Antibodies Targeting Clec9a Promote Strong Humoral Immunity Without Adjuvant in Mice and Non-Human Primates

854 Jessica Li et al. DOI: 10.1002/eji.201445127 Eur. J. Immunol. 2015. 45: 854–864

Antibodies targeting Clec9A promote strong humoral immunity without adjuvant in mice and non-human primates

Jessica Li1,2, Fatma Ahmet1, Lucy C. Sullivan2, Andrew G. Brooks2, Stephen J. Kent2, Robert De Rose2, Andres M. Salazar3, Caetano Reis e Sousa4, Ken Shortman1,5,6, Mireille H. Lahoud∗1,7, William R. Heath∗2 and Irina Caminschi∗1,2

1 Centre for Biomedical Research, Burnet Institute, Melbourne, Victoria, Australia 2 Department of Microbiology and Immunology, Peter Doherty Institute for Infection and Immunity, The University of Melbourne, Victoria, Australia 3 OncoVir, Oncovir Inc. NW, Washington DC, USA 4 Immunobiology, Cancer Research UK, London Research Institute, London, UK 5 The Walter and Eliza Hall Institute of Medical Research, Victoria, Australia 6 Department of Medical Biology, University of Melbourne, Parkville, Victoria, Australia 7 Department of Immunology, Monash University, Melbourne, Australia

Targeting antigens to dendritic cell (DC) surface receptors using antibodies has been successfully used to generate strong immune responses and is currently in clinical trials for cancer immunotherapy. Whilst cancer immunotherapy focuses on the induction of CD8+ T-cell responses, many successful vaccines to pathogens or their toxins utilize humoral immunity as the primary effector mechanism. Universally, these approaches have used adjuvants or pathogen material that augment humoral responses. However, adjuvants are associated with safety issues. One approach, successfully used in the mouse, to generate strong humoral responses in the absence of adjuvant is to target antigen to Clec9A, also known as DNGR-1, a receptor on CD8α+ DCs. Here, we address two issues relating to clinical application. First, we address the issue of variable adjuvant-dependence for different antibodies targeting mouse Clec9A. We show that multiple sites on Clec9A can be successfully targeted, but that strong in vivo binding and provision of suitable helper T cell determinants was essential for efﬁcacy. Second, we show that induction of humoral immunity to CLEC9A-targeted antigens is extremely effective in nonhuman primates, in an adjuvant-free setting. Our ﬁndings support extending this vaccination approach to humans and offer important insights into targeting design.

Keywords: Adjuvant r Clec9A r Dendritic cells r Humoral immunity r Non-human primates r Vaccines

Additional supporting information may be found in the online version of this article at the publisher’s web-site

Correspondence: Dr. Irina Caminschi ∗These authors contributed equally to this work. e-mail: [email protected]

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu Eur. J. Immunol. 2015. 45: 854–864 Immunomodulation 855

Introduction Ag to Clec9A has been extended to non-human primates. A single dose of either of two anti-human CLEC9A mAb induced high anti- The direct delivery of antigen (Ag) to DCs has been utilized in body titers in pigtail macaques (Macaca nemestrina) equivalent to many experimental models to successfully immunize against var- those seen in non-human primates immunized with anti-CLEC9A ious agents [1–4]. Though there are many ways to deliver Ag to plus adjuvant. DCs, one common approach is to generate monoclonal antibodies (mAbs) that recognize DC-specific surface molecules and then chemically or genetically engineer them to carry antigenic cargo. Results This approach has been used to deliver Ag to different DC receptors with varying immunological outcomes [1–5]. Understanding Not all mAb to Clec9A generate humoral immunity in the rules that govern the immune responses evoked by target- the absence of adjuvant ing DCs through their various receptors underpins our capacity to rationally design vaccines and is clinically relevant as DC-targeting In a number of experimental settings, the rat anti-Clec9A mAb strategies have now entered phase I/II clinical trials [2]. 10B4 was used to deliver Ag (ovalbumin (OVA), nitrophenol (NP)) Preclinical data suggest that delivering Ag to Clec9A is an ideal to DCs and induce potent Ab responses in the absence of adjuvant strategy for DC-based immunotherapy [6–10]. Clec9A, also known [6, 10]. Since the 10B4 rat mAb is foreign in the mouse, the tar- as DNGR-1, is a DC-specific molecule expressed highly on CD8+ geting mAb delivering the Ag also acted as an Ag, eliciting an DCs and CD103+ DCs and at lower levels on pDCs [7, 8, 11–13] anti-rat Ab Ig response [6, 8]. By contrast, 7H11, another rat and DC precursors [12, 14]. In a physiological setting, Clec9A facil- anti-Clec9A mAb did not induce strong anti-rat Ig Ab responses itates the cross-presentation of dead cell associated Ag [15–17], unless injected with an adjuvant [9]. Since neither 10B4 nor 7H11 by recognizing filamentous actin (F-actin) exposed on necrotic directly activated DCs [6, 17] another property of these mAb must cells [18, 19]. Delivering mAb-targeted Ag to this receptor (in dictate their capacity to induce anti-rat humoral immunity. To the presence of adjuvant) and hijacking the inherent capacity of identify the factors that support the induction of humoral immu- Clec9A to promote crosspresentation has been successfully used nity in the absence of adjuvants, we analyzed a panel of anti- to induce potent cytotoxic CD8+ T-cell responses [6, 7] capable of Clec9A mAb correlating their physical properties with their abil- destroying established tumors [7]. ity to induce humoral immunity. We have collectively generated In addition to inducing strong CD8+ T-cell responses, targeting five anti-Clec9A mAb: two rat IgG2a mAb (10B4 and 397) and Clec9A can generate potent humoral immunity [6, 8–10, 12], an three rat IgG1 mAb of which two express the kappa light chain effector mechanism associated with many successful vaccines to (7H11, 1F6) and one expresses the lambda light chain (42D2) agents such as viruses and bacterial toxins [20]. The novel and (Supporting Information Fig. 1). Each mAb was injected into a important advantage of targeting Clec9A is that strong humoral group of C57/BL6 (B6) mice with or without addition of poly I:C responses can be induced in the absence of adjuvant or any signs of as an adjuvant and the anti-rat Ab responses were measured by DC activation [6, 8, 10, 12]. Since the use of adjuvants in vaccines ELISA at weeks 2 (Fig. 1A) and 4 (Fig. 1B). This parallel com- can be associated with safety issues, development of adjuvant-free parison of the various anti-Clec9A mAb confirmed our individual vaccination approaches is highly attractive. Furthermore, adju- original observations that not all mAb were capable of inducing vant independence would greatly simplify production and clinical humoral responses in the absence of adjuvant. As shown previ- application. ously, anti-Clec9A mAb 10B4 induced anti-rat Ab responses in Here, we address two major hurdles in the development of an the absence of adjuvant [6, 8], while 7H11 required adjuvant to adjuvant free vaccine targeted to Clec9A. The first is to explain the induce humoral responses [9]. In addition, 42D2 was effective contradictory reports of adjuvant-dependence for targeting Clec9A without adjuvant, while 1F6 only induced anti-rat responses in to generate humoral immunity in the murine model [6, 8, 9, 12]. the presence of adjuvant. 397 was poorly immunogenic under any To address this issue, the two groups reporting contradictory find- circumstance, though in the presence of adjuvant a small response ings have here collaborated in a comparative analysis of their var- could be detected but this was no different from that elicited by ious anti-Clec9A mAb. This collaborative study has confirmed all control nontargeting antibody. These findings confirmed that not original observations and clarified the basis for variation. The sec- all Clec9A-specific mAb are able to induce adjuvant-free humoral ond issue is to demonstrate that this strategy works in non-human immunity, despite immunogenicity in the presence of adjuvant. primates thereby enabling translation into the clinical setting. We show that for Clec9A to promote potent humoral immunity, the targeting mAb must efficiently bind Clec9A on DCs in vivo and Adjuvant-free immunogenicity is not dictated by the provide adequate helper epitopes to support the B cell responses. specific region of Clec9A targeted Our data cautions that in vitro binding assays may not always predict in vivo targeting capacity and reiterates that strong helper Several physical attributes of the mAb could account for their epitopes are an important component of designing targeted vac- immunogenicity. Some antireceptor mAb are known to directly cines for humoral responses. Importantly, the capacity to induce a activate DCs [21], but this is not the case with the anti-Clec9A potent humoral response in the absence of adjuvant by targeting mAb [6, 17]. Clec9A on DCs does not function as an activatory

