Adjuvant Effect of Type I Interferon Induced by Many but Not All Commercial Influenza Vaccines

Abstract

Background

Seasonal influenza vaccines approved and offered in the United States have varying reported degrees of effectiveness year over year and between manufacturers. Influenza vaccines produced from live virus may include single stranded RNA (ssRNA) that is a potent activator of the innate Toll-like receptor 7 (TLR-7) ligand. Plasmacytoid dendritic cells (pDC) can be activated by ssRNA to produce type I interferons such as IFN-α, which has been shown to have an adjuvant-like effect.

Objective

Our aim was to determine if IFN-α induction in peripheral blood mononuclear cells (PBMCs) exposed to eight different commercial influenza vaccines is a pDC-dependent process mediated through TLR-7 signaling.

Results

We demonstrate the ability of multiple vaccines to induce IFN-α in a TLR-7-dependent fashion. A number of vaccines however lacked IFN-α induction. The significance of these differences between vaccines is unclear, since all the approved vaccine formulations offer some degree of protection.

Introduction

Influenza, a single-stranded RNA (ssRNA) virus, shown to induce interferon-α (IFN-α) production from plasmacytoid dendritic cells (pDCs) [1]. pDCs and their ability to activate other immune cells assist in bridging the innate and adaptive immunity. They are the main producers of type I IFNs including IFN-α, and they produce more than half the circulating IFN-α [2]. pDCs have a distinctive expression of the innate toll-like receptor 7 (TLR-7) that is activated by ssRNA [3].

Most of the influenza vaccines are generated from live influenza produced in eggs or cells. This can allow various amounts of viral constituents to enter the vaccine, including ssRNA. In most of the live-virus derived vaccine formulations therefore there is potential for pDCs to be activated by the viral ssRNA to produce type I IFNs. This group of cytokines is mainly involved in the innate immune response to viral infections. Studies have shown that IFN-α can effectively promote the differentiation and activation of subsets of T- and B-lymphocytes, and DCs, in mice and humans [4]. Data from mouse models and humans suggest that IFN-α can act as an effective vaccine adjuvant for inducing protective antiviral and antitumor immunity [4]. In this study, we will determine which of the currently licensed seasonal influenza vaccines induce IFN-α and what cell type produces it.

Methods

Subjects

The subjects were recruited from Case Western Reserve University, University Hospitals of Cleveland and Louis Stokes Cleveland VA Medical Center according to IRB approved protocols. A total of thirteen healthy donors took part in the study with an age range of 22–60 years, with 69% female, 85% Caucasian, 8% Asian, and 8% African American.

Cell isolation

Blood obtained in heparin containing Vacutainer tubes (BD Bioscience) was processed within 1 hour of drawing. PBMCs were isolated by Ficoll Plaque Plus (GE Healthcare/Amersham Bioscience) density gradient centrifugation according to the manufacturer’s instructions. In several experiments, pDCs were depleted from PBMCs using a CD304 (BDCA-4/Neuropilin-1) Microbead Kit from Miltenyi Biotec (Germany) according to the manufacturer’s instructions. For assessment of the pDC frequency and the effectiveness of the depletion method, PBMCs were stained before and after the pDC depletion using the following antibodies: anti-HLA DR-Pacific Blue and CD11c-APC (BioLegend, San Diego, CA, USA), Lineage Cocktail 1 (Lin1)-fluorescein isothiocyanate (FITC, BD, Franklin Lakes, NJ, USA), and CD123-PE (eBioscience, San Diego, CA, USA).

Materials

The vaccines used for adjuvant effect comparison were standard commercial lots from the 2018–2019 and 2019–2020 seasons: Fluzone Quadrivalent®, Fluzone High Dose®, Afluria Quadrivalent®, Flublok Quadrivalent®, Flulaval Quadrivalent®, Flucelvax Quadrivalent®, Fluad®, and Fluarix Quadrivalent®.

The following oligonucleotides were used to inhibit TLR binding: ODN 2088 Control (ODN 20958), which inhibits TLR-7-mediated signaling but not TLR-8- or TLR-9-mediated signaling and ODN 21798 Control (ODN 23098), which is an ODN control, from Miltenyi Biotec (Germany). Both used at a concentration of 5 uM; this concentration is recommended by the manufacturer’s instructions as the highest concentration for use in human immune cells.

Cell culture

PBMCs (2 × 10⁶/well) or pDC-depleted PBMCs were incubated with or without 1:50 dilution of each vaccine; or TLR-7 specific ligand (Gardiquimod at 0.625 ug/ml concentration; InvivoGen) in 96-well round-bottom plates in 200 μl X-Vivo^™ 15 Serum-free Hematopoietic Cell Medium (Lonza-BioWhittaker, Walkersville, MD, USA) at 37 °C for 20 h. The concentration of vaccines and Gardiquimod used was determined with preliminary dose-response experiments (data not shown). IFN-α production was measured using VeriKine^™ Human IFN-α Multi-Subtype ELISA Kit (detection limit 12.5 pg/ml; PBL Assay Science, Piscataway, NJ, USA) according to the manufacturer’s instructions.

