Blood was obtained from healthy adult volunteers and umbilical cord blood (UCB) was obtained from the Queensland Cord Blood Bank. Cryopreserved peripheral blood mononuclear cells (PBMCs) from patients with melanoma that had been vaccinated with NY-ESO-1 protein and ISCOMATRIX adjuvant31 or monocyte-derived DC (MoDC) pulsed with autologous tumor32 were used for some experiments. CD34⁺ hematopoeitic stem cells (HSCs) were isolated from UCB by Ficoll-Paque (GE Healthcare) density centrifugation and using a CD34⁺ isolation kit (Miltenyi Biotec) and cryopreserved. HLA typing was performed by BGI Health (Hong Kong) and or in-house PCR.33 CD34⁺ HSCs were expanded and differentiated into CD141⁺ DC using a modification of a previously described protocol34 to include 1 µM StemRegenin-1 (SR-1, Stemcell Technologies) during both expansion and differentiation stages.

Expression constructs were generated in pcDNA3.1+ plasmids encoding human IgG4 and κ constant regions and variable regions of anti-human CLEC9A clone 4C6, anti-human DEC-205 clone MMRI-7 or isotype control anti-β-galactosidase (β-gal) clone GL117.26 The antibody heavy chain sequence was fused to cDNA encoding an alanine linker and NY-ESO-1 antigenic sequence (AAAALLEFYLAMPFATPMEAELARRSLAQDAPPLPVPGVLLKEFTVSGNILTIRLTAADHRQLQLSISSCLQQLSLLMWITQCFLPVFLAQPPSGQRR) containing the HLA-A*0201-restricted _157-165epitope SLLMWITQC (SLL) and the HLA-Cw*0303_92-100-restricted epitope LAMPFATPM (LAM) (GeneArt). The resulting chimeric Abs, carrying two NY-ESO-1 antigenic sequences per Ab, were expressed in Freestyle 293F cells (Invitrogen) using 293Fectin (Invitrogen) and purified from the supernatant using Protein A sepharose (GE Healthcare). Ab specificity was confirmed by binding to 293F cells transiently transfected with full-length recombinant CLEC9A or DEC-205 as previously described.26

Binding specificity to human PBMC subsets was assessed by blocking with combined rat and mouse serum followed by incubation with chimeric Ab followed by anti-human IgG4-biotin (Life Technologies) then Streptavidin-PE (SA-PE, Life Technologies). Lymphocyte populations were identified by staining with Ab cocktails comprised of Live/Dead (Zombie) Aqua (Biolegend), anti-human CD3-BV711 or Pacific Blue (clone OTK3), CD19-Pacific Blue (HIB19), CD20-Pacific Blue (2H7), CD56-FITC or Pacific Blue (HCD56), CD14-APC or Pacific Blue (HCD14), CD16-BV785 or Pacific Blue (3G8), HLA-DR-PerCPCy5.5 or PE-Cy7 (L243), CD141-PE-Cy7 (M80), CD1c-Alexa Fluor 700 (L161), CLEC9A-PE (8F9), DEC-205-FITC (all from Biolegend), CD11c-PE-CF594 (B-LY6), CD141-BV711 (1A4) and CD123-BUV395 (7G3) (all from BD), and CADM1-Alexa Fluor 647 (3E1, Jomar/MBL). Samples were acquired on a Becton Dickinson LSR Fortessa X-20 or Beckman Coulter CytoFLEX-S flow cytometer or sorted on a MoFlo Astrios (Beckman Coulter) and analyzed using Flowjo V.9 or 10 software (Treestar).

CD8⁺ T cell lines specific for HLA-A*0201 restricted SLL and HLA-Cw*03:03 LAM NY-ESO-1 epitopes were generated and maintained as previously described.35 Purified CD141⁺ DCs were incubated with 5 µg/mL chimeric Ab, an equivalent amount of untargeted NY-ESO-1 Ag (0.7 µg/mL), SLL or LAM peptides or no Ag for 2 hours at 37°C, then washed and incubated with NY-ESO-1-specific T cell lines. Interferon-γ (IFN-γ) production was detected after 16–20 hours by ELISPOT (Mabtech). For some experiments, PBMCs from melanoma patients with known NY-ESO-1 responses were incubated with chimeric Ab or no antigen for 16 hours and IFN-γ production detected by ELISPOT.