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu 856 Jessica Li et al. Eur. J. Immunol. 2015. 45: 854–864

Figure 2. Anti-Clec9A mAb 42D2 recognizes a unique epitope of Clec9A. The capacity of 10B4 or 42D2 mAb to bind Clec9A on CHO transfectants was assessed in the presence or absence of blocking mAb. CHO- Clec9A transfectants were preincubated with isotype control mAb, or the panel of anti-Clec9A mAb (10B4, 42D2, 397, 7H11, 1F6), then stained with either 10B4-biotin or 42D2-biotin. Binding was detected with SA- PE and analyzed by ﬂow cytometry. Experiments were performed twice and representative data are presented.

the five anti-Clec9A mAbs (unlabeled), and then incubated with Figure 1. Variable antibody responses are induced by targeting biotinylated 10B4 or 42D2 mAb. Whilst unlabeled 42D2 inhibited μ Clec9A. Five B6 mice per group were injected i.v. with 0.5 g the binding of biotinylated 42D2, none of the other mAb pre- isotype control Ab (GL117, GL113) or anti-Clec9A mAb (10B4, 397, 42D2, 7H11, 1F6) in the presence or absence of poly I:C (50 μg). vented this binding (Fig. 2). Thus, immunogenic 42D2 recognized Plasma samples were taken (A) 2 and (B) 4 weeks postinjection and a distinct Clec9A epitope that was not recognized by any of the the anti-rat reactivity measured by ELISA. Experiments were per- other anti-Clec9A mAb, including the immunogenic 10B4 mAb. formed twice and pooled data are presented. Only one group of mice was injected with 0.5 μg GL117± poly I:C, but, similar results By contrast, 10B4 binding was effectively inhibited by all other were obtained using higher of doses GL117 (not shown). Each point anti-Clec9A mAb except 42D2 (Fig. 2). Thus, although both 42D2 represents the end point titer of one individual mouse and the and 10B4 promote humoral immunity, they recognize different inserted line represents the geometric mean. Statistical analysis was performed on the log-transformed data using unpaired two-tailed epitopes of Clec9A. Conversely, although 397, 7H11, and 1F6 rec- t-test and Welch correction. ognize an epitope that is proximal to the one recognized by 10B4, these three mAb are nonimmunogenic in the absence of adjuvant. pathogen recognition receptor and even binding its native ligand We further confirmed these binding properties by ELISA (Support- (dead cells) does not result in activation of DCs [17]. Consis- ing Information Fig. 2) and conclude that the site engaged by the tent with this, anti-Clec9A mAb 10B4 and 7H11 do not activate anti-Clec9A mAb does not correlate with their ability to promote DCs [6, 17]. However, Clec9A was shown to control endocytic humoral immunity. handling of necrotic cell antigens [17] and it is conceivable that the site at which the mAb bind Clec9A affects endocytic traffick- ing or some related attribute for immunogenicity. To determine Adjuvant-free immunogenicity is not dictated by mAb whether the immunogenicity of each anti-Clec9A mAb related to persistence in the serum the specific region of Clec9A recognized, we tested whether the five mAbs bound overlapping epitopes by competitive inhibition We had previously shown that the capacity of anti-Clec9A mAb assays. Clec9A transfectants were preincubated with either one of 10B4 to drive strong humoral responses correlated with its

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu Eur. J. Immunol. 2015. 45: 854–864 Immunomodulation 857

the longest period of time (7 days) (Fig. 3). Thus, all anti-Clec9A mAb persisted within blood of immunized mice comparably and this factor did not explain the divergence in immunogenicity.

