of influenza H5N1 viruses [
8].
Most therapeutic and diagnostically valuable mAbs are immunoglobulins (Igs) of the G class (IgG). These are homodimeric glycoproteins of ~150 kDa. Each IgG molecule contains two identical heavy (H) chains and two identical (L) light chains with molecular weights of ~50 kDa and ~25 kDa, respectively [
9]. The chains are connected by disulfide bonds to form a Y-shaped structure. Both the H and L chains of IgG antibodies consist of variable (V) and constant (C) domains referred to as VL and CL and VH, CH1, CH2 and CH3, respectively. The variable parts of the H and L chains, especially the complementarity-determining regions (CDRs), comprise the antigen-binding sites of the Ig molecules. They are responsible for the target epitope recognition, antigen-antibody reaction and the diversity of antibody specificity. Variations in mAbs may also result from post-translational modifications such as alternative disulfide pairings, deamidation, methionine oxidation or pyroglutamate formation. The heterogeneity of the glycosylation potently affects both pharmacokinetics and stability of various isoforms, which alter the clinical efficacy and safety of the therapeutic proteins [
10].
Monoclonal antibodies can be generated using a hybridoma technique based on spleen/myeloma fusion. The concept was developed in the 1970s [
11]. Another method involves the transformation of human B lymphocytes with the Epstein-Barr virus [
12,
13]. Irrespective of the production method, the mAbs secreted by the immortalized cells need to be selected in terms of the desired function and/or specificity. In the case of mAbs against viral HAs, the hemagglutination inhibition (HI) and virus neutralization (VN) tests are used to ascertain the ability of established antibodies to confer protection against influenza. Identification of antibodies with the desired specificity is frequently accomplished by immunological techniques such as the enzyme-linked immunosorbent assay (ELISA). Alternative protocols that optimize the screening for antibody-producing cells using flow cytometry are also available [
14,
15]. These techniques do not, however, provide complete information regarding the heterogeneity of the hybridoma clones.
In addition to the methods and techniques described above that have been applied to mAb production, mass spectrometry (MS) can be used. MALDI-TOF/TOF MS has been widely recognized in the fields of the development of therapeutic antibodies, determination of their structural features, glycan characterization and profiling [
16‐
18]. The technique enables evaluation of the recombinant protein sequence and structure and provides information on amino acid modifications and sequence alterations. Although the resolution of the mass spectrometer is insufficient to differentiate between large, intact proteins, accurate measurements can be achieved at the peptide level [
19].
To facilitate the structural analysis of antibodies, well known methods may be applied to selectively cleave the Ig molecules into fragments that have discrete characteristics and functions. If the variable regions of IgG antibodies are of primary interest, it is possible to generate their antigen-binding fragments, including the F(ab’)2, Fab’, Fab and Fv. For example, the monovalent fragment denoted Fab is composed of one constant and one variable domain of each of the H and the L chain, i.e., CH1, VH and CL, VL, respectively. The two variable domains, VH and VL, specifically bind the epitope of their respective antigen [
20]. Preparation of the Fab fragments of IgG antibodies is usually accomplished by digestion of Igs with papain in the presence of a reducing agent [
21]. In the case of mouse IgG1 antibodies, the enzyme of choice is ficin, applied with the optimized reductant concentration [
22]. The Fab fragment does not contain any part of the crystallizable portion of the constant region of the Ig (Fc). Formed entirely from the H chain constant domains, the Fc fragment does not bind antigen and is responsible for the effector functions of antibodies.
In this paper, we describe newly generated, highly specific mAbs with a broad range of activities against the H5 HA of influenza viruses. As obtained, the mAbs were indistinguishable on the basis of the range of immunoreactivities determined by ELISA. We therefore employed MALDI-TOF/TOF MS as a structural tool for their differentiation. These mass measurements enabled assessment of the heterogeneity of peptide maps, obtained for the mAb-derived Fab and Fc fragments. Moreover, they provided clear distinctions among the analyzed clones and the antibody-secreting hybridoma cell lines. Thus, our results show for the first time that peptide mapping of antibody fragments with MS is an efficient alternative method for differentiation of antibody clones and the relevant antibody-producing cell lines.
