Nanoparticle conjugation of CpG enhances adjuvancy for cellular immunity and memory recall at low dose Alexandre de Tittaa,1, Marie Ballestera,1, Ziad Juliera, Chiara Nembrinia, Laura Jeanbarta,b, André J. van der Vliesa, Melody A. Swartza,b,c,2, and Jeffrey A. Hubbella,c,2 a Institute of Bioengineering, School of Life Sciences and School of Engineering, bSwiss Institute of Experimental Cancer Research, School of Life Sciences, and cInstitute of Chemical Science and Engineering, School of Basic Sciences, Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland
Edited* by Mark E. Davis, California Institute of Technology, Pasadena, CA, and approved October 24, 2013 (received for review July 12, 2013)
In subunit vaccines, strong CD8+ T-cell responses are desired, yet they are elusive at reasonable adjuvant doses. We show that targeting adjuvant to the lymph node (LN) via ultrasmall polymeric nanoparticles (NPs), which rapidly drain to the LN after intradermal injection, greatly enhances adjuvant efficacy at low doses. Coupling CpG-B or CpG-C oligonucleotides to NPs led to better dual-targeting of adjuvant and antigen (codelivered on separate NPs) in cross-presenting dendritic cells compared with free adjuvant. This led to enhanced dendritic cell maturation and T helper 1 (Th1)-cytokine secretion, in turn driving stronger effector CD8+ T-cell activation with enhanced cytolytic profiles and, importantly, more powerful memory recall. With only 4 μg CpG, NP-CpG-B could substantially protect mice from syngeneic tumor challenge, even after 4 mo of vaccination, compared with free CpG-B. Together, these results show that nanocarriers can enhance vaccine efficacy at a low adjuvant dose for inducing potent and long-lived cellular immunity.
M
ost current subunit vaccines base their success on driving durable antibody responses that prevent pathogen infections. However, therapeutic vaccines against cancer as well as chronic infections must elicit the activation of cytotoxic T lymphocytes (CTLs) to eliminate infected or malignant cells. Induction of long-lasting CTLs with subunit vaccines requires, first, delivery of antigen to cross-presenting dendritic cells (DCs) and, second, potent activation of these DCs by the vaccine adjuvant, which instructs DCs how to activate T cells. Nanocarriers can be used to modulate the immune response induced by antigens and adjuvants by modifying their characteristics, such as stability, tissue and cell targeting, and DC-activating capacity (1–12). We have previously described the targeting benefits of conjugating antigens to ultrasmall Pluronic-stabilized poly(propylene sulfide) (PPS) nanoparticles (NPs) in terms of CD8+ T-cell and Th1 responses (13, 14). The hydrophobic PPS core is hydrolytically stable, allowing wet storage, yet is oxidized to products that are soluble in vivo (15, 16). The Pluronic stabilizing agent [a block copolymer of poly(ethylene glycol) and poly(propylene glycol), produced by BASF] results in a corona of poly(ethylene glycol), which prevents aggregation before and after antigen or adjuvant conjugation. The ∼25–30-nm ultrasmall size allows efficient delivery to and accumulation within the draining lymph node (LN) (17). Finally, we have demonstrated the advantages of antigen conjugation to the NP surface via a disulfide bond, which can be reversibly cleaved within the reductive environment within the endosomal compartment of DCs, leading to more efficient crosspresentation than when antigen is irreversibly conjugated to the NP surface (15, 18). Here, we explore conjugation of adjuvant to the NP surface, focusing on CpG oligonucleotides, with the objective of delivering both antigen and adjuvant into the same LNresident DCs, with the concept that even if the antigen and adjuvant are on separate particles (a distinct formulation advantage), the two payloads are targeted to the same cells because their carriers have the same size and clearance characteristics. Strong adjuvants, such as Toll-like receptor (TLR) agonists, are required for subunit vaccine formulations to drive effective 19902–19907 | PNAS | December 3, 2013 | vol. 110 | no. 49
immune responses, as they instruct DCs how to activate cognate T cells. CpG, a TLR-9 agonist, has been shown both in preclinical and clinical studies to promote Th1-responses (19). Three main types of CpGs have been described: the multimeric CpG-A (or D-type) stimulates plasmacytoid DCs to secrete IFN-α; CpG-B (or K-type) induces B cells and DCs to mature and secrete cytokines such as IL-12p70, IL-6, and TNF-α; and CpG-C has the ability to induce both CpG-A and CpG-B types of responses. In turn, IFN-α secretion from plasmacytoid DCs promotes natural killer T cells (20), CD8+ T-cell activation and cytotoxicity (21), and maturation of conventional DCs. Secretion of IL-12p70 induces Th1 and CTL responses, IL-6 leads to maturation of B cells and induces Th17 responses, and TNF-α stimulates DC maturation and T-cell activation (22–24). In addition to these different activities, the three forms of CpG have different sizes, which could strongly affect their transport on interstitial injection (1) and distribution within the LN (25), which could account for some of their in vivo differences. We and others have previously shown that antigen size affects LN targeting and ultimate immune response after intradermal injection, which can be optimized by coupling peptides or proteins to ultrasmall (∼25–50 nm) polymeric NPs (26). CpGA naturally forms multimeric structures of ∼20–100 nm at physiological pH and temperature (27), but CpG-B and CpG-C do not. We hypothesize that conjugation of CpG-B and CpG-C to Significance High adjuvant doses are generally required to induce strong CD8+ T-cell immunity with subunit vaccines. Here we codeliver an antigen and an adjuvant coupled on separate ultrasmall polymeric nanoparticles. Because both payloads are attached to similarly sized nanoparticles, and as size is the principle determinant of nanoparticle drainage, this enhanced the dual uptake of antigen and adjuvant by cross-presenting dendritic cells resident in the draining lymph nodes. This cotargeting induced potent effector CD8+ T cells and a more powerful memory recall of these cytotoxic T cells compared with nanoparticle-conjugated antigen with free adjuvant. As such, nanoparticle conjugation enhanced the immunogenicity of adjuvants while maintaining a low dose, and thus limiting toxicity, affecting the design of future subunit vaccine formulations. Author contributions: A.d.T., M.B., C.N., M.A.S., and J.A.H. designed research; A.d.T., M.B., Z.J., and L.J. performed research; A.d.T., M.B., L.J., and A.J.v.d.V. contributed new reagents/analytic tools; A.d.T., M.B., and Z.J. analyzed data; M.A.S. and J.A.H. directed the project; and A.d.T., M.B., M.A.S., and J.A.H. wrote the paper. Conflict of interest statement: The Ecole Polytechnique Fédérale de Lausanne has filed for patent protection on the nanoparticle platform described herein, and some of the authors are named as inventors on those filings. *This Direct Submission article had a prearranged editor. Freely available online through the PNAS open access option. 1
A.d.T. and M.B. contributed equally to this work.
2
To whom correspondence may be addressed. E-mail: melody.swartz@epfl.ch or jeffrey.hubbell@epfl.ch.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1313152110/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1313152110
NPs could increase their immunogenic potential, while allowing a minimal dose, by targeting both the adjuvant and antigen to the same LN-resident DCs. We conjugated CpG-B and CpG-C to NPs via a reducible bond and designated the NP-conjugates as NP-CpG-B and NPCpG-C, respectively. CpG-A was not conjugated, as it naturally forms a multimeric structure of similar size as our NPs (27). We administered NP-CpG-B and NP-CpG-C in conjunction with ovalbumin-conjugated NPs (NP-OVA) intradermally in the footpads in mice. Adjuvant conjugated NPs enhanced uptake of antigen by DCs and activation of cross-presenting DCs and led to the induction of potent effector and memory antigen-specific T cells. By conjugating CpG-B to NPs, we induced a more potent cellular response at an early time, characterized by increased degranulation and secretion of IFN-γ by CD8+ T cells and enhanced proportion of multifunctional CD4+ T cells, even at a low dose of 4 μg. At a later time, conjugation of CpG-B and CpG-C enhanced the clonal expansion of antigen-specific memory CD8+ T cells and differentially enhanced their cytotoxic abilities. Mice immunized with either conjugated CpG variants were able to control an IFN-γ-sensitive tumor challenge (E.G7-OVA, a tyhmoma line), but NP-CpG-B was superior at dampening the growth of a more aggressive tumor (B16-F10-OVA, a melanoma line). Results Reducible Conjugation of 5′-End Modified CpG to NPs. We first modified CpG-B and CpG-C with a reactive thiophosphate group, allowing coupling to our previously developed pyridyl-cysteamine functionalized NPs (16). This reaction occurs through a disulfide exchange between the functionalized NP and the thiophosphate terminal group of CpG (Fig. S1A). The covalent bond can be
reduced, by mimicking the reducing environment of the endosomal compartment of the cell, with 2-mercaptoethanol (Fig. S1B). NP-CpG Induces Increased Activation of Dendritic Cells. To determine the direct activity of NP-conjugated CpG on DCs, we first evaluated our formulations in vitro. Although we observed increasing CD86 expression on DCs with increasing doses of each CpG, levels were enhanced by NP-coupled formulations (Fig. 1 A–C). Similarly, the proportion of mature DCs (CD86+ MHC-II+) was enhanced after stimulation with NP-conjugated compared with free CpG-B (Fig. 1B). The response to NP-CpG-B induced significantly higher levels of IL-12p70 secretion by DCs than did free CpG-B, whereas the response to NP-CpG-C was similar to that of free CpG-C (Fig. 1D). Enhancement of immune potentiation properties as a result of NP conjugation of CpG-B were also observed in vivo. Indeed, NPCpG-B coinjected with NP-OVA more potently activated crosspresenting (CD11c+MHC-II+CD11b−) DCs (28) in the draining LNs than did free CpG-B (Fig. 1 E and F). Importantly, at least one mechanism for this enhanced in vivo efficacy of NP-CpG was improved, cotargeting with NP-OVA to the same LN-resident DCs (Fig. 1G). Although we observed an increase in activation of DCs by our NP-conjugated formulation, this did not lead to an increase in blood levels of the inflammatory cytokine TNF-α, leading to only minor elevations in IL-12p70 and IL-6 (Fig. S2). Intradermal Immunization with CpG-B-Conjugated NPs Induces Increased Polyfunctional Th1 Cells and Cytotoxic Activity of CD8+ T Cells. Next, we evaluated T-cell responses 19 d after intradermal
vaccination (Fig. 2A). Interestingly, at the low (4 μg) adjuvant dose used, NP-CpG-B with NP-OVA was the only formulation that could induce significantly higher amounts of Th1-leaning
in vitro DC Fig. 1. CpG conjugation to NPs enhances crosspresenting dendritic cell activation. (A) RepresentaPBS CpG-B (free) NP-CpG-B PBS CpG-A (free) tive CD86 mean fluorescence intensity (MFI) histo15% 1.8% 3.1% + gram of CD11c DC activated with 0.5 μM of CpG-A PBS and free or NP-conjugated CpG-B and CpG-C for 6 h CpG-B (free) in vitro. (B) Representative flow cytometry plot of NP-CpG-B DCs activated for 6 h in the presence of 1 μM free or PBS NP-conjugated CpG-B. Numbers represent percentCpG-C (free) CD86 NP-CpG-C age of DCs (CD11c+MHC-II+) expressing both CD86 CD86 and MHC-II. (C) MFI of CD86 signal indicates DC activation after 6 h exposure in vitro to increasing ## ### 3 30 (0.005–1 μM) concentrations of CpG-A, CpG-B, or CpG-A (free) *** # CpG-B (free) ns CpG-C or NP-CpG-B or NP-CpG-C, showing stronger * NP-CpG-B activation by NP-CpG-B and NP-CpG-C than by free CpG-C (free) 2 20 ** CpG-B and CpG-C. Dashed bar shows PBS control. * NP-CpG-C * Symbol legend is in D. (D) DCs secreted higher ns amounts of IL-12p70 when stimulated for 12 h with 1 10 NP-CpG-B than with free CpG-B. (E and F) Mice were injected with 4 μg soluble or NP-conjugated CpG-B 0 0 and 10 μg OVA-conjugated NP in the four footpads. 0.01 0 0.1 0.5 1 1 0.01 0.1 LNs were harvested 20 h after injection. LN cells, concentration μM concentration μM + + gated for cross-presenting DCs (CD11c MHC-II CD11b−), were analyzed by flow cytometry for CD80 in vivo LN and CD86 expression. Measurements of MFI of CD86 *** 35 100 signal (E) and fraction of DCs that are activated as naïve 30 ** indicated by CD80 and CD86 coexpression (F) show 80 25 * stronger DC activation by NP-CpG-B than by free NP-OVA 60 CpG-B. (G) Mice were injected with 4 μg soluble or 20 NP-CpG-B fluorescently labeled with Dy633 (CpG15 CpG-B (free) 40 + NP-OVA B*) and 10 μg NP-conjugated OVA fluorescently la10 beled with Alexa488 (OVA”). Draining LNs were 20 NP-CpG-B 5 + NP-OVA harvested 20 h after injection. LN cells, gated for nd + + 0 0 DCs (CD11c MHC-II ), were analyzed by flow cytomnaïve CpG-B* NPnaïve PBS CpG-B NP0 5 10 15 (free) CpG-B* (free) CpG-B MFI CD86 etry for the OVA” and CpG-B*. Measurements show + + + NP-OVA + NP-OVA’’ (CD11c MHC-II CD11b ) a significantly higher fraction of CpG-B*+ DCs associated with OVA” when CpG was conjugated to NPs. Data in C, D, and E represent mean ± SEM (n = 3 independent experiments) for in vitro study. (F and G) Median ± 95% confidence interval (n = 8 mice, 2 independent experiments) for in vivo study. *P < 0.05; #P < 0.05; **P < 0.01; ##P < 0.01; ***P < 0.001; ###P < 0.001; ns, not significant; nd, not detectable. Asterisks indicate statistics between CpG-B and NP-CpG-B; pound signs indicate statistics between CpG-C and NP-CpG-C. MHC-II
D
F
***
de Titta et al.
