Metabolic labeling and targeted modulation of dendritic cells

Abstract

Targeted immunomodulation of dendritic cells (DCs) in vivo will enable manipulation of T-cell priming and amplification of anticancer immune responses, but a general strategy has been lacking. Here we show that DCs concentrated by a biomaterial can be metabolically labelled with azido groups in situ, which allows for their subsequent tracking and targeted modulation over time. Azido-labelled DCs were detected in lymph nodes for weeks, and could covalently capture dibenzocyclooctyne (DBCO)-bearing antigens and adjuvants via efficient Click chemistry for improved antigen-specific CD8⁺ T-cell responses and antitumour efficacy. We also show that azido labelling of DCs allowed for in vitro and in vivo conjugation of DBCO-modified cytokines, including DBCO–IL-15/IL-15Rα, to improve priming of antigen-specific CD8⁺ T cells. This DC labelling and targeted modulation technology provides an unprecedented strategy for manipulating DCs and regulating DC–T-cell interactions in vivo.

Metabolic glycoengineering of unnatural sugars provides a powerful tool to label cell membranes with chemical tags^1–3, for subsequent targeted delivery of molecules of interest via efficient chemistries⁴. This technology has been used for the development of cancer-targeted chemotherapy^5,6, photothermal therapy⁷ and photoacoustic therapy^7,8, and recent efforts have extended it to the field of cancer immunotherapy⁹. However, direct metabolic labelling and targeted modulation of immune cells, dendritic cells (DCs) in particular, has not been explored so far. Owing to their role as mediators of adaptive immune responses, DCs are an important target in cancer immunotherapies¹⁰. DCs are produced in bone marrow, migrate to lymphatic and peripheral tissues and mature in the context of pathogens or tumour antigens^11–13. DCs have been fluorescently or radio-labelled ex vivo to follow their biodistribution after adoptive transfer^14–16, but technologies to specifically label DCs in situ for subsequent tracking and targeted modulation have been lacking. Targeting DCs in vivo via unique labels may allow for tracking of their migration, and targeted delivery of immunomodulatory agents to improve effector T-cell responses and overall antitumour efficacy.

Here we show that DCs can be metabolically labelled with chemical tags in vitro and in vivo, allowing for their subsequent tracking and targeted modulation over time in the body. Unnatural sugars such as azido-sugars can metabolically label cell-surface glycoproteins with azido groups^17–20. For specific labelling and tracking of DCs in vivo, we use an injectable pore-forming alginate gel releasing granulocyte-macrophage colony-stimulating factor (GM-CSF) to concentrate DCs at the vaccination site^21,22, and adapt it for encapsulation and controlled release of azido-sugar materials. In addition to DC tracking, the azido-label also enables targeted delivery of dibenzocyclooctyne (DBCO)-bearing tumour antigens, adjuvants, cytokines and other immunomodulatory agents via efficient Click chemistry^3,23. Tumour-specific immune responses could be readily generated with this approach, and DC–T-cell interactions manipulated to enhance T-cell priming. This technology provides an unprecedented ability to manipulate DCs and regulate their interactions in the body, and extends the use of metabolic labelling to immunotherapy.

Results

Metabolic labelling of DCs with azido groups in situ.

We first explored the ability of unnatural sugars, azido-sugars, to metabolically label DCs in vivo. An injectable pore-forming alginate gel releasing GM-CSF was used to concentrate DCs^21,22, and subsequently release azido-sugar materials to the accumulated DCs to label and target the cells (Fig. 1). The ability of tetraacetyl-N-azidoacetylmannosamine (Ac₄ManAz), a commonly used metabolic labelling agent, to label DC2.4 cells in vitro with azido groups was first confirmed (Supplementary Fig. 1a,b). However, Ac₄ManAz has a number of limitations for in vivo use, including low water-solubility, poor encapsulation and burst release from hydrogels. To address these issues, the C1 site of Ac₄ManAz was functionalized with an acrylate bond, followed by reversible addition-fragmentation chain transfer polymerization to yield poly(azido-sugar)_n (n = 25 (G25) or n = 400 (G400)) (Supplementary Fig. 1c,d). G25 and G400 nanoparticles (NPs) with an average diameter of 70 and 130 nm (Supplementary Fig. 1e,f), were then prepared via nanoprecipitation of G25 and G400, respectively. G25 and G400 NPs were able to enter and metabolically label DC2.4 cells and bone marrow-derived DCs (BMDCs) in a concentration-dependent manner (Fig. 2a–c, Supplementary Fig. 1g–i). Uptake of G400 NPs did not affect activation of BMDCs (Supplementary Fig. 1j,k). As azido-sugar NPs released from gels are to be used for DC labelling in vivo, their passive and active release from the gels was next studied. In contrast to G25 NP (~70 nm), which showed high baseline release from gels, G400 NP (~130 nm) showed minimal baseline release and ultrasound-triggered on-demand release from gels in vitro (Fig. 2d)²⁴. Considering the comparable DC labelling efficiency of G400 NP and G25 NP (Fig. 2e and Supplementary Fig. 1i) and ultrasound-triggered release of G400 NP from gels in vivo (Fig. 2f,g), G400 NP was used in subsequent studies.

Fig. 1 |. Strategy for DC labelling and targeting in vivo.

a, Schematic illustration of in situ recruitment and metabolic labelling of DCs in pore-forming alginate gels. Macropores within alginate gels are formed via hydrolysis over time, enabling the homing of recruited DCs. Azido-sugar materials in the bulk phase of alginate gels are then burst released using ultrasound, endocytosed and metabolized by DCs, resulting in azido-labelled glycoproteins on cell membranes. b, Schematic illustration of the migration of azido-labelled DCs from the gel scaffold to LNs and subsequent targeting of immunomodulatory agents via Click chemistry.