Methods
Hemagglutinin antigens
This work was performed with the use of recombinant H5 HA proteins and inactivated AIVs of the H1-H16 subtypes, which are listed in Additional file
1: Tables S2 and S4, respectively. The ectodomain- or the HA1 subunit-based HA proteins (rHA, rHA1, respectively) were produced in a mammalian expression system (Immune Technology Corp., New York, NY, USA), except for one rHA protein, which was of baculovirus-expression system origin (Oxford Expression Technologies Ltd., Oxford, England, UK). Prior to use, the recombinant antigens were characterized by MS, ELISA for antigenicity and oligomerization and/or the hemagglutination test, performed as described in Additional file
1. The proteins were used to immunize mice, for plasma antibody titer determination, in preliminary or further specificity testing of hybridoma culture supernatants and/or reactivity studies of the finally selected, purified mAbs. The applications of each antigen are shown in Additional file
1: Table S2. Influenza viruses of the H5 (4 strains) and non-H5 subtypes (21 strains) were certified by Istituto Zooprofilattico Sperimentale delle Venezie (Legnaro, Padova, Italy) and originated from x-OvO Ltd. (Dunfermline, Scotland, UK). The viral strains of the H5-subtype were used to test the culture supernatants from the hybridoma clones, as shown in Additional file
1: Table S4. Both H5 and non-H5 AIVs were used in the studies of the finally selected, purified mAbs (Additional file
1: Table S4). The H5 HA antigens intended for antibody screening were chosen from those commercially available to obtain the panel of antigens with diverse amino acid sequences of the HA1 subunit. The antigenic diversity was determined by homology searches against the immunogen’s HA1 subunit using the BLAST program on NCBI. Complete information on the HA antigens, including their abbreviated names and relevant viral strains, is provided in Additional file
1.
Hybridoma production and screening
Female, 6-week-old BALB/c mice (Mossakowski Medical Research Centre PAS, Warsaw, Poland) were first immunized subcutaneously with 10 μg of rHA - A/H5N1/Qinghai, emulsified with an equal volume of Complete Freund’s Adjuvant (Sigma-Aldrich, St. Louis, MO, USA) using the two-syringe method. Two weeks later, two subsequent 10 μg doses of the same immunogen were given by intraperitoneal injection in the absence of an adjuvant at 3-week interval. Thereafter, the mice received an additional intravenous dose of 10 μg of rHA protein in PBS and were euthanized 3 days later.
The mouse was chosen for fusion on the basis of the antibody titers against rHA - A/H5N1/Qinghai and rHA - A/H5N1/Poland using an ELISA of the plasma samples collected after the third immunization. The splenocytes were fused with mouse myeloma cells of SP2/0 line (ATCC, Rockville, MD, USA) in the presence of 50% PEG 1500 and 5% DMSO. The fused hybrid cells were cultured in RPMI-1640 medium containing FBS, L-glutamine, sodium pyruvate, and antibiotics (streptomycin, penicillin), with hypoxanthine, aminopterin and thymidine (HAT) as the selecting agents. The hybridomas were subcloned by the limited dilution method. To subclone, cells of each hybridoma were suspended in 5 mL of complete RPMI-1640 medium, counted and diluted to 10 or 5 cells per mL. The obtained suspension was transferred into 96-well plates (100 μL per well equivalent to 1 or 0.5 cell per well). The resulting hybridoma cell lines were grown in RPMI-1640 medium with the same supplements as the selection culture medium except for HAT. The reagents used for fusion and hybridoma culture were purchased from Sigma-Aldrich.
