Cross-protection against divergent strains of influenza virus is an objective of various vaccination approaches. B cells cross-neutralizing several influenza A heterosubtypes have been isolated from cultured human memory B cells (MBCs) and plasmablasts early after influenza vaccination or infection. However, a systematic assessment of the frequency of MBCs and plasmablasts in the blood of healthy individuals is lacking. Here, we show that under resting conditions about 45% of human adults never vaccinated nor exposed to avian A/H5N1 influenza have detectable circulating MBCs cross-reacting with H5N1. This proportion rises to 63.3% among subjects with a large pool of MBCs specific for seasonal H1N1 (i.e. frequency ≥1% of total IgG MBCs). Moreover, subjects with high baseline frequencies of H1N1-specific MBCs had an expansion of H5N1-specific MBCs producing H5-neutralizing antibodies already after the first dose of an MF59-adjuvanted H5N1 vaccine. These results suggest that H1N1-specific MBCs contain a subset of cells cross-reacting to H5. We propose that a proportion of human adults have a pool of H5/H1 cross-reactive MBCs that contribute to the rapid rise of the antibody response to divergent influenza strains. This may have implications on vaccination strategies aimed at counteracting future influenza pandemics.
The influenza A virus surface proteins hemagglutinin (HA) and neuraminidase (NA) are highly susceptible to mutations that favor the selection of variants able to escape neutralizing antibodies. This phenomenon, known as antigenic drift, necessitates the production of new influenza vaccines every year. More rarely, reassortment of genome segments from animal and human viruses might also occur, leading to the emergence of new viruses to which the human population is immunologically naïve (antigenic shift). These viruses, like avian H5N1, have the potential to cause pandemics [1-3].
The immunological naivety of the human population to H5N1 is inferred by the low proportion of adults with hemagglutination-inhibiting antibodies to H5N1 virus . In addition, clinical trials with H5N1 and H9N2 vaccines have shown that two doses of adjuvanted vaccine are needed in order to increase neutralizing antibody titers to potentially protective levels [5-10]. In spite of this, it has been observed that vaccination with seasonal strains or natural influenza infection can transiently induce antibodies neutralizing the H5N1 virus [11, 12]. Heterosubtypic monoclonal antibodies able to cross-react across group 1 or group 2 influenza virus subtypes have been isolated from MBCs and plasma cells of adults early after seasonal vaccination [13-17]. These observations demonstrate the existence of cross-neutralizing epitopes on influenza HA and lead to the idea of developing a universal vaccine able to protect people from broadly divergent influenza strains. Remarkably, however, most cross-reactive B-cell clones identified produce antibodies whose cross-neutralizing activity is restricted either to group 1 or to group 2 influenza strains. So far, B-cell clones producing antibodies with broadly neutralizing activity across group 1 and 2 influenza strains have been successfully isolated in only one case .
Overall, these observations support the notion that B-cell clones neutralizing all influenza A strains are rare, while MBCs broadly cross-reactive across group 1 or group 2 might circulate at detectable frequencies in the blood. The aim of this study was to quantify the circulating IgG-switched MBCs capable of binding group 1 HA molecules from seasonal H1N1 and avian H5N1 influenza in the blood of healthy adults, and to evaluate the impact of these MBCs on the response to avian H5N1 vaccines.
To this end, we analyzed healthy adult subjects with no known history of exposure to avian influenza for (i) the presence of detectable numbers of circulating MBCs reactive to H5N1 from the A/Vietnam/1194/04 influenza strain; (ii) the existence of an association between the frequency of circulating MBCs specific for seasonal H1N1 and H5N1; (iii) the capacity of ex vivo isolated H1N1-specific MBCs to produce antibodies cross-reactive to H5 molecules; (iv) the impact of baseline frequencies of H1N1-MBCs on the B-cell response to vaccination against H5N1.
