Research ArticleVaccines

Influenza vaccines differentially regulate the interferon response in human dendritic cell subsets

See allHide authors and affiliations

Science Translational Medicine  22 Mar 2017:
Vol. 9, Issue 382, eaaf9194
DOI: 10.1126/scitranslmed.aaf9194

Influenz-ing interferon responses in dendritic cells

Seasonal influenza vaccines have been produced and marketed for decades but are not always protective. Athale et al. tested a trivalent vaccine that outperformed the monovalent vaccine made by the same manufacturer for the ability to activate human dendritic cell subsets, which are crucial for launching adaptive immune responses. They discovered that both vaccines could activate plasmacytoid dendritic cells, but only the trivalent vaccine could induce antiviral interferon responses in other types of dendritic cells. Moreover, people immunized with the monovalent vaccine did not show early interferon responses in the blood, which were induced by trivalent vaccination. These intriguing results may help explain vaccine underperformance that is not due to antigenic mismatch.

Abstract

Human dendritic cells (DCs) play a fundamental role in the initiation of long-term adaptive immunity during vaccination against influenza. Understanding the early response of human DCs to vaccine exposure is thus essential to determine the nature and magnitude of maturation signals that have been shown to strongly correlate with vaccine effectiveness. In 2009, the H1N1 influenza epidemics fostered the commercialization of the nonadjuvanted monovalent H1N1 California vaccine (MIV-09) to complement the existing nonadjuvanted trivalent Fluzone 2009–2010 vaccine (TIV-09). In retrospective studies, MIV-09 displayed lower effectiveness than TIV-09. We show that TIV-09 induces monocyte-derived DCs (moDCs), blood conventional DCs (cDCs), and plasmacytoid DCs (pDCs) to express CD80, CD83, and CD86 and secrete cytokines. TIV-09 stimulated the secretion of type I interferons (IFNs) IFN-α and IFN-β and type III IFN interleukin-29 (IL-29) by moDC and cDC subsets. The vaccine also induced the production of IL-6, tumor necrosis factor, and the chemokines IFN-γ–inducible protein 10 (IP-10) and macrophage inflammatory protein–1β (MIP-1β). Conversely, MIV-09 did not induce the production of type I IFNs in moDCs and blood cDCs. Furthermore, it inhibited the TIV-09–induced secretion of type I IFNs by these DCs. However, both vaccines induced pDCs to secrete type I IFNs, indicating that different influenza vaccines activate distinct molecular signaling pathways in DC subsets. These results suggest that subtypes of nonadjuvanted influenza vaccines trigger immunity through different mechanisms and that the ability of a vaccine to induce an IFN response in DCs may offset the absence of adjuvant and increase vaccine efficacy.

INTRODUCTION

Vaccination is the most efficient way to protect humans against influenza. Several influenza vaccines are currently available, including split subunit vaccines and attenuated viral vaccines (1). Split vaccines, which are most commonly used, are prepared by culture of the influenza virus in hen eggs or cell cultures, followed by purification and treatment with detergents (2). In the United States, split vaccines are used without adjuvants, whereas in Europe, some vaccines contain adjuvants such as MF59 or AS03 (3, 4). Because the composition of influenza vaccines varies annually on the basis of results from global influenza surveillance data, the determination of vaccine effectiveness remains a continuous challenge.

The need for rapid implementation of a vaccination program during the outbreak of the pandemic influenza A(H1N1)pdm09 virus in 2009 led initial studies to focus on immunogenicity rather than overall efficacy of the vaccine. Several influenza vaccines were then marketed, including the trivalent nonadjuvanted seasonal flu vaccine Fluzone 2009–2010 (hereafter termed as TIV-09) containing A/Brisbane/59/2007 (H1N1)–like, A/Brisbane/10/2007 (H3N2)–like, and B/Brisbane/60/2008 viral strains, and the monovalent H1N1 vaccine (hereafter termed as MIV-09) containing A/California/7/2009 (H1N1)–like virus. The monovalent vaccine licensed in Europe and Canada was adjuvanted, whereas the vaccine licensed in Australia and the United States was not adjuvanted. Subsequent vaccine trials performed with TIV-09 or the adjuvanted forms of MIV-09 demonstrated efficacy (59). In contrast, the nonadjuvanted form of MIV-09 displayed lower effectiveness, low seroprotection in healthy children (35%), and suboptimal response in solid organ transplant recipients and children with systemic lupus erythematosus (913). It is unclear why nonadjuvanted TIV-09 and MIV-09, two vaccines produced by the same manufacturer (Sanofi Pasteur), display such disparate efficacy levels.