The hybridoma culture supernatants were screened for the presence of anti-H5 HA antibodies using ELISA. Both before and after subcloning, preliminary testing was performed using the rHA - A/H5N1/Qinghai and the rHA - A/H5N1/Poland as antigens. To select broadly reacting antibodies, the analyses were completed using the various H5 HA antigens as shown in Additional file
1: Tables S2 and S4. The recombinant H5 HA proteins from mammalian and baculovirus expression systems were coated on Ni-NTA strips (Qiagen, Hilden, Germany) and MediSorp plates (Nunc, Roskilde, Denmark), respectively, and the H5-subtype AIVs were coated on MaxiSorp plates (Nunc). The hybridoma culture supernatants were analyzed in the antigen-coated and also in the non-coated wells to control for non-specific binding. Commercial antibodies against H5 HA (mAb 8 in Additional file
1: Table S1) were used as a positive control. The blank control was the culture medium. Anti-mouse IgG (γ-chain specific) antibodies labeled with HRP (Sigma-Aldrich) were applied to detect the antigen-antibody complexes. At all stages of the procedure, the antibodies were isotyped using a commercial kit: “Mouse Monoclonal Antibody Isotyping Reagents” (ISO-2; Sigma-Aldrich).
The selected mAbs in the final set, a total of 7 clones, were purified from the hybridoma culture supernatants using “HiTrap Protein G HP” (GE Healthcare, Uppsala, Sweden) according to the manufacturer’s instructions. The purified mAbs were stored in PBS that contained sodium azide.
Determination of monoclonal antibody immunoreactivity by ELISA
The reactivity of the finally selected and purified mAbs was studied using all of the HA antigens depicted in Additional file
1: Tables S2 and S4. To perform the tests, Ni-NTA strips (Qiagen) were coated with the rHA and rHA1 proteins (1 μg/mL in 1% BSA/PBS) and MaxiSorp plates (Nunc) with AIVs of H1-H16 subtypes (4000 hemagglutination units/mL in PBS), all by overnight incubation at 2–8 °C. Because they were supplied pre-blocked, the coated Ni-NTA strips were used without the blocking step. The coated MaxiSorp plates were blocked with 2% BSA/PBS. Thereafter, the mAbs diluted in 2% BSA/PBS were applied to the antigen-coated wells and also to the non-coated wells to control for non-specific binding. The assay was performed in the presence of other control samples. Commercial antibodies against H5 HA (mAb 8 in Additional file
1: Table S1) were used in antibody testing with H5 HA antigens and non-H5 subtype AIVs to serve as positive and negative controls, respectively. The blank control was the dilution buffer. In the assays for cross-reactivity, additional control was provided by testing the mAbs with non-H5 subtype AIVs in parallel with H5N3 and H5N9 viruses. The plates with tested and control samples were incubated overnight at 2–8 °C.
Detection of signals was accomplished using HRP-labeled, anti-mouse IgG (γ-chain specific) antibodies (Sigma-Aldrich). The secondary antibodies, diluted 1:1000 in 2% BSA/PBS, were incubated with the test plates for 1 h at 37 °C. The reactions were developed with TMB (Sigma-Aldrich) at room temperature for 30 min and subsequently stopped by adding a solution of H2SO4. The absorption was read at 450 nm using a μQuant microplate spectrophotometer (BioTek Instr. Inc., Winooski, VT, USA). For each antibody sample, the mean absorbance value for blank control samples was subtracted.
Immobilized ficin digestion
The monoclonal antibodies were concentrated with the use of VivaSpin6 MWCO 10000 units (Sartorius Stedim Biotech GmbH, Goettingen, Germany) with simultaneous buffer exchange to PBS. The digestion of the mAbs with Immobilized Ficin (Pierce™ Mouse IgG1 Fab and F(ab’)2 Micro Preparation Kit; Thermo Scientific, Waltham, MA, USA) was performed according to the manufacturer’s protocol. For each mAb, a sample containing 250 μg of the protein was applied to the spin column tube containing the equilibrated enzyme. The digestion was conducted for 5 h. After separation by affinity chromatography, the two fractions, which contained the Fab and the mixture of Fc and undigested IgG, were concentrated using VivaSpin6 MWCO 5000 units (Sartorius Stedim Biotech GmbH) with simultaneous buffer exchange to PBS.
