Targeting Antigens to Different Receptors on Conventional Type 1 Dendritic Cells Impacts the Immune Response

Dendritic cells (DCs) play a crucial role in initiating immune responses to pathogens because of their capacity to efficiently capture and process foreign Ag for presentation to T cells. Since their initial description by Steinman and Cohn (1), DCs are now divided into three populations based on functional differences and ontogeny. Plasmacytoid DCs efficiently respond to viral infections by secreting type I IFN (2), whereas conventional type 1 DCs (cDC1s) and conventional type 2 DCs (cDC2s) are the main APCs responsible for initiating T cell responses (3). Both cDC1s and cDC2s are capable of presenting Ag in complex with MHC class II (MHC-II) to CD4⁺ T cells, although cDC2s have been reported to be more efficient at this process (4). Interestingly, cDC1s and cDC2s have been observed to differentially polarize CD4⁺ T cell responses, with cDC1s inducing Th1 cells, whereas cDC2s preferentially induce Th2 cells (5, 6). cDC1s, however, are superior at cross-presenting extracellular Ag to CD8⁺ T cells (7–10) and have therefore garnered much interest as a therapeutic target for enhancing cytotoxic T cell responses.

With the increased focus on delivering Ag to cDC1s, there is a need for a deeper understanding of how targeting Ag to different surface receptors on cDC1s influences the resulting immune response. A number of studies have focused on delivering Ag to the lectin receptor DEC-205 on cDC1s and have observed enhanced induction of T cells (11) in addition to protective efficacy in both tumor and infectious models (12–14). DEC-205 is, however, also expressed on other cells types, such as germinal center B cells (15), skin resident cDC2 (16), and at low levels on splenic cDC2 (17), which likely affect the resulting immune responses. By contrast, the surface receptors Clec9A and Xcr1 have emerged as being more specific for both lymphoid resident and migratory cDC1s (18–21). Clec9A is a C type lectin receptor that recognizes exposed F-actin on necrotic cells (22, 23) and can enhance cross-priming of cell-associated Ag for presentation to CD8⁺ T cells (24). In addition, delivering Ag to Clec9A has been reported to induce strong Ab responses in mice and macaques (19, 25) and to enhance the induction of T follicular helper cells (T_FH) (26). Xcr1 is a receptor for the chemokine Xcl1 (27). Although the function of Xcr1–Xcl1 interaction is not fully elucidated, the chemokine Xcl1 has been reported to induce chemotaxis of cDC1s and to play a role in the induction of CD8⁺ T cell responses (20, 21). Targeting Ag to Xcr1 induces a more Th1-polarized Ab response dominated by IgG2a (28, 29) in addition to CTL responses (28, 30, 31).

Although there are a number of studies exploring delivery of Ag to these surface receptors, the direct comparison of their findings is problematic because of the use of different mouse strains and different immunization strategies across studies (i.e., immunization route and use of adjuvant and protein versus DNA vaccination). To address this issue, in this study, we present a comprehensive comparison of targeting the Ag hemagglutinin (HA) from influenza A virus to the surface receptors DEC-205, Clec9A, or Xcr1 using a dimeric vaccine molecule (vaccibody) delivered as an intradermal (i.d.) DNA vaccine (32).

HEK293E cells (obtained from American Type Culture Collection) were used for transient transfections. Anti-HA (H-36-4-52) and anti-NP (HB65; American Type Culture Collection) were purified in the laboratory. For serum IgG ELISA, anti-mouse IgG (Fc specific), anti-mouse IgG1^a-bio (clone 10.9), anti-mouse IgG1^b-bio (clone B68-2), anti-mouse IgG2a-bio (clone 8.3), anti-mouse IgG2c-bio (clone 5.7), and anti-mouse IgG2b-bio (clone R12-3) were used. For flow cytometric analysis, anti-CD3e (145-2C11; Tonbo Biosciences), anti-CD19 (1D3; Tonbo Biosciences), anti-CD49b (DX5; eBioscience), anti–Ly-6G (1A8; Tonbo Biosciences), CD45R/B220 (RA3-6B2; Tonbo Biosciences), anti–MHC-II (M5/114.15.2; BioLegend), anti-CD11c (N418; Tonbo Biosciences), anti-CD11b (M1/70; Tonbo Biosciences), anti-CD24 (M1/69; BioLegend), anti-CD8α (53-6.7; BioLegend), anti-CD4 (GK1.5; BioLegend), anti-DO11.10 (KJ1-26; BioLegend), anti-CD45.1 (A20; Tonbo Biosciences) anti-CD14 (rmC5-3; BD Biosciences), anti–IFN-γ (XMG1.2; Tonbo Biosciences), anti–T-bet (eBio4B10; eBioscience), anti-GATA3 (TWAJ; Invitrogen), and anti-RORγt (AFKJS-9; eBioscience) were used. Influenza virus strain A/PR/8/34(H1N1) were kindly provided by Dr. A. Germundsson-Hauge at The National Veterinary Institute, Oslo, Norway.

All animal experiments were approved and performed in compliance with the regulations set by the Norwegian National Animal Research Authority. Female BALB/c and C57BL/6 mice aged 6–10 wk were obtained from Janvier Labs. Mice challenged with influenza A virus were euthanized if they lost more than 20% of their starting weight as a humane clinical end point, according to the guidelines of Norwegian National Animal Research Authority. For serum transfer experiments, naive mice received 200 μl serum obtained from immunized mice 1 d before challenge with 5xLD₅₀ influenza A/PR/8/34.