Traditionally, vaccine efficacy has been measured by hemagglutinin-inhibiting antibody response as a correlate of humoral immunity (14). Some studies suggested that assessment of T cell responses might provide a better correlate of vaccine protection against influenza, especially in the elderly (15). Other studies suggested that influenza vaccines might differentially activate subpopulations of immune cells (16, 17). We and others demonstrated that administration of seasonal influenza vaccines to healthy volunteers induces global transcriptional changes in whole blood that can be followed at discrete time points (1823). In particular, a transcriptional signature of interferon (IFN)–inducible genes was observed between 12 and 36 hours after vaccination, followed by a plasmablast signature at day 7 (23). This early signature may represent an important correlate of vaccine efficacy because type I IFNs enhance antigen cross-presentation to CD8+ T cells, foster T helper 1 differentiation, support B cell differentiation into antibody-producing plasmablasts, and induce dendritic cell (DC) maturation (2426). DCs play an essential role in vaccination by detecting and presenting foreign antigens to adaptive immune cells (27, 28). We recently showed that different DC subsets, including monocyte-derived DCs (moDCs) and blood conventional DCs (cDCs), respond differently to various bacterial and viral vaccines (29).

To better understand the biological mechanisms underlying the difference in the effectiveness of nonadjuvanted TIV-09 and MIV-09, we compared their ability to activate human DC subsets. Our results suggest that, in the absence of adjuvant, TIV-09 activates a broader range of DC subsets than MIV-09 in a type I IFN–dependent manner, possibly explaining the increased efficacy of the former vaccine.

RESULTS

TIV-09, but not MIV-09, activates moDCs and induces type I IFN secretion

To assess the effects of TIV-09 and MIV-09 on DC activation, moDCs were obtained by culturing isolated human monocytes in the presence of granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4) for 6 days and were subsequently exposed to each vaccine for 18 hours. TIV-09 up-regulated the surface expression of hallmark DC maturation markers CD40, CD80, CD83, CD86, human leukocyte antigen (HLA)–DR, programmed death–ligand 1 (PD-L1), PD-L2, and CCR7 (n = 7; Fig. 1A). Compared to TIV-09, MIV-09 induced significantly (P < 0.01) lower expression of these markers. moDC activation was further assessed by measuring the release of cytokines at different time points after vaccine challenge. IFN-β reached its peak level at 12 hours and decreased to near baseline values after 48 hours, suggesting its consumption. Macrophage inflammatory protein–1β (MIP-1β) reached its peak at 12 hours and slightly decreased thereafter. Other cytokines such as IFN-α, IL-29, IL-6, and IFN-γ–inducible protein 10 (IP-10) attained peak levels at 12 to 24 hours and remained stable at 48 hours. Conversely, MIV-09 did not induce cytokine secretion, with the exception of MIP-1β (n = 10; Fig. 1B). Because TIV (45 μg) contains three times more viral protein than MIV-09 (15 μg), we conducted a dose-dependent analysis. Even at high concentration (greater than fourfold), MIV-09 failed to induce the secretion of IFN-β and IP-10 by moDCs, suggesting that protein concentration cannot explain the observed differences (fig. S1). Staining with PI and annexin V at 4, 8, 12, and 24 hours also revealed that the lack of cytokines observed with MIV-09 was not due to vaccine-induced cell death (n = 6; Fig. 1C). We further assessed IP-10 production by moDCs in the presence of other inactivated viral vaccines such as human papillomavirus (Merck), hepatitis A and hepatitis B [GlaxoSmithKline (GSK)], and rabies (Sanofi Pasteur) (table S1). Only TIV-09 induced secretion of IP-10 by moDCs (fig. S2).

Fig. 1. TIV-09, but not MIV-09, activates moDCs.