Fig. 2 |. Azido-sugar NPs metabolically label DCs, and show on-demand release from gels.

a, Schematic of metabolic labelling of DCs with azido-sugar NPs. Azido-sugar NPs enter cells through endocytosis, disassemble and degrade into sugar-azide via hydrolysis or enzymatic degradation and the released sugar-azide is metabolically presented on the cell surface as glycoproteins. b, Percentage of azide⁺ DC2.4 cells after 3 d of incubation with Ac₄ManAz or G25 NP (n = 5 biologically independent samples, one-way ANOVA and Tukey’s test). Cell-surface azido groups were detected by DBCO/efluor660-antibody. The molar concentrations were normalized in azide equivalents. c, Confocal images of DC2.4 cells after 3 d of incubation with G25 NP and 30-min staining with DBCO/efluor660-antibody. This experiment was repeated twice independently with similar results. Scale bars, 10 μm. d, In vitro release profiles of G25 NP and G400 NP from pore-forming alginate gels (n = 4 biologically independent samples, ***compared to the same time points of G400 NP group (red line), two-tailed t-test). Green arrows indicate the time of ultrasound application. e, Percentage of azide⁺ BMDCs after 3 d of incubation with Ac₄ManAz, G25 NP and G400 NP, respectively (n = 5 biologically independent samples, one-way ANOVA and Tukey’s test). The molar concentrations were normalized in azide equivalents. f, representative IVIS images of C57BL/6 mice showing release of Cy5-labelled G400 NP from gels in vivo in the absence or presence of ultrasound (US) treatment. G400 NP-loaded gels were subcutaneously injected, and mice were imaged at designated time points. Ultrasound was applied at 72 h postgel injection. This experiment was repeated once independently with similar results. g, In vivo release profiles of G400 NP, as quantified from f (n = 4 biologically independent animals, two-tailed t-test). All the numerical data are presented as mean ± s.d. (0.01 < *P ≤ 0.05; 0.001 < **P ≤ 0.01; 0.0001 < ***P ≤ 0.001, ****P ≤ 0.0001).

We next investigated whether G400 NP-loaded gels are able to metabolically label DCs, recruited to the gel site, with azido groups. Mice were injected with gels containing G400 NP and GM-CSF on day 0, followed by ultrasound treatment on day 3 to release G400 NP when the number of recruited DCs approaches a maximum²¹. GM-CSF was conjugated to gold NPs for all the following experiments, as it enables controlled release of GM-CSF and improved DC recruitment²¹. On day 6, a high number of CD11b⁺CD11c⁺ DCs were azide positive (Fig. 3a,b and Supplementary Fig. 2a–c). In comparison, many fewer DCs in the gels without GM-CSF incorporation or ultrasound treatment were azide positive (Fig. 3a,b and Supplementary Fig. 2c), supporting the importance of concentrating DCs and triggering release of G400 NP at the time of maximum cell accumulation. Confocal images also showed increased azide density in gels containing G400 NP and GM-CSF with ultrasound treatment, in comparison to other groups (Supplementary Fig. 2d). Although G400 NP did not show preference for DC labelling (Fig. 3a and Supplementary Fig. 2e–i), as expected, the pore-forming alginate gels primarily recruited DCs compared to other types of immune cell including CD11b⁺Gr1⁺ cells (neutrophils) and CD11b⁺F4/80⁺ cells (macrophages) (Supplementary Fig. 2e). As a result, many fewer CD11b⁺Gr1⁺ cells and CD11b⁺F4/80⁺ cells in the gels were azide positive, in comparison to azide-positive DCs (Fig. 3c and Supplementary Fig. 2f–i).

Fig. 3 |. G400 NP-containing gels recruit and metabolically label DCs with azido groups in vivo.

a–g, Gels containing GM-CSF and G400 NP were subcutaneously injected into C57BL/6 mice on day 0, followed by ultrasound treatment on day 3 and analyses on day 6. Control groups include: gels containing GM-CSF and G400 NP without ultrasound treatment (G400 + GM), gels containing G400 NP alone with ultrasound treatment (G400 + US), gels containing GM-CSF alone (GM) and blank gels (blank) (n = 5 biologically independent animals). a,b, Percentage (a) and total number (b) of azido-labelled DCs (among CD11b⁺CD11c⁺) in gels on day 6 (one-way ANOVA and Tukey’s test). c, Total number of azido-labelled DCs, CD11b⁺F4/80⁺ cells and CD11b⁺Gr1⁺ cells in gel scaffolds on day 6 in the G400 + GM + US group (one-way ANOVA and Tukey’s test). d,e, Percentage (d) and total number (e) of azido-labelled DCs in dLNs on day 6 (one-way ANOVA and Tukey’s test). f, Total number of azido-labelled DCs, CD11b⁺F4/80⁺ cells and CD11b⁺Gr1⁺ cells in dLNs on day 6 in the G400 + GM + US group (one-way ANOVA and Tukey’s test). g, representative confocal images of dLN sections that were stained with DBCO/efluor660-antibody (red), pacific blue-conjugated anti-CD3 (blue), FITC-conjugated anti-CD11c (green) and PE-conjugated anti-B220 (cyan). Scale bars, 300 μm (top row) and 50 μm (bottom row). h,j, After injection of gels loaded with GM-CSF and G400 NP on day 0 and ultrasound treatment on day 3, LNs and gels were excised and analysed on days 6, 10 and 14, respectively, or on days 6, 14 and 21, respectively (n = 4 biologically independent animals). h, Percentage of azido-labelled DCs in the gels over time (one-way ANOVA and Tukey’s test). i, Percentage of azido-labelled DCs in dLNs over time. j, Percentage of azido-labelled F4/80⁺ macrophages over time. All the numerical data are presented as mean ± s.d. (0.01 < *P ≤ 0.05; 0.001 < **P ≤ 0.01; 0.0001 < ***P ≤ 0.001, ****P ≤ 0.0001).

Tracking of azido-labelled DCs.

To study whether azido labelling enables DC tracking, azido-labelled DCs in the gel-draining lymph nodes (dLNs) and non-draining LNs (NdLNs) were quantified. Azide⁺ DCs were found in the dLN of all mice treated with gels releasing G400 NP, and gels containing G400 NP and GM-CSF with ultrasound treatment resulted in the highest number of azide⁺ DCs in dLNs (Fig. 3d,e and Supplementary Fig. 3a,b). Within this group, a notably lower number of azide⁺ DCs were observed in the NdLNs compared to dLNs (Supplementary Fig. 3c,d), as expected. Also, a negligible number of azide-positive CD11b⁺Gr1⁺ and CD11b⁺F4/80⁺ cells were detected in dLNs (Fig. 3f and Supplementary Fig. 3e,f). Confocal imaging also revealed an increased azide density in dLNs of mice injected with gels containing G400 NP and GM-CSF and subjected to ultrasound treatment, in comparison to other groups (Fig. 3g and Supplementary Fig. 3a).