Gel electrophoresis, MS measurements and data analysis
Non-reducing and non-boiled SDS-PAGE was performed using a 5% stacking gel, pH 6.8 and a 12.5% resolving gel, pH 8.8. The wells were loaded with 40 μL of the concentrated fractions mixed with an equal volume of sample buffer (62.5 mM Tris-HCl, pH 6.8; 25% glycerol; 1% Bromophenol Blue). Coomassie Brilliant Blue G-250 was used to visualize the proteins. The protein bands corresponding to the Fab, Fc and undigested IgG were excised from the gel, washed, reduced in the presence of 50 μL of 10 mM dithiothreitol at 60 °C for 45 min and alkylated with 50 μL of 50 mM iodoacetamide at room temperature in the dark for 60 min. The samples were further incubated with 30 μL of trypsin buffer (10 ng/mL; Promega, Madison, WI, USA, Cat. No. V5111) at 37 °C for 18 h. Prior to the MS analysis, the samples were evaporated to dryness and dissolved in 10 μL of 0.1% trifluoroacetic acid (TFA). All other general-use reagents were from Sigma-Aldrich.
The MALDI-TOF/TOF measurements were performed using reflector mode with a 4800 Plus instrument (Applied Biosystems, Waltham, MA, USA). α-Cyano-4-hydroxy-cinnamic acid dissolved in 50:50 water/acetonitrile (J.T. Baker, Deventer, The Netherlands) with 0.1% TFA (final concentration) was the matrix used. External calibration was achieved with a 4700 proteomics analyzer calibration mixture provided by Applied Biosystems. Each sample was spotted 5 times onto a 384 Opti-TOF MALDI plate and analyzed. Data Explorer Software, Version 4.9 (Applied Biosystems) was applied to process the acquired spectra. The peptide identification was accomplished using the Mascot search engine (Matrix Science Inc., Boston, MA, USA) against the Swiss-Prot and NCBInr sequence databases. A mass tolerance of 25 ppm and one missing cleavage site for the peptide mass fingerprinting (PMF) and a tolerance of 0.6 Da and one missing cleavage site for the MS/MS search were allowed. Carbamidomethyl was set as a fixed modification, glycosylation was considered as variable modification. The monoisotopic masses of the peptides unambiguously identified by Mascot are given as [M + H]+. For the peptides not assigned to a known protein, the masses of represent the average experimental monoisotopic mass and are given as [M + H]+. The data are presented as mass-to-charge ratio (m/z) values.
Discussion
Infection with HPAIVs of the H5 subtype leads to multi-organ disease and death in domestic birds [
2,
4‐
7]. In addition, the H5N1 viral strains pose a persistent pandemic threat [
3]. To prevent and treat H5N1 influenza virus infections and for surveillance of H5N1 and other H5-subtype AIVs, mAbs against H5 HA have been developed by many research groups (e.g., [
23‐
28]; for review, see [
8]). Our work responded to the demand for diagnostically valuable mAbs with broad strain specificity against AIVs of the H5 subtype. These antibodies were produced with hybridoma technology using recombinant, ectodomain-based H5 HA protein with native-HA characteristics to immunize mice.
The hybridomas generated by this process were screened for the production of IgG antibodies against the H5 HA using ELISA. The screening was performed against several forms of the HA antigen that had various properties. The use of conformational rHA1 proteins enabled the identification of antibodies that bound to the highly variable HA1 subunit, which determines the HA subtype. Distinguishing between conformation sensitive and non-sensitive antibodies was achieved using a misfolded rHA protein. In addition to the variations in the forms, the sequences of the HA antigens that were used originated from highly divergent H5-subtype influenza viruses. As a consequence, the H5 HA antigens demonstrated substantial antigenic diversity, which was confirmed by a homology search against the immunogen.
From our screening strategy, we obtained a total of 64 hybridoma cell lines. These cell lines secreted antibodies that were reactive with all of the H5 HA antigens that were used for the specificity testing except for the non-conformational antigen. A final set of 7 hybridoma clones was selected. Specifically, the G-1-31-22, G-2-14-10, G-5-32-5, G-6-42-42, G-6-42-71, G-7-24-17 and G-7-27-18 mAbs, all of IgG1 isotype, were further analyzed. In the preliminary immunoreactivity studies, we were able to show that the newly established mAbs specifically recognized epitopes in the properly folded HA1 subunit of H5 HAs from multiple strains of the H5-subtype influenza viruses (Table
1). Importantly, they did not cross-react with influenza viruses of H1-H4 and H6-H16 subtypes (Table
1). However, these studies did not allow for clear discrimination among the finally selected mAbs and the relevant hybridoma cell lines. For this reason, the exact number of the unique antibody and hybridoma clones could not be inferred.
