Dual stimulation of antigen presenting cells using carbon nanotube-based vaccine delivery system for cancer immunotherapy

Although anti−cancer immuno−based combinatorial therapeutic approaches have shown promising results, efficient tumour eradication demands further intensification of anti−tumour immune response. With the emerging field of nanovaccinology, multi−walled carbon nanotubes (MWNTs) have manifested prominent potentials as tumour antigen nanocarriers. Nevertheless, the utilization of MWNTs in co−delivering antigen along with different types of immunoadjuvants to antigen presenting cells (APCs) has not been investigated yet. We hypothesized that harnessing MWNT for concurrent delivery of cytosine−phosphate−guanine oligodeoxynucleotide (CpG) and anti-CD40 Ig (αCD40), as immunoadjuvants, along with the model antigen ovalbumin (OVA) could potentiate immune response induced against OVA−expressing tumour cells. We initially investigated the effective method to co−deliver OVA and CpG using MWNT to the APC. Covalent conjugation of OVA and CpG prior to loading onto MWNTs markedly augmented the CpG−mediated adjuvanticity, as demonstrated by the significantly increased OVA−specific T cell responses in vitro and in C57BL/6 mice. αCD40 was then included as a second immunoadjuvant to further intensify the immune response. Immune response elicited in vitro and in vivo by OVA, CpG and αCD40 was significantly potentiated by their co−incorporation onto the MWNTs. Furthermore, MWNT remarkably improved the ability of co−loaded OVA, CpG and αCD40 in inhibiting the growth of OVA−expressing B16F10 melanoma cells in subcutaneous or lung pseudo−metastatic tumour models. Therefore, this study suggests that the utilization of MWNTs for the co−delivery of tumour−derived antigen, CpG and αCD40 could be a competent approach for efficient tumours eradication.


Introduction
Cancer therapeutic vaccines rely on the ability of professional antigen presenting cells (APCs), specifically dendritic cells (DCs), to detect, process and then present administered tumourÀantigens via the major histocompatibility complex molecules class (MHC) II or I to CD4 þ or CD8 þ T cells, respectively, leading to antiÀtumour immune responses induction [1]. However, tumourÀinduced immunosuppression and abundance of immunosuppressive regulatory T cells in the tumour microÀenvironment hinder the immune system to effectively eradicate established tumours [2]. This could be overcome by the use of combinatorial immunotherapeutic approaches, for instance by administration of tumour antigens along with different types of immunoadjuvants, rather than the single immunotherapeutic ones [3].
Carbon nanotubes (CNTs) have been developed as needleÀlike nanoscopic carriers capable of improving therapeutic agents delivery to the intracellular compartments via energyÀdependent and/or passive mechanisms of cellular uptake [4e6]. We have previously demonstrated that altering the surface chemistry of multiÀwalled CNTs (MWNTs) conjugated to the model antigen ovalbumin (OVA) can affect the extent of their cellular internalization into APCs, and thus the intensity of the resulting immune responses elicited in vitro and in vivo [7]. As a delivery vector for tumour antigens, MWNTs have markedly improved antitumour immune response against breast or liver cancerederived tumour proteins in vitro [8] or in vivo [9], respectively. SingleÀwalled CNTs (SWNTs) conjugation with the immunoadjuvant cytosineÀphos-phateÀguanine oligodeoxynucleotide (CpG) enhanced the CpGÀinduced stimulatory activities in vitro [10] and amplified the antiÀtumour response in gliomaÀbearing mice [11,12]. Despite the demonstrated proficiencies of CNTs as an efficient delivery vector for antigen or immunoadjuvant, the utilization of CNTs to coÀdeliver antigen along with various types of immunoadjuvants has not been studied yet. Agonist for TollÀlike receptors (TLRs) expressed by APCs have shown marked capabilities in augmenting the antigenÀspecific immune response via various mechanisms, including the ability to enhance antigen presentation by APCs [13]. CpG, an agonist for endosomal TLR9, has been included in various cancer vaccine formulations tested in clinical trials [14]. The coÀinternalization of both antigen and TLR agonist by the same APC has been shown to influence the TLR agonistÀmediated improvement of antigen presentation, thus the induced T cell responses. Yarovinsky et al. showed that potent induction of antigenÀspecific CD4 þ T cell response in mice required the activation of TLR11 and antigen presentation via MHC II to occur "in cis" in the same DC instead of separate DCs [15]. Wilson et al. demonstrated the significance of APC activation by TLR agonists at the time of antigen uptake showing that preÀtreatment of mice with CpG reduces the ability of DCs to take up and present viral antigens to the CD8 þ T cells [16]. Posing additional complexity, Blander et al. reported that more efficient antigen presentation by DCs could be achieved following the internalization of antigen and TLR4 agonist into the same rather than separate phagosome(s) in vitro [17]. In light of these studies, we hypothesized that designing an efficient method to coÀincorporate antigen and CpG onto MWNT could improve their concomitant delivery to APC, and thus the induction of an antigenspecific immune response.