H5N1-cross-reactive MBCs are more frequent in H5N1-naïve subjects with a large pool of H1N1-MBCs
To assess whether MBCs able to react to H5N1 virus are present at detectable frequencies in the peripheral blood of H5N1 naïve subjects, we analyzed prevaccination samples from a cohort of 60 healthy adults enrolled in a clinical study to receive two doses of an MF59-adjuvanted prepandemic A/H5N1 vaccine in combination with one dose of seasonal influenza vaccination. At baseline 98% (59/60) of the analyzed subjects had undetectable titers of H5N1-neutralizing antibodies by either hemagglutination inhibition (HI) or microneutralization (MN) assays (Fragapane et al., unpublished observations). To assess whether H5N1-cross-reactive MBCs were present at detectable frequencies in the blood of these subjects before vaccination, we measured the frequencies of MBCs specific for H5N1 or H1N1 in peripheral blood mononuclear cells (PBMCs) obtained at baseline. By using a serial limiting dilution assay (sLDA) previously described , we determined that the lowest measurable frequency of H5N1- and H1N1-specific MBCs was 0.02% of total IgG-producing MBCs. The analysis of the whole cohort shows (Fig. 1) that IgG MBCs specific for seasonal H1N1 antigens are detectable in 75% of adults (45/60) at frequencies ranging from 0.02 to 25% of total IgG MBCs (median value 1.47%). The unexpectedly high presence of H1N1-IgG MBCs in a minority of subjects (four subjects have baseline H1N1-IgG MBCs frequencies >10%) may be due to the fact that the enrollment was done over a period in which the seasonal virus was already circulating. On the other hand, H5N1-IgG MBCs are present at detectable frequencies in the blood of 45% (27/60) of healthy adults not exposed to H5N1 antigens, ranging from 0.06 to 14% of total IgG MBCs (median value 0.48%). A significant correlation was found between pair data of frequencies of H1N1- and H5N1-IgG MBCs at baseline (Spearman's ρ = 0.52, p-value < 0.0001).
To investigate whether PBMCs from subjects with a relatively large pool of H1N1-IgG MBCs have a higher probability to contain H5N1-cross-reactive IgG MBCs, the cohort of subjects was split into two groups based on the baseline frequency of H1N1-IgG MBCs. The cut-off was set at 1% of total IgG MBCs that identifies two groups of 30 subjects each with frequency of H1N1-IgG MBCs ≥1% (H1N1-High) or <1% (H1N1-Low). In the absence of established reference values to define what could be a meaningful proportion of H5N1-IgG MBCs, we arbitrarily set this endpoint at 0.1% of total IgG MBCs, which is a value fivefold higher than the lowest detectable frequency measured in this study. Remarkably, we found that the presence of circulating H5N1-IgG MBCs at frequencies equal or above the established endpoint of 0.1% significantly associates with the presence of H1N1-IgG MBCs at a frequency equal or above 1% of total IgG MBCs (p = 0.0089, accuracy 68%, two tails Fisher exact test) (Fig. 1). Indeed, while in the whole population the proportion of subjects having a meaningful proportion of H5N1 reactive MBCs was 40% (24/60), this percentage reached 60% (18/30) in subjects with frequencies of H1N1-MBCs ≥1% of total IgG MBCs (H1N1-High) and decreased to 20% (6/30) in subjects having H1N1-MBCs below 1% (H1N1-Low). Of note, about 40% (12/30) of the H1N1-High subjects had H5N1-IgG MBCs at frequency <0.1%.
Together these results show that IgG MBCs reactive to H5N1 circulate in sizeable numbers in the blood of subjects not exposed to H5N1 antigens. In addition, they suggest that a pool of H5N1-IgG MBCs of a certain size arises in the majority, but not all, of subjects who have acquired a relatively large pool of MBCs specific for seasonal H1N1.