NOD.Cg-Prkdc^scid IL2rg ^tm1Wjl Tg (HLA-A/H2-D/B2M) 1Dvs/SzJ transgenic for human HLA-A*0201 (NSG-A2) mice were purchased from The Jackson Laboratory mice (stock no: 014570). Humanized mice were generated following reconstitution with human CD34⁺ HSC transduced with lentivirus encoding the HLA-A*0201-restricted NY-ESO-1 SLL T cell receptor (TCR) according to previously published protocols.36 37

Following human CD45⁺ reconstitution, humanized mice received 2×50 µg subcutaneous injections of Flt3L 4 days apart to expand DC followed by vaccination with 10 µg of chimeric Ab or no antigen with 50 µg poly I:C (InVivogen) and mice were harvested 1 week post vaccination. Spleens were digested in collagenase IV (Worthington Biochemical) and DNase I (Roche/Sigma) followed by Percoll density gradient as previously described36 and enriched for human leukocytes using a Mouse/Human Chimera EasySep Kit (Stemcell). Expression of the NY-ESO-1 SLL TCR was confirmed by staining with NY-ESO-1 SLL dextramer-APC (Immudex), anti-mouse CD45-V500 (30-F11, BD), anti-human CD45-BUV395 (HI30, BD), CD3-Pacific Blue or BV711, CD8-PE-Cy7 (RPA-T8), CD197-BV711 (3D12, BD) and CD45RA-PE (H130, Biolegend).

For priming of naïve T cells in vitro, splenocytes from non-immunized humanized mice expressing the NY-ESO-1 SLL TCR were stimulated with SLL peptide or control-pulsed HLA-A*0201⁺ allogeneic irradiated lymphoblastoid cell lines (LCLs). IFNγ was measured in the supernatants after 3 days by ELISA (Thermo Fisher) and cultures expanded in media containing 100 U/mL IL-2, 10 ng/mL IL-7 and 20 ng/mL IL-15 for 20 days. For reactivation of in vivo-primed NY-ESO-1-specific T cells, PBMCs from vaccinated patients with melanoma or splenocytes from immunized humanized mice were incubated with 10 µg/mL chimeric Abs, SLL peptide or no Ag in the presence of poly I:C and R848 (InvivoGen) for 2 hours at 37°C, then washed and expanded in media containing IL-2, IL-7 and IL-15 for 9–14 days. Expansion of NY-ESO-1 SLL-specific CD8⁺ T cells was measured by SLL dextramer staining as described above.

Cytokine secretion was assessed by restimulation of the cultures for 6 hours in the presence or absence of SLL peptide, Brefeldin A, Monensin and CD107a-BV785, followed by staining with Live/dead Aqua, CD8-PerCpCy5.5 and CD3-BUV737. Cells were fixed and permeabilized then stained with MIP1β-PE, IFNγ-FITC, TNFα-PE-Cy7 and IL-2-APC or isotype controls for flow cytometry analysis. Cytotoxic activity of the T cells was assessed against SLL or control peptide (HLA-A2 restricted CMV pp65 NLVPMVATV) pulsed T2 targets, and melanoma cell lines LM-MEL 44 (HLA–A*0201⁺, NY-ESO-1⁺) or SK-MEL 28 (HLA-A*0201^-, NY-ESO-1^-) at an effector:target ratio of 10:1 using a Cytotox 96 Kit (Promega). Specific lysis of target cells was calculated as:

(Experimental-Effector_Spontaneous–Target_Spontaneous)/(Target_Maximum–Target_Spontaneous)×100.