The efﬁciency of binding to Clec9A contributes to adjuvant-free immunogenicity

Since we have shown that mAb were present at similar levels within the serum of vaccinated mice, we reasoned that their capacity to bind Clec9A may affect delivery of Ag to DCs. To this end, we measured the capacity of the mAb to bind Clec9A by surface plasmon resonance (Supporting Information Fig. 3) and found all five anti-Clec9A mAb had comparable binding affinities for soluble Clec9A. Since Clec9A is expressed on the cell surface of DCs, the various anti-Clec9A mAb might differ in their capacity to recog- Figure 3. In vivo persistence of anti-Clec9A mAb. Three B6 mice were injected i.v. with 2 μg of an anti-Clec9A mAb. Plasma samples were nize membrane-bound Clec9A. To test this hypothesis we stained collected 1 h later (day 0), then 2, 4, and 7 days later. The concentration CHO-Clec9A transfectants with graded concentrations of purified of mAb in the plasma was measured by ELISA. Each bar represents anti-Clec9A mAb. When comparing the two rat IgG2a mAb 10B4 pooled mean ± SEM, from 12 biological replicates over four independent experiments. Statistical analysis of antibody persistence on day 4 was and 397, the immunogenic 10B4 mAb clearly bound Clec9A more performed using unpaired two-tailed t-test. efficiently (Fig. 4A). Similarly, when comparing the IgG1 mAb, 42D2, 1F6, and 7H11, it was the immunogenic 42D2 that bound Clec9A most efficiently (Fig. 4B). Next, we confirmed this binding persistence within the serum, providing a constant source of Ag for pattern on freshly isolated DCs. As seen with the transfectants, over 4 days [6]. Thus, we assessed the persistence of the various 10B4 was significantly more effective at binding DCs than 397 anti-Clec9A mAb. All five Clec9A-specific mAb could be detected in (Fig. 4C). Similarly, although all of the IgG1 mAb bound Clec9A the plasma 4 days after injection and remained competent to bind on DCs well, 42D2 appeared particularly efficient (Fig. 4D). ligand ex vivo (Fig. 3). There was no clear correlation between the Ultimately, the ability of the anti-Clec9A mAb to bind DCs in levels of mAb persisting in the blood and the capacity to induce vivo and deliver antigenic cargo is the critical parameter. To assess humoral immunity without adjuvant; indeed the 397 mAb, least this, biotinylated or Alexa-488-conjugated anti-Clec9A mAb were capable of inducing humoral immunity, persisted in the blood for injected into mice and the DCs from these mice were isolated 1 h

Figure 4. Assessing the capacity of anti-Clec9A mAb to bind cell surface Clec9A. Graded doses of (A) IgG2a and (B) IgG1 anti-Clec9A mAbs were used to stain CHO-Clec9A transfectants. Bound antibodies were detected with anti-rat IgG PE by flow cytometry. Data are presented as the mean fluorescence intensity (MFI) obtained over a range of antibody concentrations. Experiments were performed at least twice and representative data are presented. (C, D) Enriched freshly isolated DCs cell preparations were stained with graded concentrations of biotinylated anti-Clec9A mAbs (10B4 and 397: 83, 42, 21, 10 μg/mL; 42D2, 7H11, and 1F6: 42, 10, 5, 2.6 μg/mL), then counterstained with SA-PE, anti-CD11c-FITC, and anti-CD8-APC mAb. The level of Clec9A staining on CD8+ DCs was analyzed by flow cytometry and is presented as MFI. Two independent experiments were performed and representative data are shown.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu 858 Jessica Li et al. Eur. J. Immunol. 2015. 45: 854–864

Figure 5. Persistence of anti-Clec9A mAbs on CD8+ DCsinvivo.TwoB6micewereinjectedi.v.with2μg of (A, C) A488 anti-Clec9A or (B, D) biotinylated mAb. Spleens were harvested 1 h or 4 days postinjection, pooled low density cells were enriched by density centrifugation and stained with (A) CD11c-PE and anti-CD8-APC or (B) SA-PE, anti-CD11c-FITC, and anti-CD8-APC. The level of Clec9A labeling of CD8+ DCs (CD11chiCD8+) was measured by flow cytometry. Black histograms depict Clec9A fluorescence, gray histograms depict background fluorescence of uninjected controls. Experiments were performed twice and representative histograms are presented. The mean fluorescence intensity (MFI) of each experiment is presented in panels C and D. The background fluorescence of uninjected controls was subtracted from the MFI of injected mice to obtain the MFI.

or 4 days later (Fig. 5). Strikingly, nonimmunogenic 397 did not hypervariable segments involved in Ag recognition, there is sig- target CD8+ DCs effectively in vivo, suggesting its failure to elicit nificant variation in the sequence of different heavy chain isotypes anti-rat reactivity was a direct result of its inability to bind Clec9A and the kappa or lambda light chains. In this respect, we noticed in vivo (Fig. 5A, B). Indeed, immunization with 397 even in the that in B6 mice, the rat IgG2a isotype control mAb appeared presence of adjuvant did not elicit strong Ab responses. Rather, to be more immunogenic than the rat IgG1 isotype control mAb in 397 behaved like the nontargeting isotype control inducing equiv- the presence of adjuvant (Fig. 1). To test whether some isotypes alently low Ab responses (Fig. 1). All other anti-Clec9A mAb effi- of the mAb were inherently more immunogenic, we immunized ciently targeted CD8+ DCs. When mice were injected with Alexa- B6 mice with nontargeting rat IgG2a-κ, IgG1-κ, and IgG1-λ in the 488 conjugated 42D2, 7H11, 1F6, and 10B4, Alexa-488-labeling presence of alum. The rat IgG2a-κ and IgG1-λ elicited anti-rat Ig was evident even 4 days after injection, representing internalized responses whereas the IgG1-κ was poorly immunogenic and did and cell surface bound mAb (Fig. 5A, C). To determine whether not elicit anti-rat Ig responses (Fig. 6). Interestingly, both 7H11 injected mAb were entirely internalized or also present on the and 1F6, which require adjuvant for priming humoral responses, cell surface, mice were injected with biotinylated mAb and the were IgG1-κ mAb, whereas 42D2, which primes in an adjuvant- presence of these mAb on the cell surface then determined using free manner, was a IgG1-λ, suggesting poor Ab backbone immuno- SA-PE (Fig. 5B, D). Surprisingly, three of the mAb (10B4, 1F6, genicity may underlie dependence on adjuvant. and 42D2) were still on the surface of DCs 4 days after injection. In summary, all anti-Clec9A mAb that induced potent humoral responses had the capacity to home to CD8+ DCs, but some of Altering the host mouse strain can convert these mAb (1F6, 7H11) required adjuvant for efficient priming, adjuvant-dependent responses to adjuvant whilst others did not (10B4, 42D2). Thus, whilst the ability to independent target to Clec9A in vivo was absolutely required to elicit humoral immunity, the capacity to bind Clec9A alone was insufficient to We speculated that changing the mouse strain immunized might induce humoral immunity in the absence of adjuvant. facilitate adjuvant-free vaccination by providing different helper epitopes not available in B6 mice. To test this hypothesis, we exam- ined the immunogenicity of each of the five mAb in BALB/c mice (Fig. 7), which express comparable levels of Clec9A (Supporting Antibody isotype immunogenicity affects Information Fig. 4). This revealed efficient (7H11) and interme- adjuvant-free humoral immunity diate (1F6) capacity to generate adjuvant free humoral immunity by mAb that were not effective in B6 mice. In BALB/c mice, 7H11 One remaining aspect to be explored was the immunogenic- was just as effective as 42D2. Similarly, the immunogenicity of ity of the targeting rat Ig antibody itself. Irrespective of the 397 was greater in BALB/c mice than in B6 mice (Fig. 7) although