H5/H1 cross-reactive MBCs can be isolated from individuals not exposed to nor vaccinated with H5N1
To verify whether H1/H5 cross-reactive MBCs in steady-state conditions could be detected in the bloodstream, we sorted B cells capable of ex vivo binding to purified recombinant H1 or H5 from PBMCs of three healthy donors with no known history of exposure to H5N1, and analyzed their capacity to produce cross-reactive antibodies by ELISpot. Figure 2 shows that, after polyclonal activation in vitro, H1Sol+CD20+ cells sorted from two healthy subjects (Subjects I and II), generated enriched populations of cells secreting antibodies specific for H1 and H5. These antibodies did not cross-react with human serum albumin (HSA). Consistently, H5Viet+CD20+ cells sorted from PBMCs of another healthy subject (Subject III) generated a population of cells secreting antibodies specific for H5 as well as for H1. These data show that MBCs sorted for their ability to bind to H1 can generate cells secreting antibodies capable of binding H5 and vice versa. Of note, in one of the three subjects (Subject II) the frequency of H1-specific antibody secreting cells detected by ELISpot in cultures of sorted H1+ B cells was sufficiently high to allow to determine that antibodies produced by at least 60% of them cross-reacted with H5. All together, these results show that H5-cross-reactive IgG MBCs circulate in steady-state conditions at a sizeable frequency in the blood of subjects with unknown history of exposure to H5N1 influenza antigens.
H1N1-High subjects have enhanced MBC responses to both H5N1 and H1N1
To evaluate the impact of the size of the H1N1-MBC pool at baseline on the response to vaccination against H5N1, we analyzed the frequency of H5N1- and H1N1-specific IgG MBCs in PBMC samples from healthy adults before and after vaccination with MF59-adjuvanted H5N1 administered in combination with seasonal influenza vaccine. This study was performed in a cohort of 60 subjects randomized to receive one dose of a monovalent MF59-H5N1 vaccine (A) at day 1 and 3 weeks after (at day 21) one dose of a tetravalent (T) formulation containing MF59-H5N1 and the seasonal vaccine (Group A/T, n = 30), or to receive the same vaccines but administered in the opposite sequence (Group T/A, n = 30). To verify the impact of baseline H1N1-MBCs in the response to vaccination, subjects were stratified in two groups based on the frequency of H1N1-IgG MBCs measured at baseline in their blood (H1N1-High and H1N1-Low). We found that, irrespective of the vaccination schedule, H1N1-High subjects developed a significantly larger pool of H5N1-specific IgG MBCs as compared with that of H1N1-Low subjects (Fig. 3A). Indeed, already after the first dose of H5N1 vaccine (day 21) H1N1-High subjects had frequencies of H5N1-IgG MBCs approximately threefold higher than H1N1-Low subjects (day 21 median values: 4.7 and 6.1% in H1N1-High versus 1.8 and 2.1% in H1N1-Low subjects from the A/T and T/A group, respectively). Following the second vaccination (day 43) these differences still persisted although less pronounced (day 43 median values: 7.3 and 6.1% in H1N1-High versus 2.5 and 4.5% in H1N1-Low subjects from the A/T and T/A group, respectively) (Fig. 3A). H1N1-High subjects had also a significantly more pronounced MBC response to H1N1 as compared with that of H1N1-Low subjects at all time-points after vaccination. Day 21 median values of H1N1-IgG MBCs were: 11.8 and 21.3% in H1N1-High versus 2.0 and 5.5% in H1N1-Low subjects from the A/T and T/A groups, respectively. Day 43 median values were: 22.9 and 12.6% in H1N1-High versus 5.9 and 7.5% in H1N1-Low subjects from the A/T and T/A groups, respectively (Fig. 3B). It is worth noting that most of the subjects in the A/T group experienced an expansion of their H1N1-MBC pool already in response to the first vaccination with monovalent MF59-H5N1 (Fig. 3B). This observation suggests that a proportion of preexisting H1N1-IgG MBCs cross-reacted to H5N1 in vivo.