CD80+CD86+ (% of CD11c+MHC-II+CD11b-)
E
G
OVA’’+ (% of CD11c+MHC-II+CpG*+)
MFI CD86 (CD11c+)
IL-12p70 (ng/ml)
C
PNAS | December 3, 2013 | vol. 110 | no. 49 | 19903
IMMUNOLOGY
B
A
phenotypes in CD8+ T cells, these cells must recognize and kill their antigen-bearing target cell to be effective. To determine the functionality of CTLs induced by our vaccine formulations, we challenged vaccinated mice with E.G7-OVA tumor cells (Fig. 3A). Consistent with the previous data, we found significantly delayed tumor onset in mice vaccinated with NP-CpG-B compared with free CpG-B (Fig. 3B) and longer tumor-free survival (Fig. 3C). This correlated with much higher levels of antigen-specific (SIINFEKL-MHC-I-pentamer+, the peptide
B day 0
14
19
boost
immunization NP-OVA CpG variants
IFN-γ+TNF-α+ (% of CD3ε+CD4+)
A NP-OVA CpG variants
0.15
naïve
1.04
CD107a+ (% of CD3ε+CD8+)
ns
ns
0
CpG-B (free) + NP-OVA
NP-CpG-B + NP-OVA
3.56
8.63
2.77
F 9.41
NP-OVA CpG variants
tumor
boost NP-OVA CpG variants
106 E.G7-OVA
naïve PBS + NP-OVA CpG-B (free) + NP-OVA NP-CpG-B + NP-OVA
D *
25 0 0
10
600 400 200
20
30
days after tumor inoculation
0
10
20
30
days after tumor inoculation
10
SIINFEKL-MHCI specific (% of CD3ε+CD8+B220-)
75 50
800
0
100
tumor free (%)
C
tumor volume (mm3)
immunization
B
19
8 6 4
**
2 0
ns 13
15
17
19
days after tumor inoculation
21
Fig. 3. CpG-B conjugation to NPs enhances protection against a tumor challenge. (A) Immunization schedule; 10 μg OVA conjugated to NPs was injected with 4 μg free or NP-conjugated CpG-B in the footpads of C57BL/6J mice on days 0 and 14. Control mice were injected at the same time with the same dose of NP-OVA only. On day 19, 106 E.G7-OVA lymphoma tumor cells were injected s.c. on the back. (B and C) Tumor growth was significantly delayed in animals vaccinated using NP-CpG-B compared with free CpG-B, both in terms of tumor volume (B) and proportion of tumor-free animals (C). (D) 13 d posttumor inoculation mice vaccinated with NP-CpG-B had significantly more SIINFEKL-MHC-I-specific CD8+ T cells than mice in the other groups. This advantage was lost on day 21. Growth curves (n = 8) represent mean ± SEM, and SIINFEKL-MHC-I-specific T cells curves (n = 4) represent median ± 95% confidence interval; tumor-free curves represent Kaplan-Meier survival-curves (n = 8). Statistics are between NP-OVA + CpG-B and NP-OVA + NP-CpG-B. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.
0.05
naïve
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
**
25
**
20
10 ns
5 naïve
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
400
*
*
300 200 100
ns
IFN-γ 0
sequence SIINFEKL being the MHC-I epitope of OVA) CD8+ T cells at day 13 after tumor inoculation. This difference had diminished by 21 d (Fig. 3D). Conjugation of CpG-B and CpG-C to NPs Enhances Recall of Memory CD8+ T Cells. One of the features expected by a prophylactic
15
0
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
IFN-γ (pg/ml)
IFN-γ+ (% of CD3ε+CD8+)
5
PBS + NP-OVA
CD107a
D
*
10
0
E
*
15
**
0.10
0.00
C
***
14
**
Effector CD8+ T Cells Induced By NP-Conjugated CpG-B Are Able to Efficiently Kill Target Tumor Cells. In addition to inducing effector
A day 0
*** *** ***
CD4+ T cells capable of secreting both IFN-γ+ and TNF-α+ (Fig. 2B), reflecting potent CTL help (29). In accordance with this, only NP-CpG-B could significantly induce cytotoxic (IFN-γ+) and degranulating (CD107a+) CD8+ T-cell responses (Fig. 2 C and D), including 5.6 ± 1.6% of CD8+ T cells that were double-positive compared with 2.1 ± 0.3% induced by free CpG-B (Fig. 2E). In contrast, free CpG-B induced weaker but detectable effector CD8+ T cells, whereas the same doses of CpG-A, CpG-C, or NP-CpG-C could not induce CTL responses in vivo. Moreover, the total amount of IFN-γ secretion on splenocyte restimulation was increased 6.5-fold in mice vaccinated with NP-CpGB compared with about a twofold increase in mice vaccinated with free CpG-B (Fig. 2F).
naïve
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
Fig. 2. CpG-B conjugation to NPs enhances antigen-specific Th1 and CD8+ Tcell responses. (A) Immunization schedule; 10 μg OVA conjugated to NPs was injected with 4 μg free or NP-conjugated CpG in the footpads of C57BL/6J mice on days 0 and 14. Mice were killed on day 19. Control mice were injected at the same time with the same dose of NP-OVA only. (B) Stimulation of splenocytes ex vivo for 6 h in the presence of OVA shows enhanced polyfunctional response in CD4+ T cells after vaccination using NP-conjugated CpG-B compared with free CpG-B and using NP-conjugated CpG-B compared with NP-conjugated CpG-C. (C and D) Stimulation of splenocytes ex vivo for 6 h in the presence of the MHC-I peptide SIINFEKL shows an enhanced CD8+ T-cell activation profile after vaccination using NP-conjugated CpG-B compared with free CpG-B, and with NP-CpG-B compared with NP-CpG-C, as indicated IFN-γ expression (C) and the presence of the degranulation marker CD107a (D). (E) Representative flow cytometry plots of dual CD107,a surface accumulation, and IFN-γ content by CD8+ T cells. Numbers represent percentages of cells per quadrants. (F) Assays of IFN-γ production by ELISA after stimulation with SIINFEKL confirm the flow cytometry results. Box plots represent median ± 95% confidence interval (n = 4–8 mice/group); *P < 0.05; **P < 0.01; ***P < 0.001.
19904 | www.pnas.org/cgi/doi/10.1073/pnas.1313152110
cellular immunity-inducing vaccine is induction of memory T cells that can be clonally expanded on recall (30). We first primed and boosted mice with our formulations to generate a pool of functional memory CTLs (14). Two months later, we evaluated the activation of these memory CTLs in the spleen (Fig. 4 A–C) and found that both NP-CpG-C and NP-CpG-B were equally capable of expanding and inducing cytotoxicity in memory CD8+ T cells, more so than their free CpG counterparts. Free CpG-C induced fourfold more IFN-γ+ CD107a+ CD8+ T cells than did NP-OVA alone, whereas for free CpG-A and CpG-B, the percentage of double-positive cells was equal or double, respectively (Fig. 4B). However, this CD8+ memory recall was more than doubled by NP coupling of either CpG-B or CpG-C (Fig. 4B), correlating with increased amounts of IFN-γ secreted on restimulation (Fig. 4 C and D). In addition to systemic activation in the spleen, NP-CpG-B also drove higher IFN-γ secretion locally in the skin-draining LN on recall compared with free CpG-B after ex vivo restimulation (Fig. 4 E–G). Memory Recall by NP-CpG-B and NP-CpG-C Delays Tumor Onset and Confers Enhanced Protection to Tumor Challenges. To determine
whether NP-coupled CpG was more effective at recalling memory than free CpG, we challenged mice with E.G7-OVA or the more aggressive B16-F10-OVA tumors 4–6 mo after vaccination with free CpG-B and NP-OVA and 5 d after memory cells were recalled with various formulations (Fig. 5A). In E.G7-OVAbearing mice, all recall formulations except free CpG-C drove complete protection against tumor challenge (Fig. 5B), even though similarly high numbers of SIINFEKL-MHC-I pentamer+specific CD8+ T cells were detected systemically with all formulations (Fig. 5C). In the more aggressive B16-F10 model, the de Titta et al.