Data sets were tested for normal distribution using the Kolmogorov-Smirnoff test. Multigroup comparisons were performed by using repeated measures one-way analysis of variance (ANOVA) or non-parametric equivalent (Freidmann’s) followed by appropriate post multiple comparison post-tests (Tukey’s/Dunn’s). Paired comparisons were performed using a paired t-test or non-parametric Wilcoxon’s signed rank test. Statistical significance was defined as a p value equal to or less than 0.05. GraphPad Prism V.7 was used for graphical representation and statistical analysis.

We previously developed constructs encoding human chimeric IgG4 Ab specific for CLEC9A, DEC-205 and β-gal (untargeted isotype control). The NY-ESO-1 sequence, encoding epitopes across multiple HLA alleles, was genetically fused to the heavy chain of each Ab via an alanine linker so that each Ab contained two NY-ESO-1 antigenic sequences (figure 1A). Following expression in 293F cells and purification by affinity chromatography, Ab binding specificity to target receptors was confirmed on CLEC9A or DEC-205 transfected cell lines and PBMC and by ELISA (figure 1B,C and online supplementary figure 1). As expected, the CLEC9A-NY-ESO-1 bound specifically to CD141⁺ DC, whereas the DEC-205-NY-ESO-1 bound to most human leukocyte subsets, with the highest staining intensity observed on CD1c⁺ DC and CD141⁺ DCs. The β-gal control Ab did not bind human leukocytes, with the exception of some non-specific binding on B cells.

As CD141⁺ DCs are extremely rare in human blood, we derived these cells from human cord blood CD34⁺ HSC using a previously validated in vitro culture system34 (online supplementary figure 2). The transcriptome, phenotype and function of the in vitro derived CD141⁺ DCs closely resembled those of their human blood counterparts34 and this was confirmed in our hands (data not shown). None of the targeting Ab induced activation of DC measured by expression of costimulatory molecules CD83 and CD86 (online supplementary figure 2B). CLEC9A-NY-ESO-1 and DEC-205-NY-ESO-1 Ab, but not control-NY-ESO-1, bound to in vitro-derived CD141⁺ DCs (figure 2A). Notably, the staining intensity of DEC-205-NY-ESO-1 was higher than that of CLEC9A-NY-ESO-1. Following uptake by HLA-A*0201⁺ CD141⁺ DC, only CLEC9A-NY-ESO-1 mediated cross-presentation of the HLA-A*0201-restricted SLL epitope to specific CD8⁺ T cells (figure 2B and online supplementary figure 2C). By contrast, both CLEC9A-NY-ESO-1 and DEC-205-NY-ESO-1 mediated cross-presentation of the HLA-Cw*0303-restricted NY-ESO-1 LAM epitope (figure 2C and online supplementary figure 2C). Thus, CLEC9A-NY-ESO-1 Abs are effective delivery vehicles to CD141⁺ DCs for cross-presentation of multiple NY-ESO-1-specific CD8⁺ T cell epitopes.

