Tetanus toxoid and CCL3 improve dendritic cell vaccines in mice and glioblastoma patients

Abstract

After stimulation, dendritic cells (DCs) mature and migrate to draining lymph nodes to induce immune responses. As such, autologous DCs generated ex vivo have been pulsed with tumour antigens and injected back into patients as immunotherapy. While DC vaccines have shown limited promise in the treatment of patients with advanced cancers including glioblastoma, the factors dictating DC vaccine efficacy remain poorly understood. Here we show that pre-conditioning the vaccine site with a potent recall antigen such as tetanus/diphtheria (Td) toxoid can significantly improve the lymph node homing and efficacy of tumour-antigen-specific DCs. To assess the effect of vaccine site pre-conditioning in humans, we randomized patients with glioblastoma to pre-conditioning with either mature DCs or Td unilaterally before bilateral vaccination with DCs pulsed with Cytomegalovirus phosphoprotein 65 (pp65) RNA. We and other laboratories have shown that pp65 is expressed in more than 90% of glioblastoma specimens but not in surrounding normal brain, providing an unparalleled opportunity to subvert this viral protein as a tumour-specific target. Patients given Td had enhanced DC migration bilaterally and significantly improved survival. In mice, Td pre-conditioning also enhanced bilateral DC migration and suppressed tumour growth in a manner dependent on the chemokine CCL3. Our clinical studies and corroborating investigations in mice suggest that pre-conditioning with a potent recall antigen may represent a viable strategy to improve anti-tumour immunotherapy.

Conflict of interest statement

The authors declare the following competing financial interests: D.A.M., K.A.B., and J.H.S. have filed provisional patents related to the use of Td pre-conditioning as a method to improve immunization efficacy. D.A.M. has served as a paid member of the Schering Plough North American Investigators Advisory Board. S.K.N. is a co-inventor on a patent that describes the use of DCs transfected with tumor antigen encoding RNA that has been licensed by Argos Therapeutics (Durham, NC) through Duke University. S.K.N. has no financial interests in Argos Therapeutics and is not compensated by Argos Therapeutics. D.A.R. has served as paid speaker for Schering/Merck and Genentech/Roche. The remaining authors declare no competing financial interests.

Figures

Extended Data Figure 1

Schema of clinical trial.

Extended Data Figure 2. Recall responses induced by other CD4+ T cell-dependent protein antigens increase DC migration to VDLNs

Primary immunization and vaccine site preconditioning with CD4+ T cell-dependent protein antigens increase DC migration to VDLNs. Mice were immunized with either Haemophilus b conjugate (Hib) or pneumococcal 13-valent conjugate (PCV) intramuscularly and two weeks later received vaccine site pre-conditioning with the recall antigen (Recall) or saline (Control). A separate cohort of mice received saline only throughout the immunization schedule (Saline). Scatter plot shows biological replicates of individually processed right and left iLN per mouse (4 mice per group). Percent migration of RFP+ DCs to VDLNs (one-way ANOVA, P < 0.0001; post-hoc Tukey t test, PCV Control vs. Recall, P < 0.05, Hib Control vs. Recall, P < 0.05,). Representative of n = 3 experiments; mean ± s.e.m.

Extended Data Figure 3. Bilateral migration of OVA-DCs following Td pre-conditioning or Td-activated CD4+ T cell transfer

Uptake of injected DCs to right and left iLN 48 hours following DC vaccination in Td-immune mice receiving Td pre-conditioning or naïve mice administered Td-activated CD4+ T cells. Scatter plot shows biological replicates of individually processed right and left iLN per mouse (5 mice per group). CD4Act ipsilateral vs. contralateral, paired t test, P = 0.41). Representative of n = 4 experiments; mean ± s.e.m.

Extended Data Figure 4. Unilateral pre-conditioning with unpulsed DCs or TNF-α results in increased DC homing to ipsilateral draining inguinal lymph nodes

Td-immune mice preconditioned with Td or saline prior to administration of OVA RNA-pulsed DC vaccine. Separate cohorts of naïve mice received either 1×106 unpulsed DCs or 30 ng TNF-α on one side of the groin 24 hours prior to the bilateral RFP+ DC vaccine. DC migration was quantified 24 hours post-vaccination. a, DC migration to ipsilateral lymph nodes (one-way ANOVA, P = 0.0018; post-hoc Tukey t test, Saline i.d. vs. Td i.d., P = 0.007, Saline i.d. vs. TNF-α, P < 0.05, Saline i.d. vs. DCs, P < 0.05; Td i.d. vs. TNF-α, P = 0.042; DCs vs. Td i.d. and DCs vs. TNF-α, N.S.). b, DC migration to contralateral lymph nodes (one-way ANOVA, P = 0.003; post-hoc Tukey t test, Saline i.d. vs. DCs or TNF-α, P > 0.05; Td i.d. vs. TNF-α, DCs, or Saline i.d., P < 0.05). n values are biological replicates of individually processed right and left iLN per mouse (4 mice per group). Representative of n = 3 experiments; mean ± s.e.m.

