Type I IFN as a Natural Adjuvant for a Protective Immune Response: Lessons from the Influenza Vaccine Model¹

The immune-promoting activity of any given vaccination strategy is determined not only by the presence of the relevant antigenic components in the vaccine formulation, but also by the addition of suitable adjuvants capable of activating and promoting an efficient immune response against the infectious agents (1, 2). A desirable feature of an adjuvant is that it should specifically enhance the immune response to the vaccine Ag with which it is coadministered, without causing toxic effects. Currently, special attention is being given to adjuvants capable of efficiently promoting a Th1 type of immune response, which is considered the best correlate of a protective immune response against infections (3). However, the most powerful Th1-promoting adjuvants exhibit some toxicity, which limits their clinical use. As these adjuvants induce the production of several cellular factors, the characterization of the natural mediator(s) essential for their important biologic activity not only would increase our knowledge of the mechanisms responsible for a protective immune response, but could also lead to novel and more effective strategies in vaccine development.

Type I IFNs are cytokines endowed with multiple biological activities (4). Although low levels of type I IFN are detected under physiological conditions (5), its production is markedly enhanced during infections (4). For a long time, the importance of the effects of type I IFN on the immune system remained poorly considered (6). In recent years, however, some reports have shown that these cytokines affect the differentiation, survival, and function of immune cells, including T cells (7, 8, 9, 10, 11, 12, 13, 14, 15) and dendritic cells (DCs)³ (16, 17, 18, 19, 20) and efficiently enhance a primary Ab response (21). In a recent study it was shown that immunization of mice with chicken γ-globulin in the presence of type I IFN results in the generation of a potent primary immune response, characterized by an isotype switching toward IgG2a Abs (21). In this study a defective production of IgG2a Abs was observed in type I IFN receptor knockout (IFN-IR KO) mice immunized in the presence of CFA. Although this observation could suggest a role of type I IFN in the CFA-induced immune response, it was unclear whether the most currently used adjuvants induced type I IFN and acted through the production of these cytokines for promoting a potentially protective Th1 immune response. Moreover, as the effects of type I IFN on the immune response remained controversial, with some authors emphasizing the importance of the immunosuppressive activities of these cytokines (22), it was essential to evaluate whether type I IFN, when administered at the time of the vaccine injection, could act as an effective natural adjuvant in an experimental setting in which a human vaccine and the relevant infectious agent could be used for testing vaccine efficacy. In the present study we used IFN-IR KO mice for investigating whether endogenous type I IFN itself is an essential mediator in the immune response induced after immunization with a reference protein Ag in the presence of some of the most currently used adjuvants. Moreover, we have evaluated the adjuvant activity of type I IFN compared with typical adjuvants by using human influenza vaccine as a model. We found that endogenous type I IFN is the main mediator for the promotion of Th1-type immune responses by a wide range of adjuvants. Furthermore, type I IFN itself was an unexpectedly powerful vaccine adjuvant for achieving immune protection against influenza virus. These findings can open new perspectives for vaccine development.

IFN-IR KO C3H/HeN mice were generated at the Institut Curie (Orsay, France) as follows. The mutated allele of the original IFN-IR KO strain (IFN-αβ R^0/0A129) (23) was transferred into the C3H/HeN background. Briefly, a male IFN-αβ R^0/0A129 mouse was crossed with C3H/HeN females. F₁ females were crossed with C3H/HeN males. Backcross progeny were crossed with C3H/HeN mice for nine generations. From the 10th generation backcross, brother/sister mating for 10 generations produced the IFN-IR KO mice that were used for comparative studies with identical background control C3H/HeN mice. In vitro aged peritoneal macrophages from IFN-IR KO C3H/HeN mice were not responsive to 800 U/ml type I IFN, as revealed by measuring inhibition of vesicular stomatitis virus yield using procedures described previously (5, 24), while IFN-treated peritoneal macrophages from control mice showed a 3 log₁₀ inhibition of virus yield. C57BL/6 mice were purchased from Charles River (Calco, Italy). Mice were housed in the facilities of the Department of Virology at Istituto Superiore di Sanità and were used at the age of 7–8 wk. Wild-type and IFN-IR KO mice were kept under specific pathogen-free conditions. All work with animals conformed to European Community guidelines.