Statistical analysis

Univariate descriptive statistics were used to identify means and standard deviations levels of cellular responses to influenza vaccines. In comparing the cellular responses a Wilcoxon signed-rank test or paired t-test were used.

Results

Induction of IFN-α in PBMCs incubated with influenza vaccines

The participants’ PBMCs were incubated for 20 hours with media, a TLR-7 specific ligand (Gardiquimod) as a positive control, and the panel of 2018–19 or 2019–20 commercial influenza vaccines. Most of both seasons’ influenza vaccines are egg produced split virus vaccines (Fluzone®, Fluzone HD®, Afluria®, Flulaval®, and Fluarix®), one is cell line produced split virus vaccine (Flucelvax®), one is egg produced inactivated virus vaccine with adjuvant (Fluad®), and only one is recombinant hemagglutinin antigen vaccine produced in an insect cell line (Flublok®). IFN-α production was measured in the supernatant by ELISA. No IFN-α was detected from PBMC incubated only in media. Gardiquimod stimulated the cells to produce detectable IFN-α in all subjects (Fig. 1, mean 419 pg/ml). Vaccines were studied before the expiration date for each season. Fig. 1 demonstrates that Fluzone®, Fluzone HD®, and Afluria® had significantly higher IFN-α (all p≤0.01) than the media negative control group. Flulaval did not induce significant IFN-α production, it was trending higher (p=0.06). Vaccines from two sucessive seasons were tested and verified that these results were consistent in each vaccine from year to year (data not shown).

Figure 1.

PBMCs incubation with vaccines. PBMCs (2 million cells/well) were incubated for 20 hours with media only, the TLR-7 specific ligand (Gardiquimod), or the different commercial 2018–19 influenza vaccines (Fluzone, Fluzone High Dose, Afluria, Flulaval) or 2019–20 influenza vaccine (Fluarix), IFN-α was measured in the supernatant by ELISA. Limit of detection of ELISA 12.5 pg/ml. Cells from 12 individual subjects were tested in these experiments.

Even though the sensitivity of the ELISA is 12.5 pg/ml, three of the vaccines, Flublok®, Flucelvax® and Fluad®, did not induce IFN-α (Fig.1). In a control experiment, Gardiquimod was added to the aforementioned vaccines in culture to make sure that they were not toxic. The vaccines did not reduce the level of IFN-α that Gardiquimod could induce, supporting lack of toxicity as a cause of no IFN-α induction by those vaccines (data not shown).

Vaccine induced IFN-α is produced by pDCs and TLR-7 dependent

The activation of pDCs’ by the influenza vaccines involves the recognition of viral components by both cytoplasmic receptors and TLR-7, leading to IFN-α production [1]. To demonstrate that the pDCs are the main producer of the IFN-α observed after exposure to vaccine, pDC depleted PBMCs were incubated with the IFN-α-inducing vaccines and Gardiquimod. Results showed that without pDCs, IFN-α production was lost (Fig. 2).

Figure 2.

pDC depleted PBMCs and untouched PBMCs were incubated with vaccines. Depletion of pDCs leads to vaccine induced IFN-α production below the limit of detection of the ELISA assay (< 12.5 pg/ml). Cells from 5 individual subjects were tested in these experiments.

Experiments were then performed to confirm that the IFN-α production was TLR-7 dependent. We pre-incubated the cells from 4 different donors with an oligodeoxynucleotide (ODN) sequence that specifically inhibits TLR-7, or an ODN control. The cells were then incubated with either Gardiquimod, Afluria®, or Fluzone HD®. The TLR-7-inhibiting ODN significantly reduced the induction of IFN-α compared to the no ODN and the control ODN groups (p<0.02 in Fig. 3A,B,C). This result supports that IFN-α induction is TLR-7 signaling dependent.

Figure 3.

PBMCs were pre-incubated for 45 minutes with the TLR-7 antagonist or control ODN, and/or A. Gardiquimod, B. Afluria, or C. Fluzone HD was added for an additional 24 hours incubation. IFN-alpha was determined by ELISA. The graphs show IFN alpha % reduction from the IFN alpha production induced by A. Gardiquimod, B. Afluria, or C. Fluzone HD non-preincubated with TLR-7 antagonist or control ODN, respectively. Cells from 4 individual subjects were tested in these experiments.

Discussion

Only one of the formally approved seasonal influenza vaccines includes an adjuvant, yet all provide some protection. We previously noted that several seasonal influenza vaccines potentially induce IFN-α [5]. Since type I IFNs can serve as adjuvants, we hypothesized that even “non-adjuvanted” seasonal influenza vaccines may display features of an adjuvanted vaccine.

Type I IFNs have anti-viral properties and mediate innate and adaptive immune responses. Type I IFN is a powerful adjuvant when administered in association with the human influenza vaccine. It induces strong IgG2α, a characteristic of Th1 immunity, IgA production, and also grants protection from virus challenge [6]. The benefit was obtained by injecting IFN-α as an adjuvant to the hepatitis B vaccine in immunocompromised patients [7]. Recombinant IFN-α plus hepatitis B vaccine in prior non- or low-responders also tended to generate better responses [8]. As potential mechanisms for an adjuvant-like effect, type I IFNs stimulate cross-priming, enhance presentation of viral peptides, and are proinflammatory [9].