day 0
A
84
14
immunization
boost
recall
NP-OVA CpG-B (free)
NP-OVA CpG-B (free)
NP-OVA CpG variants
spleen
IFN-γ (ng/ml)
+ NP-OVA 25
**
20 15 10 5
LN
MFI IFN-γ (CD107a+IFN-γ+CD3ε+CD8+)
0
F
LN
E
*
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
1000
500
CpG-A (free) CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS naïve
0 0.1
1 CD107a+IFN-γ+ (% of CD3ε+CD8+)
10
600 400 200
0 0.05 0.1
*
10
*
4 2
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C
LN
+ NP-OVA 100
ns
ns
80 60 40 20 0
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C
advantage of NP-CpG in recalling memory 6 mo after vaccination (Fig. 5A) was more evident in delaying tumor onset and improving survival (Fig. 5D). Tumor growth in mice that received NP-OVA only for memory recall was not delayed compared with naive controls and the no-memory recall, whereas tumor growth was slightly impaired in mice receiving NP-OVA together with CpG-B, CpG-C, or NP-CpG-C. Notably, NP-CpG-B was the only adjuvant formulation capable of killing B16-F10-OVA tumors and prolonging survival (Fig. 5E). Discussion The development of potent, yet safe, subunit vaccines that promote effector as well as memory CD8+ and Th1 responses remains of great importance. At this time, one of the main foci of vaccine research is the use of novel adjuvant formulations based on TLR agonists to enhance adaptive immunity through the activation of DCs (31, 32). CpG oligonucleotides have been shown to enhance Th1 immune responses, and formulations containing them are under investigation as adjuvants for vaccines against certain types of cancer, as well as infectious diseases (19). The different CpG types are used in clinical trials with different goals, with CpG-C being mostly used for prophylactic vaccines, CpG-A for treating certain diseases such as asthma, and CpG-B used in both therapeutic and prophylactic applications (19). In this study, we describe the enhancement of adjuvanticity of CpG-B and CpG-C sequences through conjugation, in a reducible manner, to ultrasmall Pluronic-stabilized PPS NPs. Conjugation of CpG-B to NPs enhanced activation of DCs and promoted a cytolytic phenotype of effector and memory CD8+ T cells, resulting in protection against syngeneic tumor challenges. In contrast, NP-CpG-C induced de Titta et al.
1 CD107a+IFN-γ+ (% of CD3ε+CD8+)
6
0
G
0
800
CpG-A (free) CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS naïve
8
CD107a+IFN-γ+ (% of CD3ε+CD8+)
spleen
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C
1000
+ NP-OVA
Fig. 4. CpG-B and CpG-C conjugation to NPs enhances both local and systemic CD8+ T-cell memory recall. (A) Immunization schedule; C57BL/ 6J mice were immunized with 4 μg CpG-B and 10 μg OVA conjugated to NPs on days 0 and 14. Two months later, memory was recalled with 4 μg of the different CpG formulations and 10 μg NPconjugated OVA. Five days after recall, mice were killed, and splenocytes (B–D) and LN cells (E–G) restimulated ex vivo for 6 h with the MHC-I immunodominant peptide SIINFEKL showed enhanced cytotoxic T-cell recall responses induced by NP-conjugated CpG-B or CpG-C compared with free CpG-B or CpG-C, respectively, as indicated by presence of the degranulation marker CD107a (B and E) or IFN-γ production, determined by flow cytometry (C and F) or ELISA (D and G). NPconjugated CpG-B and -C induced comparable cytotoxic T-cell recall responses. Box plots represent median ± 95% confidence interval (n = 5–8 mice/ group). *P < 0.05; **P < 0.01; ns, not significant.
maturation of DCs, but not an elevated secretion of IL-12p70 and a weakened promotion of early effector CD8+ T cells. However, NP-CpG-C was still a potent inducer of clonal expansion of memory CD8+ T cells. As described earlier, our NP-CpG-B formulation induced enhanced maturation and up-regulation of costimulatory molecules on the surface of DCs in vitro and in vivo. Importantly, upregulation of CD80 and CD86 was observed on LN CD11b− DCs, which are known to preferentially cross-present antigens and induce Th1 responses (28) (Fig. 1 A–C, E, and F). At the same time, we could detect the secretion of IL-12p70 in vitro by DCs on stimulation, and for CpG-B, it was enhanced by conjugation to NPs, suggesting a higher potential to stimulate Th1 and CD8+ T cells (Fig. 1D). Moreover, we observed that the proportion of adjuvant-positive DCs that were also positive for the antigen was significantly enhanced by NP-CpG-B (Fig. 1G). Together, these results indicate that CpG-B conjugation to NPs led to a higher frequency of activated cross-presenting DCs in the draining LNs, as well as a higher fraction of DCs able to present OVA in a Th1 and CD8+ T-cell-promoting context, for a given dose compared with free CpG-B. Notably, it was previously shown that modifications and conjugation of peptides to the 5′ end of CpG had a tendency to decrease their immunostimulatory effect (33, 34). Our 5′ end conjugation strategy between NP and CpG, which allows the latter to be released in the reductive endosomal environment, did not negatively influence the activity of the bound CpG, which were even more stimulatory when conjugated than their free counterparts. Free NPs have a very moderate adjuvant capacity on their own, as they only trigger a slight up-regulation of CD86 on the surface of DCs in vitro and no secretion of the proinflammatory cytokines IL-12p70, PNAS | December 3, 2013 | vol. 110 | no. 49 | 19905
IMMUNOLOGY
5
MFI IFN-γ (CD107a+IFN-γ+CD3ε+CD8+)
10
0
D
*
+ NP-OVA
CD107a+IFN-γ+ (% of CD3ε+CD8+)
*
C
+ NP-OVA
spleen 15
IFN-γ (ng/ml)
B
89
t
t+5
boost
recall
tumor
NP-OVA CpG-B (free)
NP-OVA CpG-B (free)
6
C
80 60 CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS naïve
20 0 0
20
40
ns
30
SIINFEKL-MHCI specific (% of CD3ε+CD8+B220-)
100
40
t = 4 months for E.G7-OVA t = 6 months for B16-F10-OVA
10 E.G7-OVA NP-OVA 5 CpG variants 5x10 B16-F10-OVA
+ NP-OVA
E.G7-OVA tumor free mice (%)
B
14
immunization
ns 20
10
0
naïve PBS
60
CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS no recall naïve
80 60
**
40
** ns
20 0
0
20
40
60
days after tumor inoculation
80
CpG-B NPCpG-C NP(free) CpG-B (free) CpG-C + NP-OVA
100
B16-F10-OVA survival (%)
E
100 + NP-OVA
B16-F10-OVA tumor free mice (%)
days after tumor inoculation
D
80 60 ns
*
40
CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS no recall naïve
ns 20 0
0
20
40
60
days after tumor inoculation
IL-6, and IL-1β (13). Moreover, by conjugating CpG to NPs, we could use a low dose of CpG to potently activate DCs in the LNs without substantially raising blood levels of the inflammatory cytokines TNF-α, IL-12p70, and IL-6 (Fig. S2); thus, the low dose of CpG used on the NP appears to limit toxicity that would be associated with higher doses of free CpG (9, 35). Vaccination with NP-CpG-B and NP-OVA increased the proportion of CD8+ T cells capable of degranulating and secreting IFN-γ compared with vaccination with soluble CpG-B and NP-OVA. Frequencies of CD4+ T cells secreting both IFN-γ and TNF-α were also increased, which could have enhanced the cytotoxic potential of the CD8+ T cells (29). Interestingly, at the adjuvant dose used, NP-CpG-B with NP-OVA was the only formulation inducing both CD8+ and CD4+ T-cell responses. CpG-B administered free induced weaker, but detectable, effector CD8+ T cells, whereas neither CpG-A nor CpG-C, either free or conjugated to NP, were able to induce effector T-cell responses in vivo (Fig. 2). In this study, we could be at a low-dose limit, where CpG-B, but not CpG-A or CpG-C, benefits from the coinjection of NP-OVA, and even more from the conjugation to NP. Interestingly, NP conjugation allowed low doses (4 μg) of CpG-B to drive potent CTL responses comparable to those elicited by 10–100 μg free CpG-B observed in other studies (4, 5). In addition to the induction of effector phenotypes in CD8+ T cells, a successful vaccine has to induce functional CTLs that recognize and kill their antigen-bearing target cell. We show here that the response induced by NP-CpG-B with NP-OVA formulation is effective at delaying the growth of E.G7-OVA cells, thus suggesting the killing of specific tumor cells (Fig. 3 B and C). On day 21, the levels of antigen-specific CD8+ T cells had diminished, likely under the selective pressure of vaccination as a result of antigenic selection, as well as natural elimination of the OVA plasmid (5, 36), and thus the tumors started to grow. Together, these results indicate that CD8+ T cells elicited by NP-CpG-B with NP-OVA displayed an enhanced cytotoxic potential. As a consequence, mice could resist a syngeneic tumor challenge longer. Heterologous prime-boost regimens have been shown to induce long-lived CD8+ T effector memory cells, which are critical for protective immunity, although the mechanism remains largely unknown (37). Moreover, CpG-A and CpG-B classes have been suggested to exert different functions on naive versus memory 19906 | www.pnas.org/cgi/doi/10.1073/pnas.1313152110
ns
+ NP-OVA
day 0
A
80
Fig. 5. Cytotoxic effect on tumor cells upon memory CD8+ T-cell recall of conjugation of CpG-B and CpG-C to NPs. (A) Immunization schedule, where 10 μg of OVA conjugated to NPs was injected intradermally with 4 μg free CpG-B in the footpads of C57BL/6J mice on days 0 and 14. Four or six months later (for E.G7-OVA or B16-F10-OVA, respectively), Tcell memory was recalled with 4 μg CpG in various formulations and 10 μg OVA conjugated to NPs. Five days after memory recall, 106 E.G7-OVA lymphoma tumor cells or 5 × 105 B16-F10-OVA melanoma cells were injected s.c. into the back. (B) All E.G7-OVAchallenged mice were tumor-free when memory had been recalled using NP-CpG-B, NP-CpG-C, or CpG-B but not with free CpG-C or CpG-A. (C) NPCpG-B induced more SIINFEKL-MHC-I-specific CD8+ T cells, even 44 d after tumor inoculation; shown are median ± 95% confidence interval (n = 4). (D and E) B16-F10-OVA tumor growth was significantly delayed in animals receiving NP-CpG-B compared with naive and memory controls, both in terms of (D) tumor onset and (E) survival. SIINFEKL-MHC-I-specific T cells curves (n = 4 mice) represent median ± 95% confidence interval, and tumor-free curves and survival curves represent Kaplan-Meier survival-curves (n = 5). Statistics are between indicated groups and naive control. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.