IFA and CFA (Sigma, St. Louis, MO) were each mixed with Ag solution at a 1/1 (v/v) ratio and emulsified, using two glass syringes and luer lock connectors, until a stable emulsion was formed. Alum (aluminum hydroxide gel, Sigma) was dissolved in the Ag solution at a ratio 1/20 (w/v), and the pH was adjusted to 6.5. After 1-h incubation at room temperature, the solution was centrifuged, and the pellet resuspended in the previous volume in saline. CpG was synthesized by Roche Diagnostic (Milan, Italy) with a phosphorothioate backbone (sequence: 5′-TsGsAsCsTsGsTsGsAsAsCsGsTsTsCsGsAsGsAsTsGsA-3′). Two hundred micrograms of CpG was dissolved in 1 ml of a solution containing 200 μg OVA. Polyinosinic-polycytidylic acid (poly(I:C); Sigma) was dissolved in saline at a concentration of 10 mg/ml, and 0.15 mg was injected i.p. in mice. MF59 (an emulsion consisting of 5% (v/v) squalene, 0.5% (v/v) Tween 80, and 0.5% (v/v) Span 85 in H₂O) was provided by Chiron Vaccines (Siena, Italy) (25). MF59 was mixed with Ag solution at a 1/1 (v/v) ratio and was emulsified by pipetting.

The biological activity of serum IFN was determined as described previously (26). One of the units, as expressed in the text, is the equivalent of four IFN reference units.

OVA (grade V, Sigma) was dissolved in 0.15 M NaCl and filter-sterilized before injection. The subunit influenza vaccine Agrippal, used for 1999/2000 vaccination campaign (supplied by Chiron), was prepared from influenza virus A/Beijing X127, a reassortant of influenza viruses A/Beijing/262/95 (H1N1) and A/PR/8/34 (H1N1). The hemagglutinin (HA) concentration in the subunit vaccine, calculated by single radial diffusion, was 470 μg/ml.

High titer IFN-α/β (2 × 10⁷ U/mg protein) was prepared in the C243-3 cell line following a method adapted from Tovey et al. (26). IFN was concentrated and partially purified by ammonium sulfate precipitation and dialysis against PBS as described previously (21).

OVA (10 μg/mouse) with or without adjuvants was injected intradermally in a volume of 50 μl/mouse. Ten and 17 days after the first treatment, mice were boosted with OVA alone. For delayed-type hypersensitivity (DTH) assays, mice were injected with OVA and adjuvant on days 0, 10, and 17. For systemic immunization, 100 μl influenza vaccine (150 μg HA/ml) mixed with 100 μl saline, type I IFN, or adjuvant were injected i.m. into the mouse thigh. For mucosal immunization, mice were anesthetized and instilled into alternate nostrils in dropwise fashion with 50 μl of a solution containing 25 μl vaccine (470 μg HA/ml) and 25 μl saline, type I IFN, or adjuvant. Vaccination was performed on days 0 and 14.

To measure OVA- or influenza-specific Ab levels, standard direct ELISAs were performed. Ninety-six-well, flat-bottom microtiter plates (Immulon 4HBX, Dynatech, Chantilly, VA) were coated with 100 μl of a 1 μg/ml (for total anti-OVA IgG detection) or 4 μg/ml (for anti-OVA IgG2a and IgG1 detection) of a solution of OVA or with 3.6 μg HA/ml influenza vaccine. The following dilutions of peroxidase-conjugated secondary Abs were used for anti-OVA Ab detection: anti-mouse IgG (Fc-specific), (Sigma), 1/1,000; anti-mouse IgG2a (Cappel Research Products, Durham, NC), 1/200; and anti-mouse IgG1 (Cappel Research Products, Durham, NC), 1/400. For influenza-specific ELISA, peroxidase-conjugated secondary Abs were used as follows: anti-mouse IgG (H + L chain; Pierce, Rockford, IL), 1/75,000; anti-mouse IgG2a (Cappel), 1/200; anti-mouse IgG1 (Cappel), 1/400; and anti-mouse IgA (Kirkegaarde & Perry, Guilford, U.K.), 1/1,500. Ortho-phenylenediamine (Sigma) was used as enzymatic substrate, and plates were read in a microplate autoreader at 490 nm wavelength.

Serum hemagglutination inhibition (HAI) titers were measured according to standard procedures (27, 28).

Preparation of spleen cell suspensions and [³H]thymidine uptake assay were performed as described previously (29). The spleen cell concentration was 5 × 10⁵ in 0.2 ml/well of 10% FCS RPMI medium containing different concentrations of OVA (0, 50, 100, and 200 μg/ml).

Thirty-five days after the first injection, control and IFN-IR KO mice were challenged with OVA (20 μg/50 μl) intradermally into the right footpad, while the left footpad was injected with saline as a control. Foot swelling was measured with a microcaliper 48 h later. Data represent the mean of OVA-challenged minus saline-challenged (contralateral) foot size of three mice per group.

The original H1N1 A/Beijing/262/95 influenza virus (supplied by National Institute for Biological Standards and Controls, Hertfordshire, U.K.) was adapted to mouse after seven blind intranasal (i.n.) passages. The virus titer was 64 HAU/or 1.1 × 10⁷ PFU/ml. The LD₅₀ corresponded to a dilution of 1/1000. For virus challenge, anesthetized mice were instilled i.n. with 50 μl of a virus suspension containing 10 LD_50.

Data were analyzed by the Wilcoxon rank-sum test.

We first investigated whether injection of mice with currently used adjuvants, IFA, CFA, alum, or CpG oligonucleotides (CpG), could induce IFN and whether the integrity of the type I IFN system was essential for the immune-promoting activity after immunization with a protein Ag (OVA). C3H/HeN mice were treated with the different adjuvants or poly(I:C) as a reference type I IFN inducer. Biologically active IFN was detected in the sera of mice 24 and 48 h after the injection of CFA or IFA. Mice injected with CpG showed increasing serum IFN levels between 24 and 72 h, whereas no serum IFN activity was detectable in alum-injected mice (Fig. 1 a).

To assess the role of type I IFN in the biological activity of these adjuvants, control and IFN-IR KO C3H/HeN mice were immunized with OVA with or without the different adjuvants. As expected, control mice immunized with the Ag together with adjuvants exhibited a marked increase in the overall OVA-specific Ab response compared with animals immunized with the Ag alone (Fig. 1, b and c). IFA and CFA acted as effective adjuvants for both IgG1 and IgG2a subclasses (typically associated with Th2 and Th1 types of Ab response, respectively), whereas CpG and alum specifically enhanced IgG2a (Th1 type) and IgG1 (Th2 type) responses, respectively. Of interest, IFN-IR KO mice showed a marked defect in the generation of anti-OVA Abs when CFA, IFA, or CpG was used as adjuvant. This defect was especially pronounced with regard to the IgG2a Abs (Fig. 1, b and c). In mice immunized with alum as adjuvant, a weak Ab response, with prevalence of IgG1 Abs, was detected in both control and IFN-IR KO mice.

As expected, CFA, IFA, and CpG also enhanced T cell priming. Thus, spleen cells derived from control mice immunized with OVA in the presence of CFA, IFA, or CpG showed a higher in vitro proliferative response to OVA compared with splenocytes from mice immunized with the Ag alone; no significant increase in lymphocyte proliferation was observed in animals immunized in the presence of alum (Fig. 2,a). Importantly, as for the Ab response, greatly reduced priming for lymphocyte proliferation was detected in IFN-IR KO mice immunized with the Ag in association with CFA, IFA, or CpG. Furthermore, while control mice immunized with the Ag in the presence of adjuvants showed a clear-cut DTH response, no significant response was detected in IFN-IR KO mice (Fig. 2 b). These results indicate that endogenous type I IFN is essential for the promotion of both IgG2a Ab responses (a typical humoral hallmark of the Th1 type of immune response) and T cell responses by a variety of adjuvants.

To evaluate the adjuvant activity of type I IFN, we used a commercially available influenza vaccine obtained from an H1N1 influenza virus circulating in 1999–2000 (A/Beijing/262/95) and, as an infectious agent, the corresponding virus.

A single injection of vaccine together with IFN into C57BL/6 mice resulted in a clear-cut seroconversion of all the animals. In contrast, only a limited portion (6 of 14) of mice inoculated with the vaccine alone seroconverted, showing Ab titers lower than those of animals treated with vaccine and IFN (Fig. 3,a, top panel). This weak Ab response was not associated with any protection from influenza virus infection, since all the animals died within 10 days after challenge. In contrast, all the mice injected with the vaccine and type I IFN were protected from virus infection (Fig. 3,b, top panel). After two injections with the vaccine alone, the majority of mice were still not protected from virus challenge (Fig. 3,b, bottom panel). Two immunizations with the vaccine mixed with IFN resulted in a homogeneous increase in Ab titers in all the mice (Fig. 3 a, bottom panel).

Qualitative analysis of the Ig response at 2 wk after the first immunization with IFN revealed high titers of IgG2a Abs, not detectable in mice injected with the vaccine alone (Fig. 3,c, top panel). Two weeks after the second immunization, mice immunized with the vaccine together with IFN showed higher levels of serum IgA than animals injected with vaccine alone. Moreover, only sera from IFN-treated mice exhibited detectable HAI titers (Fig. 3 c, bottom panel).

The adjuvant effect of type I IFN on the Ab response to influenza vaccine (total Igs and IgG2a) was dose dependent (Fig. 4,a) and paralleled a dose-dependent protection from virus challenge (Fig. 4,b). Of interest, in mice immunized with the vaccine together with the highest dose of IFN (2 × 10⁵ U), there was complete protection, as evaluated not only by the survival data, but also by the lack of any virus-induced decrease in mouse weight. Mice injected with 2 × 10⁴ U IFN survived after virus challenge, but a transient decrease in body weight after infection was observed. Only a low level of protection was detected in animals immunized with the vaccine together with the lowest dose of IFN (2 × 10³ U; Fig. 4,b). Additional IFN injections on days 1 and 2 markedly enhanced the Ab response (total IgGs and IgG2a) with respect to levels found in mice only coinjected with IFN and vaccine on day 0 (Fig. 4,c). Notably, mice subjected to a single IFN injection either before or after vaccine inoculation developed much lower Ab titers than animals coinjected with the vaccine and the cytokine (Fig. 4,d) and were not protected from virus challenge (data not shown). The adjuvant potency of type I IFN in protecting mice from a lethal challenge with influenza virus was comparable to that obtained with two of the best currently available adjuvants (CFA and MF59), while alum was ineffective (Fig. 4 e).

Further comparative experiments using control or IFN-IR KO C3H/HeN mice allowed us to verify the specificity of the IFN effect and the role of endogenous type I IFN in this vaccine model. As a reference adjuvant, we used MF59, recently considered the best adjuvant for influenza vaccine in humans (25). At all time points, the adjuvant effect of exogenous type I IFN was clearly detected in control mice, but not in IFN-IR KO animals (Fig. 5). After a first immunization, while induction of total Igs and IgG1 Abs was similar in animals immunized with either MF59 or IFN as adjuvants, only IFN-treated mice showed high levels of serum IgG2a Abs (Fig. 5,a, top). Of interest, control mice immunized with the vaccine alone showed detectable levels of IgA Abs, which were not revealed in IFN-IR KO mice, suggesting that endogenous IFN played a role in the induction of IgA in the early phase of the immune response. After the second immunization, there was an increase in Ab production in all vaccinated control mice. Interestingly, while IgG1 levels were higher in mice treated with MF59, considerable IgG2a titers were only detected in IFN-treated animals (Fig. 5, bottom). Likewise, IFN-IR KO mice immunized with MF59 as an adjuvant did not show any level of IgG2a Abs, further indicating a selective role of type I IFN for IgG2a induction. Notably, no adjuvant effect on IgA Abs was observed in IFN-IR KO mice immunized in the presence of either type I IFN or MF59 (Fig. 5, bottom), suggesting that type I IFN is important for a sustained IgA production.

Identification of mucosal adjuvants is an important task of vaccine research, since induction of protective mucosal immunity is crucial for achieving local immune protection at the pathogen entry site. In a first set of experiments we immunized C57BL/6 mice by giving two i.n. administrations, 14 days apart, of influenza vaccine alone or mixed with type I IFN; Ab levels were measured 2 wk after each immunization (Fig. 6,a). A general increase in Ab production (especially IgG2a) was detectable in IFN-treated animals after the first immunization. Two weeks after the second immunization, there was a further increase in Ab titers in IFN-treated mice compared with animals injected with the vaccine alone. Notably, at this time point, an impressive increase in IgG2a and IgA titers (1000- and 100-fold, respectively) was observed in animals immunized with the vaccine mixed with IFN compared with mice injected with vaccine alone. Mice immunized with IFN as an adjuvant also showed higher levels of secretory pulmonary IgA than control animals. Of interest, all mice given the IFN-adjuvanted vaccine i.n. were protected from influenza virus infection, as revealed by both survival values and lack of decrease in mouse weight after challenge, while only a partially protective effect was found in animals immunized with vaccine alone (Fig. 6,b). In a similar immunization experiment in IFN-IR KO and control C3H/HeN mice, type I IFN proved to be superior to MF59 in inducing IgG2a and IgA in control animals at both time points, while MF59 was more effective in inducing IgG1 Abs after two immunizations (Fig. 6,c, bottom). As expected, no significant Ab response for all Ig subclasses was observed in IFN-IR KO animals immunized i.n. with IFN as adjuvant. In contrast, MF59 was still capable of inducing IgG1 Abs in IFN-IR KO mice, but the induction of IgG2a and IgA was largely abrogated compared with the response detected in control animals (Fig. 6 c).

Although type I IFNs are the most used cytokines in patients, their clinical use as modulators of the immune response has received poor consideration (6). Recently, some studies have described new effects of type I IFN on DC differentiation and function (16, 17, 18, 19, 20), while other reports have emphasized the importance of immunosuppressive activity of these cytokines (22). The most remarkable finding reported in the present study is the demonstration that type I IFN, coadministered with a human vaccine (influenza), represents an unexpectedly powerful adjuvant, inducing a Th1 type of immune response and protection against virus challenge. The finding that widely used adjuvants, such as IFA, CFA, and CpG, induce the expression of type I IFN together with the demonstration of clear-cut defects in the production of IgG2a Abs and in the Ag-specific T cell response (T cell proliferation in vitro and DTH response in vivo) specifically occurring in IFN-IR KO mice demonstrate that the endogenous IFN system is essential for full expression of a Th1-type immune response. Intriguingly, defective production of IgG1 Abs was observed in IFN-IR KO mice when alum, an adjuvant promoting a Th2 type of immune response, was injected together with OVA. Thus, we can assume that endogenous IFN may play a different role in shaping the immune response, depending on the type of adjuvant and Ag. Recently, Van Huden and colleagues have reported that IFN-IR KO mice exhibit a defective IgG2a response with respect to control animals when immunized with β-galactosidase together with CpG as adjuvant, suggesting that type I IFN is required to mount an adaptive response to immunostimulatory DNA (30). Notably, our results obtained in the influenza vaccine model show that type I IFN was unexpectedly effective in inducing rapid seroconversion in all the animals, characterized by selective induction of high levels of IgG2a even after a single immunization, resulting in full protection from virus challenge.

IgG2a Abs are characteristic of the response to virus infection, are often protective, and demonstrate neutralizing activity (31). Thus, we may assume that in the course of a virus infection the expression of type I IFN and generation of IgG2a Abs are linked events of biological relevance for the subsequent generation of protective immunity.

When given i.m. as adjuvant, type I IFN was far superior to alum and was equivalent to CFA, considered one of the most powerful adjuvants in animal models, and to MF59, a new adjuvant recently considered the best candidate for anti-influenza vaccination in elderly individuals (32). Of interest, while these typical adjuvants have been used at the maximal tolerated dose, IFN dosages higher than those used in our experiments could result in even more potent adjuvant activity. Notably, two subsequent treatments with IFN on days 1 and 2 further increased the level of IgG2a Ab, while a single IFN injection at a time different from that of vaccine administration was ineffective. Thus, the optimal adjuvant activity is obtained under conditions mimicking the natural response to viral infections, often resulting in long term immunity, where considerable levels of type I IFN are produced at early times after virus contact with specific host cells, such as the so-called natural IFN-producing cells (also named pDC2 or plasmocytoid DCs), considered professional cells for producing high IFN levels in response to viral challenge (33, 34).

One of the major issues in vaccine research is the definition of strategies for inducing mucosal immunity and IgA production, which are important for immune protection against infectious agents transmitted through the respiratory system (35). In this regard it is of special interest that i.n. injection of influenza vaccine with IFN was particularly effective in inducing serum IgG2a and IgA Abs and full protection from virus challenge. Type I IFN was superior to MF59 in inducing IgG2a and IgA Abs when used as an i.n. adjuvant. This study reports the first evidence indicating that type I IFN is important for IgA production and for the establishment of mucosal immunity. Recent studies have shown that the i.n. administration of type I IFN represents an effective delivery system for inducing therapeutic effects with these cytokines (36). However, the mechanisms of action are still unclear (37). Thus, the present finding showing that the i.n. administration of IFN is unusually effective in enhancing vaccine efficacy and inducing immune correlates of protection encourages further studies for understanding whether type I IFN can represent a crucial factor in breaking tolerance and inducing mucosal immunity.

With regard to the role of endogenous type I IFN in the immune response to influenza vaccine, our results indicate that the induction of IgG2a and IgA Abs is mostly mediated by type I IFN itself, since no or poor production of these Abs was observed in IFN-IR KO mice. Early studies had shown that low levels of spontaneous type I IFN can be responsible for the natural antiviral state of macrophages (5) as well as for host-mediated restriction of tumor growth in mice transplanted with syngeneic IFN-resistant tumor cells (38), supporting the concept that even basal IFN levels can play important in vivo roles. Thus, we may argue that even in the absence of specific IFN induction, the defective response to the i.n. immunization of IFN-IR KO mice with influenza vaccine is indicative of the importance of type I IFN basal levels in the generation of a specific humoral immune response.

Identification of new adjuvants is an urgent need for vaccine development and especially for subunit vaccines, which are poorly immunogenic. The use of adjuvants effective in animal models is often restricted by safety concerns. The finding that type I IFNs, cytokines with a long record of clinical use (39, 40), are necessary and sufficient to induce a protective immune response to vaccines is not only important for the comprehension of the mechanisms underlying the immune response to infections, but can also exhibit practical implications for new strategies in vaccine development. As differences in the type I IFN-mediated regulation of the Th1 responses between mice and humans have been reported (41), the possible transference to humans of data obtained in mouse models should be regarded with caution. Selected clinical trials are needed to establish whether type I IFN can represent valuable natural adjuvants for human vaccines.

We thank Dr. I. Gresser for helpful discussion. We are grateful to Anna Ferrigno and Cinzia Gasparrini for excellent secretarial assistance.