(A) moDCs were stimulated with TIV-09 (6 μl/ml), MIV-09 (6 μl/ml), or media for 18 hours. Surface expression of moDC activation markers was assessed by flow cytometry (n = 7). DC-SIGN, DC-specific intercellular adhesion molecule grabbibng nonintegrin. (B) moDCs were cultured in the presence of TIV-09 or MIV-09 (6μl/ml) for 0, 6, 12, 24, and 48 hours. Supernatants were harvested for quantification of cytokines by Luminex (n = 10). Statistics shown represent the comparison between TIV-09 and MIV-09 at each time point. (C) moDCs were stained with annexin V and propidium iodide (PI) to examine cell death at 4, 8, 12, and 24 hours after vaccine treatment (n = 6). (D) Transcriptional analysis of IFNs and other cytokines in moDCs cultured with TIV-09 or MIV-09 for 6 hours. Data were normalized to medium control (n = 6) (*P < 0.05; **P < 0.01; ***P < 0.001, Mann-Whitney U test).

Gene expression profiling was performed by microarray on moDCs from six healthy individuals cultured with TIV-09 or MIV-09 for 6 hours. TIV-09 up-regulated all known IFN-α transcripts, IFN-β, IFN-ω, and IL-29 (n = 6; Fig. 1D). Consistent with the protein data, MIV-09 failed to increase the transcription of type I and type III IFN genes. In addition, TIV-09 up-regulated the expression of other proinflammatory and regulatory cytokines, including tumor necrosis factor (TNF), IL-6, IL-1β, and IL-10, which were also slightly up-regulated in response to MIV-09. Together, these results indicate that the seasonal influenza vaccine TIV-09, but not MIV-09, potently activates moDCs, including induction of type I and type III IFNs.

Cell-intrinsic type I IFN production contributes to moDC activation

Upon activation, several cell types first transcribe the IFN-β gene, which results in the prompt secretion of IFN-β. IFN-β then binds to the IFN-α receptor (IFNAR), consequently initiating IFN-α production (30, 31). To determine whether the early secretion of IFN-β by moDCs might contribute to their further activation, we exposed moDCs to TIV-09 in the presence of a blocking anti–IFN-α/βR2 monoclonal antibody. Activation was assayed by measuring the secretion of the IFN-inducible cytokine IP-10. The addition of anti–IFN-α/βR2 antibody abolished IP-10 production and strongly reduced production of IFN-α and IL-29 without affecting that of IFN-β (n = 4; Fig. 2A). In addition, IFNAR blockade resulted in the up-regulation of IL-1β, TNF, and IL-10, suggesting an autocrine inhibitory loop (32). The addition of anti–IFN-α/βR2 antibody also partially inhibited the up-regulation of CD80, CD83, CD86, CD70, and CD40 (n = 4; Fig. 2B), suggesting that moDC maturation is dependent on type I IFN secretion. These data indicate that the early burst of IFN-β production contributes to TIV-09–induced activation of moDCs.

Fig. 2. TIV-09 is a strong inducer of type I IFNs.

(A) moDCs were treated with increasing doses of TIV-09 with or without anti-IFNAR antibody (10 μg/ml) for 8 hours. Supernatants were assessed by Luminex for the presence of type I IFNs and inflammatory cytokines. Statistics shown represent the comparison between TIV-09 and TIV-09 with IFN-α/βR2 antibody (Ab) at each concentration (*P < 0.05; **P < 0.01; ***P < 0.001, Mann-Whitney U test). IgG, immunoglobulin G; LPS, lipopolysaccharide. (B) moDCs were stimulated with TIV-09 (6 μl/ml) with or without anti-IFNAR antibody for 18 hours. Surface expression of DC markers was assessed by flow cytometry. The figure is representative of experiments in four independent donors.

MIV-09 inhibits the IFN response in moDCs

We next analyzed the effects of adding the H1N1 strain A/California/07/2009 X-179A in vaccines by testing several Fluzone batches prepared over different years on moDCs. Fluzone vaccines from years 2006–2007, 2008–2009, and 2009–2010, composed of three closely related influenza strains (see Methods), induced moDCs to secrete similar levels of IP-10 (n = 10; Fig. 3A). However, Fluzone 2010–2011, which contains the H1N1 strain A/California/07/2009 X-179A, failed to induce IP-10 secretion by moDCs.

Fig. 3. MIV-09 inhibits type I IFNs.