To investigate the ability of CLEC9A-NY-ESO-1 Ab to reactivate NY-ESO-1-specific T cell responses, we used PBMC from melanoma patients that had been previously vaccinated with NY-ESO-1 protein and ISCOMATRIX adjuvant31 or MoDC pulsed with autologous tumor32 and had detectable responses to NY-ESO-1. The percentage of CLEC9A-expressing cells ranged from 0.015% to 0.11% (mean 0.05%) of PBMC and was restricted to the CD141⁺ DC population (figure 3A), consistent with healthy donors.26 By contrast, DEC-205 was expressed by 5.35%–41.9% (mean 25.8%) of melanoma patient PBMC, with the highest expression seen on monocytes, CD141⁺ DC and CD1c⁺ DC, consistent with healthy donors. Despite only a small percentage of PBMC expressing CLEC9A, these were potent at reactivating NY-ESO-1-specific T cell responses following a short stimulation with CLEC9A-NY-ESO-1 Ab ex vivo (figure 3B). In contrast, DEC-205-NY-ESO-1, control-NY-ESO-1 and unconjugated CLEC9A Ab did not induce responses (figure 3B and online supplementary figure 3). Consistent with previous findings,31 responses to the HLA-A*0201-restricted SLL epitope were confirmed in only a few (2/8) of the HLA-A*0201⁺ patients. Three of the donors had previously identified responses to a very broad range of MHC I and MHC II restricted epitopes (patients 7, 9, 2131). In another donor, the NY-ESO-1-specific response was mapped to a novel HLA-B*5601 -restricted NY-ESO-1_93-102 AMPFATPMEA epitope (Dr Volker Lennerz and Dr Thomas Wölfel, University Medical Center, Mainz, Germany, personal communication). Responses to known Cw3-restricted epitopes were not detectable in any of the 7 HLA-Cw*03⁺ patients (online supplementary figure 3 and not shown). Thus, the majority of the responses appeared to be restricted to other, potentially as yet undescribed CD8 and/or CD4 epitopes. For one patient that had pre-existing responses to the HLA-A*0201-restricted SLL epitope, we examined whether the chimeric Ab could also expand NY-ESO-1-specific responses with effector function. For these experiments, patient PBMC were incubated with chimeric Ab or no antigen and then expanded for 2 weeks in the presence of T cell cytokines. Both CLEC9A-NY-ESO-1 and DEC-205-NY-ESO-1 Ab could expand a small percentage of NY-ESO-1 SLL-specific CD8⁺ T cells as observed by NY-ESO-1 SLL dextramer staining (figure 3C). Although the control-NY-ESO-1 Ab was also able to expand NY-ESO-1 SLL-specific CD8⁺ T cells, there were insufficient cell numbers from this and the no antigen control cultures to assess in killing assays. T cells expanded by CLEC9A-NY-ESO-1 Ab, but not DEC-205-NY-ESO-1 Ab, could kill SLL peptide coated T2 tumor targets (figure 3D). T cells expanded by CLEC9A-NY-ESO-1 Ab were also more effective at lysis of HLA-A*0201⁺NY-ESO-1⁺ tumor targets (figure 3E).

Humanized mice were generated following reconstitution with human CD34⁺ HSC transduced with lentivirus encoding NY-ESO-1 SLL TCR (figure 4A). Introduction of the TCR transgene into HSC inhibits endogenous TCR gene rearrangement in recombinant T cells, resulting in exclusive expression of the transgenic TCR on the cell surface via a process known as allelic exclusion.38 Engraftment in neonatal NSG-A2 mice enabled thymic selection of NY-ESO-1 SLL-specific CD8⁺ T cells on human HLA-A*0201. The mean percentage of human CD45⁺ cell engraftment in blood ranged from 0.04% to 94% (mean 17.53%±23.56 SD, n=66). Engrafted mice developed human CD141⁺ DC, CD1c⁺ DC, plasmacytoid DC, monocytes and T cells in similar proportions to those previously described in NSG-A2 mice36 37 (figure 4B,C). No SLL-specific (dextramer⁺) CD8⁺ T cells were detectable in mice engrafted with mock transduced CD34⁺ HSC, suggesting an absence or very low frequency of NY-ESO-1 SLL-specific CD8⁺ T cell precursors (not shown). In mice engrafted with TCR lentiviral transduced CD34⁺ HSC, the NY-ESO-1 SLL TCR was restricted to 0%–47.1% of splenic CD8⁺ T cells (mean 5.932%±11.05 SD, figure 4D). Importantly, all of the CD8⁺ T cells expressing the SLL TCR coexpressed CD45RA and CCR7, indicative of a naïve phenotype (figure 4D). Ex vivo coculture of splenocytes from these mice with SLL peptide-pulsed HLA-A*0201⁺ allogeneic irradiated LCLs led to the expansion of NY-ESO-1 SLL-specific CD8⁺ T cells (figure 4E), and induced IFNγ production on restimulation with cognate SLL peptide (figure 4F). The expanded NY-ESO-1 SLL-specific CD8⁺ T cells were capable of lysing T2 target cells pulsed with cognate SLL peptide, and HLA-A*0201⁺NY-ESO-1⁺ (LM-MEL 44), but not HLA-A*0201^-NYESO-1^- (SK-MEL 28) tumor cell lines (figure 4G). Collectively, these data demonstrate successful generation of humanized mice comprised of both human DC subsets and a detectable repertoire of naïve NYESO-1-specific CD8⁺ T cells that can be primed to proliferate and develop effector function.