Preexisting H1N1-IgG MBCs may favor a functional response to H5N1 after vaccination
To gain insight on whether neutralizing epitopes were included among those targeted by H5/H1 cross-reactive MBCs, we looked for H5-neutralizing antibodies in plasma samples and supernatants from MBC cultures with a pseudoparticle neutralization assay (PNA)  using pseudoviral particles expressing the HA from the A/Vietnam/1194/04 H5N1 strain. This assay not only provides information about the impact of a large H1N1-MBC preexistent pool on the development of neutralizing antibodies, but also restricts the effect to cross-reactivity to HA molecule, since the pseudoparticles only express the H5 HA, but not the NA.
As already described in previous studies, owing to the poor immunogenicity of the HA of A/H5N1 and the immunologic naivety of the population, H5N1-specific antibodies can be measured by HI and MN assays mainly after the second vaccine dose of MF59-H5N1 vaccine, and only rarely functional antibodies are measurable at baseline [4, 21]. However, using PNA, which is more sensitive than HI and MN assays , we detected neutralizing antibodies already at baseline in plasma of 42% of H1N1-High subjects, even though at low titers. On the contrary, plasma-neutralizing antibodies were present at baseline only in 8% of H1N1-Low subjects. Titers were also significantly different in the two groups (p < 0.004) (Fig. 4A). Remarkably, following the first vaccination (day 21) H5 PNA neutralizing antibodies (but not HI or MN antibodies, data not shown) were detected in the plasma of all subjects and still at significantly higher titers in the H1N1-High than in the H1N1-Low subjects (median values 600 versus 300 for H1N1-High and H1N1-Low subjects, respectively; p < 0.005). Following the second dose of vaccine (day 43), no significant difference in H5 PNA neutralizing titers was observed between the two groups (median values 1300 versus 800 for H1N1-High and H1N1-Low subjects, respectively). The presence of low amounts of H5-neutralizing antibodies already at baseline suggested that at least part of the H5 cross-reactive MBCs detected at this time point could be directed against neutralizing epitopes. To verify this hypothesis, we performed the PNA assay on supernatants of MBC cultures (SN). At baseline these antibodies were present only in a minority of SN from either H1N1-High (14%) or Low (17%) subjects (Fig. 4B). However, after the first vaccination (day 21) 86% of H1N1-High subjects had MBCs producing detectable H5-neutralizing antibodies in culture as compared with 44% of H1N1-Low subjects (Fig. 4B). At day 43 the number of subjects with measurable H5 neutralizing antibodies in SN was similar in H1N1-High or Low subjects (79% versus 76%). Comparable results were obtained if considering the two vaccination groups (AT and TA) separately. Overall these data provide evidence that although human population is naive for the H5N1 virus, the exposure to seasonal H1N1 might predispose individuals to a more rapid response to the vaccination against H5 viruses.
Multiple studies [12-14, 17] suggest that B-cell clones cross-reactive to H1 and H5 antigens might circulate in adults’ blood at detectable frequencies. This observation raises the need of understanding their actual representation in larger cohorts of subjects, as well as their cross-reactive capacity in vivo. The aim of our study was to evaluate quantitatively and functionally MBCs cross-reactive to H5N1 and H1N1 influenza antigens circulating in the blood of healthy adults with no history of exposure to avian H5N1 influenza virus or vaccines. To this end, we measured the frequency of H5N1 and H1N1 MBCs in a cohort of healthy adults and analyzed whether the presence of H5N1-specific MBCs was significantly associated with that of H1N1-specific MBCs. Our results show that in resting conditions detectable frequencies of H5N1-IgG MBCs are present in the blood of almost half (45%) of healthy adults and correlate with the frequency of H1N1-MBCs. Remarkably, the detection of circulating H5N1-IgG MBCs at frequencies equal or above the threshold of 0.1% significantly associates with the concomitant presence of a relatively high number of H1N1-IgG MBCs, i.e. corresponding to a frequency equal or above 1% of total IgG MBCs. While we cannot formally exclude that the sLDA used to enumerate H5N1- and H1N1-IgG MBCs also detects a certain number of neuraminidase-specific MBCs, the fact that by sorting H1+ MBCs using a recombinant HA as bait we were able to isolate B cells also binding H5 supports the notion that H1/H5 cross-reactive cells were included in the H1+ MBC subset. Although it is not possible to rule out that among the sorted H1+MBCs there might be some secreting polyreactive antibodies, sorted cells do not recognize HSA. Moreover, if polyreactive MBCs would be a meaningful proportion of MBCs detected by sLDA assay a parallel expansion of the H5N1- and H1N1-MBC pools would be expected. Conversely, the diverse expansion of the H1N1- and H5N1-MBC pools in subjects receiving monovalent MF59-adjuvanted H5N1 vaccine as first dose suggests that this phenomenon, even if present, might be of limited relevance.