CD8+ T cells in vitro (38). To evaluate whether our formulations were inducing different phenotypes of memory CD8+ T cells in vivo, we first primed and boosted mice with NP-OVA with CpGB to generate a pool of functional memory CTLs, as previously shown in our laboratory (14). At least 2 mo later, we then evaluated the activation and phenotype of these memory CTLs by the different formulations when coinjected with NP-OVA. Our results show that on memory recall, both NP-CpG-C and NPCpG-B were equally capable of expanding and inducing cytotoxicity in memory CD8+ T cells, but more so than their free CpG counterparts. This enhancement was observable not only in terms of percentage of cells that were degranulating, secreting IFN-γ, or both (Fig. 4 B, D, E, and G), but also in terms of the intensity of the secretion of IFN-γ (Fig. 4 C and F). This trend was observed both systemically in the spleen (Fig. 4 B–D) and locally in the draining LN (Fig. 4 E and F). The intensity of IFNγ secretion has been associated with multifunctional capacity of CTLs and correlated with enhanced effector functions, as these cells are able to act on infected target cells by secreting a combination of effector cytokines or acting directly via ligands (39). Considering that pathogens can either create local or systemic infections, the observation that NP-CpG-B enhanced both the frequency of effector memory CD8+ T cells as well as their capacity to display a potent cytotoxic phenotype on memory recall, both locally in the draining LN and systemically, suggests that our NP-conjugated formulation might have enhanced protective potential compared with NP-OVA with free CpG. To demonstrate the cytotoxic phenotype of the CD8+ T cells induced by our memory-recalling formulations, a vaccination regimen that would be relevant in the context of seasonal virus infections, we first challenged mice with 106 E.G7-OVA thymoma cells (Fig. 5 B and C). However, this experimental set-up was not suited to answer our question, as most of the mice, by having been immunized twice with a strong prime-boost regime of NP-OVA with CpG-B, remained tumor-free. We then challenged another set of mice, using a similar schedule, with a more aggressive tumor model (B16-F10-OVA melanoma cells). Tumor growth was mainly dampened in mice whose memory was recalled with NP-CpG-B compared with control, nontreated mice. Strikingly, NP-CpG-C was not as effective in this model and produced an outcome similar to that of free CpG-B and de Titta et al.
1. Swartz MA, Hirosue S, Hubbell JA (2012) Engineering approaches to immunotherapy. Sci Transl Med 4(148):rv9. 2. Moon JJ, Huang B, Irvine DJ (2012) Engineering nano- and microparticles to tune immunity. Adv Mater 24(28):3724–3746. 3. Li WA, Mooney DJ (2013) Materials based tumor immunotherapy vaccines. Curr Opin Immunol 25(2):238–245. 4. Bourquin C, et al. (2008) Targeting CpG oligonucleotides to the lymph node by nanoparticles elicits efficient antitumoral immunity. J Immunol 181(5):2990–2998. 5. Beaudette TT, et al. (2009) In vivo studies on the effect of co-encapsulation of CpG DNA and antigen in acid-degradable microparticle vaccines. Mol Pharm 6(4): 1160–1169. 6. Li AV, et al. (2013) Generation of effector memory T cell-based mucosal and systemic immunity with pulmonary nanoparticle vaccination. Sci Transl Med 5(204):ra130. 7. Hafner AM, Corthésy B, Merkle HP (2013) Particulate formulations for the delivery of poly(I:C) as vaccine adjuvant. Adv Drug Deliv Rev 65(10):1386–1399. 8. Leleux J, Roy K (2013) Micro and nanoparticle-based delivery systems for vaccine immunotherapy: An immunological and materials perspective. Adv Healthc Mater 2(1):72–94. 9. Rozenfeld JHK, Silva SR, Ranéia PA, Faquim-Mauro E, Carmona-Ribeiro AM (2012) Stable assemblies of cationic bilayer fragments and CpG oligonucleotide with enhanced immunoadjuvant activity in vivo. J Control Release 160(2):367–373. 10. Alving CR, Rao M (2008) Lipid A and liposomes containing lipid A as antigens and adjuvants. Vaccine 26(24):3036–3045. 11. Andrews CD, Provoda CJ, Ott G, Lee K-D (2011) Conjugation of lipid and CpG-containing oligonucleotide yields an efficient method for liposome incorporation. Bioconjug Chem 22(7):1279–1286. 12. Storni T, et al. (2004) Nonmethylated CG motifs packaged into virus-like particles induce protective cytotoxic T cell responses in the absence of systemic side effects. J Immunol 172(3):1777–1785. 13. Ballester M, et al. (2011) Nanoparticle conjugation and pulmonary delivery enhance the protective efficacy of Ag85B and CpG against tuberculosis. Vaccine 29(40): 6959–6966. 14. Nembrini C, et al. (2011) Nanoparticle conjugation of antigen enhances cytotoxic T-cell responses in pulmonary vaccination. Proc Natl Acad Sci USA 108(44):E989–E997. 15. Hirosue S, Kourtis IC, van der Vlies AJ, Hubbell JA, Swartz MA (2010) Antigen delivery to dendritic cells by poly(propylene sulfide) nanoparticles with disulfide conjugated peptides: Cross-presentation and T cell activation. Vaccine 28(50):7897–7906. 16. van der Vlies AJ, O’Neil CP, Hasegawa U, Hammond N, Hubbell JA (2010) Synthesis of pyridyl disulfide-functionalized nanoparticles for conjugating thiol-containing small molecules, peptides, and proteins. Bioconjug Chem 21(4):653–662. 17. Kourtis IC, et al. (2013) Peripherally administered nanoparticles target monocytic myeloid cells, secondary lymphoid organs and tumors in mice. PLoS ONE 8(4):e61646. 18. Foster S, Duvall CL, Crownover EF, Hoffman AS, Stayton PS (2010) Intracellular delivery of a protein antigen with an endosomal-releasing polymer enhances CD8 T-cell production and prophylactic vaccine efficacy. Bioconjug Chem 21(12):2205–2212. 19. Krieg AM (2012) CpG still rocks! Update on an accidental drug. Nucleic Acid Ther 22(2):77–89. 20. Marschner A, et al. (2005) CpG ODN enhance antigen-specific NKT cell activation via plasmacytoid dendritic cells. Eur J Immunol 35(8):2347–2357. 21. Huber JP, Farrar JD (2011) Regulation of effector and memory T-cell functions by type I interferon. Immunology 132(4):466–474.
de Titta et al.
cells; thus, other mechanisms that might be involved would require further investigation. Materials and Methods NPs were synthesized and functionalized with either antigen or CpG adjuvant conjugated via a reduction-sensitive link. Two murine tumor models were utilized, namely the E.G7-OVA model (a thymoma expressing OVA as a model tumor antigen) and the B16-F10 model (a melanoma expressing OVA). The details of materials and methods are described in SI Materials and Methods. ACKNOWLEDGMENTS. We thank K. Y. Dane, A. Stano, P. S. Briquez, K. Ciapala, A. W. Lund, E. Simeoni, D. K. Bonner, S. Hirosue, S. N. Thomas, X. Quaglia, and P. Corthésy-Henrioud for technical assistance and useful discussions. This work was supported by the European Research Council (NanoImmune), the Bill and Melinda Gates Foundation, and the Swiss National Science Foundation.
22. Krug A, et al. (2001) Identification of CpG oligonucleotide sequences with high induction of IFN-alpha/beta in plasmacytoid dendritic cells. Eur J Immunol 31(7): 2154–2163. 23. Krieg AM (2002) CpG motifs in bacterial DNA and their immune effects. Annu Rev Immunol 20:709–760. 24. Vollmer J, et al. (2004) Characterization of three CpG oligodeoxynucleotide classes with distinct immunostimulatory activities. Eur J Immunol 34(1):251–262. 25. Gonzalez SF, et al. (2011) Trafficking of B cell antigen in lymph nodes. Annu Rev Immunol 29:215–233. 26. Reddy ST, et al. (2007) Exploiting lymphatic transport and complement activation in nanoparticle vaccines. Nat Biotechnol 25(10):1159–1164. 27. Kerkmann M, et al. (2005) Spontaneous formation of nucleic acid-based nanoparticles is responsible for high interferon-alpha induction by CpG-A in plasmacytoid dendritic cells. J Biol Chem 280(9):8086–8093. 28. Pulendran B, Tang H, Denning TL (2008) Division of labor, plasticity, and crosstalk between dendritic cell subsets. Curr Opin Immunol 20(1):61–67. 29. Agnellini P, et al. (2008) Kinetic and mechanistic requirements for helping CD8 T cells. J Immunol 180(3):1517–1525. 30. Sad S, Krishnan L (2003) Maintenance and attrition of T-cell memory. Crit Rev Immunol 23(1-2):129–147. 31. Schwarz K, et al. (2003) Role of Toll-like receptors in costimulating cytotoxic T cell responses. Eur J Immunol 33(6):1465–1470. 32. Duthie MS, Windish HP, Fox CB, Reed SG (2011) Use of defined TLR ligands as adjuvants within human vaccines. Immunol Rev 239(1):178–196. 33. Kandimalla ER, et al. (2002) Conjugation of ligands at the 5′-end of CpG DNA affects immunostimulatory activity. Bioconjug Chem 13(5):966–974. 34. Putta MR, et al. (2010) Peptide conjugation at the 5′-end of oligodeoxynucleotides abrogates toll-like receptor 9-mediated immune stimulatory activity. Bioconjug Chem 21(1):39–45. 35. Purswani MU, Eckert SJ, Arora HK, Noel GJ (2002) Effect of ciprofloxacin on lethal and sublethal challenge with endotoxin and on early cytokine responses in a murine in vivo model. J Antimicrob Chemother 50(1):51–58. 36. Breckpot K, et al. (2003) Lentivirally transduced dendritic cells as a tool for cancer immunotherapy. J Gene Med 5(8):654–667. 37. Vasconcelos JR, et al. (2012) Relevance of long-lived CD8(+) T effector memory cells for protective immunity elicited by heterologous prime-boost vaccination. Front Immunol 3:358. 38. Rothenfusser S, et al. (2004) CpG-A and CpG-B oligonucleotides differentially enhance human peptide-specific primary and memory CD8+ T-cell responses in vitro. Blood 103(6):2162–2169. 39. Seder RA, Darrah PA, Roederer M (2008) T-cell quality in memory and protection: Implications for vaccine design. Nat Rev Immunol 8(4):247–258. 40. Brincks EL, Katewa A, Kucaba TA, Griffith TS, Legge KL (2008) CD8 T cells utilize TRAIL to control influenza virus infection. J Immunol 181(7):4918–4925. 41. Zhang Y, et al. (2007) Th1 cell adjuvant therapy combined with tumor vaccination: A novel strategy for promoting CTL responses while avoiding the accumulation of Tregs. Int Immunol 19(2):151–161. 42. Hollenbaugh JA, Reome J, Dobrzanski M, Dutton RW (2004) The rate of the CD8dependent initial reduction in tumor volume is not limited by contact-dependent perforin, Fas ligand, or TNF-mediated cytolysis. J Immunol 173(3):1738–1743.