The construction of Xcr1-targeted and DEC-205–targeted fusion proteins have been described elsewhere (28, 33). For targeting Clec9A, a single-chain variable fragment (scFv) (Clec9A-specfic scFv [scFv^Clec9A]) was generated based on the 10B4 clone (19), with added 5′ BsmI and 3′ BsiWI sites, and inserted into the pLNOH2 vector containing a dimerization unit consisting of the hinge and CH₃ domain from human IgG3 (32). Insertion of mCherry (34), OVA (28), and HA from the influenza virus strain A/PR/8/34 (35) has been previously described. A fusion vaccine containing an scFv specific for the hapten 5-iodo-4-hydroxy-3-nitrophenacetyl (NIP) included a nontargeted control and has previously been described in more detail (32).

Xcl1, DEC-205–specific scFv (scFv^DEC-205), scFv^Clec9A, or scFv specific for the hapten NIP (scFv^NIP) fusion proteins containing mCherry or OVA were purified as previously reported (28) with some modifications. HEK293E cells were seeded in five-layer tissue culture flasks (Falcon Multi-Flasks) and transfected using polyethylenimine (1 mg/ml stock) at a ratio of 500 μg polyethylenimine to 250 μg DNA. Supernatant was harvested after 4–5 d and applied on a CaptureSelect FcXL Affinity Matrix column (Life Technologies) connected to an Äktaprime Plus (GE Healthcare). Bound proteins were washed with PBS, eluted in 0.1 M glycin–HCl (pH 2.7), and immediately dialyzed twice against PBS. Purified proteins were concentrated using 10-Kd cutoff Vivaspin columns (Sartorius Stedim Biotech), aliquoted, and stored at −80°C until use.

DCs from spleens of BALB/c mice were prepared using the GentleMACS dissociator (Miltenyi Biotec) according to the manufacturers protocol. Briefly, spleens were dissociated in GentleMACS C tubes in RPMI medium supplemented with 10% FBS. Erythrocytes were lysed by incubation with ACT buffer for 5 min on ice. Finally, cells were filtered through a 75-μm Nylon cell strainer. The following Abs were used for subsequent flow cytometry analysis: anti-CD3e, anti-CD19, anti-CD49b, anti–Ly-6G, CD45R, anti–MHC-II, anti-CD11c, anti-CD11b, and anti-CD24. After staining, cells were analyzed using an FACS Fortessa system with DIVA software (BD Biosciences). Neutrophils, B, T, and NK cells were systematically excluded using a Lin⁺ gate. For in vivo staining with Xcl1-, scFv^DEC-205–, scFv^Clec9A-, or scFv^NIP-mCherry, 20 μg purified proteins were injected i.v., and spleens were harvested 2 h later and processed and stained as indicated above.

BALB/c mice were anesthetized with 150 μl of 250 mg/ml Zolentil Forte (Virbac), 20 mg/ml Rompun (Bayer Animal Health), and 50 μg/ml Fentanyl (Actavis). After shaving the lower back, 25 μl of DNA vaccine (0.5 μg/μl in 0.9% NaCl) was injected i.d. on the left flank, followed by electroporation using the DermaVax (Cyto Pulse Sciences) system with two pulses of 450 V/cm × 2.5 μs and eight pulses of 110 V/cm × 8.1 ms. The procedure was repeated on the right flank.

Serum ELISAs were performed as previously described (28). Ninety-six-well ELISA plates (Costar) were coated with 1.25 μg/ml inactivated PR8 influenza virus (Charles River Laboratories), and incubated overnight at 4°C. After blocking with 1% (w/v) BSA in PBS (with 0.02% [w/v] Na-azide) for 1 h at room temperature (RT), plates were washed and serum samples added serially diluted in ELISA buffer (0.1% [w/v] BSA in PBS with 0.2% Tween 20 and 0.02% [w/v] Na-azide). ELISA plates were incubated with serum samples overnight at 4°C. Next, the ELISA plates were incubated with 1 μg/ml biotinylated IgG-specific Abs and incubated for 1 h 37°C. After washing, the plates were incubated with streptavidin-ALP (GE Healthcare [RPN1234V], 1:3000) for 45 min at RT and developed by adding 100 μl/well of substrate buffer (1 mg/ml phosphate substrate (P4744; Sigma-Aldrich), After 30 min, OD_{405 nm} was measured on a Tecan Sunrise spectrophotometer. For analyzing total serum Ig, plates were incubated with ALP-conjugated anti-mouse IgG (Fc specific, 1:3000). Ab titer was defined as the highest dilution of a serum sample with OD values >(mean + 3 × SD) of NaCl-vaccinated mice. If OD values did not exceed that of the NaCl-vaccinated mice (mean + 3 × SD), the sample was given an end point titer of 1.

For evaluating the avidity of serum IgG responses, one additional washing step with 7 M urea was included after incubation with serum, as previously described (29). In short, the serum ELISA was performed as detailed above until the wash after serum incubation. The ELISA plates were then incubated with either PBS or 7 M urea for 10 min at RT, before the plates were washed and incubated with ALP-conjugated anti-mouse IgG (Fc specific, 1:3000). The plates were developed as described above. Avidity index was calculated by dividing the area under the curve of for each sample on the plates incubated with urea, with the area under the curve of the same sample from the plate incubated with PBS (36).

Neutralization assays were performed as previously published (35). In short, individual serum samples from immunized mice were treated with receptor-destroying enzyme (II) and diluted 2-fold in virus diluents (DMEM with 1% bovine albumin fraction V, antibiotics, and 0.02 M HEPES). Fifty microliters of 50% tissue culture infective dose was added to each well, and the plates were incubated for 2 h at 37°C with 5% CO₂ before adding 2 × 10⁴ Madin–Darby canine kidney cells and incubating for 20 h at 37°C with 5% CO_2. Cells were fixed with 80% acetone for 10 min, and infection was evaluated by ELISA using biotinylated mAb against the viral nucleoprotein (HB65).