To probe the fate of the azido-labelled DCs, azido-labelled immune cells were monitored over a longer period. The percentage of azide⁺ DCs in gels and dLNs decreased from days 6 to 14 (Fig. 3h–i). The percentage of azide⁺ DCs in dLNs remained unchanged from days 14 to 21, along with a slight increase in the percentage of azide⁺ macrophages (Fig. 3i,j). Given the similar cellular composition in gels from days 6 to 14 (Supplementary Fig. 4a,b) and the dramatic decrease in the percentage of azide⁺ macrophages in gels by day 14 (Supplementary Fig. 4c), the data are not consistent with azide⁺ macrophages gradually migrating to lymph nodes (LNs) and increasing the number of azide⁺ macrophages in LNs. When gels loaded with G400 NP only were used, the percentage of azide⁺ macrophages in LNs did not show any increase over time (Supplementary Fig. 4d), indicating little leakage of G400 NP to LNs for metabolic labelling of macrophages at that site. Since the azide density decreases on cell divisions and is expected to fall below the detection limit over time, a decrease of azide⁺ DCs and macrophages in LNs was expected. The surprising increase of azide⁺ macrophages might indicate that macrophages and LN-resident DCs can phagocytose apoptotic azido-labelled DCs in the LNs^25,26 and remetabolize the azido-sugars.

Targeting of azido-labelled DCs via Click chemistry.

As azido-labelled DCs are present in dLNs after three weeks, the ability of these azido-labelled cells to mediate targeted delivery of DBCO-modified agents via Click chemistry was examined. Mice with azido-labelled DCs were intravenously injected with DBCO-Cy5, resulting in substantially enhanced Cy5 fluorescence intensity in the dLNs, as compared to NdLNs (Fig. 4a,b). The initial fluorescence intensity ratio was 2.7 ± 0.8 (Fig. 4a,b and Supplementary Fig. 5a), and was 2.3 ± 0.8 when DBCO-Cy5 was injected at 15 d (Fig. 4c,d and Supplementary Fig. 5b). Fluorescence-activated cell sorting (FACS) analyses and confocal imaging confirmed a substantially increased number of Cy5⁺ DCs in dLNs (Supplementary Fig. 5c,d). In contrast, no difference in Cy5 fluorescence intensity between dLNs and NdLNs of mice treated with control gels without G400 NP or a PBS solution of G400 NP was observed (Fig. 4a–d and Supplementary Fig. 5a–d).

Fig. 4 |. Azido-labelled DCs mediate targeted conjugation of DBCO-molecules via Click chemistry.

a,b, Gels with GM-CSF and G400 NP were subcutaneously injected (day 0), followed by ultrasound treatment (day 3) and intravenous injection of DBCO-Cy5 (day 8). Mice injected with G400 NP in PBS (G400 SQ) or control gels with GM-CSF only were controls. Ultrasound was applied in gel groups. a, Quantification of Cy5 fluorescence in LNs at 24 h post DBCO-Cy5 injection; the dashed line indicates baseline fluorescence intensity (two-tailed t-test). b, dLN/NdLN Cy5 fluorescence intensity ratio (one-way ANOVA and Tukey’s test). c,d, After injection of gels (day 0) and ultrasound treatment (day 3), DBCO-Cy5 was intravenously injected (day 15). c, Quantification of Cy5 fluorescence in LNs at 24 h post DBCO-Cy5 injection (two-tailed t-test). d, dLN/NdLN Cy5 fluorescence intensity ratio (one-way ANOVA and Tukey’s test). e, Schematic illustration of conjugation of DBCO-cytokines to azido-labelled DCs and subsequent T-cell priming. f–h, Percentages of IL-2 (f), IFN-γ (g) and IL-15/IL-15rα-displaying BMDCs (h) (30-min in vitro incubation with 200 ng ml−¹ DBCO- and Cy5-modified cytokines; one-way ANOVA and Tukey’s test). BMDCs were pretreated with G400 NP or PBS. i, Division index of OT1 cells after incubation with IL-15/IL-15rα-displaying BMDCs in presence of SIINFEKL (one-way ANOVA and Tukey’s test). Azido-labelled DCs were pre-incubated with DBCO–IL-15/IL-15rα or IL-15/IL-15rα (20 ng ml−¹). DC-OT1 cocultures in continuous presence of IL-15/IL-15rα (20 ng ml−¹) were controls. j,k, After DC labelling (gel injection on day 0 and ultrasound day 3), DBCO/Cy5–IL-15/IL-15rα or Cy5–IL-15/IL-15rα was subcutaneously injected (day 6). j,k, The percentage (j) of Cy5⁺ DCs and mean Cy5 fluorescence intensity (k) of DCs in gels 16 h later (two-tailed t-test). All numerical data represent mean ± s.d. (0.01 < *P ≤ 0.05; 0.001 < **P ≤ 0.01; 0.0001 < ***P ≤ 0.001, ****P ≤ 0.0001); n = 4–5 biologically independent samples in all experiments.