APCs express a number of receptors known as tumour necrosis factor receptors (TNFRs) such as CD40. Anti-CD40 Ig (aCD40), an agonist for the CD40 co-stimulation molecule, has exhibited potential benefits in amplifying antigenÀspecific immune responses [18,19]. It has been reported that DC stimulation with aCD40 chemically conjugated to peptide antigens increased the DC capacity to induce antigenÀspecific CD8 þ T cell response in vitro [20,21]. This has been attributed to the demonstrated aCD40 ability to intracellularly target the conjugated antigen to the early endosomes of DCs. Antigen routing to the early endosomes of DCs has shown to facilitate antigen proteasomal degradation, loading onto MHC I and, subsequently, presentation to CD8 þ T cells [22]. Thereby, by utilizing the aCD40Àmediated enhancement of antigen presentation, stimulation of DCs with MWNT loaded with aCD40 in addition to OVA could further improve the induction of OVAÀspecific CD8 þ T cell response [23]. CD40 interaction with aCD40 has been found to provide APCs with the CD4 þ T cellÀderived licensing signals required for CD8 þ T cell stimulation. This has been demonstrated by the ability of administered aCD40 to restore antigen specific CD8 þ T cell response in CD4 þ T cellÀdepleted mice [24,25]. In addition, for an efficient CD8 þ T cell response induction, the process of antigen recognition by both CD4 þ and CD8 þ T cells has to occur via the same APC [26]. Hence, theoretically, higher immune response intensity could be achieved using delivery approaches that coÀdeliver the antigen and aCD40 signal to the same APC.
We hypothesized that CpG and aCD40 coÀincorporation onto MWNT carrying the model antigen OVA would synergistically and significantly improve the OVAÀspecific immune responses, and effectively retard the growth of OVAÀexpressing B16F10 melanoma cells in solid or pseudoÀmetastatic tumour models.

Mice
All the experiments involving the animal use were carried out in accordance with the project and personal license authorized by the UK Home Office and UKCCCR Guidelines (1998). The C57BL/6 mice were purchased from Harlan (UK). The OT1 Rag À/À and OT2 Rag À/À mice were maintained at Charles River (UK). All experiments use were performed using female 6e8 weeks old mice.

Synthesis of S
Synthesis of chemically functionalized MWNT has been described before and is shown in Scheme 1 [7]. Briefly, pristine MWNTs (pÀMWNTs) (20e30 nm diameter, 0.5e2 mm length, Nanostructured and Amorphous Materials, USA) were oxidized using acidic mixture, followed by incorporation of amineÀterminated spacer using amide coupling reaction yielding a functionalized MWNT named S À/þ . The synthesis of OVAÀCpG is illustrated in Scheme S1 and described in Supplementary Information [27,28]. For the synthesis of (OVA)S À/þ (CpG), 0.5 ml of PBS (PAA Laboratories Ltd, UK) containing 1 mg OVA (EndoGrade ® Ovalbumin, Hyglos GmbH, Germany) and 1.1 mg CpG (phosphorothioate ODN CpG 1668 (5 0 À(TCCATGACGTTCCTGATGCT)À3 0 ), Eurogentec, Belgium) were mixed with a dispersion of 2 mg S À/þ in 2 ml PBS. For the synthesis of S À/þ (OVAÀCpG), OVAÀCpG containing 1 mg OVA and 1.1 mg CpG in 0.5 ml PBS was mixed with a dispersion of 2 mg S À/þ in 2 ml PBS. Both reactions were mixed for 8 h at 4 C. The reaction mixtures were briefly sonicated then vacuum filtered through 0.22 mm polycarbonate membrane filter (Isopore™ Membrane, Merck Millipore, Germany). The solids recovered were reÀdispersed in 2.5 ml PBS and the obtained dispersion was briefly sonicated and then vacuum filtered. Unreacted OVA and CpG contained in the collected filtrates were quantified using bicinchoninic acid protein (BCA) assay reagent (Fisher Scientific, UK) and Nano-Drop (NDÀ1000 spectrophotometer, NanoDrop Technologies, USA), respectively, as described in Supplementary Information. The recovered S À/þ (OVAÀCpG) or (OVA)S À/þ (CpG) solids were further washed with methanol (Fisher Scientific, UK) and then vacuum filtered through 0.22 mm polycarbonate membrane filter, dried and recovered. Synthesized conjugates were characterized using thermogravimetric analysis (TGA) and polyacrylamide gel electrophoresis (PAGE) that were performed as described before [7].

Synthesis of (aCD40)S À/þ (OVAÀCpG)
To a dispersion of 3 mg S À/þ in 2 ml PBS, 1 mg of aCD40 (purified rat antiÀmouse CD40 monoclonal antibody, BD Biosciences, USA) in 0.5 ml PBS was added. The reaction was mixed for 8 h at 4 C. The reaction mixture was briefly sonicated then vacuum filtered through 0.22 mm polycarbonate membrane filter. The solids recovered were reÀdispersed in 2.5 ml PBS and the obtained dispersion was briefly sonicated and then vacuum filtered.