Taken together, our results demonstrate that MBCs cross-reactive to H5 circulate in the blood of healthy adults at frequencies associated with the size of the H1-specific MBC pool. Furthermore, our data suggest that repeated exposures to seasonal H1N1 influenza virus antigens (via natural infections and/or vaccinations) increase the probability that at least one of ten of the acquired H1N1-specific MBCs recognizes epitopes conserved across H5 and H1 molecules. In line with this hypothesis, we found that subjects with larger pools of H1N1-IgG MBCs developed faster immune responses to an MF59-adjuvanted H5N1 vaccine, as compared with subjects with lower numbers of H1N1-specific MBCs. Indeed, subjects with high baseline frequencies of H1N1-specific MBCs show an expansion of H5N1-specific MBCs already after the first dose of the monovalent MF59-adjuvanted H5N1 vaccine. In addition, by using the PNA we also found that, once reactivated in culture, at least part of the rapidly expanded H5N1-IgG MBCs produced H5 neutralizing antibodies, providing an advantage to subjects with high baseline frequencies of H1N1-specific MBCs. The vaccination regimen did not influence the results, since subjects with a larger pool of H1N1-specific MBCs at baseline had a higher percentage of MBCs producing detectable H5-neutralizing antibodies in culture, after receiving either one dose of the monovalent MF59-adjuvanted H5N1, or the adjuvanted H5N1 vaccine combined with the seasonal vaccine. Although H1N1/H5N1 cross-reactive MBCs do not necessarily produce neutralizing antibodies the presence of a large pool of H1N1-IgG MBCs was associated with detectable circulating H5-neutralizing antibodies at baseline and with a fast increase in the titer after the first vaccine dose. Due to the particular vaccination schedule, this study does not allow to determine whether H1N1-Low subjects truly require a boost to reach titers equivalent to H1N1-High subjects or if they only experience a slower response. Further studies will be needed to evaluate whether the kinetic advantage shown by H1N1-High individuals in developing H5-neutralizing antibodies actually translates into a better protection.
Previous studies also reported that subjects with no known history of exposure to the H5N1 virus had H5-binding antibodies able to neutralize H5 pseudoparticles . Further characterization at clonal level of baseline MBCs would add information on their capability to produce H1/H5-neutralizing antibodies, however, this study design did not allow to perform such in-depth analysis. Besides the benefit conferred by the presence of a large H1N1-IgG MBC pool at baseline, we observed that also the presence of detectable H5N1-IgG MBCs had an impact on the response to H5N1 vaccination. Indeed, in our cohort of 60 subjects, 27 had measurable frequencies of H5N1-IgG MBCs at baseline. The presence of detectable H5N1-IgG MBCs at baseline associated with a larger expansion of the H5N1-MBC pool at day 21 (p = 0.09) and with a significant expansion at day 43 postvaccination (p = 0.04). It also associated with the presence of a larger pool of H1N1-MBCs at baseline (p = 0.0018), but not with a larger expansion of H1N1-MBCs after vaccination (Supporting Information Fig. 1). This observation is not unexpected, since subjects are not naive for H1N1 at baseline, due to previous contacts with H1N1 antigens via natural influenza infections and/or vaccinations. Therefore the response to the seasonal vaccination mainly depends on the baseline frequency of H1N1-MBCs, and should be less influenced by the baseline frequency of H5/H1 cross-reactive cells, representing only a subset of H1N1-MBCs. On the contrary, being subjects naive to H5N1, the contribution of H1/H5 cross-reactive cells at baseline becomes more relevant in the response to H5N1 vaccination. We also analyzed the response in terms of fold increase in the frequencies of H5N1- and H1N1-MBCs after vaccination, above the baseline frequencies in H1N1-High and