As certain DBCO agents, once covalently captured by azido-labelled cells, are stable on the cell surface for extended times²⁷, we hypothesized that DBCO-cytokines can be targeted to azido-labelled DCs to supplement natural paracrine signalling and regulate T-cell priming (Fig. 4e). While cytokine therapies can be potent, treating patients with these pleiotropic agents in a non-targeted manner typically leads to severe complications^28–30. The conjugation of DBCO-modified cytokines, including IL-2, IFN-γ and IL-15/IL-15Rα, onto azido-labelled DCs in vitro was demonstrated first (Fig. 4f–h and Supplementary Figs 6, 7 and 8a–c). As IL-15/IL-15Rα on the surface of antigen-presenting cells can induce the proliferation of CD8⁺ T cells and natural killer cells^31–33, we next studied the ability of IL-15/IL-15Rα–displayed by DCs to improve the activation and proliferation of antigen-specific CD8⁺ T cells. IL-15/IL-15Rα–displaying DCs were cocultured with OT1 cells in the presence of SIINFEKL, and notably increased pSTAT5 expression and proliferation of OT1 cells, as compared to DCs without surface cytokine conjugation, including control DCs with continuous IL-15/IL-15Rα supplementation (Fig. 4i and Supplementary Figs. 8d–f and 9a,b). The proliferation index of OT1 cells also increased with the concentration of G400 NP used to metabolically label DCs before DBCO–IL-15/IL-15Rα conjugation (Supplementary Fig. 9c,d). These experiments demonstrated that the targeted display of IL-15/IL-15Rα on the surface of DCs could improve the activation and proliferation of antigen-specific CD8⁺ T cells, and this was dependent on the concentration of IL-15/IL-15Rα and antigen (Supplementary Fig. 9e,f). In vivo, azido-labelled DCs in the gels and dLNs, generated through previous gel treatment, were also able to capture subsequently administered DBCO/Cy5–IL-15/IL-15Rα (Fig. 4j,k and Supplementary Fig. 10). We next studied whether targeting cytokines to DCs in vivo would improve vaccine-induced neoantigen-specific CD8⁺ T-cell responses and antitumour efficacy. A vaccine consisting of G400 NP, GM-CSF, CpG and two B16F10 neoantigens, M27 and M30 (ref.³⁴), was administered to generate immunity against the neoantigens, while labelling recruited DCs with azido groups. Compared to non-targeted IL-15/IL-15Rα, DBCO–IL-15/IL-15Rα resulted in improved M27-specific CD8⁺ T-cell responses (Supplementary Fig. 11a,b). In a therapeutic study, DBCO–IL-15/IL-15Rα treatment (1 μg kg⁻¹, three doses) resulted in more persistent tumour control and longer survival than other groups (Supplementary Fig. 11c,d). DBCO–IL-15/IL-15Rα treatment resulted in complete tumour regression in 25% of mice and a 57% increase in median survival (Supplementary Fig. 11c,d). It is noteworthy that this dose of DBCO–IL-15/IL-15Rα is much lower than that used in past studies (125–750 μg kg⁻¹)^32,35. No side effects were noted with the cytokine therapy, likely due to the small doses needed here to yield effective responses.

DC-targeted cancer vaccines.

Targeted delivery of tumour antigens and adjuvants to DCs in the LNs has been a key challenge in developing effective cancer vaccines^36–38, so we next explored whether azido-labelled DCs in LNs could capture DBCO-bearing antigens and adjuvants. The ability of azido-labelled DCs to covalently capture DBCO-ovalbumin (OVA) and DBCO-CpG in vitro was first studied. DBCO/FITC-OVA was taken up by azido-labelled BMDCs to a greater extent than FITC-OVA (Supplementary Fig. 12a,b), indicating successful targeting. Similarly, azido-labelled BMDCs were able to capture more DBCO/Cy5-CpG than control BMDCs (Supplementary Fig. 12c,d). Mice with labelled DCs in the dLN, via previous gel treatment, were subsequently subjected to subcutaneous injection of DBCO/Alexa Fluor 647 (A647)-OVA or A647-OVA. At 6 h postinjection, DBCO/A647-OVA showed substantially improved accumulation in dLNs compared to A647-OVA (Fig. 5a,b). FACS analyses confirmed a substantially higher number of A647-OVA⁺ DCs in the dLNs of DBCO/A647-OVA group, in comparison to dLNs of A647-OVA group and NdLNs of both groups (Fig. 5c and Supplementary Fig. 13). At 24 or 48 h postinjection, azido-labelled DCs in dLNs still mediated targeted delivery of DBCO/A647-OVA, as compared to A647-OVA (Fig. 5b,d and Supplementary Fig. 13). The ability to target antigens and adjuvants to DCs in LNs may enable effective vaccination^38,39. To study this possibility, mice with azido-labelled DCs in the dLN, due to previous gel treatment, were subjected to subcutaneous injection of DBCO-OVA/DBCO-CpG. At 8 and 14 d postvaccination, a substantially higher number of OVA-specific CD8⁺ T cells (SIINFEKL tetramer⁺ CD8⁺ T cells and IFN-γ⁺ CD8⁺ T cells) was present systemically in the DC-targeted vaccine group in comparison to other groups (Supplementary Fig. 14).

Fig. 5 |. Azido labelling of DCs mediates targeted delivery of DBCO-antigens and DBCO-adjuvants, which generates potent cellular immune responses.

a–d, After injection of gels (day 0) and ultrasound treatment (day 3), Alexa Fluor 647 (A647)-conjugated DBCO-OVA or OVA was subcutaneously injected (day 6). a, IVIS imaging of dLNs and NdLNs (6 h postinjection of A647-conjugated DBCO-OVA or OVA). Experiment was repeated once independently with similar results. b, Quantification of A647 fluorescence in LNs at 6, 24 and 48 h postinjection of A647-conjugated DBCO-OVA or OVA, respectively. c, Total number of A647-OVA⁺ DCs in LNs (6 h postinjection of A647-conjugated DBCO-OVA or OVA). d, Total number of A647-OVA⁺ DCs in LNs (24 h postinjection of A647-conjugated DBCO-OVA or OVA). e–h, Gels loaded with G400 NP and GM-CSF were subcutaneously injected (day 0), ultrasound was applied (day 3) and DBCO-E7 and DBCO-CpG were subcutaneously injected (days 6, 8 and 10). e,f, representative FACS plots (e) and percentage of E7 tetramer⁺ cells (f) among CD8⁺ T cells in PBMCs (day 16). g,h, representative FACS plots (g) and percentage of IFN-γ⁺ cells (h) among CD8⁺ T cells in PBMCs (day 16). i,j, TC-1 tumours were inoculated (day 0), followed by subcutaneous injection of gels with G400 NP and GM-CSF (day 4), ultrasound treatment (day 7) and subcutaneous injection of DBCO-E7 and DBCO-CpG (days 10, 12 and 14). i, Average tumour volumes over therapeutic study (statistical comparisons on day 35 given). j, Kaplan-Meier plots. All numerical data are presented as mean ± s.d. except for i where mean ± s.e.m. is used; one-way ANOVA and Tukey’s test for b–i and log-rank (Mantel-Cox) test for j (0.01 < *P ≤ 0.05; 0.001 < **P ≤ 0.01; 0.0001 < ***P ≤ 0.001, ****P ≤ 0.0001), n = 6–7 (b–g) and 8–10 (i,j) biologically independent animals.