Unreacted aCD40 contained in the collected filtrates was quantified using BCA assay as described in Supplementary Information. The recovered (aCD40)S À/þ solids were washed with methanol, vacuum filtered through 0.22 mm polycarbonate membrane filter, dried and reÀdispersed in 2 ml PBS. To the (aCD40)S À/þ dispersion, OVAÀCpG containing 1 mg OVA and 1.1 mg CpG in 0.5 ml PBS was added. The reaction was mixed for 8 h at 4 C. The reaction mixture was briefly sonicated then vacuum filtered through 0.22 mm polycarbonate membrane filter. The solids recovered were reÀdispersed in 2.5 ml PBS and the obtained dispersion was briefly sonicated and then vacuum filtered. Unreacted OVA and CpG contained in the collected filtrates were quantified using BCA assay and NanoDrop, respectively, as described in Supplementary Information. The recovered (aCD40)S À/þ (CpG) solids were further washed with methanol and then vacuum filtered through 0.22 mm polycarbonate membrane filter, dried and recovered.
2.4. Assessment of OVA presentation induced by (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) treated BMÀDCs in vitro DCs were generated from the bone marrow of C57BL/6 mice and characterized for their purity as previously described [7]. Bone marrowÀderived DCs (BMÀDCs) were incubated for 24 h with OVA, mixture of unconjugated OVA and CpG (referred to as OVA þ CpG), OVAÀCpG, (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) each containing 5 mg/ml OVA. The used doses were determined from the optimization studies described in Supplementary Information (Figure S1). BMÀDCs were incubated, as a control, with S À/þ alone at concentrations equivalent to those contained in (OVA)S À/þ (CpG) or S À/ þ (OVAÀCpG) (20e38 mg/ml). Treated BMÀDCs were harvested, washed several times with RPMI 1640 and gammaÀirradiated using CesiumÀ137 at 3000 Gy for 10 min. CD4 þ and CD8 þ T cells were isolated from the OTÀII and OTÀI mice spleen, respectively, and characterized for their purity as described before [7]. In 96Àwell roundÀbottom plate, 25 Â 10 3 of CD4 þ or CD8 þ T cells were coÀcultured with the BMÀDCs at 1:4 ratio in a total volume of 200 ml complete medium per well. The BMÀDCs: T cell coÀculture ratio was determined from previous optimization studies [7]. As a control, CD4 þ or CD8 þ T cells were cultured alone or with naïve BMÀDCs. Cultured cells were maintained for 3 days at 37 C. For the last 18 h of incubation, 50 ml of the supernatants were removed and replaced with a fresh 50 ml of complete medium containing 1 mCi of 3 HÀthymidine (Thymidine (MethylÀ 3 H), Perkin Elmer, USA). T cell proliferation was assessed by measuring the incorporated 3 HÀthymidine emitted radiation using liquid scintillation counter (Wallac 1205 Betaplate) [7]. The levels of IFNÀg in the supernatants collected from BMÀDCs coÀcultured with CD4 þ or CD8 þ T cells were quantified using antiÀmouse IFNÀg sandwich ELISA kit (eBioscience, USA) following the manufacturer's protocol. The absorbance of each well was measured at 450 nm using a plate reader (FLUOstar Omega, BMG LABTECH, Germany).
The in vivo cytotoxic T lymphocyte (CTL) assay was performed following previously described method [7]. Briefly, a 1:1 splenocytes mixture consisting of 0.5 mM CFSE (carboxyfluorescein diacetate succinimidyl ester, eBioscience, USA) Àlabelled and SIINFEKLÀpulsed splenocytes (referred to as 0.5 mM CFSE SIINFEKL ) and 5 mM CFSEÀlabelled unÀpulsed splenocytes (referred to as 5 mM CFSE no SIINFEKL ), were administered in immunized mice or mice injected with PBS via the tail vein at 10 Â 10 6 cells per 200 ml per mouse, on the 8th day post immunization. At 18 h post-Àinjection, mice were scarified; spleens were harvested and digested in collagenase/DNase solution. The percentage of SIIN-FEKLÀpulsed and unÀpulsed splenocytes, induced by each treatment, in the harvested splenocytes was determined using flow cytometry. AntigenÀspecific killing was calculated using the following equation: Assessment of the antiÀtumour response induced by (aCD40) S À/þ (OVAÀCpG) in subcutaneous tumour model LuciferaseÀtransfected melanoma B16F10 cells were obtained from Perkin Elmer (USA) and were transduced with vesicular stomatitis virus G pseudotyped retrovirus encoding green fluorescence protein (GFP)Àtagged OVA [27,29]. GFP positive cells were then sorted as single cells using GFP filter. C57BL/6 mice were subcutaneously inoculated in both flanks with 2.5 Â 10 5 OVA-Àexpressing and luciferaseÀtransfected B16F10 (OVAÀB16F10ÀLuc) cells. On the 7th day post tumour inoculation, mice were randomly assigned to 5 groups (n ¼ 7). On the 7th and 14th days post tumour inoculation, mice were immunized via footpad injection with S À/þ (OVAÀCpG), aCD40 þ OVAÀCpG, aCD40 þ S À/þ (OVAÀCpG) or (aCD40)S À/þ (OVAÀCpG), each containing 6 mg OVA, 6 mg CpG and/or 21 mg aCD40 in 50 ml PBS. PBS injected mice were used as untreated controls. A calliper was used to measure the tumour length (L) and width (W), and the tumour volume was calculated using the following equation: Tumour volume ¼ 0.52 Â W 2 Â L. Mice were sacrificed when the tumour volume reached 1000 mm 3 .