Low subjects. H1N1-High and H1N1-Low subjects exhibit comparable expansion of H5N1-MBCs (Supporting Information Fig. 2, panel A). The fact that both H1N1-High and H1N1-Low groups show comparable expansion of H5N1-MBCs following vaccination supports the notion that both groups are likely to harbor MBCs specific for epitopes conserved across H1N1 and H5N1 influenza antigens. However, the absolute numbers of these cells would be related to the baseline levels of H1N1-MBCs. Therefore the higher frequencies of H5N1-MBCs observed in H1N1-High subjects after vaccination may not be due to an enhanced response experienced by these subjects but to the higher baseline absolute numbers of H5N1 cross-reactive cells. Conversely, a clear inverse relationship was found between the fold increase of H5N1-MBCs at day 21 over baseline and the baseline frequencies of H5N1-MBCs, as well as between the fold increase of H1N1-MBCs and the baseline frequencies of H1N1-MBCs (Supporting Information Fig. 2, panels B and C).
It is known that most of the epitopes conserved across different HA subserotypes map in the HA2 region, and that, aside from the few targeted by antibodies that freeze the fusion peptide in its not functional conformation, most of these epitopes are not neutralizing [12, 13, 22-24]. Using phage-display libraries of H5N1 and of pandemic H1N1 antigens, prevaccination sera from healthy adults and children have been shown to contain antibodies that predominantly bind to the HA2 region of HA [24, 25]. Interestingly, in those same studies it was also shown that MF59-adjuvanted vaccines are particularly efficient both at inducing affinity maturation of preexisting antibodies and at expanding the repertoire of epitopes recognized by HA-specific antibodies [24, 25]. Since the vaccines administered in this study were given in presence of MF59, it is reasonable to assume that similar processes also occurred in the population of subjects analyzed here. Further studies specifically designed for mapping clonal relationships and fine specificities of representative numbers of H1-specific MBC clones obtained before and after vaccination with H5N1 influenza vaccine will be necessary to verify whether the enhanced B-cell responses to H5 observed in subjects with a large number of H1N1-MBCs is mostly driven by H1/H5 cross-reactive MBCs already present at baseline or by newly generated clones. Also, a finer definition of the binding specificity toward the H5 stalk or globular head region would represent an important piece of information. However, such deeper analysis was not feasible due to the design of the study and to the lack of sufficient volume of blood.
To our knowledge, this is the first report providing quantitative information on the actual representation of H1/H5 cross-reactive MBCs in a large population of healthy adults and on their range of frequencies. Moreover, the results from this study provide additional information on concomitant vaccination against seasonal and H5N1 influenza. Previous reports from clinical studies already showed no interference in safety and seroconversion rates following this strategy of vaccination . Our data provide evidence that such strategies might be also effective at priming the population's immune memory against a broad variety of influenza serotypes and reduce the risk of new pandemics. Clearly, further studies are required to understand why H5-cross-reactive MBCs do not develop in all subjects who have been exposed to seasonal H1N1 and developed large H1N1-MBC pool. Nevertheless, the data presented provide clear evidence that a strong preexisting immunity to seasonal H1N1 influenza viruses, acquired via previous vaccinations and/or natural influenza infections, can favor the formation of a pool of potentially cross-reactive MBCs that could be rapidly expanded upon pandemic vaccination. In this respect, regular yearly vaccinations with seasonal influenza vaccine can offer substantial benefits against future pandemics.