To demonstrate the potency of this DC/LN-targeted cancer vaccine system, we first studied its capability to amplify E7-specific CD8⁺ T-cell responses. E7 peptide is derived from human papillomavirus (HPV) E7 oncoprotein, and can be processed and presented in the complex of the major histocompatibility complex class I. Mice with azido-labelled DCs in the dLN were subjected to subcutaneous injection of DBCO-E7/DBCO-CpG. The DC-targeted cancer vaccine was again able to generate substantially higher numbers of E7 tetramer⁺ CD8⁺ T cells and IFN-γ⁺ CD8⁺ T cells, as compared to non-targeting groups (Fig. 5e–h). This potent T-cell response translated to full protection from a challenge of E7-expressing TC-1 tumours in the subsequent prophylactic study (Supplementary Fig. 15). In a therapeutic study, a DC-targeted vaccine was able to eradicate established TC-1 tumours, and resulted in the slowest tumour growth and highest tumour-free survival (Fig. 5i,j). In comparison, gel loaded with G400 NP alone plus DBCO-E7/DBCO-CpG showed substantially reduced antitumour efficacy (Supplementary Fig. 16), demonstrating the importance of DC recruitment mediated by GM-CSF. DC-targeted cancer vaccines also clearly outperformed anti-PD-1 and anti-CTLA-4 in treating TC-1 tumours, although anti-PD-1 and anti-CTLA-4 showed improved antitumour efficacy in comparison to the untreated group (Supplementary Fig. 16). Altogether, these experiments demonstrated the potency and broad applicability of this DC-targeted cancer vaccine strategy.

Discussion

We describe a powerful strategy to metabolically label DCs in vivo for subsequent tracking and modulation purposes. This ability can provide both a better understanding of the role of these cells in immunity, and new therapeutic strategies for immunomodulation. Labelling DCs with fluorescent dyes or radioactive tags in vitro allows for the distribution of adoptively transferred DCs to be tracked^14–16, but does not allow labelling and tracking of endogenous DCs in vivo, let alone targeted modulation. Transgenic mice expressing photoconvertible Kaede protein have also been used to monitor the migration of DCs^40–42. However, the poor penetration of the light mediating the labelling (350–400 nm), potential skin damage and unclear impact of Kaede expression on the function of immune cells remain concerns^42,43. In comparison, gels loaded with G400 NP and GM-CSF in our approach enable metabolic labelling of endogenous DCs, tracking of the DCs migrating from scaffolds to LNs and targeted modulation of these DCs with tumour antigens, adjuvants and other immunomodulatory agents. To more precisely label DCs in the future, one could design DC-specific azido-sugars by incorporating linkers cleavable only by DCs^27,44. This metabolic labelling and targeting approach can also be applied to other types of biomaterial scaffold. The pore-forming gels used in this study recruit a high percentage of DCs, making them suitable to initially explore this approach for in vivo labelling, tracking and targeting of DCs.

A new strategy for DC-targeted cancer vaccines was generated by coupling G400 NP-loaded gels with systemic delivery of DBCO-modified antigens/adjuvants. Targeted delivery of antigens and adjuvants to DCs, especially in LNs, has long been a goal of therapeutic cancer vaccines. Nanomaterials that encapsulate antigens and adjuvants can passively accumulate in LNs and release them slowly for DC uptake^45,46. Antigens and adjuvants have also been modified with albumin-binding lipids for LN-trafficking property³⁶. However, these strategies do not directly target DCs in LNs. The use of antibodies that bind to protein receptors on the surface of DCs such as CD11c, DEC 205, anti-CD40 and DC-SIGN has also been explored^{36,37,47–49}. Our DC-targeting technology enables specific targeting of antigens/adjuvants/cytokines to DCs within the LNs without involving endogenous cell-surface receptors, while the impact of hijacking endogenous receptors remains unclear for antibody approaches. DBCO-cytokines are displayed on the surface of DCs with a tuneable turnover rate, while antibody-cytokine conjugates are internalized rapidly. Also, antigens/adjuvants/cytokines or NPs can be easily modified with DBCO with a much higher density than antibodies. The azido-labelled DCs enable targeted delivery of DBCO-antigens and DBCO-adjuvants for improved antigen-specific CD8⁺ T-cell responses and antitumour efficacy (Fig. 5). Nanomaterial cancer vaccines could also be combined with our DC labelling and targeting technology, via simple modification of those materials with DBCO. This DC-targeting approach is highly adaptable to different types of tumour, antigen and adjuvant, including neoantigens or RNAs encoding neoantigens that have shown potential for clinical translation⁵⁰.

The ability to target and present cytokines from antigen-presenting DCs in vivo will likely be broadly useful for priming of antigen-specific T-cell responses. Cytokines are commonly used to regulate the activation and differentiation of T cells, and display of certain cytokines on the surface of antigen-specific DCs could reshape the T-cell priming process. Subcutaneously injected gels were able to generate azido-labelled DCs in LNs, which captured subsequently administered DBCO–IL-15/IL-15Rα for potential, improved priming of antigen-specific CD8⁺ T cells. This cytokine-display approach might also change the phenotype of effector T cells, including the generation of more memory T cells, which should be explored in future studies. In principle, any molecule of interest, including a variety of cytokines, after simple DBCO modification, can be targeted to azido-labelled DCs via efficient Click chemistry to manipulate the interaction between DCs and other types of immune cell.

The targeting approach described here, which involves subcutaneous injection of gels loaded with chemokines and azido-sugar materials, ultrasound treatment and administration of DBCO-modified cancer vaccines or immunomodulatory agents, has notable potential for clinical translation. The biocompatibility of alginate gels and ultrasound treatment have been well demonstrated. A variety of cancer vaccines that have been or are being evaluated in clinical trials showed good safety but modest antigen-specific CD8⁺ T-cell responses and limited therapeutic benefit^51,52. The DC labelling and targeting approach can potentially improve the antitumour efficacy of these cancer vaccines. The potential of biomaterial-based therapeutic cancer vaccines to be used clinically has also been established⁵³, and future optimization of the gels, sugar materials, ultrasound conditions and treatment regimen used here will further facilitate the clinical translation of this technology.

Online content

Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41563-020-0680-1.