The differentiation of antibody clones and relevant antibody-producing cell lines is of special importance for comprehensive assessment of their possible applications. When used in diagnostics or basic research, the set of mAbs that recognize different epitopes in the H5 HAs potentially extends the range of the target AIVs among the formerly and currently circulating viral strains. It can also facilitate the identification of the novel emerging H5-subtype AIVs. In addition, availability of different antibody clones enables the choice of the ones that will be best suited to specified method or technique. For example, two distinct mAbs can be successfully used as detection and capture antibodies in virus detection by a sandwich ELISA or immunochromatography.
Insight into the antibody heterogeneity could be provided by a comparison of the sequences encoding their variable regions, especially the CDRs [
29]. Sequencing is routinely used to identify antibodies. However, it may be perceived as challenging if the presence of pseudogenes and mRNAs encoding non-functional antibody chains in hybridoma cells is considered [
30]. Another method for differentiating mAbs is based on cross-inhibition experiments (e.g., [
31]). Antibodies that do not compete for binding to the target antigen are considered to recognize distinct, non-overlapping epitopes. Competition between mAbs is interpreted as indicating that the tested antibody clones bind to the same or to closely related epitopes. Thus, cross-inhibition experiments may not give conclusive results. As the first mass spectra characterizing the generated antibody clones showed some differences between selected antibodies, we decided to expand them with peptide mapping of the Fab and Fc fragments. It was assumed that antibody examination at the protein level would allow to avoid some possible drawbacks related with their analyses at the genetic and functional levels.
Digestion with the immobilized ficin produced Fc and Fab fragments of the G-1-31-22, G-2-14-10, G-5-32-5, G-6-42-42, G-6-42-71, G-7-24-17 and G-7-27-18 mAbs. Subsequently, tryptic peptide maps of these fragments were generated. Based on the resulting MS and MS/MS spectra, Mascot searches against the Swiss-Prot and NCBInr sequence databases were performed. This enabled identification of some peptides derived from both Fc and Fab fragments, all of which belonged to the Ig class of proteins (Tables
2,
3, and
4).
Most of the peptides detected in the Fc fragments were “common” for the analyzed mAbs (69%; Table
4). The majority of the amino acid sequences of these fragments were identified within the protein databases. This is consistent with the widely accepted view that the Fc fragments are species- and isotype-conserved components of the Igs, which have no significance for their specificity [
9]. In contrast, “discriminatory” peptides dominated the Fab fragment maps (61%; Table
3). Within these antibody fragments, very few sequences could be identified with the database searches: many fewer than for the conserved Fc fragments (17% vs. 69%; Tables
2,
3 and
4). These different proportions can be explained by the fact that the Fab fragments exhibit considerable variation in the specificity-determining sequences. For this reason, the protein databases are incomplete in this area.
A close inspection of the peptide maps of the Fab fragments revealed that the analyzed antibodies differed in the profiles of their “discriminatory” peptides (Table
3). Accordingly, 6 different clones were distinguished among the 7 selected mAbs. Presumably, these mAbs target distinct epitopes in the H5 HA molecule. For the G-6-42-42 and G-6-42-71 clones, identical peptide maps of the Fab and Fc fragments were obtained (Tables
2,
3 and
4). This indicates that these two mAbs are the same antibody clone. Interestingly, the G-6-42-42 and G-6-42-71 antibodies were the only clones among selected mAbs that originated from subcloning of the same hybridoma. Conclusions from the mass spectrometry approach are consistent with those from the advanced immunoreactivity studies (Additional file
6: Figures S10-S12).
Addendum
On 9 June 2016, G-1-31-22, G-2-14-10, G-5-32-5, G-6-42-42, G-7-24-17, G-7-27-18 hybridoma cell lines were given the following Accession Numbers by the International Depositary Authority: DSM ACC3292, DSM ACC3293, DSM ACC3294, DSM ACC3295, DSM ACC3296 and DSM ACC3297, respectively. They are all held by the Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany).
Acknowledgements
Not applicable.