2.8. Assessment of antiÀtumour response induced by (aCD40)S À/ þ (OVAÀCpG) in lung pseudoÀmetastatic tumour model C57BL/6 mice were intravenously inoculated with 2.5 Â 10 5 OVAÀB16F10ÀLuc cells. On the 4th day post tumour inoculation, mice were randomly assigned to 3 groups (n ¼ 6e8). On the 4th and 9th days post tumour inoculation, mice were immunized via footpad injection with S À/þ (OVAÀCpG) or (aCD40)S À/þ (OVAÀCpG) containing 6 mg OVA, 6 mg CpG and/or 21 mg aCD40. PBS injected mice were used as untreated controls. Tumour growth was monitored by detecting the bioluminescence emitted from the inoculated OVAÀB16F10ÀLuc cells following DÀLuciferin (Perkin Elmer, USA) injection. Every 3e4 days post tumour inoculation, mice were anesthetized and subcutaneously injected with DÀLuciferin (150 mg/kg) in PBS. Imaging was performed using IVIS Lumina III and images analysis was conducted with Living Image ® 4.3.1 Service Pack 2 software (Perkin Elmer, USA).

Histological analysis
Heart, lung, kidney, spleen and lymph nodes were isolated from mice at sacrifice. Isolated tissues were fixed using 10% neutral buffer formalin (Sigma, UK) and tissue sections were stained using haematoxylin and eosin (H & E) or Neutral Red (NR) following the standard staining protocols of the Royal Veterinary College (UK). Images of the stained histological specimens were captured using Leica DM 1000 LED Microscope (Leica Microsystems, UK) connected to CDD digital camera (Qimaging, UK).

Statistical analysis
Results are expressed as mean value ± standard deviation (S.D.), unless otherwise stated. Statistical analysis was performed using GraphPad Prism version 5.01 (USA). Statistical differences were determined using oneÀway ANOVA followed by Bonferroni postÀtest.

Synthesis and characterization of (OVA)S
As depicted in Scheme 1A, the length of pÀMWNTs was shortened by oxidation reaction using sulphuric and nitric acids mixture and bath sonication, yielding MWNT 1. This step was followed by partial neutralization of the incorporated negatively charged carboxylic acid moieties using an amine terminated spacer, yielding S À/þ . We have previously reported that this functionalization approach, compared to other chemical functionalization methods, yields a functionalized MWNT (S À/þ ) capable of significantly improving the loaded antigen internalization by APCs in vitro and in vivo [7].
OVA, CpG and/or aCD40 loading onto S À/þ was also confirmed using TGA [32]. TGA was performed under inert gas (nitrogen) by exposing the tested samples to gradually increasing temperature (up to 800 C). The graphitic structure of the pristine CNT (p-MWNT) is stable against sublimation within the applied temperatures. However, surface defects and impurities such as amorphous carbon (that constitute approximately 2% of p-MWNT) are less stable and thermally decompose by sublimation at 600 C [33,34]. Organic functional groups decomposition also occurs at temperatures lower than 600 C. It was previously reported that thermally degraded functional groups or biomolecules decompose mainly into carbon dioxide, carbon monoxide, ammonia [35].
The density of the loaded functional groups and biomolecules is directly related to the sample weight loss 600 C, as a result of thermal decomposition. As demonstrated in Fig. 1B, a greater reduction in the thermal stability was observed for S À/þ compared to pÀMWNT as a result of decomposition of the functional groups. Expectedly, (OVA)S À/þ (CpG), (aCD40)S À/þ (OVAÀCpG) and S À/ þ (OVAÀCpG) achieved higher weight losses than S À/þ , in the same order. This observation could be assigned to the fact that (OVA)S À/ þ (CpG) possessed the highest content of incorporated biomolecules followed by (aCD40)S À/þ (OVAÀCpG) then S À/þ (OVAÀCpG) ( Table 1). TGA confirmed the success of chemical modification of S À/ þ and loading of OVA, CpG and aCD40 onto S À/þ . PAGE electrophoresis was employed to visualize the loaded OVA and/or aCD40. Similar to unconjugated OVA, the OVA contained in (OVA)S À/þ (CpG) appeared as an intense band of~45 kDa (Fig. 1C). In case of OVAÀCpG or S À/þ (OVAÀCpG), an increase in OVA molecular weight was observed (>45 kDa) due to successful conjugation with CpG. Exposing (aCD40)S À/þ (OVAÀCpG) to gel electrophoresis confirmed the presence of aCD40 as a main intense band of  ~150 kDa and CpGÀconjugated OVA bands (Fig. 1D).

3.2.