Methods

Materials, cell lines and animals.

d-Mannosamine hydrochloride and other chemicals were purchased from Sigma-Aldrich unless otherwise noted. DBCO-NHS and DBCO-sulfo-maleimide were purchased from Click Chemistry Tools. PRONOVA UP MVG sodium alginate (endotoxin-free) was purchased from Fmc Biopolymer AS. GM-CSF, IL-2 and IFN-γ were purchased from PeproTech. SIINFEKL, E7, M27 and M30 peptides were obtained from Peptide 2.0. Endotoxin-free OVA, FITC-OVA and A647-OVA were purchased from Invivogen. CpG, CpG-amine and CpG-Cy5 were purchased from Integrated DNA Technologies. Anti-PD-1 (RMP1–14) and anti-CTLA-4 (9D9) were purchased from BioXCell. SIINFEKL and E7_49–57 tetramers were obtained from the Emory National Institutes of Health Tetramer Core Facility. Primary antibodies used in this study include antimouse CD8-α (BioLegend, 100712, 53–6.7), antimouse IFN-γ (BioLegend, 505810, XMG1.2), antimouse CD3-ε (BioLegend, 100320, 145–2C11), antimouse B220 (BioLegend, 103208, RA3–6B2), antimouse F4/80 (BioLegend, 123110, BM8), antimouse CD11b (BioLegend, 101206, M1/70), antimouse CD11c (BioLegend, 117310, N418), antimouse CD86 (BioLegend, 105008, GL-1), antimouse CD4 (BioLegend, 116008, RM4–4), antimouse Gr1 (Biolegend, 108417, RB6-BC5) and antimouse MHCII (Biolegend, 109908, 10–3.6). TC-1 cells were generated in the laboratory of T. C. Wu (Johns Hopkins University, tested to be mycoplasma free) and maintained in RPMI supplemented with 10% foetal bovine serum (FBS), 1% penicillin/streptomycin and 50 μg ml⁻¹ G418. B16F10 cells (ATCC) were authenticated and tested to be mycoplasma free by ATCC and were maintained in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. C57BL/6 mice were purchased from the Jackson Laboratory. Feed and water were available ad libitum. Artificial light was provided in a 12 h/12 h cycle. All procedures involving animals were done in compliance with National Institutes of Health and Institutional guidelines with approval of Harvard University’s Institutional Animal Care and Use Committee.

Synthesis of poly(azido-sugar).

Ac₄ManAz was first synthesized following the reported procedure²⁷. Ac₄ManAz (1 mmol) was dissolved in methanol/tetrahydrofuran (1/2, v/v), followed by the addition of ammonium carbonate (1.2 mmol). The reaction mixture was stirred at room temperature for 24 h. After removal of the solvent under reduced pressure, the crude product was purified by silica gel column chromatography to yield Ac₃ManAzOH^6,27. Ac₃ManAzOH (1.0 mmol) was then dissolved in dry dichloromethane, followed by the addition of acryloyl chloride (3.0 mmol) and triethylamine (1.0 mmol). The reaction mixture was stirred at room temperature for 24 h. After removal of the solvent and residual acryloyl chloride, the crude product was redissolved in dichloromethane, washed with deionized water three times and dried to yield Ac₃ManAzAL. The crude product was further purified via silica column chromatography using hexane/ethyl acetate as the eluent. Mass spectrometry (liquid chromatography-mass spectrometry) m/z: calculated for C₁₇H₂₃N₄O₁₀ [M + H]⁺ 443.5, found 443.4. ¹H NMR (CDCl₃, 600 MHz): δ (ppm) 6.55 (d, 1H, C(O)NHCH), 6.19 (dd, 1H, C(O)CHCH₂), 6.11 (d, 1H, NHCHCHO), 6.01 (m, 1H, C(O)CHCH₂), 5.69 (m, 1H, C(O)CHCH₂), 5.37 (dd, 1H, CH₂CHCHCH), 5.19 (t, 1H, CH₂CHCHCH), 4.70 (ddd, 1H, NHCHCHO), 4.29 (m, 2H, CH₂CHCHCH), 4.06 (m, 2H, C(O) CH₂N₃), 3.80 (ddd, 1H, CH₂CHCHCH), 2.10, 2.09 and 2.07 (s, 9H, CH₃C(O)). Ac₃ManAzAL (1.0 mmol), azobisisobutyronitrile (AIBN, 0.008 or 0.0005 mmol) and poly(ethylene glycol) methyl ether 2-(dodecylthiocarbonothioylthio)2-methylpropionate (PEG DDMAT, 0.04 or 0.0025 mmol) were dissolved in anhydrous dimethylformamide, followed by three freeze-thaw cycles and stirring at 65 °C for 48 h. Poly(azido-sugar) (G25 or G400) was obtained via precipitation in cold diethyl ether, washing with diethyl ether three times and drying under reduced pressure. Fluorescently labelled G25 and G400 were prepared via conjugation of DBCO-dyes to G25 and G400, respectively. Molecular weights of G25 and G400 were characterized by ¹H NMR spectrometry and gel permeation chromatography.

Preparation of G25 NP and G400 NP.

G25 or G400 polymer was dissolved in dimethylformamide at a concentration of 40 mg ml⁻¹, and dropwise added to ultrapure water (20-fold volume) on vigorous stirring. After stirring for 4 h, G25 NP or G400 NP solutions were dialysed against deionized water for 48 h, sterilized and then stored at 4 °C for use. Dye-labelled G25 NP and G400 NP were prepared similarly using dye-labelled G25 and G400, respectively.

General procedures for flow cytometry analysis of in vitro azido-labelled cells.

Cells were seeded in a 24-well plate at a density of 1 × 10⁵ cells per well. Azido-sugar materials were added and incubated with cells for 72 h. After washing, cells were incubated with DBCO/efluor660-antibody for 30 min along with other antibody stains on ice, before flow cytometry analyses. For BMDCs, bone marrow cells were extracted from the tibia and femur and cultured in RPMI medium containing GM-CSF. Cells on days 6 or 7 were used.

Preparation of pore-forming alginate gels.

GM-CSF loaded gold NPs and porogen beads were prepared following the reported method²¹. A solution of 3% MVG alginate in DMEM was mixed with G400 NP and GM-CSF (antigens and adjuvants were incorporated in some studies), resulting in a final concentration of 2% w/v alginate. This mixture, which constituted the bulk phase of the gels, was then mixed with porogen beads. Finally, the bulk phase alginates were cross-linked by mixing with a sterile CaSO₄ slurry (0.2 g ml⁻¹). For in vitro studies, the gels were immediately cast between two silanized glass plates separated by 2 mm spacers. After allowing the gels to cross-link for 40 min, gel disks were punched out using a sterile 8-mm biopsy punch. For in vivo studies, gels (100 μl) containing 3 μg of GM-CSF were freshly prepared and subcutaneously injected via an 18G needle.