Using this model, we examined the capacity of CLEC9A-NY-ESO-1 to prime naïve NY-ESO-1-specific CD8⁺ T cells. For these experiments, humanized mice were injected with Flt3L to expand DC subsets, then mice were immunized with chimeric Ab (or HBSS no Ag) along with poly I:C adjuvant (figure 5A). Increased percentages of NY-ESO-1 SLL-specific CD8⁺ T cells were observed in some of the CLEC9A-NY-ESO-1- and DEC-205-NY-ESO-1-immunized animals compared with the control-NY-ESO-1 vaccinated mice but this was not statistically significant (figure 5B). However, splenocytes from mice immunized with CLEC9A-NY-ESO-1 demonstrated significantly higher capacity to produce IFNγ compared with DEC-205-NY-ESO-1 and control-NY-ESO-1 immunized mice following a short ex vivo reactivation with cognate SLL peptide (figure 5C). This suggested that CLEC9A-NY-ESO-1 could prime naïve CD8⁺ T cells in vivo.

To investigate the effector function of CLEC9A-NYESO-1-primed CD8⁺ T cells, splenocytes were harvested 1 week after vaccination and restimulated ex vivo with no Ag or the same chimeric Ab they were vaccinated with. NY-ESO-1 SLL-specific CD8⁺ T cells were expanded from the splenocytes of CLEC9A-NY-ESO-1 or DEC-205-NY-ESO-1 vaccinated mice, with a significantly higher degree of expansion after CLEC9A-NY-ESO-1 immunization (figure 6A,B). Splenocytes from non-immunized control mice failed to expand in vitro and minimal expansion was observed following restimulation with recombinant NY-ESO-1 protein (not shown). CD8⁺ T cells primed by either CLEC9A-NY-ESO-1 or DEC-205-NY-ESO-1 did not produce IL-2, but expressed cytotoxic degranulation marker CD107a, and produced IFNγ, MIP-1β and TNF following reactivation with SLL cognate peptide (figure 6C, online supplementary figure 3). Boolean gating analysis showed that CD8⁺ T cells primed with CLEC9A-NY-ESO-1 had a significantly higher proportion of polyfunctional Ag-specific CD8⁺ T cells than those primed with DEC-205-NY-ESO-1, defined as the percentage of CD8⁺ T cells coexpressing IFNγ, MIP-1β, TNF and CD107a (figure 6D). Effector function of the CLEC9A-NYESO-1 primed CD8⁺ T cells was further demonstrated by their capacity to lyse T2 target cells pulsed with cognate SLL peptide and HLA-A*0201⁺ NY-ESO-1⁺ melanoma cells (LM-MEL 44), but not T2 cells pulsed with irrelevant peptide (CMV pp65) or HLA-A*0201^-NYESO-1^- melanoma cells (figure 6E,F). CD8⁺ T cells primed by DEC-205-NY-ESO-1 were also capable of lysing target cells but were less when compared with those primed with CLEC9A-NY-ESO-1. The percentage of cytokine-producing effector cells and target cell lysis correlated with the percentage of SLL-specific T cells in the cultures, suggesting that CLEC9A-NY-ESO-1 priming stimulates increased expansion rather than T cell quality on a per cell basis. These data demonstrate that CLEC9A-NY-ESO-1 Ab can prime naïve CD8⁺ T cells to expand and develop into polyclonal effectors.