PNAS | December 3, 2013 | vol. 110 | no. 49 | 19907
IMMUNOLOGY
CpG-C (Fig. 5 D and E). Different mechanisms could explain the differences between memory recall with NP-CpG-B and NP-CpG-C, considering that the proportion of IFN-γ-secreting and degranulating CD8+ T cells induced was similar for both formulations. Cytotoxic T cells can act on infected cells by different mechanisms requiring either soluble factors such as cytokines, granzyme B, and perforin or cell-contact-dependent mechanisms such as TNF-related apoptosis-inducing ligand (TRAIL) and FasL (40). Depending on the pathological cell type or organ, certain responses are more desirable than others. E.G7-OVA cells are known to be sensitive to CD8+ T cells (41), and especially to their IFN-γ secretion (42), making our vaccine formulation more potent for complete eradication of these tumor cells. However, the exact mechanism leading to B16-F10-OVA melanoma cells elimination is less obvious. In our studies, we focused on the induction of cytotoxic CD8+ T
Edited* by Mark E. Davis, California Institute of Technology, Pasadena, CA, and approved October 24, 2013 (received for review July 12, 2013)
In subunit vaccines, strong CD8+ T-cell responses are desired, yet they are elusive at reasonable adjuvant doses. We show that targeting adjuvant to the lymph node (LN) via ultrasmall polymeric nanoparticles (NPs), which rapidly drain to the LN after intradermal injection, greatly enhances adjuvant efficacy at low doses. Coupling CpG-B or CpG-C oligonucleotides to NPs led to better dual-targeting of adjuvant and antigen (codelivered on separate NPs) in cross-presenting dendritic cells compared with free adjuvant. This led to enhanced dendritic cell maturation and T helper 1 (Th1)-cytokine secretion, in turn driving stronger effector CD8+ T-cell activation with enhanced cytolytic profiles and, importantly, more powerful memory recall. With only 4 μg CpG, NP-CpG-B could substantially protect mice from syngeneic tumor challenge, even after 4 mo of vaccination, compared with free CpG-B. Together, these results show that nanocarriers can enhance vaccine efficacy at a low adjuvant dose for inducing potent and long-lived cellular immunity.
M
ost current subunit vaccines base their success on driving durable antibody responses that prevent pathogen infections. However, therapeutic vaccines against cancer as well as chronic infections must elicit the activation of cytotoxic T lymphocytes (CTLs) to eliminate infected or malignant cells. Induction of long-lasting CTLs with subunit vaccines requires, first, delivery of antigen to cross-presenting dendritic cells (DCs) and, second, potent activation of these DCs by the vaccine adjuvant, which instructs DCs how to activate T cells. Nanocarriers can be used to modulate the immune response induced by antigens and adjuvants by modifying their characteristics, such as stability, tissue and cell targeting, and DC-activating capacity (1–12). We have previously described the targeting benefits of conjugating antigens to ultrasmall Pluronic-stabilized poly(propylene sulfide) (PPS) nanoparticles (NPs) in terms of CD8+ T-cell and Th1 responses (13, 14). The hydrophobic PPS core is hydrolytically stable, allowing wet storage, yet is oxidized to products that are soluble in vivo (15, 16). The Pluronic stabilizing agent [a block copolymer of poly(ethylene glycol) and poly(propylene glycol), produced by BASF] results in a corona of poly(ethylene glycol), which prevents aggregation before and after antigen or adjuvant conjugation. The ∼25–30-nm ultrasmall size allows efficient delivery to and accumulation within the draining lymph node (LN) (17). Finally, we have demonstrated the advantages of antigen conjugation to the NP surface via a disulfide bond, which can be reversibly cleaved within the reductive environment within the endosomal compartment of DCs, leading to more efficient crosspresentation than when antigen is irreversibly conjugated to the NP surface (15, 18). Here, we explore conjugation of adjuvant to the NP surface, focusing on CpG oligonucleotides, with the objective of delivering both antigen and adjuvant into the same LNresident DCs, with the concept that even if the antigen and adjuvant are on separate particles (a distinct formulation advantage), the two payloads are targeted to the same cells because their carriers have the same size and clearance characteristics. Strong adjuvants, such as Toll-like receptor (TLR) agonists, are required for subunit vaccine formulations to drive effective 19902–19907 | PNAS | December 3, 2013 | vol. 110 | no. 49
immune responses, as they instruct DCs how to activate cognate T cells. CpG, a TLR-9 agonist, has been shown both in preclinical and clinical studies to promote Th1-responses (19). Three main types of CpGs have been described: the multimeric CpG-A (or D-type) stimulates plasmacytoid DCs to secrete IFN-α; CpG-B (or K-type) induces B cells and DCs to mature and secrete cytokines such as IL-12p70, IL-6, and TNF-α; and CpG-C has the ability to induce both CpG-A and CpG-B types of responses. In turn, IFN-α secretion from plasmacytoid DCs promotes natural killer T cells (20), CD8+ T-cell activation and cytotoxicity (21), and maturation of conventional DCs. Secretion of IL-12p70 induces Th1 and CTL responses, IL-6 leads to maturation of B cells and induces Th17 responses, and TNF-α stimulates DC maturation and T-cell activation (22–24). In addition to these different activities, the three forms of CpG have different sizes, which could strongly affect their transport on interstitial injection (1) and distribution within the LN (25), which could account for some of their in vivo differences. We and others have previously shown that antigen size affects LN targeting and ultimate immune response after intradermal injection, which can be optimized by coupling peptides or proteins to ultrasmall (∼25–50 nm) polymeric NPs (26). CpGA naturally forms multimeric structures of ∼20–100 nm at physiological pH and temperature (27), but CpG-B and CpG-C do not. We hypothesize that conjugation of CpG-B and CpG-C to Significance High adjuvant doses are generally required to induce strong CD8+ T-cell immunity with subunit vaccines. Here we codeliver an antigen and an adjuvant coupled on separate ultrasmall polymeric nanoparticles. Because both payloads are attached to similarly sized nanoparticles, and as size is the principle determinant of nanoparticle drainage, this enhanced the dual uptake of antigen and adjuvant by cross-presenting dendritic cells resident in the draining lymph nodes. This cotargeting induced potent effector CD8+ T cells and a more powerful memory recall of these cytotoxic T cells compared with nanoparticle-conjugated antigen with free adjuvant. As such, nanoparticle conjugation enhanced the immunogenicity of adjuvants while maintaining a low dose, and thus limiting toxicity, affecting the design of future subunit vaccine formulations. Author contributions: A.d.T., M.B., C.N., M.A.S., and J.A.H. designed research; A.d.T., M.B., Z.J., and L.J. performed research; A.d.T., M.B., L.J., and A.J.v.d.V. contributed new reagents/analytic tools; A.d.T., M.B., and Z.J. analyzed data; M.A.S. and J.A.H. directed the project; and A.d.T., M.B., M.A.S., and J.A.H. wrote the paper. Conflict of interest statement: The Ecole Polytechnique Fédérale de Lausanne has filed for patent protection on the nanoparticle platform described herein, and some of the authors are named as inventors on those filings. *This Direct Submission article had a prearranged editor. Freely available online through the PNAS open access option. 1
A.d.T. and M.B. contributed equally to this work.
2
To whom correspondence may be addressed. E-mail: melody.swartz@epfl.ch or jeffrey.hubbell@epfl.ch.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1313152110/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1313152110
NPs could increase their immunogenic potential, while allowing a minimal dose, by targeting both the adjuvant and antigen to the same LN-resident DCs. We conjugated CpG-B and CpG-C to NPs via a reducible bond and designated the NP-conjugates as NP-CpG-B and NPCpG-C, respectively. CpG-A was not conjugated, as it naturally forms a multimeric structure of similar size as our NPs (27). We administered NP-CpG-B and NP-CpG-C in conjunction with ovalbumin-conjugated NPs (NP-OVA) intradermally in the footpads in mice. Adjuvant conjugated NPs enhanced uptake of antigen by DCs and activation of cross-presenting DCs and led to the induction of potent effector and memory antigen-specific T cells. By conjugating CpG-B to NPs, we induced a more potent cellular response at an early time, characterized by increased degranulation and secretion of IFN-γ by CD8+ T cells and enhanced proportion of multifunctional CD4+ T cells, even at a low dose of 4 μg. At a later time, conjugation of CpG-B and CpG-C enhanced the clonal expansion of antigen-specific memory CD8+ T cells and differentially enhanced their cytotoxic abilities. Mice immunized with either conjugated CpG variants were able to control an IFN-γ-sensitive tumor challenge (E.G7-OVA, a tyhmoma line), but NP-CpG-B was superior at dampening the growth of a more aggressive tumor (B16-F10-OVA, a melanoma line). Results Reducible Conjugation of 5′-End Modified CpG to NPs. We first modified CpG-B and CpG-C with a reactive thiophosphate group, allowing coupling to our previously developed pyridyl-cysteamine functionalized NPs (16). This reaction occurs through a disulfide exchange between the functionalized NP and the thiophosphate terminal group of CpG (Fig. S1A). The covalent bond can be
reduced, by mimicking the reducing environment of the endosomal compartment of the cell, with 2-mercaptoethanol (Fig. S1B). NP-CpG Induces Increased Activation of Dendritic Cells. To determine the direct activity of NP-conjugated CpG on DCs, we first evaluated our formulations in vitro. Although we observed increasing CD86 expression on DCs with increasing doses of each CpG, levels were enhanced by NP-coupled formulations (Fig. 1 A–C). Similarly, the proportion of mature DCs (CD86+ MHC-II+) was enhanced after stimulation with NP-conjugated compared with free CpG-B (Fig. 1B). The response to NP-CpG-B induced significantly higher levels of IL-12p70 secretion by DCs than did free CpG-B, whereas the response to NP-CpG-C was similar to that of free CpG-C (Fig. 1D). Enhancement of immune potentiation properties as a result of NP conjugation of CpG-B were also observed in vivo. Indeed, NPCpG-B coinjected with NP-OVA more potently activated crosspresenting (CD11c+MHC-II+CD11b−) DCs (28) in the draining LNs than did free CpG-B (Fig. 1 E and F). Importantly, at least one mechanism for this enhanced in vivo efficacy of NP-CpG was improved, cotargeting with NP-OVA to the same LN-resident DCs (Fig. 1G). Although we observed an increase in activation of DCs by our NP-conjugated formulation, this did not lead to an increase in blood levels of the inflammatory cytokine TNF-α, leading to only minor elevations in IL-12p70 and IL-6 (Fig. S2). Intradermal Immunization with CpG-B-Conjugated NPs Induces Increased Polyfunctional Th1 Cells and Cytotoxic Activity of CD8+ T Cells. Next, we evaluated T-cell responses 19 d after intradermal
vaccination (Fig. 2A). Interestingly, at the low (4 μg) adjuvant dose used, NP-CpG-B with NP-OVA was the only formulation that could induce significantly higher amounts of Th1-leaning
in vitro DC Fig. 1. CpG conjugation to NPs enhances crosspresenting dendritic cell activation. (A) RepresentaPBS CpG-B (free) NP-CpG-B PBS CpG-A (free) tive CD86 mean fluorescence intensity (MFI) histo15% 1.8% 3.1% + gram of CD11c DC activated with 0.5 μM of CpG-A PBS and free or NP-conjugated CpG-B and CpG-C for 6 h CpG-B (free) in vitro. (B) Representative flow cytometry plot of NP-CpG-B DCs activated for 6 h in the presence of 1 μM free or PBS NP-conjugated CpG-B. Numbers represent percentCpG-C (free) CD86 NP-CpG-C age of DCs (CD11c+MHC-II+) expressing both CD86 CD86 and MHC-II. (C) MFI of CD86 signal indicates DC activation after 6 h exposure in vitro to increasing ## ### 3 30 (0.005–1 μM) concentrations of CpG-A, CpG-B, or CpG-A (free) *** # CpG-B (free) ns CpG-C or NP-CpG-B or NP-CpG-C, showing stronger * NP-CpG-B activation by NP-CpG-B and NP-CpG-C than by free CpG-C (free) 2 20 ** CpG-B and CpG-C. Dashed bar shows PBS control. * NP-CpG-C * Symbol legend is in D. (D) DCs secreted higher ns amounts of IL-12p70 when stimulated for 12 h with 1 10 NP-CpG-B than with free CpG-B. (E and F) Mice were injected with 4 μg soluble or NP-conjugated CpG-B 0 0 and 10 μg OVA-conjugated NP in the four footpads. 0.01 0 0.1 0.5 1 1 0.01 0.1 LNs were harvested 20 h after injection. LN cells, concentration μM concentration μM + + gated for cross-presenting DCs (CD11c MHC-II CD11b−), were analyzed by flow cytometry for CD80 in vivo LN and CD86 expression. Measurements of MFI of CD86 *** 35 100 signal (E) and fraction of DCs that are activated as naïve 30 ** indicated by CD80 and CD86 coexpression (F) show 80 25 * stronger DC activation by NP-CpG-B than by free NP-OVA 60 CpG-B. (G) Mice were injected with 4 μg soluble or 20 NP-CpG-B fluorescently labeled with Dy633 (CpG15 CpG-B (free) 40 + NP-OVA B*) and 10 μg NP-conjugated OVA fluorescently la10 beled with Alexa488 (OVA”). Draining LNs were 20 NP-CpG-B 5 + NP-OVA harvested 20 h after injection. LN cells, gated for nd + + 0 0 DCs (CD11c MHC-II ), were analyzed by flow cytomnaïve CpG-B* NPnaïve PBS CpG-B NP0 5 10 15 (free) CpG-B* (free) CpG-B MFI CD86 etry for the OVA” and CpG-B*. Measurements show + + + NP-OVA + NP-OVA’’ (CD11c MHC-II CD11b ) a significantly higher fraction of CpG-B*+ DCs associated with OVA” when CpG was conjugated to NPs. Data in C, D, and E represent mean ± SEM (n = 3 independent experiments) for in vitro study. (F and G) Median ± 95% confidence interval (n = 8 mice, 2 independent experiments) for in vivo study. *P < 0.05; #P < 0.05; **P < 0.01; ##P < 0.01; ***P < 0.001; ###P < 0.001; ns, not significant; nd, not detectable. Asterisks indicate statistics between CpG-B and NP-CpG-B; pound signs indicate statistics between CpG-C and NP-CpG-C. MHC-II
D
F
***
de Titta et al.