(A) moDCs were treated with increasing doses of influenza vaccines used during the 2006–2007, 2008–2009, 2009–2010, and 2010–2011 seasons. The graph summarizes data from 10 donors. (B) moDCs were treated with TIV-09 (25 μl/ml) combined with increasing doses of MIV-09 for 8 hours. Supernatants were harvested for enzyme-linked immunosorbent assay (ELISA). The graph summarizes data from seven donors. (C) moDCs were activated with TIV-09 (6 μl/ml), LPS (50 ng/ml), CD40L (1 μg/ml), R848 (3 μg/ml), or CL097 (5 μg/ml) alone or with MIV-09 (10 μl/ml) with or without recombinant IFN-α (2000 U/ml). Supernatant cytokine concentrations were assessed by Luminex. The graph summarizes data from three donors. (D) moDCs were stimulated with TIV-09, LPS, CD40L, R848, or CL097 with or without MIV-09. The surface expression of moDC activation markers was examined by flow cytometry. The figure is representative of experiments in four independent donors. (E) IPA of the transcripts unique to each vaccine and most highly expressed after vaccine exposure. The top three canonical pathways enriched in the transcripts induced by each vaccine are shown (n = 3) (***P < 0.001, Mann-Whitney U test). eIF2, eukaryotic initiation factor 2; mTOR, mechanistic target of rapamycin; PRRs, pattern recognition receptors; JAK1, Janus kinase 1; TYK2, tyrosine kinase 2.

To determine whether MIV-09 could inhibit moDC activation, we activated moDCs with TIV-09 or selected Toll-like receptor (TLR) agonists, including LPS (TLR4 ligand), R848 (TLR7/8 ligand), CL097 (TLR7/8 ligand), and CD40L, with or without MIV-09. At a concentration of 25 μg/ml, MIV-09 not only completely abrogated the secretion of IP-10 induced by TIV-09 but also the secretion of other IFN-related cytokines, including IFN-β, IL-29, and IL-6 [Fig. 3, B (n = 7) and C (n = 3)]. MIV-09 also inhibited the secretion of cytokines by moDCs activated by CD40L and all TLR agonists tested. Because MIV-09 could inhibit the early IFN-β response, we wondered whether the addition of exogenous recombinant IFN-α might reverse the inhibitory effects of MIV-09 on the secretion of other cytokines, which was not the case (n = 7; Fig. 3C). MIV-09 also prevented the expression of CD80, CD83, CD86, CD70, and CD40 induced by TIV-09 and all DC-activating stimuli tested (n = 4; Fig. 3D). These results indicate that MIV-09 acts as a potent inhibitor of cytokine secretion and maturation in activated moDCs.

To further understand the molecular alterations induced by both vaccines at the pathway level, we conducted Ingenuity Pathway Analysis (IPA) on the transcriptional fingerprints of vaccine-treated cells (Fig. 3E). The top pathways associated with TIV-09 exposure included IFN signaling (yellow and blue), protein ubiquitination (blue), IFN regulatory factor (IRF) signaling (yellow), and retinoic acid–mediated signaling (yellow). These pathways were suppressed when moDCs were exposed to MIV-09. On the other hand, MIV-09 enhanced the extracellular signal–regulated kinase (ERK)/mitogen-activated protein kinase (MAPK) signaling pathway (purple), which can inhibit IFN responses (33). These results suggest that MIV-09 inhibits IFN signals and moDC maturation by skewing the transcriptional cascade of vaccine-exposed moDC toward ERK-mediated proinflammation.

TIV-09 elicits a type I IFN signature in blood cDCs

Because moDCs and blood cDCs differ phenotypically and functionally, we next assessed the effects of TIV-09 and MIV-09 on sorted ex vivo blood cDCs. As observed in moDCs, the surface expression of CD80, CD83, CD86, and HLA-DR was increased on cDCs exposed to TIV-09 for 18 hours. Exposure to MIV-09 was associated with lower induction of CD80, CD83, and CD86 (n = 4; Fig. 4A). Although cDCs express significantly lower levels of type I IFNs than pDCs (34), cDCs exposed to TIV-09, but not to MIV-09, displayed up-regulation of both IFN-inducible transcripts and IP-10 secretion [Fig. 4, B (n = 3) and C (n = 6)]. As observed with moDCs, MIV-09 prevented the TIV-09–dependent secretion of IP-10 by cDCs (n = 6; Fig. 4D) and down-regulated TIV-09–induced surface expression of CD83, CD86, and HLA-DR (n = 4; Fig. 4E). These data suggest that MIV-09 also blocks IFN signals in ex vivo cDCs.

Fig. 4. TIV-09 activates cDCs.