Bone marrow (BM) cells were harvested by flushing tibiae and femur with medium. The cell suspension was filtered through a 75-μm Nylon cell strainer and seeded at a concentration of 2 × 10⁶ cells/ml, 5 ml/well in a six-well plate. A total of 0.1 μg/ml of Flt3L (PeproTech) was added, and the cells were incubated for 9 d at 37°C 5% CO₂ (37). Semiadherent cells were subsequently harvested and analyzed by flow cytometry after staining with anti-CD45R/B220, anti-CD11c, anti-CD11b, and anti-CD24 for 20 min on ice.

OVA-specific CD4⁺ T cells were isolated from DO11.10 TCR transgenic mice by harvesting spleens and generating single-cell suspensions as described for DC isolation from spleen. DO11.10 CD4⁺ T cells were then purified using a CD4 T cells isolation kit (Miltenyi Biotec), according to the manufacturer’s protocol. Purified DO11.10 CD4⁺ T cells were seeded at a concentration of 5 × 10⁴ cells in 48-well plates together with 2.5 × 10⁵ BM DCs and 0.5 μg Xcl1-, scFv^DEC-205–, scFv^Clec9A-, or scFv^NIP-OVA fusion proteins in RPMI 1640 with 10% FCS. The cell were incubated for 72 h at 37°C with 5% CO₂ before they were harvested and analyzed by flow cytometry after staining for anti-CD4, anti-DO11.10, anti–T-bet, anti-GATA3, and anti-RORγt. Data were acquired on a Fortessa (BD Biosciences) flow cytometer and analyzed using FlowJo software (FlowJo). Supernatants were used for measuring cytokines by ELISA.

BM-derived DCs were differentiated for 9 d as described above, and cDC1s and cDC2s were sorted by using FACSAria IIIu (BD Biosciences). After sorting, the purity of the DC populations was observed to be >99% pure. For proliferation assays, OVA-specific CD4⁺ T cells from DO11.10 mice and CD8⁺ T cells from OT-I mice were enriched using mouse CD4⁺ or CD8⁺ T cell magnetic bead isolation kits (Miltenyi Biotec), respectively. Isolated T cells were stained with 5 μM CellTrace Violet (CTV; Molecular Probes) before being incubated with sorted DCs at a ratio of DC/T cells of 1:3. DC–T cell cocultures were incubated in the presence of 1 μg/ml purified Xcl1-, scFv^DEC-205–, scFv^Clec9A-, or scFv^NIP-OVA for 4 d at 37°C and 5% CO₂ before being analyzed for proliferation by flow cytometry.

OVA-specific DO11.10 CD4⁺ T cells were isolated as for in vitro Th polarization and 1 × 10⁶ cells transferred to naive mice 1 d before i.d. immunization with 25 μg DNA encoding Xcl1-, scFv^DEC-205-, scFv^Clec9A-, or scFv^NIP-OVA. Inguinal and axillary lymph nodes (LN) were harvested 5 d after immunization, and single-cell suspensions were generated using GentleMACS dissociator (Miltenyi Biotec). In short, LN were placed in C MACS tubes containing RPMI medium and dissociated according to the manufacturer’s recommendations. After filtration, single-cell suspensions from LN were washed in PBS and analyzed by flow cytometry after staining for anti-CD19, anti-CD14, anti-CD3, anti-CD4, anti-DO11.10, anti–T-bet anti-GATA3, and anti-RORγt. Data were acquired on a Fortessa (BD Biosciences) flow cytometer and analyzed using FlowJo software (FlowJo). Next, single-cell suspensions of splenocytes from the same mice were stimulated with a DO11.10 specific peptide for 48 h ex vivo, and cytokines were determined by ELISA on cell supernatants.

For intracellular IFN-γ staining, spleens and lungs were harvested 9 d postvaccination, and single-cell suspensions were obtained as described for DC isolation from spleen. Splenocytes were seeded at 2 × 10⁶ cells/ml in 500 μl RPMI in a 48-well plate and stimulated with the MHC-I–restricted peptide IYSTVASSL peptide (2 μg/ml). Cells were stimulated for 20 h at 37°C with GolgiStop being added for the last 4 h. Cells were subsequently stained with anti-CD3, anti-CD4, anti-CD8, anti-CD19, anti-CD14, and anti–IFN-γ.

In vivo cytotoxicity assays were modified from Durward et al. (38). Splenocytes were harvested and incubated with the influenza HA (PR8) peptide IYSTVASSL (BALB/c) or the OVA-derived peptide SIINFEKL (C57BL/6) or a negative control peptide at a density of 5 × 10⁷ cells/ml for 1 h at 4°C before being transferred to 37°C for an additional 30 min. Peptide-loaded cells were washed twice in PBS and subsequently stained with 1.25 (negative control) or 12.5 μM (IYSTVASSL or SIINFEKL) CTV at a density of 5 × 10⁷ cells/ml for 20 min at 37°C. After being washed twice in PBS, the cells were resuspended in PBS at a density 5 × 10⁷ cells/ml, and the two populations mixed 1:1. A total of 1 × 10⁷ mixed cells were injected i.v. into BALB/c or C57BL/6 mice immunized 9 d earlier with HA or OVA-containing fusion vaccines, respectively. Spleens were harvested 18 h later, and the ratio of CTV^low to CTV^high cells was determined by flow cytometric analysis. Cytotoxicity calculated as % specific lysis = [1− (nontransferred control ratio/experimental ratio)] × 100.