Materials and methods
Subjects and vaccinations
This was a randomized, placebo-controlled, observer-blind, multicenter, phase II study in a population of subjects aged 18 years and above. This study was conducted in accordance with the latest Declaration of Helsinki guidelines and German regulatory requirements. The protocol received the approval of the local ethical authorities and was registered with the ClinicalTrials.gov Identifier: NCT00620815. All volunteers provided written informed consent.
The details of this study and the complete set of results including safety and antibody responses will be reported elsewhere (Fragapane et al., unpublished observation). Briefly, 601 subjects were enrolled and randomized to receive the seasonal inactivated influenza vaccine consisting of A/Solomon Islands/3/2006 (H1N1), A/Wisconsin/67/2005 (H3N2), and B/Malaysia/2506/2004 (15 micrograms each) plus the H5N1 A/Vietnam/1194/04 vaccine (7.5 micrograms) adjuvanted with MF59. At the first visit enrolled subjects received the two vaccines either given concomitantly in the same arm (tetravalent vaccine: T) or in different arms. An additional group received monovalent MF59-adjuvanted H5N1 vaccine alone (A). At a second visit, 21 days later, these groups received a second vaccination with either MF59-adjuvanted H5N1 or tetravalent. All vaccines were produced at Novartis Vaccines and Diagnostics, Siena, Italy. For the purpose of the present study, PBMCs were obtained from 60 subjects representing a subcohort of 30 subjects/group, randomly chosen with an age range from 18 to 58 years. Of these subjects, 30 received one dose of the tetravalent influenza vaccine and at day 21 they received one dose of MF59-adjuvanted H5N1 (T/A group). Other 30 subjects received one dose of MF59-adjuvanted H5N1 and at day 21 one dose of the tetravalent vaccine (A/T group).
Enumeration of H5N1- and H1N1-specific MBCs
MBCs were determined by a previously described the ELISA-coupled limiting dilution assay . Briefly, PBMCs were plated in 0.2 mL of RPMI with 5% FBS in serial twofold dilutions, six replicates per dilution, starting from 8 × 105 PBMCs/well, in 96-well U-bottom plates containing 2.5 μg/mL of a phosphorothioate CpG oligonucleotide (tcg tcg ttt tgt cgt ttt gtc gtt; Primm, Milan, Italy) and 1000 units/mL rhIL-2 (Proleukin, Novartis Pharmaceuticals, Basel, Switzerland). Parallel control cultures of PBMCs were run in medium alone. On day ten individual supernatants were collected and kept at –20°C until tested in ELISA for their content in H5N1- and H1N1-specific and total IgG. The antigens used for coating were: H5N1 (A/Vietnam/1194/2004) 5 μg/mL in PBS, pH 7.5; H1N1 (A/Solomon Islands/3/2006, both from Novartis Vaccines and Diagnostics, Siena, Italy) 5 μg/mL in PBS, pH 7.5; or a polyclonal sheep IgG fraction against whole human IgG (Sigma, St. Louis, MO, US) (2 μg/mL in PBS, pH 7.5). Wells displaying at 405 nm an optical density (OD) ≥0.4 (total IgG) or an OD ≥0.45 (H5N1-IgG and H1N1-IgG) (fivefold higher than the blank OD) were considered positive. Frequencies of H5N1-IgG secreting cell precursors (H5N1-IgG MBCs) and H1N1-IgG secreting cell precursors (H1N1-IgG MBCs) were expressed as percentage of the total IgG MBC precursors measured.
Pseudoparticle neutralization assay
Plasma samples were tested by PNA as previously described starting from a dilution 1:100, for 5 1:3 dilutions in duplicates . Neutralization titers were determined as the dilution required to obtain 80% reduction in relative light units compared with control wells with virus alone.
Supernatants from PBMCs (2 × 106/mL) activated with 2.5 μg/mL CpG and 1000 IU/mL of IL-2 for 10 days in vitro were also tested in PNA assay in duplicate in serial 1:2 dilutions (starting from 1:2.5). Supernatants that, at a dilution ≥2.5 or above, caused 50% inhibition of pseudoparticles infectivity were assigned a positive score. Medium with equal amount of CpG and IL-2 but not cells were used as negative control and never caused inhibition of pseudoparticles infectivity.