(A) Sorted blood cDCs were stimulated in vitro with media alone, TIV-09, or MIV-09 for 18 hours. Surface expression of activation markers was examined by flow cytometry. The figure is representative of experiments in four independent donors. (B) Transcriptional analysis of IFNs in cDCs stimulated with TIV-09 or MIV-09 for 6 hours. Data were normalized to media control (n = 3). (C) cDCs were activated in vitro with media, TIV-09, or MIV-09 for 0, 6, 24, and 48 hours. IP-10 levels were measured by ELISA. The graph summarizes data from six donors. (D) cDCs were treated with TIV-09 (6 μl/ml) combined with different doses of MIV-09 for 8 hours. Supernatants were harvested, and IP-10 was measured by ELISA. The graph summarizes data from six donors. (E) cDCs were stimulated with TIV-09 in the presence or absence of MIV-09. Surface expression of CD80, CD86, CD83, and HLA-DR was measured by flow cytometry. The figure is representative of experiments in four independent donors (*P < 0.05; Mann-Whitney U test).

Both vaccines elicit a type I IFN signature in blood pDCs

Plasmacytoid dendritic cell (pDCs) promptly secrete large amounts of type I IFNs in response to a number of stimuli including viruses (35). Therefore, we exposed pDCs to TIV-09 or MIV-09 for 18 hours and assessed the expression of activation markers flow cytometry and IFN transcription by microarray. Both vaccines up-regulated CD80, CD86, and HLA-DR expression, with TIV-09 being consistently more potent than MIV-09 (n = 4; Fig. 5A). Transcriptional analysis revealed that both vaccines up-regulated type I IFN transcripts (n = 3; Fig. 5B). MIV-09 was less efficient than TIV-09 at inducing the secretion of IP-10 and IFN-β [Fig. 5, C (n = 6) and D (n = 6)]. When the two vaccines were combined, MIV-09 suppressed TIV-09–induced IP-10 production by about 70% (n = 6; Fig. 5E) and down-regulated TIV-09–induced surface expression of CD83 (n = 4; Fig. 5F).

Fig. 5. TIV-09 activates pDCs.

(A) Sorted plasmacytoid DCs (pDCs) were activated in vitro with TIV-09 or MIV-09 or control media for 18 hours, and surface expression of activation markers was measured by flow cytometry. The figure is representative of experiments in four independent donors. (B) Transcriptional analysis of IFNs in pDCs stimulated with TIV-09 or MIV-09 for 6 hours. Data were normalized to medium control (n = 3). (C and D) pDCs were activated in vitro with media, TIV-09, or MIV-09 for 0, 6, 24, and 48 hours. IP-10 and IFN-β levels were measured by ELISA. (E) pDCs were treated with TIV-09 (6 μl/ml), combined with different doses of MIV-09 for 8 hours. Supernatants were harvested, and IP-10 levels were measured by ELISA. Figure 5 (C to E) summarizes data from six donors. (F) pDCs were stimulated with TIV-09 in the presence or absence of MIV-09. Surface expression of CD80, CD86, CD83, and HLA-DR was measured by flow cytometry. The figure is representative of experiments in four independent donors (*P < 0.05; Mann-Whitney U test).

MIV-09 fails to induce a type I IFN signature in vivo

To better understand the early immune response to influenza vaccination in vivo, we vaccinated healthy individuals with TIV-09, MIV-09, or saline control and assessed the transcriptional profiles of whole blood at 0, 1.5, 3, 6, 9, 12, 15, 24, and 36 hours and 2, 3, and 7 days after vaccination by microarray. Data were analyzed both at the single-transcript level and using a previously described framework of modules of coexpressed transcripts, focusing on three IFN modules (36). Consistent with previous studies from our group (18), as well as others (1923), TIV-09 induced a transient type I IFN signature 24 to 48 hours after vaccination. However, this signature was not induced in response to MIV-09 (Fig. 6). These results indicate that MIV-09 fails to induce an IFN response in vivo.

Fig. 6. TIV-09, but not MIV-09, induces an IFN signature in the blood of vaccinated individuals at day 1.

Healthy individuals were vaccinated at t = 0 hour with saline, TIV-09, or MIV-09 (n = 3 per group), and blood was drawn at 0, 1.5, 3, 6, 9, 12, 15, 24, and 36 hours and 2, 3, and 7 days. Data were normalized to the median of the t = 0 hour samples across donors. Transcripts from three IFN blood modules (M1.2, M3.4, and M5.12) (36) were selected for analysis. Both transcript- and module-level analyses are displayed. Module expression is calculated as the percentage of transcripts from a specific module that are over- or underexpressed under a specific condition as compared to the t = 0 hour reference control.