All statistical analyses were performed using the GraphPad Prism 7 software. Significant differences in Ab responses were calculated using nonparametric one-way ANOVA with Dunn multiple comparison correction, whereas differences in T cell responses or T cell proliferation were calculated using parametric one-way ANOVA with Tukey multiple comparison correction. Differences in weight curves postinfection were calculated using two-way ANOVA with Tukey multiple comparison correction. Differences in survival were calculated by Mantel–Cox with Bonferroni corrected threshold (*p < 0.05, **p < 0.01, and ***p < 0.001).

To compare different strategies aimed at delivering Ag to cDC1s, the surface receptors Xcr1, Clec9A, and DEC-205 were chosen as targets. To ensure that the fusion proteins were as similar as possible, we generated an scFv based on the anti-Clec9A clone 10B4 (19) and inserted it into a bivalent vaccine molecule (vaccibody) containing mCherry as an Ag (Supplemental Fig. 1A). We have previously generated fusion proteins containing an scFv specific for DEC-205 (referred to as scFv^DEC-205–mCherry) (33) in addition to Xcl1-fusion proteins targeting the Xcr1 receptor (28). As a nontargeted control, fusion vaccines containing an scFv^NIP was included (32). The specificity of scFv^Clec9A-mCherry was confirmed on Clec9A-transfected HEK293E cells (Supplemental Fig. 1B). No binding was seen to Clec9A-transfected cells with scFv^NIP-mCherry (Supplemental Fig. 1B). The specificity of scFv^DEC-205–mCherry was confirmed by preincubating splenocytes from BALB/c mice with the mAb that the scFv were derived from (NLDC145), which resulted in a complete blocking of binding (Supplemental Fig. 1C). Injection of Xcl1-, scFv^DEC-205–, or scFv^Clec9A-mCherry i.v. into BALB/c mice resulted in specific staining of cDC1s in spleen, confirming that all three fusion vaccines targeted the same DC population in vivo (Fig. 1A).

To test immune responses induced when targeting DEC-205, Clec9A, or Xcr1, we conjugated the scFv^Clec9A to HA from influenza A/PR/34/08 (PR8). We have previously generated Xcl1-HA and scFv^DEC-205–HA (28, 33). Expression of the HA fusion vaccines were evaluated by ELISA to ensure that the different constructs were secreted at comparable levels (Supplemental Fig. 1D).

Although the rational for targeting Ag to DCs generally is to enhance T cell responses, several studies from us and others have demonstrated that targeting cDC1s via Xcr1, DEC-205, or Clec9A can also enhance Ab responses (19, 28, 29, 33, 39, 40). To evaluate Ab responses, BALB/c mice were immunized by i.d. injection of 25 μg of DNA encoding scFv^NIP-, scFv^DEC-205–, Xcl1-, or scFv^Clec9A-HA, followed by electroporation of the injection site. The i.d. delivery was chosen because this delivery route gives equal or better responses compared with i.m. DNA vaccination when targeting cDC1s (41). Serum samples were harvested after 2, 6, and 12 wk and evaluated for HA-specific IgG (Fig. 1B). Immunization with scFv^Clec9A-HA induced significantly higher anti-HA IgG titers compared with scFv^NIP-HA and scFv^DEC-205–HA 2 wk after vaccination (Supplemental Fig. 1E). Immunization with Xcl1-HA also induced significantly higher IgG titers compared with scFv^DEC-205–HA after 2 wk (Supplemental Fig. 1E). Indeed, immunization with scFv^DEC-205–HA generally induced poor Ab responses in BALB/c, and scFv^Clec9A-HA, Xcl1-HA, and scFv^NIP-HA all induced significantly higher anti-HA IgG titers after 6 and 12 wk (Fig. 1B). This was unlikely because of inefficient expression or secretion of the scFv^DEC-205–HA fusion vaccine, as all vaccine constructs expressed at similar levels in transient transfection experiments in in vitro (Supplemental Fig. 1D).

To further test the quality of the induced Ab responses, an avidity ELISA was performed on serum samples harvested from scFv^NIP-, Xcl1-, and scFv^Clec9A-HA–immunized mice. After 12 wk, serum from scFv^Clec9A-HA–immunized mice displayed significantly higher Ab avidity compared with Xcl1-HA (Fig. 1C), although no difference was seen between scFv^NIP-HA– and scFv^Clec9A-HA–immunized mice at this time point. However, immunization with scFv^Clec9A-HA induced Abs with significantly higher neutralizing capability compared with scFv^NIP-, scFv^DEC-205–, and Xcl1-HA (Fig. 1D). Together, these results suggest that scFv^Clec9A-HA immunization induces Abs with a higher quality compared with the other targeting strategies.

We have previously observed that targeting Ag to the Xcr1 receptor by i.d. DNA vaccination induces an Ab response dominated by the Th1-associated IgG subclass IgG2a (28, 29, 41). To evaluate if targeting cDC1s via Clec9A or DEC-205 induced similarly polarized Ab responses, BALB/c mice were immunized once by i.d. DNA vaccination as in Fig. 1. Serum samples were harvested after 2, 6, and 12 wk and evaluated for induction of HA-specific Abs of the IgG1, IgG2a, or IgG2b subclass (Fig. 2A, Supplemental Fig. 2A, 2B). Mice immunized with Xcl1-HA or scFv^Clec9A-HA induced significantly higher Ab titers of the IgG2a subclass compared with scFv^NIP-HA and scFv^DEC-205-HA and consequently a high IgG2a/IgG1 ratio indicative of a Th1-polarized response (Fig. 2A, 2B). Xcl1-HA had a tendency to induce higher titers of IgG2b after 2 wk, but the difference to scFv^NIP-HA and scFv^Clec9A-HA disappeared by week 6 (Supplemental Fig. 2A). By contrast, we observed highest IgG1 titers with scFv^Clec9A-HA, although the difference was only significant when compared with scFv^DEC-205–HA (Fig. 2A). No difference in Ab titer was seen between scFv^Clec9A-HA and Xcl1-HA at weeks 6 and 12 for any of the subclasses (Supplemental Fig. 2A, 2B). scFv^DEC-205–HA induced relatively low levels of all the subclasses, correlating to the total IgG results in Fig. 1 (Fig. 2A, Supplemental Fig. 2A, 2B).