Loading of OVAÀCpG conjugate onto S À/þ offers more potent in vitro antigen presentation than the loading of unconjugated OVA and CpG To determine the effect of (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) on the maturation of DC, the synthesized conjugates were incubated with BMÀDCs for 24 h and the expression of MHC as well as co-stimulatory molecules were determined using specific antibodies and flow cytometry as described in Supplementary Information. Similar to the CpG treatment alone, incubation of BMÀDCs with (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) increased the expression of MHC I, MHC II, CD40 and CD86 ( Fig. 2A and Figure S3) with no significant differences seen between groups. OVAÀtreated BMÀDCs showed no signs of maturation. Incubation of BMÀDCs with S À/þ alone has been shown previously to not affect the expression of these molecules [7]. Taken together the data suggest that maturation of BMÀDCs induced by (OVA)S À/þ (CpG) or S À/ þ (OVAÀCpG) was CpGÀdependant.
Next, the efficiency of the two approaches in enhancing OVA presentation by BMÀDCs was assessed in vitro using OVAÀspecific transgenic CD4 þ or CD8 þ T cells. After incubation of BMÀDCs with the S À/þ based conjugates (containing 5 mg/ml OVA), (OVA)S À/ þ (CpG) or S À/þ (OVAÀCpG) significantly enhanced OVAÀspecific T cells proliferation as compared to their control treatments, namely OVA þ CpG or OVAÀCpG (p < 0.001), respectively (Fig. 2B). However, treatment with S À/þ (OVAÀCpG) resulted in elevated responses compared to (OVA)S À/þ (CpG) (p < 0.05). This was further confirmed with IFNÀg cytokine production profiles (Fig. 2C). Similar T cell responses were obtained when BMÀDCs were treated with the synthesized conjugates containing a lower dose of 2.5 mg/ ml OVA (data not shown).
These observations indicated that loading OVA and CpG onto S À/ þ in the form of a conjugate can lead to enhanced OVA presentation by BMÀDCs in vitro, compared to loading unconjugated OVA and CpG.

Immunization with S À/þ loaded with OVAÀCpG elicits potent cellular and humoral immune responses
The capability of (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) to induce a cellÀmediated immune response in vivo was determined using an in vivo CTL assay [27]. In these experiments, the OVA immunization dose used in (OVA)S À/þ (CpG), S À/þ (OVAÀCpG) or their controls was 6 mg, given our observation that in C57BL/6 mice treated with various doses of OVAÀCpG, a measureable OVAÀspecific CTL immune response was detected at OVA content of 6 mg ( Figure S4).
To assess the humoral response induced by the conjugates in vivo, C57BL/6 mice were treated with the conjugates or appropriate controls, and OVAÀspecific antibodies were quantified using an OVAÀspecific ELISA 21 days post immunization (Fig. 3B). Immunization with S À/þ (OVAÀCpG) significantly boosted the production of antiÀOVA IgG and IgG2c antibodies titres compared to (OVA)S À/þ (CpG) or other treatments. Both conjugates, however, ensued comparable antiÀOVA IgG1 titres.
These findings demonstrated further the augmentation in antigen specific immune response in vivo achieved by S À/þ (OVAÀCpG) over (OVA)S À/þ (CpG). We adopted this conjugate S À/þ (OVAÀCpG) for subsequent studies in combination with aCD40 as a second immunoadjuvant.

Incorporation of aCD40 as a second immunoadjuvant improves OVA presentation in vitro and intensifies OVAÀspecific immune response in vivo even at lower OVA doses
To further intensify the antigenÀspecific immune responses observed, aCD40 antibody was loaded onto S À/þ as a second immunoadjuvant. In order to assess the effect of aCD40 contained in (aCD40)S À/þ (OVAÀCpG) on DC activation markers, BMÀDCs were stimulated with the conjugate or control treatments and known DC markers were assessed using flow cytometry as described in Supplementary Information. BMÀDC stimulated with (aCD40)S À/þ (OVAÀCpG) expressed significantly higher levels of MHC I and CD86 compared to those stimulated with S À/ þ (OVAÀCpG) or a mixture of unconjugated aCD40 and S À/ þ (OVAÀCpG) (aCD40 þ S À/þ (OVAÀCpG)) (Fig. 4A). BMÀDCs stimulated with (aCD40)S À/þ (OVAÀCpG) showed lower expression of CD40 compared to S À/þ (OVAÀCpG). This could be attributed to the cellular internalization of the CD40 receptor following its ligation with aCD40 contained in (aCD40)S À/þ (OVAÀCpG) [20,21].
Given the increase in MHC I expression, OVA presentation to CD8 þ T cell was assessed, following stimulation with (aCD40)S À/ þ (OVAÀCpG). In order to assess the synergy provided by the CpG and aCD40 loaded onto S À/þ , BMÀDCs were incubated with (aCD40)S À/þ (OVAÀCpG) containing half the OVA and CpG amounts present in S À/þ (OVAÀCpG) treatment (Fig. 4B). Higher CD8 þ T cell proliferation was induced following stimulation with (aCD40)S À/ þ (OVAÀCpG) containing 0.5 mg/ml of both OVA and CpG compared to the control treatment aCD40 þ S À/þ (OVAÀCpG), and S À/ þ (OVAÀCpG) treatment that contained 1 mg/ml of both OVA and CpG. Production of IFNÀg by the stimulated CD8 þ T cells correlated well with their pattern of proliferation. These observations reflected the significance of aCD40 conjugation with S À/þ on BMÀDC activation.