CD80+CD86+ (% of CD11c+MHC-II+CD11b-)
E
G
OVA’’+ (% of CD11c+MHC-II+CpG*+)
MFI CD86 (CD11c+)
IL-12p70 (ng/ml)
C
PNAS | December 3, 2013 | vol. 110 | no. 49 | 19903
IMMUNOLOGY
B
A
phenotypes in CD8+ T cells, these cells must recognize and kill their antigen-bearing target cell to be effective. To determine the functionality of CTLs induced by our vaccine formulations, we challenged vaccinated mice with E.G7-OVA tumor cells (Fig. 3A). Consistent with the previous data, we found significantly delayed tumor onset in mice vaccinated with NP-CpG-B compared with free CpG-B (Fig. 3B) and longer tumor-free survival (Fig. 3C). This correlated with much higher levels of antigen-specific (SIINFEKL-MHC-I-pentamer+, the peptide
B day 0
14
19
boost
immunization NP-OVA CpG variants
IFN-γ+TNF-α+ (% of CD3ε+CD4+)
A NP-OVA CpG variants
0.15
naïve
1.04
CD107a+ (% of CD3ε+CD8+)
ns
ns
0
CpG-B (free) + NP-OVA
NP-CpG-B + NP-OVA
3.56
8.63
2.77
F 9.41
NP-OVA CpG variants
tumor
boost NP-OVA CpG variants
106 E.G7-OVA
naïve PBS + NP-OVA CpG-B (free) + NP-OVA NP-CpG-B + NP-OVA
D *
25 0 0
10
600 400 200
20
30
days after tumor inoculation
0
10
20
30
days after tumor inoculation
10
SIINFEKL-MHCI specific (% of CD3ε+CD8+B220-)
75 50
800
0
100
tumor free (%)
C
tumor volume (mm3)
immunization
B
19
8 6 4
**
2 0
ns 13
15
17
19
days after tumor inoculation
21
Fig. 3. CpG-B conjugation to NPs enhances protection against a tumor challenge. (A) Immunization schedule; 10 μg OVA conjugated to NPs was injected with 4 μg free or NP-conjugated CpG-B in the footpads of C57BL/6J mice on days 0 and 14. Control mice were injected at the same time with the same dose of NP-OVA only. On day 19, 106 E.G7-OVA lymphoma tumor cells were injected s.c. on the back. (B and C) Tumor growth was significantly delayed in animals vaccinated using NP-CpG-B compared with free CpG-B, both in terms of tumor volume (B) and proportion of tumor-free animals (C). (D) 13 d posttumor inoculation mice vaccinated with NP-CpG-B had significantly more SIINFEKL-MHC-I-specific CD8+ T cells than mice in the other groups. This advantage was lost on day 21. Growth curves (n = 8) represent mean ± SEM, and SIINFEKL-MHC-I-specific T cells curves (n = 4) represent median ± 95% confidence interval; tumor-free curves represent Kaplan-Meier survival-curves (n = 8). Statistics are between NP-OVA + CpG-B and NP-OVA + NP-CpG-B. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.
0.05
naïve
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
**
25
**
20
10 ns
5 naïve
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
400
*
*
300 200 100
ns
IFN-γ 0
sequence SIINFEKL being the MHC-I epitope of OVA) CD8+ T cells at day 13 after tumor inoculation. This difference had diminished by 21 d (Fig. 3D). Conjugation of CpG-B and CpG-C to NPs Enhances Recall of Memory CD8+ T Cells. One of the features expected by a prophylactic
15
0
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
IFN-γ (pg/ml)
IFN-γ+ (% of CD3ε+CD8+)
5
PBS + NP-OVA
CD107a
D
*
10
0
E
*
15
**
0.10
0.00
C
***
14
**
Effector CD8+ T Cells Induced By NP-Conjugated CpG-B Are Able to Efficiently Kill Target Tumor Cells. In addition to inducing effector
A day 0
*** *** ***
CD4+ T cells capable of secreting both IFN-γ+ and TNF-α+ (Fig. 2B), reflecting potent CTL help (29). In accordance with this, only NP-CpG-B could significantly induce cytotoxic (IFN-γ+) and degranulating (CD107a+) CD8+ T-cell responses (Fig. 2 C and D), including 5.6 ± 1.6% of CD8+ T cells that were double-positive compared with 2.1 ± 0.3% induced by free CpG-B (Fig. 2E). In contrast, free CpG-B induced weaker but detectable effector CD8+ T cells, whereas the same doses of CpG-A, CpG-C, or NP-CpG-C could not induce CTL responses in vivo. Moreover, the total amount of IFN-γ secretion on splenocyte restimulation was increased 6.5-fold in mice vaccinated with NP-CpGB compared with about a twofold increase in mice vaccinated with free CpG-B (Fig. 2F).
naïve
PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
Fig. 2. CpG-B conjugation to NPs enhances antigen-specific Th1 and CD8+ Tcell responses. (A) Immunization schedule; 10 μg OVA conjugated to NPs was injected with 4 μg free or NP-conjugated CpG in the footpads of C57BL/6J mice on days 0 and 14. Mice were killed on day 19. Control mice were injected at the same time with the same dose of NP-OVA only. (B) Stimulation of splenocytes ex vivo for 6 h in the presence of OVA shows enhanced polyfunctional response in CD4+ T cells after vaccination using NP-conjugated CpG-B compared with free CpG-B and using NP-conjugated CpG-B compared with NP-conjugated CpG-C. (C and D) Stimulation of splenocytes ex vivo for 6 h in the presence of the MHC-I peptide SIINFEKL shows an enhanced CD8+ T-cell activation profile after vaccination using NP-conjugated CpG-B compared with free CpG-B, and with NP-CpG-B compared with NP-CpG-C, as indicated IFN-γ expression (C) and the presence of the degranulation marker CD107a (D). (E) Representative flow cytometry plots of dual CD107,a surface accumulation, and IFN-γ content by CD8+ T cells. Numbers represent percentages of cells per quadrants. (F) Assays of IFN-γ production by ELISA after stimulation with SIINFEKL confirm the flow cytometry results. Box plots represent median ± 95% confidence interval (n = 4–8 mice/group); *P < 0.05; **P < 0.01; ***P < 0.001.
19904 | www.pnas.org/cgi/doi/10.1073/pnas.1313152110
cellular immunity-inducing vaccine is induction of memory T cells that can be clonally expanded on recall (30). We first primed and boosted mice with our formulations to generate a pool of functional memory CTLs (14). Two months later, we evaluated the activation of these memory CTLs in the spleen (Fig. 4 A–C) and found that both NP-CpG-C and NP-CpG-B were equally capable of expanding and inducing cytotoxicity in memory CD8+ T cells, more so than their free CpG counterparts. Free CpG-C induced fourfold more IFN-γ+ CD107a+ CD8+ T cells than did NP-OVA alone, whereas for free CpG-A and CpG-B, the percentage of double-positive cells was equal or double, respectively (Fig. 4B). However, this CD8+ memory recall was more than doubled by NP coupling of either CpG-B or CpG-C (Fig. 4B), correlating with increased amounts of IFN-γ secreted on restimulation (Fig. 4 C and D). In addition to systemic activation in the spleen, NP-CpG-B also drove higher IFN-γ secretion locally in the skin-draining LN on recall compared with free CpG-B after ex vivo restimulation (Fig. 4 E–G). Memory Recall by NP-CpG-B and NP-CpG-C Delays Tumor Onset and Confers Enhanced Protection to Tumor Challenges. To determine
whether NP-coupled CpG was more effective at recalling memory than free CpG, we challenged mice with E.G7-OVA or the more aggressive B16-F10-OVA tumors 4–6 mo after vaccination with free CpG-B and NP-OVA and 5 d after memory cells were recalled with various formulations (Fig. 5A). In E.G7-OVAbearing mice, all recall formulations except free CpG-C drove complete protection against tumor challenge (Fig. 5B), even though similarly high numbers of SIINFEKL-MHC-I pentamer+specific CD8+ T cells were detected systemically with all formulations (Fig. 5C). In the more aggressive B16-F10 model, the de Titta et al.
day 0
A
84
14
immunization
boost
recall
NP-OVA CpG-B (free)
NP-OVA CpG-B (free)
NP-OVA CpG variants
spleen
IFN-γ (ng/ml)
+ NP-OVA 25
**
20 15 10 5
LN
MFI IFN-γ (CD107a+IFN-γ+CD3ε+CD8+)
0
F
LN
E
*
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C + NP-OVA
1000
500
CpG-A (free) CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS naïve
0 0.1
1 CD107a+IFN-γ+ (% of CD3ε+CD8+)
10
600 400 200
0 0.05 0.1
*
10
*
4 2
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C
LN
+ NP-OVA 100
ns
ns
80 60 40 20 0
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C
advantage of NP-CpG in recalling memory 6 mo after vaccination (Fig. 5A) was more evident in delaying tumor onset and improving survival (Fig. 5D). Tumor growth in mice that received NP-OVA only for memory recall was not delayed compared with naive controls and the no-memory recall, whereas tumor growth was slightly impaired in mice receiving NP-OVA together with CpG-B, CpG-C, or NP-CpG-C. Notably, NP-CpG-B was the only adjuvant formulation capable of killing B16-F10-OVA tumors and prolonging survival (Fig. 5E). Discussion The development of potent, yet safe, subunit vaccines that promote effector as well as memory CD8+ and Th1 responses remains of great importance. At this time, one of the main foci of vaccine research is the use of novel adjuvant formulations based on TLR agonists to enhance adaptive immunity through the activation of DCs (31, 32). CpG oligonucleotides have been shown to enhance Th1 immune responses, and formulations containing them are under investigation as adjuvants for vaccines against certain types of cancer, as well as infectious diseases (19). The different CpG types are used in clinical trials with different goals, with CpG-C being mostly used for prophylactic vaccines, CpG-A for treating certain diseases such as asthma, and CpG-B used in both therapeutic and prophylactic applications (19). In this study, we describe the enhancement of adjuvanticity of CpG-B and CpG-C sequences through conjugation, in a reducible manner, to ultrasmall Pluronic-stabilized PPS NPs. Conjugation of CpG-B to NPs enhanced activation of DCs and promoted a cytolytic phenotype of effector and memory CD8+ T cells, resulting in protection against syngeneic tumor challenges. In contrast, NP-CpG-C induced de Titta et al.