DISCUSSION

This study focuses on how nonadjuvanted influenza vaccines that display different effectiveness affect the biological functions of human DC subsets. We first show that the Fluzone vaccines, which do not contain the H1N1 strain A/California/07/2009 X-179A, activate moDCs and blood cDCs as revealed by secretion of type I IFNs, IL-6, IL-29, and MIP-1β and increased surface expression of several DC activation markers. Blocking IFN signaling with an anti-IFNAR antibody further revealed the partial dependence of this activation on an early wave of IFN-β production by DCs. TIV-09 also induced pDCs to transcribe and secrete type I IFNs. We then show that the monovalent nonadjuvanted MIV-09, which displayed limited clinical efficacy, (i) did not induce an IFN response and maturation of moDCs and cDCs and (ii) blocked the IFN-mediated activating effects of TIV-09 in vitro and IFN signaling in vivo. These results suggest that, in the absence of adjuvant, certain vaccines, but not others, retain immunogenic properties through IFN-mediated DC activation.

These findings might partly explain why in vivo administration of TIV-09 results in the short-term expression of an IFN signature in circulating blood cells. Type I IFNs play a pivotal role in the induction of adaptive immune responses by promoting the expansion of CD4+ and CD8+ T cells and the maturation of B cells into plasmablasts (24, 26). High vaccine responses correlate with increased early expression of IFN signaling and antigen processing and presentation (21). Furthermore, type I IFNs act as powerful adjuvants when administered with influenza vaccine to mice (37). Thus, one could speculate that the antibody response elicited against TIV-09 might at least, in part, be due to this early IFN response.

On the other hand, MIV-09 was not able to induce moDC and blood cDC activation as measured by secretion of cytokines, including type I IFNs, and induction of activation markers. Moreover, MIV-09 acted as a powerful inhibitor of the effects of TIV-09 on the production of cytokines, including IFN and induction of activation markers by moDCs and cDCs. Instead, exposure to MIV-09 was associated with a unique transcriptional signature in moDCs that was enriched for ERK/MAPK signaling. Studies with murine macrophages indicate that IFN-β production is inhibited by ERK activation (33), thereby providing a plausible mechanism for the IFN-inhibitory activity of MIV-09 in moDCs and blood cDCs. The inability of MIV-09 to completely shut down the IFN machinery in pDCs may be due to constitutive expression of IRF7 in these cells (38). Moreover, the combination of MIV-09 with TIV-09 results in production levels of IP-10 that are lower than those induced by TIV-09.

Comparative studies between TIV-09 and the 23-valent pneumococcal vaccine Pneumovax 23 demonstrated that the two vaccines induce protective antibody responses but differ in the blood transcriptional profiles they elicit (18). When assessing IP-10 production by moDCs in the presence of other inactivated viral vaccines including human papillomavirus, hepatitis A, hepatitis B, and rabies (table S1), only TIV-09 induced secretion of IP-10 by moDCs (fig. S2). Most seasonal influenza vaccines have been shown to exhibit early IFN signatures (18, 2123), suggesting that type I IFN plays an important role in shaping protective immunity against influenza. Our study shows that, unlike seasonal influenza vaccines, MIV-09 fails to exhibit the classic type I IFN signature in vitro and in vivo.

Higher effectiveness has been observed in MIV-09 supplemented with squalene-based adjuvants MF59 and AS03, as compared with its nonadjuvanted form (3, 4). Several observations support a direct effect of MF59 on DCs. MF59 rapidly localizes to intracellular DEC-205+ DC compartments after injection in mice (39), promotes recruitment of human myeloid cells to the injection site, enhances DC maturation and antigen uptake, and facilitates DC migration to draining lymph nodes (40). A recent study in mice also demonstrated that MF59 promotes the retention of unprocessed antigen within the lymph nodes, thereby facilitating antigen encounters with follicular DCs (41). Whether these two adjuvants can induce the IFN response in DCs is still unknown.