To test if the observed Ab responses were unique to BALB/c mice, the immunization experiments were repeated in C57BL/6 mice (Fig. 2C, 2D). In accordance with the observations in BALB/c, immunization with Xcl1-HA or scFv^Clec9A-HA induced strong IgG2c responses after 2 wk. Interestingly, immunization with scFv^DEC-205–HA induced stronger Ab responses in C57BL/6 mice compared with BALB/c, whereas the scFv^NIP-HA induced very low Ab responses in C57BL/6 (Fig. 2, Supplemental Fig. 2). Consequently, scFv^DEC-205–HA, Xcl1-HA, and scFv^Clec9A-HA all induced significantly higher titers of IgG2c and IgG2b compared with scFv^NIP-HA. scFv^Clec9A-HA induced significantly higher IgG1 responses compared with Xcl1-HA, whereas Xcl1–HA induced a higher IgG2c/IgG1 ratio compared with scFv^Clec9A-HA (Fig. 2D). In summary, the Ab responses obtained after Xcl1-HA or scFv^Clec9A-HA immunization were similar in BALB/c and C57BL/6.

The observed Ab responses suggest differences in how targeting Xcr1, Clec9A, or DEC-205 influence polarization of CD4⁺ T cell responses. To test this possibility, we took advantage of the TCR transgenic DO11.10 mouse strain that contains a TCR specific for OVA_323–339 presented on I-A^d (42). Xcl1-, scFv^DEC-205–, scFv^Clec9A-, and scFv^NIP-fusion proteins with OVA as an Ag were generated and purified as previously described (28). Next, BM DCs were incubated with DO11.10 CD4⁺ T cells in the presence of 0.5 μg/ml scFv^NIP-, scFv^DEC-205–, Xcl1-, or scFv^Clec9A-OVA protein for 72 h and polarization toward Th1, Th2, or Th17 evaluated by analyzing expression of the transcription factors T-bet, GATA3, or RORγt, respectively. Incubation with Xcl1-OVA induced a significantly higher percentage of T-bet⁺ DO11.10 CD4⁺ T cells compared with scFv^Clec9A-OVA and scFv^DEC-205–OVA (Fig. 3A). In addition, expression of T-bet was higher in the T-bet⁺ population after incubation with Xcl1-OVA, as determined by mean fluorescence intensity (MFI), compared with scFv^Clec9A-OVA and scFv^DEC-205-OVA (Fig. 3B). Although incubation with Xcl1-OVA also resulted in a higher percentage of GATA3⁺ cells (Supplemental Fig. 3A), the level of expression of GATA3 was not different from cells incubated with scFv^DEC-205–OVA, scFv^Clec9A-OVA, or scFv^NIP-OVA (Supplemental Fig. 3B).

Next, we tested if cytokine secretion reflected the observed expression of transcription factors in DO11.10 CD4⁺ T cells. Incubation with Xcl1-OVA and scFv^Clec9A-OVA induced a strong increase in the secretion of IFN-γ that was significantly higher than incubation with scFv^DEC-205–OVA (Fig. 3C). Interestingly, scFv^Clec9A-OVA also induced significantly higher secretion of IL-12, most likely from deriving from the DCs (43), whereas Xcl1-OVA induced higher secretion of TNF-α compared with the other fusion molecules (Fig. 3C). When considering secretion of Th2 associated cytokines, Xcl1-OVA, scFv^Clec9A-OVA, and scFv^DEC-205–OVA induced similar levels of IL-13, whereas Xcl1-OVA and scFv^DEC-205–OVA induced higher secretion of IL-5 compared with scFv^Clec9A-OVA (Supplemental Fig. 3C).

Surprisingly, we observed that incubation with scFv^DEC-205– and Xcl1-OVA induced a high percentage of RORγt⁺ DO11.10 CD4⁺ T cells in vitro, which was significantly higher than scFv^Clec9A-OVA (Fig. 3D). There was, however, no difference between scFv^DEC-205–, Xcl1-, and scFv^Clec9A-OVA when evaluating MFI of the RORγt⁺ DO11.10 CD4⁺ T cells (Fig. 3E). Nevertheless, we observed significantly higher secretion of IL-17A from cells incubated with scFv^DEC-205–OVA compared with scFv^NIP-, Xcl1-, and scFv^Clec9A-OVA (Fig. 3F).

cDC2s have been reported to express DEC-205 (16) and to induce Th17 polarization (44, 45). To test if the increased IL-17A secretion we observed with scFv^DEC-205–OVA was related to cDC2 targeting, we sorted BM DCs into cDC1s and cDC2s and incubated with CTV-stained DO11.10 CD4⁺ T cells and purified fusion protein. Proliferation of CD4⁺ T cells was evaluated by flow cytometry (Fig. 4A). Xcl1-, scFv^Clec9A-, and scFv^DEC-205–OVA all induced similar level of proliferation when incubated with cDC1s, whereas scFv^DEC-205–OVA induced significantly higher proliferation when incubated with cDC2s (Fig. 4A, Supplemental Fig. 3D). We also observed higher proliferation with Xcl1-OVA on cDC2s compared scFv^Clec9A-OVA, which may be related to the glycosaminoglycan-binding properties of Xcl1 (46). Interestingly, we observed increased secretion of IL-17A after incubation with scFv^DEC-205–OVA on both cDC1s and cDC2s (Fig. 4B). In contrast, incubation with Xcl1 induced higher secretion of IFN-γ on cDC1s (Fig. 4C). These observations confirm that the differences in polarization between especially Xcl1- and scFv^DEC-205–OVA are related to the targeted receptor and not simply due to targeting different DC subsets.