Staining and sorting of H1+MBCs and H5+MBCs
H1+ and H5+ MBCs were stained and sorted from PBMCs of healthy donors following a protocol developed in our laboratory (Bardelli et al., unpublished observation). Briefly 10 × 106 PBMCs were stained with live-dead (Molecular Probes, Eugene, OR, USA), anti-CD20 PrCPCy5.5 and anti-CD27PE mAbs (BD Biosciences, Milan, Italy) and 0.3 μg of H1 A/Solomon Island/03/2006 or H5 A/Vietnam/1203/2004 (Protein Sciences, Meridien, CT) or HSA labeled with Alexa-647 (Molecular Probes, Eugene, OR) following manufacturer instructions. Cells were stained in 100 μL volumes for 1 h at 4°C, washed, and resuspended in PBS 0.1% FCS. Cells were sorted by FACS Aria (BD Biosciences, San Jose, CA, USA). Antigen-specific B cells, sorted as H1+ or H5+ among CD20+ cells, were cultured with non-B cells, sorted as CD20− cells from the same sample, in 1:50 ratio in RPMI complete medium added with rhIL-2 (1000 UI/mL) and CpG (2.5 μg/mL) for 5 days in 96-well plates.
ELISpot: ELISpot plates (Millipore MultiScreen HTS-HA Billerica, MA, USA) were coated with 100 μL/well of PBS containing HSA, H5N1 A/Vietnam/1203/2004, H1N1 A/Solomon Island/3/2006 at 10 μg/mL or 2.5 μg/mL of purified anti-human IgG (BD Pharmingen, NJ, USA), for 16–20 h at 4°C O/N. Nonspecific binding sites were blocked with 200 μL of PBS containing 1% dried skimmed milk for 2 h at room temperature. Activated cells were plated in equal numbers in duplicate wells O/N. IgG secretion was revealed with biotinylated anti-hIgG (Southern Biotech, Birmingham, AL), followed by extravidine-HPR (TEMA Ricerca, Bologna, Italy) and developed with the AEC chromogen (Sigma). Spots were counted using the ImmunoSpot Series 5 UV Analyzer (CTL Europe, Bonn, Germany).
All the statistical analyses were performed using the JMP 8.0.1 software. The existence of correlation between log10-transformed values of H1N1-IgG MBC and H5N1-IgG MBC frequencies at baseline was analyzed by the Spearman test for not parametric correlations; a p value of 0.05 was considered significant. The Fisher exact test was applied to verify the existence of a significant association between the presence of circulating H5N1-IgG MBCs at frequency ≥0.1% of total IgG MBCs and the concomitant presence of H1N1-IgG MBCs at frequency ≥1% of total IgG. A two-tailed p-value = 0.05 was considered significant. A nonparametric Wilcoxon/Kruskal–Wallis test was performed to compare the differences between means of antigen-specific MBCs and H5-PNA plasma titers in H1N1-High and H1N1-Low subjects, as well as the differences in fold increase of H5N1-IgG MBCs between H1N1-High and H1N1-Low subjects. A p-value = 0.05 was considered significant. Differences in the proportion of MBC cultures scoring positive for the presence of H5-neutralizing antibodies between H1N1-High and H1N1-Low subjects were analyzed by the Fisher exact test; a two-tailed p-value = 0.05 was considered significant.
We thank Rino Rappuoli and Philip Dormitzer, Novartis Vaccines and Diagnostics, for helpful discussion and insightful comments; Fabio Rigat, Novartis Vaccines, and Diagnostics, for the invaluable statistical help; Giorgio Corsi, Novartis Vaccines, and Diagnostics, for artwork.
Conflict of interest
All authors are full time employees of Novartis Vaccines and Diagnostics.