Because of the presence of broadly cross-reactive antibody responses as a result of prior vaccination or infection (42, 43), it will be challenging to determine whether the lack of in vitro and in vivo activation of type I IFN signature reflects the reduced immunogenicity of MIV-09 as compared to TIV-09. Vaccines against pneumococcus or rabies are fully immunogenic in vivo despite lacking interferogenic properties in DC in vitro (29). These observations suggest the existence of IFN-dependent and IFN-independent mechanisms for mounting immune responses against specific pathogens.

Our study is limited by its in vitro nature. Our model of vaccine-exposed moDC and sorted blood DC does not fully represent molecular and cellular events occurring in situ after vaccination. In particular, it does not account for DC activation mechanisms that can be triggered indirectly through the recruitment of other inflammatory cell populations. The phenotype of the DC populations studied may also not accurately reflect the phenotype of DCs found in skin and muscle, especially that of Langerhans cells.

Together, this study sheds light on the mechanisms of DC activation by influenza vaccines. Furthermore, it highlights the differences between nonadjuvanted influenza vaccines in inducing IFN and proinflammatory responses in vitro and in vivo, further unraveling the mechanisms of action of vaccines at the systemic level and the role of early inflammatory responses in the process of vaccination.

METHODS

Subjects and study design

This study was approved by the Baylor Institute for Immunology Research Institutional Review Board (BIIR IRB 009-282). Written informed consent was obtained for all subjects. For in vitro studies, cells were obtained from apheresis of 10 healthy individuals (six females and four males) aged 27 to 59 years. For the vaccination study, healthy adults aged 18 to 64 years were enrolled to receive a single intramuscular dose of TIV-09 [A/Brisbane/59/2007 IVR-148 (H1N1), A/Uruguay/716/2007 NYMC X-175C (H3N2), and B/Brisbane/60/2008], MIV-09 [A/California/07/2009 X-179A (H1N1)], or placebo (saline) (n = 3 per group). Exclusion criteria included pregnancy, active allergy, or vaccinations within the previous 2 months. Blood was drawn by fingerprick at 0, 1.5, 3, 6, 9, 12, 15, 24, and 36 hours and 2, 3, and 7 days. Power analysis was conducted to estimate sample size for moDC experiments on the basis of preliminary results from microarray, Luminex, and FACS analyses. With n = 6, the study was adequately powered (>0.8) to detect a threefold difference when as few as 2000 genes are differentially expressed (microarray), a twofold difference in secreted proteins (Luminex), and a change ≥7.5% of positive cells for the surface markers considered (FACS). All primary data are in the Supplementary Materials (table S2).

In vitro DC culture and stimulation

moDCs were generated from monocytes isolated from healthy donor apheresis fractionated by elutriation. Monocytes were enriched by negative selection using EasySep Human Monocyte Enrichment Kit without CD16 depletion (STEMCELL Technologies) and cultured in CellGro DC medium (CellGenix) with 1% penicillin/streptomycin. Cells were cultured in sterile 50-ml tissue culture bags (American Fluoroseal Corporation) at an initial density of 1 × 106 cells/ml along with GM-CSF at 10 ng/ml and IL-4 (R&D Systems) at 50 ng/ml for 6 days. Cells were supplemented with additional cytokine doses at days 2 and 4. At day 6, DCs were harvested and further cultured in 200 μl of RPMI 1640 supplemented with 1% glutamine, 1% penicillin/streptomycin, 1% Hepes, 1% non–amino acids, 1% sodium pyruvate, 0.1% β-mercaptoethanol (Invitrogen), and 10% fetal bovine serum at a density of 1 × 105 cells/ml in 96-well plates. TIV-09 (45 μg of viral protein; 6 μl/ml) and/or MIV-09 (15 μg of viral protein; 6 μl/ml) (Sanofi Pasteur) was added, and cells were incubated at 37°C for 0, 6, 12, 24, and 48 hours. Supernatants were collected and stored at −80°C.

Another set of experiments was performed with Fluzone from different vaccination seasons. moDCs were exposed for 8 hours to one of the following vaccines (all from Sanofi Pasteur): (i) Fluzone 2006–2007 [A/New Caledonia/20/99 (H1N1), A/Wisconsin/67/05 (H3N2), and B/Malaysia/250604], (ii) Fluzone 2008–2009 [A/Brisbane/59/2007 (H1N1), A/Brisbane/10/2007 (H3N2), and B/Florida/04/2006], or (iii) Fluzone 2010–2011 [A/California/07/2009 X-179A (H1N1), A/Victoria/210/2009 X-187 (H3N2), and B/Brisbane/60/2008]. Supernatants were collected and stored at −80°C.