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu Eur. J. Immunol. 2015. 45: 854–864 Immunomodulation 859

Targeting antigen to Clec9A in non-human primates promotes strong humoral immunity

While we have extensive murine data showing Clec9A-targeting can induce robust humoral immunity in the absence of adjuvant, one major step toward clinical application is to demonstrate fea- sibility in non-human primates. To address whether targeting Ag to Clec9A could induce a robust Ab response in non-human primates in an adjuvant-free setting, we utilized previously generated rat anti-human CLEC9A mAb that were found to crossreact with Clec9A expressed in Macaca nemestrina monkeys. Using these mAb we showed that the PBMC of Macaca nemestrina monkeys, similarly to humans, had a clear population of Clec9A+ DCs [8]. In this setting, rat-specific determinants of the immunoglobulin molecules themselves act as the Ag to which monkeys can make Figure 6. The rat IgG2a isotype is inherently more immunogenic than humoral responses, so no additional antigenic cargo is required. rat IgG1. Five B6 mice were injected i.p. with 2 μg of IgG2a isotype To test immunogenicity, Macaca nemestrina were immunized with (GL117), IgG1κ isotype (GL113), and IgG1λ isotype in the presence of absence of Alhydrogel. Two weeks later, reactivity against the orig- either of two IgG2a rat anti-CLEC9A mAb or an isotype control in inal immunogen was measured by ELISA. The experiment was per- the presence or absence of polyICLC as an adjuvant and their anti- formed twice and pooled data are presented. Only one group of mice rat Ig responses measured. Monkeys were bled 2 and 4 weeks post was injected with IgG1 isotype alone to confirm lack of immunity, because this immunization protocol consistently yielded poor anti-rat immunization. By comparing responses with and without adju- responses (see Fig. 2). Each point represents the end point titer of one vant, it was possible to assess the relative efficacy of adjuvant-free individual mouse and the inserted line represents the geometric mean. vaccination. Targeting these rat IgG2a mAb to Clec9A resulted in Statistical analysis was performed on the log-transformed data using unpaired two-tailed t-test and Welch correction. potent anti-rat Ig humoral responses, up to 100-fold higher than the responses seen in monkeys injected with nontargeted isotype control mAb (Fig. 8). Indeed the level of anti-rat Ig reactivity achieved by targeting Clec9A without adjuvant was equivalent to that induced by immunizing monkeys with nontargeting mAb plus adjuvant. This shows that adjuvant-free delivery of Ag to Clec9A is an effective way to promote humoral immunity, even in non- human primates. Thus, Clec9A is an ideal candidate for exploring the delivery of Ag to DCs where the primary goal is the induction of protective Ab responses in the absence of adjuvant.

Discussion

The central role that DCs play in the priming and maintenance of immune responses has made these APCs the focus of certain immunotherapies. In vitro propagated DCs loaded with tumor Ag have had anecdotal success but have not emerged as a general treatment of malignancy [22]. A simpler and more effective way Figure 7. Targeting Clec9A results in antibody responses in BALB/c to initiate immune responses may be to deliver Ag directly to μ mice. Five BALB/c mice were injected i.v. with 0.5 g isotype control Ab the DCs in vivo using mAb that recognize speciﬁc DC cell surface (GL117, GL113) or anti-Clec9A mAb (10B4, 397, 42D2, 7H11, 1F6). Plasma samples were taken 2 weeks postinjection and the anti-rat reactivity molecules. Delivering Ag to DCs in vivo negates cumbersome and measured by ELISA. Experiments were performed two to three times expensive in vitro manipulations, and consequently, is actively and pooled data are presented. Each point represents the end point titer being pursued as a novel vaccination platform [1, 2, 23]. A lead- of one individual mouse and the inserted line represents the geometric mean. Statistical analysis was performed on the log-transformed data ing candidate for delivery of Ag to DCs is the mAb against the using unpaired two-tailed t-test and Welch correction. multilectin DEC-205, which is currently being assessed in phase I/II clinical trails [2]. DEC-205 and Clec9A both target the CD8+ DCs lineage in mice and are comparable in their ability to pro- it was still poor, consistent with the weak capacity of the mAb mote potent cellular immunity and particularly crosspriming (pre- to bind Clec9A on DCs in vivo. Together, these data point to the sentation of exogenous Ag in the class I pathway for the activa- inherent immunogenicity of the antibody backbone as critical in tion of CD8+ T-cells) when Ag is delivered in the presence of a facilitating efﬁcient adjuvant-free humoral immunity. DC-maturation agent or adjuvant [6, 24]. However, one striking

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu 860 Jessica Li et al. Eur. J. Immunol. 2015. 45: 854–864