In vitro release of G25 NP and G400 NP from gels.

Gel disks containing Cy3-labelled G25 NP or G400 NP (n = 6) were incubated in DMEM at 37 °C. At different time points, medium was collected for fluorescence measurement on a plate reader. Fresh medium was added to keep the total volume constant. Gels were subjected to ultrasound treatment (30% amplitude, 20 kHz) for 2.5 min at certain time points.

In vivo release of G400 NP from gels.

Mice were subcutaneously injected with gels containing Cy5-labelled G400 NP, and imaged via the IVIS imaging system at 1, 3, 20, 48 and 72 h postgel injection. Mice were then treated with ultrasound (30% amplitude) for 2.5 min, and imaged again at 1, 9, 24, 48 and 72 h postultrasound treatment.

In vivo DC labelling study.

C57BL/6 mice were divided into five groups: gels containing G400 NP (1 mg) and GM-CSF (3 μg) plus ultrasound treatment, gels containing G400 NP and GM-CSF, gels containing G400 NP, gels containing GM-CSF and blank gels. Gels were freshly prepared and subcutaneously injected into the right flank of mice on day 0. On day 3, the hair around the gel was removed and a layer of transmission gel was added, followed by ultrasound treatment (30% amplitude) for 2.5 min. On day 6, LNs and gel scaffolds were excised for analyses: (1) For FACS analyses, gel scaffolds were treated with EDTA for 10 min on ice and disrupted to release the encapsulated cells. LNs were treated with collagenase for 30 min and disrupted to release cells. Cells were strained using a 70-μm cell strainer and seeded into 96-well plates, incubated with DBCO/efluo660-antibody for 20 min on ice and further incubated together with other antibody stains of cell-surface markers for another 20 min on ice. After washing twice, cells were resuspended in FACS buffer and analysed by flow cytometry. Representative gating strategies are provided in Supplementary Fig. 17. (2) For confocal imaging, LNs and gel scaffolds were directly frozen in optimal cutting temperature compound and sectioned with a thickness of 8 μm. Tissue sections were rehydrated, incubated with blocking buffer (5% goat serum) for 2 h, and then stained with DBCO/efluor660-antibody and primary antibodies overnight at 4 °C. For the study of DC labelling kinetics in vivo, mice were subcutaneously injected with gels containing G400 NP and GM-CSF or gels containing G400 NP only on day 0, followed by ultrasound treatment on day 3. Gel scaffolds and LNs were harvested and analysed on days 6, 10 and 14, respectively or days 6, 14 and 21, respectively, following the abovementioned procedures.

In vivo targeting of DBCO-Cy5.

C57BL/6 mice were subcutaneously injected with gels containing G400 NP (1 mg) and GM-CSF (3 μg) or control gels containing GM-CSF alone on day 0, followed by ultrasound treatment (30% amplitude, 2.5 min) on day 3. Mice subcutaneously injected with the PBS solution of G400 NP were not treated with ultrasound. DBCO-Cy5 (5 mg kg⁻¹) was tail vein injected on days 8 or 15. At 24 h postinjection of DBCO-Cy5, gels and LNs were harvested and imaged ex vivo. Cells were then isolated from gels and LNs for flow cytometry analyses.

In vitro targeting of DBCO-OVA and DBCO-CpG.

DBCO-OVA and DBCO-CpG were synthesized via conjugation of DBCO-NHS to OVA and CpG-amine, respectively. BMDCs were incubated with G400 NP or PBS for 3 d. Cells were washed with PBS three times, and incubated with DBCO/FITC-OVA, FITC-OVA, DBCO/Cy5-CpG or Cy5-CpG for 0.5, 1 and 2 h, respectively. Cells were washed and gathered for flow cytometry analyses.

In vivo targeting of DBCO-OVA.

Mice were subcutaneously injected with gels containing G400 NP (1 mg) and GM-CSF (3 μg) or control gels containing GM-CSF alone on day 0, followed by ultrasound treatment (30% amplitude, 2.5 min) on day 3. Alexa Fluor 647-conjugated DBCO-OVA or Alexa Fluor 647-conjugated OVA was subcutaneously injected at tail base on day 6. At 6, 24 and 48 h postinjection, gels and LNs were harvested and imaged ex vivo. Cells were then isolated for flow cytometry analyses.

Vaccination study with DBCO-OVA/DBCO-CpG.

Mice were divided into four groups (n = 8 per group): gel containing G400 NP (1 mg) and GM-CSF (3 μg) + ultrasound + DBCO-OVA (100 μg) /DBCO-CpG (50 μg); gel containing G400 NP and GM-CSF + ultrasound + OVA/CpG; control gels containing GM-GSF + ultrasound + DBCO-OVA/DBCO-CpG; untreated. Gels were subcutaneously injected on day 0, followed by ultrasound treatment (30% amplitude, 2.5 min) on day 3 and tail base subcutaneous injection of DBCO-OVA/DBCO-CpG or OVA/CpG on day 6. On day 10, 14 and 20, respectively, blood was collected and peripheral blood mononuclear cell (PBMCs) were isolated for H2Kb-SIINFEKL tetramer staining. For in vitro peptide restimulation, PBMCs were pulsed with 2 μg ml⁻¹ SIINFEKL and OVA323–339 CD4 epitope for 1.5 h and incubated with GolgiPlug for 4 h, before antibody staining and flow cytometry analyses.

Vaccination study with DBCO-E7/DBCO-CpG.