To assess if the in vitro polarization was reflected in vivo, 1 × 10⁶ purified DO11.10 CD4⁺ T cells were transferred to naive mice prior to i.d. DNA vaccination with scFv^NIP-, scFv^DEC-205–, Xcl1-, and scFv^Clec9A-OVA. Draining inguinal and axillary LN were harvested 5 d after immunization, and DO11.10 CD4⁺ T cells were evaluated for expression of Th associated transcription factors. Immunization with scFv^DEC-205– and Xcl1-OVA induced a higher percentage of T-bet⁺ DO11.10 CD4⁺ T cells compared with scFv^NIP-OVA and scFv^Clec9A-OVA immunization (Fig. 5A). Xcl1-OVA also induced a higher expression of T-bet compared with scFv^DEC-205–OVA and scFv^Clec9A-OVA, as determined by MFI of the T-bet⁺DO11.10⁺ CD4⁺ T cells, although the difference did not reach significance (Fig. 5B). ELISAs on supernatants from stimulated splenocytes resulted in significantly increased secretion of IFN-γ and TNF-α after immunization with Xcl1-OVA, compared with scFv^NIP-, scFv^DEC-205–, and scFv^Clec9A-OVA, confirming the enhanced Th1 polarization seen with the Xcl1 fusion vaccine (Fig. 5C).

Corroborating the in vitro observations, immunization with scFv^DEC-205–OVA induced the highest percentage and MFI of RORγt⁺ DO11.10 CD4⁺ T cells (Fig. 5D, 5E). However, we did not observe a consistent secretion of IL-17A from stimulated splenocytes from any of the immunized groups (data not shown). It should be noted that immunization with scFv^DEC-205–OVA also induced the highest frequency of GATA3⁺ DO11.10 CD4⁺ T cells, with a higher MFI for GATA3 compared with Xcl1-OVA (Supplemental Fig. 3E, 3F).

Together, these data indicate that targeting Ag to Clec9A, and especially Xcr1, by i.d. DNA vaccination results in a Th1-polarized immune response, whereas targeting DEC-205 might induce a more mixed Th response.

We have previously observed that Xcl1-targeted fusion proteins induce cytotoxic CD8⁺ T cell responses when administered as either DNA or protein (28, 30). Similarly, enhanced CD8⁺ T cell responses have also been reported when targeting DEC-205 and Clec9A (11, 18, 19). To test if the three targeting strategies differed in their ability to induce cytotoxic T cells when given as DNA vaccines, BALB/c mice were immunized as above and evaluated for the presence of Ag-specific IFN-γ⁺CD8⁺ T cells. Lungs and spleens were harvested 9 d after immunization, and single-cell suspensions were stimulated with the MHC-I–restricted peptide IYSTVASSL (Fig. 6A, 6B). Mice immunized with the targeted Xcl1-HA, scFv^DEC-205–HA or scFv^Clec9A-HA all induced Ag-specific IFN-γ⁺CD8⁺ T cells in the lung, although only Xcl1-HA and scFv^DEC-205–HA induced significantly higher responses compared with scFv^NIP-HA (Fig. 6A). Immunization with Xcl1-HA also resulted in significantly higher number of IFN-γ⁺CD8⁺ T cells in spleen compared with scFv^NIP-HA (Fig. 6B). To evaluate the functionality of the CD8⁺ T cell responses, we performed an in vivo cytotoxicity assay 9 d after immunization with 25 μg DNA encoding scFv^NIP-, scFv^DEC-205–, Xcl1-, or scFv^Clec9A-HA. In accordance with the induction of IFN-γ⁺ CD8⁺ T cells, immunization with Xcl1-HA induced the highest cytotoxicity, significantly higher than that seen in scFv^DEC-205– and scFv^Clec9A-HA–immunized mice (Fig. 6C).

To test if the observed cytotoxicity was influenced by mouse strain, we immunized C57BL/6 mice with 25 μg DNA encoding scFv^DEC-205–OVA, Xcl1-OVA, or scFv^Clec9A-OVA and performed an in vivo cytotoxicity assay. OVA was chosen as an Ag in C57BL/6 mice because of the lack of known H2^b CD8⁺ T cell epitopes from HA (PR8). As was observed in BALB/c, Xcl1-OVA immunization induced the strongest cytotoxicity that was significantly higher than scFv^Clec9A-OVA immunization, although not significantly different to scFv^DEC-205–OVA immunization (Fig. 6D). scFv^DEC-205–OVA also induced stronger cytotoxic responses compared with scFv^Clec9A-OVA in C57BL/6. When lowering the amount of DNA used for immunization to 5 μg, Xcl1-OVA–vaccinated mice displayed a significantly higher cytotoxicity than both scFv^DEC-205–OVA and scFv^Clec9A-OVA (Fig. 6E), whereas scFv^DEC-205–OVA maintained stronger cytotoxic responses compared with scFv^Cleca9-OVA. The observation that Xcl1-OVA induced almost 100% cytotoxicity even when the immunization dose was lowered to 5 μg DNA suggests that the chemokine is highly efficient at inducing cytotoxic T cell responses in C57BL/6 mice.