Taking the in vitro and in vivo data together we can conclude that inclusion of aCD40 antibody as a second immunoadjuvant resulted in synergy of the MWNT-mediated delivery of OVAÀCpG as shown by the marked increase in antigenÀspecific immune responses at lower OVA and CpG doses.
3.5. aCD40 and OVAÀCpG loading onto S À/þ effectively delays the tumour growth in both solid and lung pseudoÀmetastatic tumour models The therapeutic efficacy of the conjugates in delaying the growth of a solid tumour was investigated. Immunization of C57BL/ 6 mice subcutaneously inoculated with LucÀB16F10ÀOVA cells with S À/þ (OVAÀCpG) containing 12 or 25 mg of both OVA and CpG led to significant tumour growth retardation compared to unimmunized mice ( Figure S6A). Furthermore, administration of S À/ þ (OVAÀCpG), containing 25 mg of both OVA and CpG, to mice subcutaneously inoculated with B16 cells (tumour cells which do not express OVA), failed to impede the B16 cells growth ( Figure S6B), indicating that the induced antiÀtumour immune response was antigenÀspecific.
To determine the ability of (aCD40)S À/þ (OVAÀCpG) to delay the solid tumour growth at reduced OVA and CpG doses, mice were subcutaneously inoculated with LucÀB16F10ÀOVA cells and then immunized with (aCD40)S À/þ (OVAÀCpG), S À/þ (OVAÀCpG) or other controls at 6 mg of both OVA and CpG. As demonstrated in Fig. 5A, immunization with S À/þ (OVAÀCpG) at 6 mg OVA did not significantly delay the tumour growth compared to unimmunized mice. However, immunization with (aCD40)S À/þ (OVAÀCpG) led to significant tumour growth retardation compared to unimmunized mice, and mice immunized with S À/þ (OVAÀCpG). Immunization with aCD40 þ S À/þ (OVAÀCpG) failed to delay the tumour growth to the same extent as (aCD40)S À/þ (OVAÀCpG), highlighting the increase in aCD40Àmediated immune enhancement achieved on incorporating aCD40 onto S À/þ in addition to OVAÀCpG. Additionally, vaccination with (aCD40)S À/þ (OVAÀCpG) prolonged the tumourÀinoculated mice survival in a significant manner compared to the other treatments (Fig. 5A).
No changes in the histological features of the excised organs between the untreated and treated tumour bearing mice were observed indicating lack of organotoxicity (Fig. 5B). Dark black Effect of (OVA)S À/þ (CpG) or S À/ þ (OVAÀCpG) on BMÀDC maturation. BMÀDCs were incubated for 24 h with 5 mg/ml CpG, OVA, (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG), each contained 5 mg/ml OVA. BMÀDCs were stained with fluorescently labelled specific antibodies against MHC I, MHC II, CD40, CD80 or CD86, and cell analysis was performed using flow cytometry. The mean fluorescence intensity (MFI) of the positive CD11cÀexpressing BMÀDCs was measured to assess the fold change in the expression of each marker with respect to the naïve BMÀDCs, results represent the mean ± S.D. (B, C) OVA presentation by (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) treated BMÀDCs. BMÀDCs were incubated for 24 h with OVA þ CpG, OVAÀCpG, (OVA)S À/ þ (CpG) or S À/þ (OVAÀCpG), each contained 5 mg/ml OVA. Treated BMÀDCs were coÀcultured with CD4 þ or CD8 þ T cells isolated from the spleen of OTÀ2 or OTÀ1 C57BL/6 mice, respectively, at 1:4 ratio for 3 days. On the last 18 h of incubation, CD4 þ T cells (B, left) or CD8 þ T cells (B, right) were pulsed with 1 mCi of 3 HÀthymidine and the proliferation was measured using 3 HÀthymidine incorporation assay. The content of IFNÀg in the supernatants of the proliferating CD4 þ T cells (C, left) or CD8 þ T cells (C, right) was quantified using ELISA. Measurements were performed in triplicates for each condition, results represent the mean ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001.
aggregates, which were absent in naïve mice, were detected in the popliteal lymph nodes from mice immunized with (aCD40)S À/ þ (OVAÀCpG), suggesting drainage of S À/þ into the popliteal lymph nodes.
Therapy studies were then performed in the more challenging pseudoÀmetastatic lung tumour model. The conjugates, containing 6 mg of both OVA and CpG, were administered to C57BL/6 mice previously intravenously inoculated with OVAÀB16F10ÀLuc cells. Smaller bioluminescence signals and lung weights were observed in mice immunized with (aCD40)S À/þ (OVAÀCpG) compared to S À/ þ (OVAÀCpG) treated mice (Fig. 6).
From our data we conclude that vaccination with (aCD40)S À/ þ (OVAÀCpG) efficiently delayed the OVAÀB16F10ÀLuc tumour growth in both solid and pseudoÀmetastatic tumour models, consistent with the immune enhancements observed in vitro and in vivo.