E7 peptide (GQAEPDRAHYNIVTFCCKCDSTLRLCVQSTHVDIR) is derived from the human papillomavirus (HPV) E7 oncoprotein. DBCO-E7 was synthesized via the coupling reaction of E7 peptide and DBCO-NHS. Mice were divided into four groups (n = 8 per group): gel containing G400 NP (1 mg) and GM-CSF (3 μg) + ultrasound + DBCO-E7 (100 μg)/DBCO-CpG (50 μg); gel containing G400 NP and GM-CSF + ultrasound + E7/CpG; control gels containing GM-CSF + ultrasound + DBCO-E7/DBCO-CpG; untreated. Gels were subcutaneously injected on day 0, followed by ultrasound treatment (30% amplitude, 2.5 min) on day 3 and tail base subcutaneous injection of DBCO-E7/DBCO-CpG or E7/CpG on days 6, 8 and 10. On days 12 and 16, blood was collected and PBMCs were isolated for E7 tetramer staining and IFN-γrestimulation. For in vitro peptide restimulation, PBMCs were pulsed with 5 μg ml⁻¹ E7 for 1.5 h and incubated with GolgiPlug for 4 h, before antibody staining and flow cytometry analyses. For prophylactic tumour study, mice were challenged with a subcutaneous injection of 2.5 × 10⁵ TC-1 cells on day 19.

In vitro display of cytokines on azido-labelled DCs.

IL-15/IL-15Rα with a pending cysteine group was synthesized and purified. DBCO–IL-15/IL-15Rα was obtained via conjugation of DBCO-sulfo-maleimide to IL-15/IL-15Rα. DBCO/Cy5–IL-15/IL-15Rα and Cy5–IL-15/IL-15Rα were synthesized via conjugation of Cy5-NHS to DBCO–IL-15/IL-15Rα or IL-15/IL-15Rα. Similarly, DBCO/Cy5–IL-2 and DBCO/Cy5–IFN-γ were synthesized via conjugation of DBCO-sulfo-NHS and Cy5-NHS to IL-2 or IFN-γ. BMDCs were incubated with different concentrations of G400 NP (2, 10, 50 and 200 μM, respectively) for 3 d, washed and incubated with DBCO-cytokines or cytokines for 30 min.

DC–T-cell coculture.

BMDCs were pretreated with different concentrations of G400 NP (2, 10, 50 and 200 μM, respectively) for 3 d and incubated with DBCO–IL-15/IL-15Rα (2, 6, 20 and 60 ng ml⁻¹, respectively) or IL-15/IL-15Rα or PBS for 30 min. For pSTAT5 analyses, these DCs were cocultured with OT1 cells for 1.5 h and fixed with cold methanol (90%, v/v) overnight, before antimouse pSTAT5 staining and FACS analyses. For OT1 proliferation analyses, these DCs were cocultured with carboxyfluorescein succinimidyl ester-labelled OT1 cells (1:1 T cell to DC ratio) in the presence of SIINFEKL peptide (1, 5, 20 and 100 nM, respectively). Cocultures with the continuous presence of IL-15/IL-15Rα were used as controls. FACS analyses were conducted 3 d later. For some studies, BMDCs were labelled with G400 NP for 3 d and pulsed with SIINFEKL peptide and CpG for 24 h, before the conjugation of DBCO–IL-15/IL-15Rα.

In vivo DC targeting of DBCO–IL-15/IL-15Rα.

Gels containing GM-CSF (1 mg) and G400 NP (3 μg) were subcutaneously injected on day 0, followed by ultrasound treatment on day 3 and tail base subcutaneous injection of DBCO/Cy5–IL-15/IL15Rα (200 ng) or Cy5–IL-15/IL-15Rα (200 ng) on day 6. Gels and LNs were excised for cell isolation and FACS analyses 16 h later.

Coculture of splenic CD8⁺ T cells and CD11c⁺ DCs.

C57BL/6 mice were subcutaneously injected with gels containing G400 NP (1 mg), GM-CSF (3 μg), M27 (50 μg), M30 (50 μg) and CpG (50 μg) on day 0, treated with ultrasound on day 3 and subcutaneously injected with DBCO–IL-15/IL-15Rα (20 ng) or IL-15/IL-15Rα (20 ng) or PBS at tail base on day 6. Spleens were harvested and disrupted on day 12. CD8⁺ T cells were isolated from splenocytes, stained with carboxyfluorescein succinimidyl ester and cocultured with pre-isolated CD11c⁺ DCs (10:1 T cell to DC ratio) from spleens of naïve mice. After 3 d, cells were stained with anti-CD3, anti-CD8, live/dead stain, anti-IFN-γ and anti-TNF-α, followed by FACS analyses.

Therapeutic tumour study.

For TC-1 tumour study, TC-1 tumours (2.5 × 10⁵ cells per mouse) were inoculated on day 0, followed by subcutaneous injection of gels loaded with G400 NP (1 mg) and GM-CSF (3 μg) on day 4, ultrasound treatment on day 7 and subcutaneous injection of DBCO-E7 (100 μg) and DBCO-CpG (50 μg) on days 10, 12 and 14. Tumour growth and body weight of animals were closely monitored. The tumour volume was calculated using the formula (length) × (width)²/2, where the long axis diameter was regarded as the length and the short axis diameter was regarded as the width. In some studies, anti-PD-1 (100 μg) or anti-CTLA-4 (100 μg) was intraperitoneally injected on day 6, 9, 12, 15, 18 and 21, respectively. Animals were euthanized for humane reasons when tumours grew to 20 mm in longest diameter. For B16F10 tumour study, mice were subcutaneously injected with B16F10 tumour cells (1 × 10⁵ per mouse) on day 0, injected with gels containing G400 NP (1 mg), GM-CSF (3 μg), M27 (50 μg), M30 (50 μg), gp100 (25 μg), TRP2 (25 μg) and CpG (50 μg) on day 5, treated with ultrasound on day 8 and subcutaneously injected with DBCO–IL-15/IL-15Rα (20 ng) or IL-15/IL-15Rα (20 ng) on day 11. Tumour growth and body weight of animals were closely monitored.

Statistical analyses.

The significance between two groups was analysed by a two-tailed, Student’s t-test. For multiple comparisons, one-way analysis of variance (ANOVA) with Tukey’s test was performed. P values that were ≤0.05 were considered statistically significant (0.01 < *P ≤ 0.05; 0.001 < **P ≤ 0.01; 0.0001 < ***P ≤ 0.001, ****P ≤ 0.0001). Sample size was empirically set at n = 3–6 for in vitro cell experiments, n = 3–6 for in vivo imaging and DC tracking studies, n = 5–8 for vaccination and prophylactic tumour studies and n = 7–10 for therapeutic tumour studies.

Reporting Summary.

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

All data supporting the results are provided with the manuscript. Raw datasets are available at https://dataverse.harvard.edu/privateurl.xhtml?token=4d292632-f6274a56-8e01-60e36d0883f5.