Canaday et al. showed that live and UV-inactivated influenza virus induced pDCs to produce IFN-α in a TLR-7 dependent manner [5]. In the current study, we demonstrated that this effect applies to many of the commercially available licensed seasonal influenza vaccine formulations. A potential mechanism behind the differences may be the different production methods used to purify the influenza antigens. Influenza vaccines derived from live virus have different processes to inactivate the virus and varying degrees of purification of the vaccine components after that inactivation. As a result of the different degrees of purification, in addition to the pre-specified amount of hemagglutinin, they may include other viral components that could modulate the immune response. One of the viral components that can be carried along is ssRNA. That appears to be the case with all but one (Flulaval®) of the split virus vaccines, including Fluzone®, Fluzone High Dose®, Afluria®, and Fluarix®. When PBMCs are incubated with these vaccines they readily produce detectable IFN-α.

Three vaccines did not induce IFN-α. Flucelvax® and Fluad®’s production both rely on live virus. Unlike for split and whole virus vaccines, both undergo an additional detergent purification step after viral inactivation that removes many other viral products. Flublok®, the only FDA approved recombinant influenza vaccine available in the US, has baculovirus vectored HA protein extracted from an insect cell line but no ssRNA [10]. Fluad® is the only influenza vaccine on the US market with an adjuvant added, yet it does not induce IFN-α [11]. The adjuvant effect in Fluad® is from MF59 and squalene that stimulates granulocytes, which release chemokines, including CCL-2, -3, -4, CCR-7 and IL-18, attracting further waves of inflammatory cells [12].

All of these influenza vaccines, whether or not they induce IFN-α, received FDA approval and licensure based on clinical effectiveness or at least non-inferiority data [13] [14]. Influenza vaccines are generally not tested in a focused head to head fashion powered properly to assess superiority and differences between them. A few studies between some of these vaccines show differential effectiveness but there are caveats to understanding what role induction of type I IFN may play in these differences. Even though both vaccines induce IFN-α, Fluzone HD® immunization provided greater protection from laboratory confirmed flu in older adults and respiratory-related hospitalizations in nursing home residents than Fluzone® [15] [16]. Fluzone HD® did induce higher IFN-α in our studies but the four-fold greater antigen concentration could also explain its superiority. A randomized clinical trial evaluating clinical effectiveness found that Flublok® (non-IFN inducer) was superior to Fluarix® (IFN-inducer) in persons over age 50 [17]. Immunogenicity data from children showed significantly improved responses to influenza A strains in Flublok® over Fluarix® as well [18]. Flublok® has three times more HA antigen than Fluarix®. This suggests that the antigen dose or some other unknown factor is more important in Flublok® than the IFN-α effect present in Fluarix®. Compared with Fluzone®, Fluad® was associated with increased immunogenicity in elderly populations but its formulation includes the extrinsic MF59 adjuvant [19].

There is a concept in murine systems of peripheral tolerance, where immunizing with an antigen in the absence of an adjuvant leads to anergy rather than immunity [20]. The presence or absence of costimulatory signals delivered by the adjuvant-activated antigen presenting cell determines whether the T-cell receptor engagement activates or inhibits the T-cell’s ability to subsequently respond to a peptide [20] [21]. Flublok® is a purified protein non-adjuvanted vaccine that has been shown to be highly effective. Why does this not induce a peripheral tolerance environment? The answer is not fully clear. It is possible that there are other adjuvant-like components in this vaccine that elicit other chemokines and cytokines with adjuvant properties besides IFN-α, which most of the other vaccines induce. For example, the trauma from delivering the shot intramuscularly elicits a modest local innate reaction [22]. A booster response, given that most who receive an influenza vaccine have some prior immunity whether from remote infections or prior immunizations, could also contribute. Such prior priming could support rapid generation of an inflammatory adjuvant-like milieu that further catalyzes a T- and B-cell immune response. However, in murine systems subcutaneous administration of peptide decreases the T-cell response to subsequent challenge with the same peptide, whether the animals are naive or specifically pre-immunized [23]. This observation undermines the argument that prior immunity gives an adjuvant-like effect. It is possible that this is one of the multiple differences between human and murine immunology or perhaps that peripheral tolerance just does not apply in influenza vaccination in people.

In summary, the significance of these differences between vaccines ability to induce type I IFN that could have an adjuvant effect is unclear, since all the approved vaccine formulations offer some degree of protection. This is an area still worthy of exploration, however as we experience clinically there is still significant room for improvement in the effectiveness of the current group of influenza vaccines, particularly in the elderly.

Conclusion

We demonstrated that among commercially available influenza vaccines, most of the egg-produced ones generate TLR-7-dependent induction of IFN-α by peripheral pDCs; cell line-produced vaccines and one adjuvanted vaccine did not. These vaccines do have some protective efficacy, but with absent comprehensive head to head testing we cannot determine the clinical importance of these differences. We need a deeper understanding of vaccine immunobiology and the associated clinical impact to help us design even better vaccines.