1 CD107a+IFN-γ+ (% of CD3ε+CD8+)
6
0
G
0
800
CpG-A (free) CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS naïve
8
CD107a+IFN-γ+ (% of CD3ε+CD8+)
spleen
naïve PBS CpG-A CpG-B NP- CpG-C NP(free) (free) CpG-B (free) CpG-C
1000
+ NP-OVA
Fig. 4. CpG-B and CpG-C conjugation to NPs enhances both local and systemic CD8+ T-cell memory recall. (A) Immunization schedule; C57BL/ 6J mice were immunized with 4 μg CpG-B and 10 μg OVA conjugated to NPs on days 0 and 14. Two months later, memory was recalled with 4 μg of the different CpG formulations and 10 μg NPconjugated OVA. Five days after recall, mice were killed, and splenocytes (B–D) and LN cells (E–G) restimulated ex vivo for 6 h with the MHC-I immunodominant peptide SIINFEKL showed enhanced cytotoxic T-cell recall responses induced by NP-conjugated CpG-B or CpG-C compared with free CpG-B or CpG-C, respectively, as indicated by presence of the degranulation marker CD107a (B and E) or IFN-γ production, determined by flow cytometry (C and F) or ELISA (D and G). NPconjugated CpG-B and -C induced comparable cytotoxic T-cell recall responses. Box plots represent median ± 95% confidence interval (n = 5–8 mice/ group). *P < 0.05; **P < 0.01; ns, not significant.
maturation of DCs, but not an elevated secretion of IL-12p70 and a weakened promotion of early effector CD8+ T cells. However, NP-CpG-C was still a potent inducer of clonal expansion of memory CD8+ T cells. As described earlier, our NP-CpG-B formulation induced enhanced maturation and up-regulation of costimulatory molecules on the surface of DCs in vitro and in vivo. Importantly, upregulation of CD80 and CD86 was observed on LN CD11b− DCs, which are known to preferentially cross-present antigens and induce Th1 responses (28) (Fig. 1 A–C, E, and F). At the same time, we could detect the secretion of IL-12p70 in vitro by DCs on stimulation, and for CpG-B, it was enhanced by conjugation to NPs, suggesting a higher potential to stimulate Th1 and CD8+ T cells (Fig. 1D). Moreover, we observed that the proportion of adjuvant-positive DCs that were also positive for the antigen was significantly enhanced by NP-CpG-B (Fig. 1G). Together, these results indicate that CpG-B conjugation to NPs led to a higher frequency of activated cross-presenting DCs in the draining LNs, as well as a higher fraction of DCs able to present OVA in a Th1 and CD8+ T-cell-promoting context, for a given dose compared with free CpG-B. Notably, it was previously shown that modifications and conjugation of peptides to the 5′ end of CpG had a tendency to decrease their immunostimulatory effect (33, 34). Our 5′ end conjugation strategy between NP and CpG, which allows the latter to be released in the reductive endosomal environment, did not negatively influence the activity of the bound CpG, which were even more stimulatory when conjugated than their free counterparts. Free NPs have a very moderate adjuvant capacity on their own, as they only trigger a slight up-regulation of CD86 on the surface of DCs in vitro and no secretion of the proinflammatory cytokines IL-12p70, PNAS | December 3, 2013 | vol. 110 | no. 49 | 19905
IMMUNOLOGY
5
MFI IFN-γ (CD107a+IFN-γ+CD3ε+CD8+)
10
0
D
*
+ NP-OVA
CD107a+IFN-γ+ (% of CD3ε+CD8+)
*
C
+ NP-OVA
spleen 15
IFN-γ (ng/ml)
B
89
t
t+5
boost
recall
tumor
NP-OVA CpG-B (free)
NP-OVA CpG-B (free)
6
C
80 60 CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS naïve
20 0 0
20
40
ns
30
SIINFEKL-MHCI specific (% of CD3ε+CD8+B220-)
100
40
t = 4 months for E.G7-OVA t = 6 months for B16-F10-OVA
10 E.G7-OVA NP-OVA 5 CpG variants 5x10 B16-F10-OVA
+ NP-OVA
E.G7-OVA tumor free mice (%)
B
14
immunization
ns 20
10
0
naïve PBS
60
CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS no recall naïve
80 60
**
40
** ns
20 0
0
20
40
60
days after tumor inoculation
80
CpG-B NPCpG-C NP(free) CpG-B (free) CpG-C + NP-OVA
100
B16-F10-OVA survival (%)
E
100 + NP-OVA
B16-F10-OVA tumor free mice (%)
days after tumor inoculation
D
80 60 ns
*
40
CpG-B (free) NP-CpG-B CpG-C (free) NP-CpG-C PBS no recall naïve
ns 20 0
0
20
40
60
days after tumor inoculation
IL-6, and IL-1β (13). Moreover, by conjugating CpG to NPs, we could use a low dose of CpG to potently activate DCs in the LNs without substantially raising blood levels of the inflammatory cytokines TNF-α, IL-12p70, and IL-6 (Fig. S2); thus, the low dose of CpG used on the NP appears to limit toxicity that would be associated with higher doses of free CpG (9, 35). Vaccination with NP-CpG-B and NP-OVA increased the proportion of CD8+ T cells capable of degranulating and secreting IFN-γ compared with vaccination with soluble CpG-B and NP-OVA. Frequencies of CD4+ T cells secreting both IFN-γ and TNF-α were also increased, which could have enhanced the cytotoxic potential of the CD8+ T cells (29). Interestingly, at the adjuvant dose used, NP-CpG-B with NP-OVA was the only formulation inducing both CD8+ and CD4+ T-cell responses. CpG-B administered free induced weaker, but detectable, effector CD8+ T cells, whereas neither CpG-A nor CpG-C, either free or conjugated to NP, were able to induce effector T-cell responses in vivo (Fig. 2). In this study, we could be at a low-dose limit, where CpG-B, but not CpG-A or CpG-C, benefits from the coinjection of NP-OVA, and even more from the conjugation to NP. Interestingly, NP conjugation allowed low doses (4 μg) of CpG-B to drive potent CTL responses comparable to those elicited by 10–100 μg free CpG-B observed in other studies (4, 5). In addition to the induction of effector phenotypes in CD8+ T cells, a successful vaccine has to induce functional CTLs that recognize and kill their antigen-bearing target cell. We show here that the response induced by NP-CpG-B with NP-OVA formulation is effective at delaying the growth of E.G7-OVA cells, thus suggesting the killing of specific tumor cells (Fig. 3 B and C). On day 21, the levels of antigen-specific CD8+ T cells had diminished, likely under the selective pressure of vaccination as a result of antigenic selection, as well as natural elimination of the OVA plasmid (5, 36), and thus the tumors started to grow. Together, these results indicate that CD8+ T cells elicited by NP-CpG-B with NP-OVA displayed an enhanced cytotoxic potential. As a consequence, mice could resist a syngeneic tumor challenge longer. Heterologous prime-boost regimens have been shown to induce long-lived CD8+ T effector memory cells, which are critical for protective immunity, although the mechanism remains largely unknown (37). Moreover, CpG-A and CpG-B classes have been suggested to exert different functions on naive versus memory 19906 | www.pnas.org/cgi/doi/10.1073/pnas.1313152110
ns
+ NP-OVA
day 0
A
80
Fig. 5. Cytotoxic effect on tumor cells upon memory CD8+ T-cell recall of conjugation of CpG-B and CpG-C to NPs. (A) Immunization schedule, where 10 μg of OVA conjugated to NPs was injected intradermally with 4 μg free CpG-B in the footpads of C57BL/6J mice on days 0 and 14. Four or six months later (for E.G7-OVA or B16-F10-OVA, respectively), Tcell memory was recalled with 4 μg CpG in various formulations and 10 μg OVA conjugated to NPs. Five days after memory recall, 106 E.G7-OVA lymphoma tumor cells or 5 × 105 B16-F10-OVA melanoma cells were injected s.c. into the back. (B) All E.G7-OVAchallenged mice were tumor-free when memory had been recalled using NP-CpG-B, NP-CpG-C, or CpG-B but not with free CpG-C or CpG-A. (C) NPCpG-B induced more SIINFEKL-MHC-I-specific CD8+ T cells, even 44 d after tumor inoculation; shown are median ± 95% confidence interval (n = 4). (D and E) B16-F10-OVA tumor growth was significantly delayed in animals receiving NP-CpG-B compared with naive and memory controls, both in terms of (D) tumor onset and (E) survival. SIINFEKL-MHC-I-specific T cells curves (n = 4 mice) represent median ± 95% confidence interval, and tumor-free curves and survival curves represent Kaplan-Meier survival-curves (n = 5). Statistics are between indicated groups and naive control. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.