Discussion
One of the purposes of using delivery vectors for antigens is to improve the antigen uptake by the antigen presenting cells (APCs) in order to increase the intracellular antigen concentration, thus the density of antigen presented by the APCs to T cells. It has been previously reported that polymeric spherical nanoparticles, e.g. PLGA nanoparticles, mainly utilise energyÀdependent mechanisms of cellular uptake rather than energyÀindependent ones [36e38]. The reported findings that demonstrated the CNTs' ability to enter the cells via more than one route i.e. energyÀdependent and/or passive routes [4e6], may suggest that CNTs can deliver higher amounts of antigens into the APCs compared to spherical nanoparticles. However, comparative studies need to be carried out to investigate the cellular uptake of CNTs versus the extensively studied spherical nanoparticles, e.g. PLGA nanoparticles and liposomes, by the APCs and the ensuing effects on the magnitude of immune response elicited against incorporated antigen.
Covalent conjugation of OVA and CpG and their loading onto S À/ þ improved OVA presentation in vitro by BMÀDCs and efficiently elevated the magnitude of OVAÀspecific immune response in vivo.
Additionally, the presence of aCD40 in S À/þ containing conjugated OVAÀCpG led to i) more advanced augmentation of the OVAÀspecific immune response in vitro and in vivo, and ii) delayed growth of OVAÀexpressing B16F10 cells effectively in both subcutaneous and lung pseudoÀmetastatic tumour models, at reduced OVA and CpG doses.
We initially proposed two distinct approaches for the concomitant delivery of the model antigen OVA and CpG using S À/þ to APCs. Mixing S À/þ with OVA and CpG was the first method, yielding (OVA) S À/þ (CpG); however, the loading of OVA and CpG onto each S À/þ was uncontrolled. In other words, the prepared (OVA)S À/þ (CpG) might possessed lower OVA and CpG coÀloading onto each S À/þ compared to S À/þ (OVAÀCpG), and the formation of OVA or CpG only Àconjugated S À/þ was also possible. Accordingly, the other approach was loading both agents in the form of the chemical conjugate, OVAÀCpG, to ensure the coÀincorporation of OVA and CpG onto the same S À/þ (S À/þ (OVAÀCpG)). PAGE gel results confirmed that OVA contained in S À/þ (OVAÀCpG) was in the CpGÀconjugated form. (OVA)S À/þ (CpG) or S À/þ (OVAÀCpG) elicited higher immune response potency in vitro and in vivo, compared to their control treatments, namely the mixture of unconjugated OVA and CpG or OVAÀCpG, respectively. This was in agreement with the previously reported benefits of CNTs as a delivery vehicle for antigens [7,9] or immunoadjuvants [11,12] in vitro and in vivo.
The coÀdelivery of antigen and CpG to APC has also been demonstrated using other particulate delivery systems. For instance, mice immunization with microparticles coÀencapsulating OVA and CpG increased the antiÀOVA antibodies [39] and CD8 þ T cell responses [40,41], and utilization of gold nanoclusters for the coÀdelivery of OVAÀderived peptide and CpG augmented the production of antiÀOVA antibodies in mice [42].
In a study by de Faria et al., OVA and CpG were coÀincorporated onto MWNTs by mixing MWNTs with unÀconjugated OVA and CpG, yielding a conjugate that induced higher immune response compared to mixture of unconjugated OVA and CpG (MWNT-free) in vivo [43]. We utilized the same approach to prepare (OVA)S À/ þ (CpG), but additionally, we introduced in this study a more robust approach for comparison, where a covalently conjugated OVA and Quantification of OVAÀspecific IgG. C57BL/6 mice (n ¼ 3) were immunized with the indicated treatments, via footpad injection, each treatment contained 6 mg of OVA and CpG. On day 21 following injection, control or immunized mice sera were collected. The OVAÀspecific IgG, IgG1 or IgG2c were determined using ELISA. Data represent the mean value ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001.
(B) Assessment of OVA presentation. BMÀDCs were incubated for 24 h with either 1 mg/ml OVA (contained in OVAÀCpG or S À/þ (OVAÀCpG)) or 0.5 mg/ml OVA (contained in OVAÀCpG þ aCD40, S À/þ (OVAÀCpG) þ aCD40 or (aCD40)S À/þ (OVAÀCpG)). S À/þ unconjugated or conjugated aCD40 was used at 1.8 mg/ml. BMÀDCs were coÀcultured with CD8 þ T cells. (Left) CD8 þ T cell proliferation. CD8 þ T were pulsed with 3 HÀthymidine and proliferation was measured. (Right) IFNÀg quantification. The content of IFNÀg in the supernatants of the proliferating CD8 þ T cell was quantified using ELISA. Measurements were performed in triplicates for each condition, results represent the mean ± S.D. (C) CTL response. C57BL/6 mice (n ¼ 3e5) were immunized, via footpad injection, with either 6 mg OVA (contained in S À/þ (OVAÀCpG)) or 3 mg OVA (contained in OVAÀCpG þ aCD40 or (aCD40)S À/þ (OVAÀCpG)). The S À/þ unconjugated or conjugated aCD40 was 10 mg. Each dot represents killing of target cells by each mouse, the mean value for each treatment is shown as a horizontal bar. *P < 0.05, **P < 0.01, ***P < 0.001. enhanced OVA presentation might account for the higher cellular and humoral immune responses elicited by vaccination of C57BL/6 mice with S À/þ (OVAÀCpG). Similar to our findings, but without the use of a delivery system, previous studies have demonstrated that mice immunization with covalently conjugated OVA and CpG induced higher OVAÀspecific immune response compared to immunization with a mixture of unconjugated OVA and CpG [15,28].