Extended Data Figure 5. Serum cytokine and chemokine profile following Td preconditioning in patients and mice

a, Serum cytokine panel of patients following vaccine site pre-conditioning with Td or unpulsed DCs (Wilcoxon rank sum, IFN-γ and IL-4, P < 0.05 (n = 6 patients). b, Similar panel in mice (Wilcoxon rank sum, all comparisons, P > 0.05; Td recall n = 5, Non-Td n = 6). c, Patient serum chemokines following vaccine site pre-conditioning. Patient CCL2 and CCL3 in Td recall (Td, n = 6) vs. Non-Td (unpulsed DC, n = 5) (one-way ANOVA and Wilcoxon rank sum, P < 0.05). d, Murine CCL22, CCL7, and CCL3 in Td recall (Td, n = 8 mice) and Non-Td (saline, n = 8 mice) (one-way ANOVA and Wilcoxon rank sum, P < 0.05). a–d, individual values represent biological replicates; mean ± s.e.m.

Extended Data Figure 6. Td vaccine site pre-conditioning results in CCL3 upregulation in Td-immune hosts

a, CCL3 production in skin site following Td pre-conditioning (Td ipsilateral vs. contralateral. Representative of four independent experiments b, CCL3 production in skin following Td recall response. c, CCL3 induction at skin site is abrogated with prior host depletion of CD4+ T cells. a-c, Bars represent CCL3 protein detected in skin sites from n = 2 mice with n = 2 technical replicates performed per mouse. d, CCL3 remains elevated at the Td pre-conditioning site in the skin following DC vaccination (24, 48, and 72 hr, one-way ANOVA, P = 0.0001, Td + Td i.d. + OVA-DC vs. Td + saline i.d. + OVA-DC and Saline + saline i.d. + OVA-DC, P < 0.05, post-hoc Tukey t test). Individual values represent biological replicates from n = 4 mice; mean ± s.e.m.

Extended Data Figure 7. Migratory DC subsets in wild-type and Ccl3−/− mice following induction of Td recall responses

Both wild-type and Ccl3−/− mice were first immunized with Td and then challenged with Td pre-conditioning. Migration of endogenous DC subsets to inguinal lymph nodes contralateral to the site of Td pre-conditioning was assessed at day 4 and day 8 following Td administration. a, Gating strategy used to quantify DC subsets in inguinal lymph nodes following skin pre-conditioning with Td (LC: Langerhans cells; MoDC: monocyte-derived DC). b, Day 8 migration of LC population to non-draining inguinal lymph nodes in Ccl3−/− hosts is reduced in absence of CCL3 (two-sample t test, P = 0.046). Representative of three experiments. Individual values represent biological replicates from n = 4 mice; mean ± s.e.m.

Extended Data Figure 8. Anti-tetanus toxoid memory responses are induced and maintained in wild-type and Ccl3−/− mice throughout Td priming and boosting

WT and Ccl3−/− mice primed and boosted with Td. Serum from immunized mice was harvested two weeks following each immunization prior to the next booster vaccine (for each boosting phase, WT vs. Ccl3−/−, two-sample t test, P > 0.05); Intramuscular (i.m.). Scatter plot showing averaged values from n = 4 mice with n = 2 technical replicates performed per mouse. Representative of three experiments; mean ± s.e.m.

Extended Data Figure 9. CCL21 levels in Td pre-conditioning skin sites and draining lymph nodes of wild type and Ccl3−/− mice

a, CCL21 levels in skin site of Ccl3−/− hosts following induction of Td recall response and CCL3 administration. Mixed model accounting for within-mouse correlation of measurements, F-test, P < 0.001; Pairwise comparisons, WT Td + Td i.d. (n = 4) vs. Ccl3−/− Td + Td i.d. (n = 3), P = 0.049; Ccl3−/− Td + Td i.d. + CCL3 i.v. (n = 4) vs. Ccl3−/− Td + Td i.d. and vs. Ccl3−/− + CCL3 i.v. (n = 4), P = 0.044 and P = 0.0045, respectively. Scatter plot shows averaged values with n = 2 technical replicates performed per mouse. b, Bilateral inguinal lymph node (iLN) CCL21 levels in WT mice following Td recall with skin site pre-conditioning. Bars represent CCL21 protein within iLN ipsilateral and contralateral to the side of Td pre-conditioning from n = 2 mice with n = 2 technical replicates performed per iLN. Representative of three experiments. c, Increased lymph node CCL21 in Ccl3−/− hosts following CCL3 reconstitution and induction of Td recall response (All two group comparisons). CCL21 levels in bilateral iLN of Ccl3−/− hosts following induction of Td recall response and CCL3 administration. Mixed model accounting for within-mouse correlation of measurements, F-test, P < 0.001; Pairwise comparisons, WT Td + Td i.d. (n = 4 iLN) vs. Ccl3−/− Td + Td i.d. (n = 3 iLN), P = 0.045; Ccl3−/− Td + Td i.d. + CCL3 i.v. (n = 4 iLN) vs. Ccl3−/− Td + Td i.d. and vs. Ccl3−/− + CCL3 i.v. (n = 4 iLN), P = 0.0066 and P = 0.026, respectively. Scatter plot shows averaged values with n = 2 technical replicates performed per LN sampled.