Figure 8. Targeting Ag to Clec9A in non-human primates elicits potent humoral responses in the absence of adjuvant. Macaca nemestrina were injected s.c. with 200 μg of IgG2a isotype (GL117), or anti-CLEC9A (rat IgG2a: clone 3A4 or 4C6) in the presence or absence of 200 μg polyICLC (Hiltonol). Blood samples were taken preimmu- nization (time = 0) and 2 and 4 weeks post immunization. Each point represents the end point titer of one individual mouse. Non-human primates immunized with anti-CLEC9A mAb 4C6 and 3A4 are represented by triangles pointing up or down, respectively. This experiment was performed once. The anti-rat Ig responses induced by immunizing with anti-CLEC9A mAb were sta- tistically signiﬁcant.

difference is that targeting Ag to Clec9A, but not DEC-205, will express IAb), we found that the rat IgG2a mAb appeared to be the result in potent humoral immunity in the absence of adjuvant. As most immunogenic, while the IgG1 isotype was poorly immuno- antibodies are the major effectors of most vaccines against infec- genic unless paired with a lambda light chain. This hierarchy tious diseases, adjuvant-free vaccination with Clec9A is a promis- matches well the immunogenicity of Clec9A-targeted mAb in B6 ing approach for the development of the next generation of such mice. Thus, the rat IgG2a mAb 10B4 is immunogenic and does not vaccines. require adjuvant to induce humoral immunity [6, 8, 12], whereas The ability to elicit antibody responses in the absence of adju- 7H11, a rat IgG1 mAb, is poorly immunogenic and requires adjuvant and without the additional consequence of inducing CTL may vant to induce humoral immunity [9]. While we have not defined be advantageous in various scenarios. For example, when humoral the precise nature underlying this immunogenicity, we favor the immunity is all that is required, or when CTL-mediated effec- view that immunogenicity relates to the presence of helper T-cell tor mechanisms might contribute to pathology, as in some viral epitopes within the immunoglobulin. Our previous studies indi- and parasitic diseases [25, 26]. By targeting Ag to Clec9A in the cate that effective generation of follicular helper T-cells induced absence of adjuvant any potential adverse effects associated with by Clec9A targeting is crucial to the resulting Ab response [6]. the adjuvant or with CTL may be avoided. Given these advantages, Since mouse Ab share a high homology with rat Ab, the number developing a Clec9A-targeting strategy is clinically relevant, and of T-cell epitopes that can be recognized as nonself likely varies this warrants elucidating the underlying mechanism. Since not all between isotypes, light chains, and even the idiotypes of the mAbs. Clec9A-specific mAb appeared to elicit strong humoral immunity, As alum was used as the adjuvant in our studies, no additional we set out to identify the critical attributes of the targeting mAb T-cell epitopes would be provided by the adjuvant. Thus, differ- that favor humoral immunity. One critical factor was the ability ences in immunogenicity would reflect the protein sequence of to effectively home to Clec9A on DCs in vivo. Although all anti- the mAb. Our hypothesis is supported by the increased immuno- Clec9A bound soluble Clec9A in vitro with comparable efficiency, genicity of anti-Clec9A mAb of the rat IgG1-κ type in BALB/c one of these mAb, 397, failed to target Clec9A in vivo and conse- mice relative to B6 mice, where alternative helper determinants quently did not induce Ab responses. Why this mAb bound soluble would be available for presentation by IEd or IAd. This aspect of Clec9A in vitro but was so ineffective in vivo can only be specu- the study suggests that targeting Ag that contain strong helper lated, but may relate to modifications to the binding site in vivo or T-cell epitopes is likely to improve humoral immunity to the tar- when Clec9A is associated with membranes. Potentially, it could geted Ag. As “humanized” Ab are likely to be used in the clinical also be caused by changes to the mAb itself in vivo, though this is setting, the helper T-cell epitopes would not be provided by the unlikely since 397 persisted well in vivo in serum and could still humanized immunoglobulin. Rather, the vaccine Ag attached to bind soluble Clec9A when recovered ex vivo. the humanized mAb will need to provide the helper epitope(s). While the ability to bind to Clec9A in vivo was essential, it was In the instance that the vaccine Ag is known to lack adequate not sufficient to ensure strong Ab responses, since mAb 7H11 and helper epitopes, the humanized monoclonal Ab can be engineered 1F6 showed strong in vivo binding but were still poorly immuno- to carry both vaccine Ag and a universal helper epitope such as the genic in B6 mice. Rather the mAb backbone, which in our experi- tetanus toxoid epitope. Another clinical consideration is whether mental setting also acted as the Ag and differed somewhat between to use adjuvant to enhance humoral immunity. In this vein, it is targeting mAb isotypes, showed variation in its immunogenicity. worth noting that although delivering Ag to Clec9A induces potent In studies comparing the immunogenicity of various heavy chain antibody responses, coadministration of adjuvant often enhances isotypes and light chains using alum as adjuvant in B6 mice (which humoral immunity. Indeed the synergistic effects of adjuvants and

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu Eur. J. Immunol. 2015. 45: 854–864 Immunomodulation 861

Clec9A targeting may be exploited in settings where it has proven and 4C6, or isotype control: GL117) together with, or without, difficult to raise Ab responses. 200 μg of polyICLC, as indicated. In summary, we provide several important advances in the development of an effective DC-targeting strategy for humoral immunity. Our study shows that targeting antigens to CLEC9A in Antibodies and fluorescent staining non-human primates is a highly efficient approach for generating humoral immunity, and that the two anti-human CLEC9A mAb we To stain mouse Ag the following fluorochrome-conjugated mAb have generated are effective. Interestingly, humoral immunity was were used: anti-CD11c (N418-PE or -FITC); anti-CD8 (YTS 169.4- not further enhanced by the administration of adjuvant, arguing APC); isotype control IgG2a-biotin (eBioscience); anti-Clec9A that maximum Ab responses had been generated. Future studies (397, 10B4, 42D2, 7H11, 1F6 as biotin or Alexa-488 conjugates). could focus on ascertaining the minimum dose of anti-Clec9A mAb Biotin was detected using streptavidin (SA) conjugated to phycoo- required to elicit maximum Ab responses and also determining erythrin (PE). Cells were incubated with mAb or SA-PE for 30 min the longevity of these responses. Furthermore, our rodent studies at 4°C. Fc-mediated binding was blocked by preincubating cells underscore the need for strong helper epitopes and efficient in (10 min 4°C) with rat Ig and anti-FcR mAb (2.4G2). Propidium vivo binding to the targeted DC determinant. Finally, the success iodide (PI) (0.5 μg/mL) was added to the final cell wash to allow of generating humoral immunity in non-human primates suggests exclusion of PI positive dead cells. that CLEC9A itself may be a preferred DC target over other cell surface molecules such as DEC-205 for generating adjuvant free priming for humoral immunity in humans. Isolation of DCs