Differences in cytotoxic T cell responses may be due to intrinsic differences in the endocytosis process of the targeted receptors, resulting in varying ability to present Ag derived peptides to CD8⁺ T cells. We therefore sorted BM-derived cDC1 and cDC2 from C57BL/6 mice and incubated cells with purified OT-I CD8⁺ T cells and scFv^DEC-205–OVA, Xcl1-OVA, or scFv^Clec9A-OVA before evaluating proliferation (Fig. 6F). No difference was observed with sorted cDC1, suggesting that delivering Ag to the DEC-205, Xcr1, or Clec9A results in equal peptide presentation on MHC-I in vitro (Fig. 6F, Supplemental Fig. 3G). Somewhat surprisingly, incubation with scFv^DEC-205–OVA with cDC2s induced strong proliferation of OT-I CD8⁺ T cells, suggesting that these cells are equally capable of cross-presenting Ag when DEC-205 is targeted (Fig. 6F, Supplemental Fig. 3G).

To test how the different cDC1 targeting strategies compared in terms of protective efficacy, BALB/c mice were immunized once with scFv^DEC-205–, Xcl1-, or scFv^Clec9A-HA DNA i.d. and subsequently challenged with a 5XLD₅₀ of influenza A/PR/34/8 H1N1 (PR8) virus after 2 wk. Weight loss was monitored as a sign of disease progression, and mice were euthanized if they lost more than 20% of the starting weight as a humane end point. Both Xcl1-HA– and scFv^Clec9A-HA–immunized mice only suffered slight weight reduction before returning to their starting weight by day 8 postinfection (Fig. 7A). Consequently, all mice survived the infection (Fig. 7B). Mice immunized with scFv^DEC-205–HA lost significantly more weight compared with Xcl1-HA and scFv^Clec9A-HA, although 60% of the mice eventually recovered and survived the challenge (Fig. 7A, 7B). Similar results were observed when C57BL/6 were immunized and challenged with 5xLD₅₀ PR8 (Supplemental Fig. 4A, 4B). When BALB/c mice were immunized as in Fig. 5A and challenged with 50xLD₅₀ after 2 wk, both scFv^Clec9A-HA– and Xcl1-HA–immunized mice lost more weight initially but recovered by day 8 postinfection, and all mice survived the challenge experiment (Supplemental Fig. 4C, 4D). By contrast, mice immunized with scFv^DEC-205–HA quickly succumbed to the infection, and 80% of the mice had to be euthanized (Supplemental Fig. 4C, 4D). Next, we evaluated the ability of the different targeting strategies to induce long-term protection by challenging BALB/c mice immunized as in Fig. 7A with 5xLD₅₀ 12 wk after vaccination. Once again, mice immunized with scFv^Clec9A-HA and Xcl1-HA were fully protected from infection (Fig. 7C, 7D). Mice immunized with scFv^DEC-205–HA succumbed to the infection, suggesting that the protection seen in Fig. 7A was only transient and did not result in long-term protective responses (Fig. 7C, 7D).

Finally, we evaluated if Abs could mediate protection after immunization with the different targeting strategies. BALB/c mice were immunized twice with DNA encoding scFv^DEC-205–, Xcl1-, or scFv^Clec9A-HA 3 wk apart. Serum was harvested 6 wk after the first vaccination (3 wk after boost) and transferred to naive mice that were subsequently challenged with 5xLD₅₀ PR8 (Fig. 7E, 7F). All immunized mice initially lost weight, whereas 6/6 mice receiving serum from Xcl1-HA–immunized mice and 3/6 receiving serum from scFv^Clec9A-HA–immunized mice eventually recovered from infection. Only 1/6 mice receiving serum from scFv^DEC-205–HA-immunized mice recovered from challenge. Nevertheless, serum from all immunized groups resulted in a significant delay in disease progression compared with NaCl mice (Fig. 7E, 7F). Surprisingly, transfer of serum from Xcl1-HA–immunized mice resulted in less weight loss than serum from scFv^Clec9A-HA–immunized mice, although the difference in survival was NS. When evaluating HA-specific IgG we observed that whereas scFv^Clec9A-HA immunization induced significantly higher Ab responses than Xcl1-HA after the first vaccination, there was no difference after boost (Supplemental Fig. 4E). Consequently, boosting was more efficient with Xcl1-HA, although we also observed an increase in Ab titer scFv^Clec9A-HA (Supplemental Fig. 4E). Nevertheless, these observations indicate that immunization with both Xcl1- or scFv^Clec9A-HA can induce Ab responses capable of protecting mice against a lethal challenge with influenza virus.

In conclusion, our study indicates that delivering Ag to DEC-205, Xcr1, or Clec9A results in different immune responses, which may impact efficacy in a disease model.

Delivering Ags to conventional DCs is an attractive strategy for enhancing immune responses that may benefit the development of vaccines against cancer and infectious diseases. In this study, we have evaluated the effect of delivering Ag to different surface receptors on cDC1s, with focus on Ab responses, CD4⁺ T cell polarization, induction of cytotoxic T cell responses, and ability to mediate protection in an influenza infection model. Our observations suggest that the choice of receptor to target significantly impacts the immune response and protection against influenza infection after i.d. DNA immunization.

Our observations differ from a previous study by Idoyaga and colleagues (47) in which targeting Ag to DEC-205, langerin, or Clec9A induced similar Th1 and CD8⁺ T cell responses. It should, however, be noted that the immunizations performed by Idoyaga et al. (47) were predominantly done in combination with adjuvant in the form of poly(I:C) and anti-CD40, which may mask differences between the targeting units. Indeed, use of different adjuvants can influence T cell polarization when delivering Ag to cDC1s (48). In our study, we observe clear differences in CD8⁺ T cell induction and Th polarization after i.d. DNA vaccination when targeting Xcr1, DEC-205, or Clec9A. Similarly, targeting Clec9A resulted in a strong induction of T_FH in the absence of adjuvant, which was not seen when targeting DEC-205 (17). Although we do not add adjuvant to the DNA vaccine, the injection site is electroporated, resulting in local tissue damage that may attract and activate APCs and therefore have an adjuvant effect (49). Nevertheless, the differences in CD4⁺ T cell polarization observed in vitro was largely maintained in vivo, suggesting that any adjuvant effect from electroporation does not overpower the effect of targeting.