Figure 1. Td pre-conditioning increases DC migration to VDLNs and is associated with improved clinical outcomes

a, DC migration in Td (n = 6) vs. unpulsed DC patients (n = 6) (two sample t test, P = 0.049). Mean ± s.e.m., n values represent biological replicates of patient bilateral inguinal lymph nodes (iLNs). b, Patient PFS and c, OS (Logrank test, P = 0.013). d, Hazard ratios (HRs): DC migration efficiency from Td and DC cohorts showing the effect of a 1 unit increase in percent migration on PFS (top) and OS (bottom) (Cox proportional hazards model, PFS HR = 0.845 P = 0.027; OS HR = 0.820 P = 0.023). b and c, n = 3 censored Td patients (no progressive disease at survival analysis).

Figure 2. Td recall response activates CD4+ T cells to increase DC migration to VDLNs

a, Control inguinal (n = 5 mice) vs. Td inguinal (n = 5 mice, two sample t test, P = 0.0001; Td popliteal (n = 5 mice) vs. Td inguinal, paired t test, P = 0.014. b, Mice primed and boosted with saline (Primary Td, n = 6) or Td (Control and Recall Td, n = 6) with Td (Primary and Recall Td) or saline (Control) pre-conditioning; one-way ANOVA, P = 0.004; post-hoc Tukey t test, Control vs. Recall Td, P = 0.006; Primary Td vs. Recall Td, P = 0.011. c, DC migration in depleted Td-immunized mice (n = 5); one-way ANOVA, P < 0.0001; post-hoc Tukey t test, Td vs. CD4, P = 0.005; Td vs. CD8, CD19, or NK1.1, P > 0.05). d, DC migration following CD4+ transfer (n = 4 mice); one-way ANOVA, P < 0.0001; post-hoc Tukey t test, CD4Act vs. CD4Naive, P < 0.05; Control vs. CD4, P > 0.05; Td vs. CD4, P > 0.05. e, Patient iLN ipsilateral (n = Naive Act 6) and contralateral (n = 6) to pre-conditioning (paired t test, P = 0.28). f, Mouse (paired t test, P = 0.37; n = 6 LN per group,). a–d, f, Representative of 4 experiments; mean ± s.e.m. a–d, biological replicates of individual right and left iLN or LN ipsilateral to Td/saline for popliteal and GFP− groups.

Figure 3. Td recall responses and induced CCL3 cooperate to facilitate DC migration to VDLNs

a, Serum CCL3 fold increase over non-Td cohorts (patient, n = 6; mouse, n = 8 biological replicates); signed rank test, P = 0.031 and P = 0.039. b, Left: DC migration in WT (n = 5) vs. Ccl3−/− mice (n = 4), two sample t test, P = 0.023). Right: Td-activated CD4+ transfer in Ccl3−/− hosts (n = 4 mice); two sample t test, P = 0.029. c, CCL3 and Td recall responses rescue migration (n = 4 mice); one-way ANOVA, P < 0.0001; post-hoc Tukey t test, Ccl3−/− Td + Td i.d. + CCL3 i.v. vs. Ccl3−/− Td + Td i.d. and vs. Ccl3−/− + CCL3 i.v., P = 0.007 and P = 0.001, respectively. b, c, biological replicates of individual right and left iLNs. Representative of 3 experiments; mean ± s.e.m.; intradermal (i.d.); intramuscular (i.m.); intravenous (i.v.).

Figure 4. Td pre-conditioning improves responses in tumor-bearing mice

a, Insert: Transformed growth curves, mixed linear effects model. Pairwise comparisons of regression line slopes (F-test, P < 0.0001). Day 22 volume (Td + OVA-DC vs. Td + GFP-DC, two sample t test, P = 0.002; n = 7). b, Antigen-specific responses with Td pre-conditioning. Day 15 volume (all groups, F-test, P = 0.016; pairwise Tukey t tests, Td + OVA-DC + B16-OVA (n = 7) vs. Td + OVA-DC + B16 (n = 5), P = 0.0004; Td + OVA-DC + B16-OVA vs. Saline + OVA-DC + B16-OVA (n = 6), P = 0.0002). Day 22 volume (Td + OVA-DC + B16-OVA vs. Td + OVA-DC + B16, two sample t test, P < 0.0001). c, Tumor growth in Ccl3−/− mice. Day 11 volume (all groups, F-test, P = 0.005). Day 27 volume (Td + OVA-DC WT vs. Td + OVA-DC Ccl3−/−, two sample t test, P = 0.042; n = 8). d, Antitumor responses in plt mice. Day 16 volume (all groups, F-test, P = 0.004). Day 24 volume (Td + OVA-DC plt (n = 7) vs. Td + OVA-DC WT (n = 6), two sample t test, P < 0.05). a–d, Representative of 3 experiments; mean ± s.e.m; autologous lymphocyte transfer (ALT).