DCs were isolated as previously described [6, 27]. Briefly, spleens Materials and methods were chopped into fine fragments and enzymatically digested with DNase/collagenase (20 min; room temperature). T-cell-DC clusters were disrupted using EDTA (0.01M) and low-density Mice cells enriched by density centrifugation (1.077 g/cm3 Nycodenz; Axis-Shield, Norway). Non-DC lineage cells were coated with C57BL/6 (B6) and BALB/c mice were bred under specific mAb (KT3-1.1; anti-CD3, RA36B2; anti-B220, T24/31.7; anti- pathogen-free conditions at the Walter and Eliza Hall Institute. Thy-1, RB68C5; anti-Ly6C/G and TER119; anti-erythrocyte) then Mice were used at 6–12 weeks of age, age-matched and gender- removed with anti-rat Ig magnetic beads (Biomag beads, Qiagen, matched for each experiment. Animals were handled according Australia). to the guidelines of the National Health and Medical Research Council of Australia. Experimental procedures were approved by the Animal Ethics Committee, Melbourne Health Research Detection of anti-Clec9A mAb in the plasma of Directorate. immunized mice

ELISA plates (Costar, Broadway, Cambridge, UK) were coated Non-human primates overnight at 4°Cwith1.5μg/mL of rat mAb GL117 (rat IgG2a), GL113 (rat IgG1), or 42D2 (anti-Clec9A rat IgG1λ). Unbound Juvenile pigtail macaques (Macaca nemestrina), 3–5 years of protein was washed away (PBS, 0.05% Tween-20). Plasma sam- age, free from HIV-1/SIV/simian retrovirus (SRV) infection, were ples, serially diluted in PBS/5% milk powder, were plated and housed under physical containment level 3 (PC3) conditions incubated at 4°C overnight. Bound mouse anti-rat IgG Abs were and anesthetised with ketamine (10 mg/kg intramuscular (IM)) detected using donkey anti-mouse-IgG-HRP (Chemicon Interna- prior to procedures. All experiments were performed according tional, Temecula, CA, USA) and visualized using ABTS (2,2- to National Institutes of Health guidelines on the care and use of Azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) diammonium laboratory animals, and were approved by the University of Mel- salt, Sigma-Aldrich). Optical density (O.D.) was measured at 405 bourne and CSIRO Livestock Industries Animal Experimentation – 490 nm. Endpoint titer was deﬁned as the highest dilution above and Ethics Committees. background. A positive control was included on each plate to mon- itor assay variation.

Immunization using anti-Clec9A mAb FACS competition assay assessing crossreactivity of Mice were injected i.v. with speciﬁed amounts of rat mAb in the Clec9A mAb absence or presence of 50 μg of poly I:C (Amersham) or i.p. with 50 μL of 1.3% Alhydrogel (Brenntag). Macaca nemestrina were CHO-Clec9A transfectants were preincubated (20 min; 4°C) with a injected s.c. with 200 μg of mAb (anti-human CLEC9A mAb: 3A4 saturating concentration of unconjugated mAb (10B4, 60 μg/mL;

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu 862 Jessica Li et al. Eur. J. Immunol. 2015. 45: 854–864

42D2, 12.4 μg/mL; 397, 100 μg/mL; 7H11, 27.8 μg/mL; 1F6, Measuring the anti-rat Ig humoral response in 22 μg/mL; GL117, 115.3 μg/mL; GL113, 100 μg/mL). Biotiny- non-human primates lated 10B4 (0.15625 μg/mL) or 42D2 (0.0039 μg/mL) were then added directly into the preincubating cell suspension and ELISA plates were coated overnight at 4°Cwith1.5μg/mL of incubated for a further 30 min. Samples were then washed and rat mAb GL117 (rat IgG2a). Unbound protein was washed away counterstained with SA-PE. PI was added to the ﬁnal cell wash (0.05% Tween-20/PBS). Plasma samples from Macaca nemest- to allow exclusion of PI positive dead cells. Flow cytometric rina, serially diluted (5% milk powder/PBS), were plated and analysis was performed on a FACSCalibur (Becton Dickinson, incubated at 4°C overnight. Bound monkey anti-rat IgG Ab were NJ, USA). detected using mouse anti-monkey-IgG-HRP (clone 1B3; NIH Non- human Primate Reagent Resource, Boston, USA) and visualized using ABTS. Measuring in vivo persistence of anti-Clec9A mAb