A separate experiment was performed to assess IP-10 production by moDCs in the presence of other inactivated viral vaccines such as human papillomavirus (Merck), hepatitis A, hepatitis B (GSK), and rabies (Sanofi Pasteur). moDCs were treated with vaccines for 18 hours. Supernatants were stored at −80°C.

DC phenotype and cytokine assay

DCs (1 × 105) were stimulated with selected vaccines for 18 hours and labeled with fluorochrome-conjugated anti-CD83, anti-CD80, anti-CD86, anti–HLA-DR, anti–PD-L1, anti–PD-L2, anti-CCR7, and anti-CD40 antibodies (BD Biosciences). Cell death was assessed by annexin V and PI staining (BD Biosciences). Labeled cells were analyzed using a FACSCanto (BD Biosciences). IP-10 (BD Biosciences) and IFN-β (R&D Systems) levels were measured by ELISA, and other cytokines were assessed using bead-based cytokine multiplex analysis (Luminex, Bio-Rad).

Purification and activation of pDCs and cDCs

Buffy coats obtained from healthy donors (Carter BloodCare) were fractionated on a Ficoll gradient. Peripheral blood mononuclear cells were further enriched for DCs through negative selection using the Pan-DC Enrichment Kit (STEMCELL Technologies). The enriched cells were sorted on a FACSAria (BD Biosciences) for LIN/HLA-DR+/CD123+ and CD11c pDCs and LIN/HLA-DR+/CD123 and CD11c+ cDCs (fig. S3). Purity was routinely >98%. DCs were stimulated at a density of 5 × 104 cells per well with vaccines (6 μl/ml) in 200-μl complete medium in 96-well plates for 0, 6, 16, 24, and 48 hours. For microarray analysis, cells (0.5 × 106) were cultured with vaccine (6 μl/ml) in a 0.5-ml complete medium for 6 hours and processed for RNA analysis.

RNA processing and transcriptional analysis

After vaccine stimulation, cells were lysed in an RLT buffer (RNeasy Plus Mini Kit, Qiagen) supplemented with 1% β-mercaptoethanol and stored at −80°C. Total RNA was isolated from cell lysates according to the manufacturer’s instructions. RNA from samples passing quality control was amplified and labeled with Illumina TotalPrep RNA Amplification Kit (Illumina). RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). Biotinylated complementary RNA was hybridized to Illumina Human-6 BeadChip Array version 2 and scanned in Illumina BeadStation 500. Fluorescent hybridization signals were assessed with BeadStudio software (Illumina). Statistical analysis and hierarchical clustering were performed with GeneSpring 7.3.1 software (Agilent Technologies). For the vaccinated healthy cohort, data were normalized to the median of the t = 0 hour samples across donors.

Statistical analysis

Data from cell culture experiments were expressed as mean of replicate experiments. Statistical significance was determined by nonparametric test such as Mann-Whitney U test. P < 0.05 was considered statistically significant. For microarray analysis, unless otherwise specified, we conducted Welch’s unequal variances t test (P < 0.01), adjusting the false discovery rate with the Benjamini-Hochberg procedure. For all figures, horizontal lines or bars represent the mean, and whiskers represent the SD.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/382/eaaf9194/DC1

Fig. S1. Dose-dependent induction of IFN-β and IP-10 by TIV-09 and MIV-09 in moDCs.

Fig. S2. IP-10 production by moDCs challenged with various inactivated viral vaccines.

Fig. S3. Gating strategy for moDC and pDC sorting.

Table S1. Vaccines used for comparative studies in moDCs.

Table S2. Primary data.

REFERENCES AND NOTES

Acknowledgments: We thank N. Baldwin for her help with the manuscript and G. Zurawski and S. Zurawski for their help with vaccine fractionation. Funding: The project was supported by NIH U19 AI057234, NIH U01AI124297 (to J.B.), and U19 AI089987 (to K.P. and V.P.). Author contributions: S.A. and L.T.-S. performed the culture experiments and S.A. wrote the manuscript. R.B. and Y.W. analyzed the microarray data and R.B. wrote the manuscript. K.P. and V.P. contributed to the overall design and writing the manuscript. J.B. designed and oversaw the study and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The data sets presented in this manuscript are deposited in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) under GEO Series accession number GSE81692.
View Abstract

Navigate This Article