Schlosser et al. demonstrated that mixing PLGA polymer with OVA and CpG yielded microparticles that were described as OVA and CpG coÀencapsulating microparticles, these microparticles induced higher CD8 þ T cell response in vitro and in vivo in contrast to a mixture of OVA onlyÀencapsulating microparticle and CpG onlyÀencapsulating microparticle [44]. Similarly, Li et al. mixed lipid polymer with HERÀ2/neu derived peptide and CpG to yield liposomes that were referred to as antigen and CpG coÀencapsulating liposomes, mice immunization with these liposomes induced higher immune response compared to a mixture of antigenÀcontaining liposome and CpGÀcontaining liposome [45]. The approach applied in these studies for antigen and CpG coÀincorporation into a delivery system by mixing polymers with unconjugated antigen and CpG is similar, in its basic principle, to the one we followed for the synthesis of (OVA)S À/þ (CpG) but not S À/ þ (OVAÀCpG). These studies highlighted the importance of antigen and CpG coÀdelivery using a delivery system by comparing coÀincorporated to separately incorporated antigen and CpG. However, our study is introducing a more advanced line of complexity by contrasting two methods for antigen and CpG concomitant delivery using a delivery vehicle as demonstrated by (OVA)S À/þ (CpG) versus S À/þ (OVAÀCpG).
Previous studies reported the use of sphericalÀshaped delivery systems to improve antigen and aCD40 coÀdelivery. Hatzifoti et al. demonstrated that coÀencapsulation of tetanus toxoid and aCD40 in liposomes augmented the antigenÀspecific antibody response in BALB/c mice [46], and Rosalia et al. reported an increase in CD8 þ T cell response following mice immunization with aCD40Àcoated PLGA nanoparticles coÀincorporating an oncoprotein and ligands for TLR2 and TLR3 [47]. Instead of harnessing the conventional spherical particulate delivery systems to coÀdeliver CpG and aCD40, we utilized an emerging cylindrical vectors, namely the MWNTs, as a nanoÀcarrier for both CpG and aCD40. In addition, the therapeutic outcome provided by the MWNT-delivered CpG and aCD40 was not only evaluated in a standard subcutaneous tumour model but also lung pseudo-metastatic tumour model. The intensified strength of OVA specificÀCTL response induced by vaccination of C57BL/6 mice with (aCD40)S À/þ (OVAÀCpG) might be assigned to the better ability of this conjugate to induce DC maturation and to further fortify OVA presentation as observed in vitro. Stimulation of APC with aCD40 has been shown to upregulate MHC I and CD86 expression [48,49]. Intracellular signalling induced by ligation of CD40 receptor with aCD40 requires CD40 receptor crossÀlinking that increases, accordingly, with the increase in the number of aCD40 interacting at the cell surface [50].
The fact that the MWNTÀbased conjugates were detected in the lymph nodes was in agreement with our previous study. Where, we were able to detect the presence of S À/þ and the processing of S À/þ conjugated OVA in the CD11c þve DCs subsets in the popliteal lymph nodes [7]. These observations reflected the proficiency of MWNT as vaccine delivery vectors. Since efficient antitumourÀimmune response induction demands antigen trafficking through the lymphatic vessels and internalization by the lymph nodeÀresiding CD8 þ DC, which is the only DC subset capable of inducing CD8 þ T cell response in vivo [53]. Efficient eradication of B16ÀOVA or B16F10ÀOVA tumours in mice has been found to be associated with the cytolytic activity of CD8 þ T cells demonstrated by OVAÀspecific CTL response [41,47,54]. The fact that lower OVA and CpG doses were required by (aCD40)S À/þ (OVAÀCpG) than S À/þ (OVAÀCpG) to induce strong antiÀtumour response indicates the higher potency and better efficacy of the former conjugate in vivo. Collectively, the results shown in this study highlighted the exploitation of MWNTs as antigen and immunoadjuvants nanocarrier for the purpose of inducing potent antiÀtumour immune response.

Conclusions
OVA incorporation onto the MWNT in the form of CpGÀconjugated OVA improved the CpGÀmediated enhancements of OVAÀspecific immune response in vitro and in vivo. Furthermore, the utilization of MWNTs as vaccine delivery vector has intensified the CpG and aCD40Àderived synergism that markedly retarded the OVAÀB16F10 growth in the tested tumour models. The MWNTÀdelivered immunoÀbased combinatorial therapeutic approach presented in this study could be exploited for potent antiÀtumour immune response induction against challenging cancer diseases.