Plasma samples were obtained from mice injected with anti- Statistical analysis Clec9A mAb. ELISA plates were coated with 0.1–1 μg/mL mouse t Clec9A overnight (4°C). Unbound protein was washed away Unpaired two-tailed -test was performed on the log-transformed (0.05% Tween 20/PBS) before plates were blocked with 1% data and Welch correction was applied, unless otherwise indi- BSA/PBS (30 min; 23°C). Serial dilution of plasma samples were cated. The significance of differences is listed as follows: not sig- p > p < p < p < added and incubated for 2 h at room temperature. Bound Ab were nificant (n/s), 0.05; *, 0.05; **, 0.01; ***, 0.001, p < detected with anti-rat IgG biotin (Biolegend) (1 h; room temper- ****, 0.0001. Analysis was performed in Prism (GraphPad ature), followed by SA-HRP (GE Healthcare) (1 h; room tempera- Sofware, Inc). ture), then visualized with ABTS. 42D2 was detected with anti-rat IgG1 biotin (BD Pharmingen). Each plate contained an internal standard curve generated using the purified mAb of known con- Competition inhibition by ELISA centration. ELISA plates were coated overnight (4°C) with 0.1 μg/mL purified Clec9A, then washed and blocked for 30 min with 1% BSA (room temperature). Plates were then incubated with graded con- Assessing the capacity of anti-Clec9A mAb to bind cell centrations of unconjugated mAb (isotype control mAb: GL117 or surface Clec9A GL113 or with anti-Clec9A mAb: 10B4, 42D2, 397, 7H11, 1F6). After 1 h, half the volume of unconjugated mAb was removed and CHO-Clec9A transfectants were incubated with graded doses of replaced with biotinylated 10B4 or 42D2 (12.35 ng/mL). After a 2 anti-Clec9A mAb. Cells were washed, counterstained with anti-rat h incubation, bound biotinylated mAb was detected with SA-HRP IgG-PE, washed again then resuspended in medium containing and visualized with ABTS substrate. PI for flow cytometric analysis (FACSCalibur, Becton Dickinson, NJ, USA). Similarly, splenic DCs were purified, preblocked with anti-FcR and rat Ig then incubated with graded concentrations Titration of mAb by ELISA of biotinylated anti-Clec9A mAb. DCs were washed and stained with SA-PE, anti-CD11c-FITC and anti-CD8-APC mAb, washed ELISA plates were coated with 0.1 μg/mL Clec9A overnight (4°C), again, resuspended in medium containing PI, then analyzed by unbound protein was washed away (0.05% Tween 20/PBS), flow cytometry. before plates were blocked with 1% BSA (30 min, room temperature). Plates were then incubated for 2 h with serial dilutions of each anti-Clec9A mAb. Purified bound mAb were detected using Persistence of anti-Clec9A mAb on CD8+ DCs in vivo anti-rat IgG biotin (Biolegend) followed by SA-HRP (1 h incuba- tions). Biotinylated mAb were directly detected using SA-HRP and B6 mice were injected i.v. with 2 μg of anti-Clec9A mAbs con- visualized with ABTS substrate. jugated to either biotin or AlexaFluor-488 (A488). Spleens were harvested 1 h or 4 days later. DCs were isolated as described above, except lineage depletion was not performed, as the anti-rat Measuring the affinities of the anti-Clec9A mAb by Ig magnetic beads would also bind and deplete cells labeled with surface plasmon resonance the Clec9A mAb. Instead, the low-density cells were stained and analyzed. Cells from mice injected with biotinylated mAb were A ProteOn XPR36 protein-interaction array system (Bio-Rad) was stained with SA-PE, anti-CD11c-FITC. and anti-CD8-APC, whilst used for surface plasmon resonance. The Clec9A mAbs were cou- cells from mice injected with A488 mAb were stained with anti- pled to a GLC Sensor Chip (Bio-Rad) by direct amine coupling CD11c-PE and anti-CD8-APC. (100 response units) and an empty flow cell that had been

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu Eur. J. Immunol. 2015. 45: 854–864 Immunomodulation 863

activated and quenched in an identical manner served as a control 8 Caminschi, I., Proietto, A. I., Ahmet, F., Kitsoulis, S., Shin Teh, J., Lo, surface. Clec9A was serially diluted (200–12.5 nM) and simul- J. C., Rizzitelli, A. et al., The dendritic cell subtype-restricted C-type taneously injected over test and control surfaces at a rate of lectin Clec9A is a target for vaccine enhancement. Blood 2008. 112: 3264– 3273. 80 μL/min for 180 s. After subtraction of data from the control ﬂow cell, interactions were analyzed with ProteOn Manager 9 Joffre, O. P., Sancho, D., Zelenay, S., Keller, A. M. and Reis E Sousa, C., Efﬁcient and versatile manipulation of the peripheral CD4(+) T-cell software version 3.0.1 as well as Prism (GraphPad), for steady- compartment by antigen targeting to DNGR-1/CLEC9A. Eur. J. Immunol. state dissociation constants from responses at equilibrium. All 2010. 40: 1255–1265. interactions were tested at least in duplicate. 10 Park, H. Y., Light, A., Lahoud, M. H., Caminschi, I., Tarlinton, D. M. and Shortman, K., Evolution of B cell responses to Clec9A-targeted antigen. J. Immunol. 2013. 191: 4919–4925.

11 Huysamen, C., Willment, J. A., Dennehy, K. M. and Brown, G. D., CLEC9A is a novel activation C-type lectin-like receptor expressed on BDCA3+ dendritic cells and a subset of monocytes. J. Biol. Chem. 2008. 283: 16693– 16701. Acknowledgments: M.H.L., K.S, W.R.H and I.C. are supported by project grants from the National Health and Medical Research 12 Caminschi, I., Vremec, D., Ahmet, F., Lahoud, M. H., Villadangos, J. A., Murphy,K.M., Heath, W. R. et al., Antibody responses initiated Council of Australia (NHMRC). K.S, S.J.K. and W.R.H. are sup- by Clec9A-bearing dendritic cells in normal and Batf3(-/-) mice. Mol. ported by an NHMRC Program grant. W.R.H. is supported by an Immunol. 2012. 50: 9–17.

NHMRC fellowship. C.R.S. is supported by Cancer Research UK 13 Desch, A. N., Randolph, G. J., Murphy, K., Gautier, E. L., Kedl, R. M., and the European Research Council. L.C.S is supported by an Lahoud, M. H., Caminschi, I. et al., CD103 +pulmonary dendritic cells NHMRC career development award. J.L. is the recipient of an preferentially acquire and present apoptotic cell-associated antigen. Australian Postgraduate Award PhD Scholarship. This work was J. Exp. Med. 2011. 208: 1789–1797. made possible through Victorian State Government Operational 14 Schraml, B. U., van Blijswijk, J., Zelenay, S., Whitney, P. G., Filby, A., Infrastructure Support and Australian Government NHMRC Inde- Acton, S. E., Rogers,N.C.etal., Genetic tracing via DNGR-1 expression pendent Research Institute Infrastructure Support Scheme. history deﬁnes dendritic cells as a hematopoietic lineage. Cell 2013. 154: 843–858.