CD8+ T cells in vitro (38). To evaluate whether our formulations were inducing different phenotypes of memory CD8+ T cells in vivo, we first primed and boosted mice with NP-OVA with CpGB to generate a pool of functional memory CTLs, as previously shown in our laboratory (14). At least 2 mo later, we then evaluated the activation and phenotype of these memory CTLs by the different formulations when coinjected with NP-OVA. Our results show that on memory recall, both NP-CpG-C and NPCpG-B were equally capable of expanding and inducing cytotoxicity in memory CD8+ T cells, but more so than their free CpG counterparts. This enhancement was observable not only in terms of percentage of cells that were degranulating, secreting IFN-γ, or both (Fig. 4 B, D, E, and G), but also in terms of the intensity of the secretion of IFN-γ (Fig. 4 C and F). This trend was observed both systemically in the spleen (Fig. 4 B–D) and locally in the draining LN (Fig. 4 E and F). The intensity of IFNγ secretion has been associated with multifunctional capacity of CTLs and correlated with enhanced effector functions, as these cells are able to act on infected target cells by secreting a combination of effector cytokines or acting directly via ligands (39). Considering that pathogens can either create local or systemic infections, the observation that NP-CpG-B enhanced both the frequency of effector memory CD8+ T cells as well as their capacity to display a potent cytotoxic phenotype on memory recall, both locally in the draining LN and systemically, suggests that our NP-conjugated formulation might have enhanced protective potential compared with NP-OVA with free CpG. To demonstrate the cytotoxic phenotype of the CD8+ T cells induced by our memory-recalling formulations, a vaccination regimen that would be relevant in the context of seasonal virus infections, we first challenged mice with 106 E.G7-OVA thymoma cells (Fig. 5 B and C). However, this experimental set-up was not suited to answer our question, as most of the mice, by having been immunized twice with a strong prime-boost regime of NP-OVA with CpG-B, remained tumor-free. We then challenged another set of mice, using a similar schedule, with a more aggressive tumor model (B16-F10-OVA melanoma cells). Tumor growth was mainly dampened in mice whose memory was recalled with NP-CpG-B compared with control, nontreated mice. Strikingly, NP-CpG-C was not as effective in this model and produced an outcome similar to that of free CpG-B and de Titta et al.
1. Swartz MA, Hirosue S, Hubbell JA (2012) Engineering approaches to immunotherapy. Sci Transl Med 4(148):rv9. 2. Moon JJ, Huang B, Irvine DJ (2012) Engineering nano- and microparticles to tune immunity. Adv Mater 24(28):3724–3746. 3. Li WA, Mooney DJ (2013) Materials based tumor immunotherapy vaccines. Curr Opin Immunol 25(2):238–245. 4. Bourquin C, et al. (2008) Targeting CpG oligonucleotides to the lymph node by nanoparticles elicits efficient antitumoral immunity. J Immunol 181(5):2990–2998. 5. Beaudette TT, et al. (2009) In vivo studies on the effect of co-encapsulation of CpG DNA and antigen in acid-degradable microparticle vaccines. Mol Pharm 6(4): 1160–1169. 6. Li AV, et al. (2013) Generation of effector memory T cell-based mucosal and systemic immunity with pulmonary nanoparticle vaccination. Sci Transl Med 5(204):ra130. 7. Hafner AM, Corthésy B, Merkle HP (2013) Particulate formulations for the delivery of poly(I:C) as vaccine adjuvant. Adv Drug Deliv Rev 65(10):1386–1399. 8. Leleux J, Roy K (2013) Micro and nanoparticle-based delivery systems for vaccine immunotherapy: An immunological and materials perspective. Adv Healthc Mater 2(1):72–94. 9. Rozenfeld JHK, Silva SR, Ranéia PA, Faquim-Mauro E, Carmona-Ribeiro AM (2012) Stable assemblies of cationic bilayer fragments and CpG oligonucleotide with enhanced immunoadjuvant activity in vivo. J Control Release 160(2):367–373. 10. Alving CR, Rao M (2008) Lipid A and liposomes containing lipid A as antigens and adjuvants. Vaccine 26(24):3036–3045. 11. Andrews CD, Provoda CJ, Ott G, Lee K-D (2011) Conjugation of lipid and CpG-containing oligonucleotide yields an efficient method for liposome incorporation. Bioconjug Chem 22(7):1279–1286. 12. Storni T, et al. (2004) Nonmethylated CG motifs packaged into virus-like particles induce protective cytotoxic T cell responses in the absence of systemic side effects. J Immunol 172(3):1777–1785. 13. Ballester M, et al. (2011) Nanoparticle conjugation and pulmonary delivery enhance the protective efficacy of Ag85B and CpG against tuberculosis. Vaccine 29(40): 6959–6966. 14. Nembrini C, et al. (2011) Nanoparticle conjugation of antigen enhances cytotoxic T-cell responses in pulmonary vaccination. Proc Natl Acad Sci USA 108(44):E989–E997. 15. Hirosue S, Kourtis IC, van der Vlies AJ, Hubbell JA, Swartz MA (2010) Antigen delivery to dendritic cells by poly(propylene sulfide) nanoparticles with disulfide conjugated peptides: Cross-presentation and T cell activation. Vaccine 28(50):7897–7906. 16. van der Vlies AJ, O’Neil CP, Hasegawa U, Hammond N, Hubbell JA (2010) Synthesis of pyridyl disulfide-functionalized nanoparticles for conjugating thiol-containing small molecules, peptides, and proteins. Bioconjug Chem 21(4):653–662. 17. Kourtis IC, et al. (2013) Peripherally administered nanoparticles target monocytic myeloid cells, secondary lymphoid organs and tumors in mice. PLoS ONE 8(4):e61646. 18. Foster S, Duvall CL, Crownover EF, Hoffman AS, Stayton PS (2010) Intracellular delivery of a protein antigen with an endosomal-releasing polymer enhances CD8 T-cell production and prophylactic vaccine efficacy. Bioconjug Chem 21(12):2205–2212. 19. Krieg AM (2012) CpG still rocks! Update on an accidental drug. Nucleic Acid Ther 22(2):77–89. 20. Marschner A, et al. (2005) CpG ODN enhance antigen-specific NKT cell activation via plasmacytoid dendritic cells. Eur J Immunol 35(8):2347–2357. 21. Huber JP, Farrar JD (2011) Regulation of effector and memory T-cell functions by type I interferon. Immunology 132(4):466–474.
de Titta et al.
cells; thus, other mechanisms that might be involved would require further investigation. Materials and Methods NPs were synthesized and functionalized with either antigen or CpG adjuvant conjugated via a reduction-sensitive link. Two murine tumor models were utilized, namely the E.G7-OVA model (a thymoma expressing OVA as a model tumor antigen) and the B16-F10 model (a melanoma expressing OVA). The details of materials and methods are described in SI Materials and Methods. ACKNOWLEDGMENTS. We thank K. Y. Dane, A. Stano, P. S. Briquez, K. Ciapala, A. W. Lund, E. Simeoni, D. K. Bonner, S. Hirosue, S. N. Thomas, X. Quaglia, and P. Corthésy-Henrioud for technical assistance and useful discussions. This work was supported by the European Research Council (NanoImmune), the Bill and Melinda Gates Foundation, and the Swiss National Science Foundation.
22. Krug A, et al. (2001) Identification of CpG oligonucleotide sequences with high induction of IFN-alpha/beta in plasmacytoid dendritic cells. Eur J Immunol 31(7): 2154–2163. 23. Krieg AM (2002) CpG motifs in bacterial DNA and their immune effects. Annu Rev Immunol 20:709–760. 24. Vollmer J, et al. (2004) Characterization of three CpG oligodeoxynucleotide classes with distinct immunostimulatory activities. Eur J Immunol 34(1):251–262. 25. Gonzalez SF, et al. (2011) Trafficking of B cell antigen in lymph nodes. Annu Rev Immunol 29:215–233. 26. Reddy ST, et al. (2007) Exploiting lymphatic transport and complement activation in nanoparticle vaccines. Nat Biotechnol 25(10):1159–1164. 27. Kerkmann M, et al. (2005) Spontaneous formation of nucleic acid-based nanoparticles is responsible for high interferon-alpha induction by CpG-A in plasmacytoid dendritic cells. J Biol Chem 280(9):8086–8093. 28. Pulendran B, Tang H, Denning TL (2008) Division of labor, plasticity, and crosstalk between dendritic cell subsets. Curr Opin Immunol 20(1):61–67. 29. Agnellini P, et al. (2008) Kinetic and mechanistic requirements for helping CD8 T cells. J Immunol 180(3):1517–1525. 30. Sad S, Krishnan L (2003) Maintenance and attrition of T-cell memory. Crit Rev Immunol 23(1-2):129–147. 31. Schwarz K, et al. (2003) Role of Toll-like receptors in costimulating cytotoxic T cell responses. Eur J Immunol 33(6):1465–1470. 32. Duthie MS, Windish HP, Fox CB, Reed SG (2011) Use of defined TLR ligands as adjuvants within human vaccines. Immunol Rev 239(1):178–196. 33. Kandimalla ER, et al. (2002) Conjugation of ligands at the 5′-end of CpG DNA affects immunostimulatory activity. Bioconjug Chem 13(5):966–974. 34. Putta MR, et al. (2010) Peptide conjugation at the 5′-end of oligodeoxynucleotides abrogates toll-like receptor 9-mediated immune stimulatory activity. Bioconjug Chem 21(1):39–45. 35. Purswani MU, Eckert SJ, Arora HK, Noel GJ (2002) Effect of ciprofloxacin on lethal and sublethal challenge with endotoxin and on early cytokine responses in a murine in vivo model. J Antimicrob Chemother 50(1):51–58. 36. Breckpot K, et al. (2003) Lentivirally transduced dendritic cells as a tool for cancer immunotherapy. J Gene Med 5(8):654–667. 37. Vasconcelos JR, et al. (2012) Relevance of long-lived CD8(+) T effector memory cells for protective immunity elicited by heterologous prime-boost vaccination. Front Immunol 3:358. 38. Rothenfusser S, et al. (2004) CpG-A and CpG-B oligonucleotides differentially enhance human peptide-specific primary and memory CD8+ T-cell responses in vitro. Blood 103(6):2162–2169. 39. Seder RA, Darrah PA, Roederer M (2008) T-cell quality in memory and protection: Implications for vaccine design. Nat Rev Immunol 8(4):247–258. 40. Brincks EL, Katewa A, Kucaba TA, Griffith TS, Legge KL (2008) CD8 T cells utilize TRAIL to control influenza virus infection. J Immunol 181(7):4918–4925. 41. Zhang Y, et al. (2007) Th1 cell adjuvant therapy combined with tumor vaccination: A novel strategy for promoting CTL responses while avoiding the accumulation of Tregs. Int Immunol 19(2):151–161. 42. Hollenbaugh JA, Reome J, Dobrzanski M, Dutton RW (2004) The rate of the CD8dependent initial reduction in tumor volume is not limited by contact-dependent perforin, Fas ligand, or TNF-mediated cytolysis. J Immunol 173(3):1738–1743.
PNAS | December 3, 2013 | vol. 110 | no. 49 | 19907
IMMUNOLOGY
CpG-C (Fig. 5 D and E). Different mechanisms could explain the differences between memory recall with NP-CpG-B and NP-CpG-C, considering that the proportion of IFN-γ-secreting and degranulating CD8+ T cells induced was similar for both formulations. Cytotoxic T cells can act on infected cells by different mechanisms requiring either soluble factors such as cytokines, granzyme B, and perforin or cell-contact-dependent mechanisms such as TNF-related apoptosis-inducing ligand (TRAIL) and FasL (40). Depending on the pathological cell type or organ, certain responses are more desirable than others. E.G7-OVA cells are known to be sensitive to CD8+ T cells (41), and especially to their IFN-γ secretion (42), making our vaccine formulation more potent for complete eradication of these tumor cells. However, the exact mechanism leading to B16-F10-OVA melanoma cells elimination is less obvious. In our studies, we focused on the induction of cytotoxic CD8+ T