In this study, we have used the chemokine Xcl1 or scFvs specific for DEC-205 or Clec9A for targeting cDC1s. It is possible that the chemokine and the scFvs differ in how they interact with their respective receptors, although neither Xcl1, anti–DEC-205 (clone NLDC145), nor anti-Clec9A (clone 10B4) activate cDC1s (17, 41, 50). Because Xcl1 is chemotactic, it may enhance internalization of the Xcr1 receptor resulting in increased Ag loading. However, when incubating OT-I CD8⁺ T cells with BM DCs and purified Xcl1-, scFv^DEC-205–, or scFv^Clec9A-OVA, we did not see any difference in proliferation of CD8⁺ T cells, suggesting that the different targeting receptors can result in efficient processing and presentation of peptides on MHC-I. Consequently, receptor activation does not appear to have had a major impact on the CD8⁺ T cell responses. Nevertheless, the Xcl1 fusion vaccines induces stronger cytotoxic T cell responses in vivo, irrespective of Ag or mouse strain. Because Xcl1 is a chemokine, it is possible that it attracts more cDC1s to the immunization site, which in turn potentiate the CD8⁺ T cell responses.

The results obtained with the Xcl1- and scFv^Clec9A-OVA correlated well with previous reports, demonstrating that cDC1s preferentially induce Th1 responses (5, 6, 51). There was a tendency for Xcl1-OVA to induce higher expression of T-bet and a more Th1-polarized Ab response 2 wk after vaccination compared with scFv^Clec9A-OVA. scFv^Clec9A-OVA did, however, induce higher secretion of IL-12 in vitro and induced higher titers of IgG2a at 12 wk compared with Xcl1-OVA. Whether these differences were due to efficacy of cDC1 targeting when using Xcl1-OVA versus scFv^Clec9A-OVA or if the targeted receptors influence Th1 polarization is still unclear. However, it should be noted that Xcl1 has previously been reported to enhance Th1 polarization (52).

By contrast, immunization with scFv^DEC-205–OVA resulted in a more mixed Th polarized response. Previous studies have reported that cDC1s also can polarize T cells toward Th17 (48), which fits well with our in vitro observations in which scFv^DEC-205–OVA induced increased IL-17A secretion when incubated with DO11.10 CD4⁺ T cells and sorted cDC1s. Interestingly, targeting Ag to Clec9A in the presence of the adjuvant curdlan can also induce Th17 cells (48). Because all vaccines were delivered as DNA i.d. followed by electroporation, the mixed polarization should be related to targeting the DEC-205 receptor and not the mode of delivery. Collectively, our data strongly suggest that the choice of receptor to target, even on the same DC population, can influence the resulting immune response.

Targeting Clec9A resulted in the strongest Ab responses after one vaccination, as determined by titer, avidity, and virus neutralization. Although not explored in this study, this could be due to enhanced induction of T_FH, as previously reported (17, 26). Somewhat surprisingly, we observed that transfer of serum from Clec9A-immunized mice to naive mice after two immunizations only provided partial protection against infection. However, previous studies have similarly seen that there is a limited effect of boosting with targeting Clec9A (53), which may explain why we see a better boosting effect and overall protection with serum from Xcl1-HA–immunized mice.

In contrast, targeting DEC-205 generally resulted in poorer Ab responses, which were particularly striking in BALB/c mice. Although delivery of Ag to DEC-205 have been reported to enhance Ab induction in BALB/c (39), other studies in both BALB/c and primates have yielded more modest Ab responses (33, 54). It is currently unclear why targeting DEC-205 results in poor Ab responses, although it may be due to rapid clearance of the Ag. We have previously observed that targeting Ag to cDC1s without inducing active endocytosis enhances Ab responses (29, 40). Interestingly, a study by Reuter and colleagues (55) observed that delivering Ag to DEC-205 was associated with rapid endocytosis of the Ag. Broader expression on other cell types may also result in more rapid clearance of the Ag.

Much of our current understanding of the conventional DC subsets stems from experiments performed in mice. However, based on transcriptional and functional profiling, recent studies have identified the homologous cDC1 and cDC2 subsets in humans (reviewed in Ref. 3). Although expression of surface receptors can differ between mouse and human, both XCR1 and CLEC9A are predominantly found on human cDC1 (18, 19, 21, 56–58). DEC-205 is also expressed by human cDC1s but has been reported to have a broader expression profile in humans and can also be found on monocytes, B cells, and to a lesser degree on NK cells, plasmacytoid DCs, and T cells (59). So far, only DEC-205–targeted vaccines have been evaluated in clinical trials and have been found to induce Ag-specific immune responses against the melanoma Ag NY-ESO-1 (60). Enhanced priming of CD4⁺ and CD8⁺ T cells have, however, also been observed in vitro when targeting Ag to human CLEC9A in vitro (61). In addition, targeting CLEC9A or XCR1 has been shown to enhance Ab responses in macaques (25) and pigs (62), respectively. These observations indicate that the strategy of targeting cDC1s can enhance immune responses in a number of species, including humans. However, it is yet unclear if targeting cDC1s also can impact polarization of the immune response in species other than mice.

We thank Peter Hofgaard for technical assistance with experiments and animal handling and the Center for Comparative Medicine at Oslo University Hospital.

B.B. is inventor on patents on the vaccine molecules described in this article (vaccibodies). He is leader of the scientific panel of Vaccibody AS and has shares in the company. The other authors have no financial conflicts of interest.