15 Iborra, S., Izquierdo, H. M., Martinez-Lopez, M., Blanco-Menendez, N., Reis e Sousa, C. and Sancho, D., The DC receptor DNGR-1 mediates cross- Conflict of interest: The authors declare no financial or commer- priming of CTLs during vaccinia virus infection in mice. J. Clin. Investig. 2012. 122: 1628–1643. cial conflict of interest. 16 Sancho, D., Joffre, O. P., Keller, A. M., Rogers, N. C., Martinez, D., Hernanz- Falcon, P., Rosewell, I. et al., Identification of a dendritic cell receptor that couples sensing of necrosis to immunity. Nature 2009. 458: References 899–903. 17 Zelenay, S., Keller, A. M., Whitney, P. G., Schraml, B. U., Deddouche, 1 Caminschi, I., Maraskovsky, E. and Heath, W. R., Targeting dendritic S., Rogers,N.C., Schulz, O. et al., The dendritic cell receptor DNGR- cells in vivo for cancer therapy. Front. Immunol. 2012. 3: 13. 1 controls endocytic handling of necrotic cell antigens to favor cross- 2 Trumpfheller, C., Longhi, M. P., Caskey, M., Idoyaga, J., Bozzacco, L., priming of CTLs in virus-infected mice. J. Clin. Investig. 2012. 122: 1615– Keler, T., Schlesinger, S. J. et al., Dendritic cell-targeted protein vaccines: 1627. a novel approach to induce T-cell immunity. J. Intern. Med. 2012. 271: 18 Ahrens, S., Zelenay, S., Sancho, D., Hanc, P., Kjaer, S., Feest, C., Fletcher, 183–192. G. et al., F-actin is an evolutionarily conserved damage-associated molec- 3 Palucka, K., Banchereau, J. and Mellman, I., Designing vaccines based ular pattern recognized by DNGR-1, a receptor for dead cells. Immunity on biology of human dendritic cell subsets. Immunity 2010. 33: 464– 2012. 36: 635–645. 478. 19 Zhang, J. G., Czabotar, P. E., Policheni, A. N., Caminschi, I., Wan, S. S., 4 Tacken, P. J., de Vries, I. J., Torensma, R. and Figdor, C. G., Dendritic- Kitsoulis, S., Tullett, K. M. et al., The dendritic cell receptor Clec9A binds cell immunotherapy: from ex vivo loading to in vivo targeting. Nat. Rev. damaged cells via exposed actin filaments. Immunity 2012. 36: 646–657. Immunol 2007. 7: 790–802. 20 Plotkin, S. A., Correlates of protection induced by vaccination. Clin. Vac- 5 Caminschi, I. and Shortman, K., Boosting antibody responses by target- cine Immunol. 2010. 17: 1055–1065. ing antigens to dendritic cells. Trends Immunol. 2011. 33: 71–77. 21 Hodges, A., Sharrocks, K., Edelmann, M., Baban, D., Moris, A., 6 Lahoud, M. H., Ahmet, F., Kitsoulis, S., Wan, S. S., Vremec, D., Lee, C. N., Schwartz, O., Drakesmith, H. et al., Activation of the lectin DC- Phipson, B. et al., Targeting antigen to mouse dendritic cells via Clec9A SIGN induces an immature dendritic cell phenotype triggering Rho- induces potent CD4 T cell responses biased toward a follicular helper GTPase activity required for HIV-1 replication. Nat. Immunol. 2007. 8: phenotype. J. Immunol. 2011. 187: 842–850. 569–577.

7 Sancho, D., Mourao-Sa, D., Joffre, O. P., Schulz, O., Rogers, N. C., Pen- 22 Lesterhuis, W. J., Aarntzen,E.H., De Vries, I. J., Schuurhuis, D. H., nington, D. J., Carlyle, J. R. et al., Tumor therapy in mice via antigen Figdor, C. G., Adema, G. J. and Punt, C. J., Dendritic cell vaccines in targeting to a novel, DC-restricted C-type lectin. J. Clin. Invest. 2008. 118: melanoma: from promise to proof? Crit. Rev. Oncol. Hematol 2008. 66: 118– 2098–2110. 134.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu 864 Jessica Li et al. Eur. J. Immunol. 2015. 45: 854–864

23 Figdor, C. G., de Vries, I. J., Lesterhuis, W. J. and Melief, C. J., Den- Abbreviations: pDC: plasmacytoid dendritic cell · PI: propidium iodide · dritic cell immunotherapy: mapping the way. Nat. Med. 2004. 10: SA: streptavidin 475–480. Full correspondence: Dr. I. Caminschi, Centre for Biomedical Research, 24 Idoyaga, J., Lubkin, A., Fiorese, C., Lahoud, M. H., Caminschi, I., Huang, Burnet Institute, 85 Commercial Rd, Melbourne, Victoria, Australia, Y., Rodriguez, A. et al., Comparable T helper 1 (Th1) and CD8 T-cell immu- 3000 nity by targeting HIV gag p24 to CD8 dendritic cells within antibodies to e-mail: [email protected] Langerin, DEC205, and Clec9A. Proc. Natl. Acad. Sci. USA 2011. 108: 2384– 2389. Additional correspondence: William R. Heath, Department of Microbiology and Immunology, The University of Melbourne, at the 25 Belnoue, E., Kayibanda, M., Vigario, A. M., Deschemin, J. C., van Rooi- Peter Doherty Institute for Infection and Immunity, Victoria 3010, jen, N., Viguier, M., Snounou, G. et al., On the pathogenic role of Australia brain-sequestered alphabeta CD8+ T cells in experimental cerebral e-mail: [email protected] malaria. J. Immunol. 2002. 169: 6369–6375.

26 Dixon, J. E., Allan, J. E. and Doherty, P. C., The acute inﬂammatory pro- Additional correspondence: Mireille H. Lahoud, Centre for Biomedical cess in murine lymphocytic choriomeningitis is dependent on Lyt-2+ Research, Burnet Institute, 85 Commercial Rd, Melbourne, Victoria, immune T cells. Cell. Immunol. 1987. 107: 8–14. Australia, 3000 e-mail: [email protected] 27 Vremec, D., Pooley, J., Hochrein, H., Wu, L. and Shortman, K.,CD4and CD8 expression by dendritic cell subtypes in mouse thymus and spleen. Received: 13/8/2014 J. Immunol. 2000. 164: 2978–2986. Revised: 4/11/2014 Accepted: 3/12/2014 Accepted article online: 9/12/2014

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu