Journal of Controlled Release 195 (2014) 99–109

Contents lists available at ScienceDirect

Journal of Controlled Release

j ourna l homepage: www.e lsev ie r .com/ locate / jconre l

Nanoporous polyelectrolyte vaccine microcarriers. A formulationplatform for enhancing humoral and cellular immune responses

Stefaan De Koker a,b, Kaat Fierens c,d, Marijke Dierendonck a, Riet De Rycke c,e, Bart N. Lambrecht c,d,Johan Grooten b, Jean Paul Remon a, Bruno G. De Geest a,⁎a Department of Pharmaceutics, Ghent University, Ghent, Belgiumb Department of Biomedical Molecular Biology, Ghent University, Zwijnaarde, Ghent, Belgiumc VIB Inflammation Research Center, University of Ghent, Ghent, Belgiumd Department of Respiratory Medicine, University Hospital Ghent, Ghent, Belgiume Department of Plant Systems Biology, VIB, and Department of Plant Biotechnology and Bioinformatics, Ghent University, Ghent, Belgium

⁎ Corresponding author at: Department of PharmHarelbekestraat 72, 9000 Ghent, Belgium. Tel: +32 9 264

E-mail address: br.degeest@ugent.be (B.G. De Geest).

http://dx.doi.org/10.1016/j.jconrel.2014.07.0430168-3659/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 6 May 2014Accepted 20 July 2014Available online 28 July 2014

Keywords:MicroparticlesVaccinesAntigensPolyelectrolytesSpray drying

In this paper we report on the design, characterization and immuno-biological evaluation of nanoporous poly-electrolyte microparticles as vaccine carrier. Relative to soluble antigen, formulation of antigen as a sub-10 μmparticle can strongly enhance antigen-specific cellular immune responses. The latter is crucial to confer protectiveimmunity against intracellular pathogens and for anti-cancer vaccines. However, a major bottleneck inmicroparticulate vaccine formulation is the development of generic strategies that afford antigen encapsulationunder benign and scalable conditions. Our strategy is based on spray drying of a dilute aqueous solution of anti-gen, oppositely charged polyelectrolytes and mannitol as a pore-forming component. The obtained solid micro-particles can be redispersed in aqueous medium, leading to leaching out of the mannitol, thereby creating ahighly porous internal structure. This porous structure enhances enzymatic processing of encapsulated proteins.After optimizing the conditions to process these microparticles we demonstrate that they strongly enhancecross-presentation in vitro by dendritic cells to CD8 T cells. In vivo experiments in mice confirm that this vaccineformulation technology is capable of enhancing cellular immune responses.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Formulating vaccine antigens into microcarriers has emerged as anattractive strategy to amplify antigen-specific immune responses, inparticular the cellular arm of the immune response [1–3]. Whereas sol-uble antigen is predominantly presented by dendritic cells (DCs; themost potent class of antigen presenting cells of our immune system)to CD4 T cells, antigen in the form of nano- andmicroparticles becomespresented to both CD4 and CD8 T cells, a process termed cross-presentation of exogenous antigen [1,4,5]. The reason for this is that an-tigens in a particulate form mimic the morphology of micro-organismsand are thus far better recognized by DCs as being potentially danger-ous, thereby altering the route of internalization and presentation ofthe antigen [5–9]. The functional relevance of these processes lays inthe fact that in the presence of the correct cytokine stimuli, CD8 T cellscan differentiate into cytotoxic T cells (CTLs) that can recognize andeliminate infected or malignant cells. The latter is of great importance

aceutics, Ghent University,80 55; fax: +32 9 222 82 36.

for the development of vaccines against intracellular pathogens suchas HIV, tuberculosis, malaria… and for therapeutic anti-cancer vaccines[10,11].

Many nano- and microparticulate vaccine formulation strategieshave been reported in literature [1–3]. However, there remains a clearneed for simple and scalable formulation strategies involving a minimalnumber of batch steps and avoiding organic solvents and reactive chem-istries. Additionally, when envisioning vaccines for the developingworld and for pandemic vaccines, there is a particular interest forformulations that avoid the cold chain and do not require refrigeratedconditions for transportation and long term storage [12]. In this regard,a dry powder formulation that can be reconstituted in aqueousmediumprior to administration would be highly beneficial.

Recently we have reported on a novel type of vaccine microcarriersbased on oppositely charged polyelectrolytes that form stable micropar-ticles via electrostatic interaction as schematically shown in Fig. 1[13–15]. These microparticles were assembled by atomizing a dilutedaqueous solution of the polyelectrolytes into a hot air stream. Thisspray drying process evaporates the water to yield solid microparticles.By adding protein antigen to the polyelectrolyte mixture prior to spraydrying the antigen becomes encapsulatedwithin the polyelectrolytema-trix and, interestingly, remains stably encapsulated upon redispersion of

Fig. 1. Schematic representation of the encapsulation of protein antigen in porous polyelectrolyte microparticles. As pore-former mannitol is used which instantaneously dissolves uponredispersion in aqueous medium, thereby creating a highly porous matrix.

100 S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

themicroparticles in aqueousmedium. Using dextran sulfate and poly-L-arginine as degradable polyanion/polycationwe previously demonstratethe pivotal role of usingmannitol as an excipient during the spray dryingprocess. This is also illustrated in Fig. 1. Mannitol is a hydrophilic compo-nentwhich is added in excess to the polyelectrolyte/proteinmixture andbecomes incorporated as well into themicroparticles. On the one hand itplays an important in the spray drying process itself by enhancing parti-cle recovery and protecting proteins from denaturation. The latter wasdemonstrated for retaining up to 85% of the catalytic activity of horserad-ish peroxidase that was encapsulated by this technology [15]. On theother hand it allows to form a nanoporous matrix upon redispersion ofthe microparticles in aqueous medium. Indeed, in aqueous mediummannitol dissolves, leaches out, thereby creating a highly porous internalstructure within the microparticles. After cellular uptake, intracellularproteases can diffuse through the nanoporous network and process en-capsulated antigen. Poly-L-arginine can also be degraded by proteaseswhich will further facilitate the intracellular antigen processing. Forboth the encapsulation of vaccine antigens and enzymes into porouspolyelectrolyte microparticles, the use of mannitol appeared crucial topreserve the bioactivity of these proteins in vitro. Indeed, mannitol is aknown stabilizer in spray drying applications [16] and the porous struc-ture within the microparticles allows enzyme substrates or intracellularproteases to access encapsulated proteins much more efficiently.

We demonstrated that nanoporous microparticles based on dex-tran sulfate and poly-L-arginine, and loaded with the model antigenovalbumin (OVA), are efficiently internalized by DCs in vitro and pro-mote presentation of the OVA CD8 epitope as MHCI complex on thesurface of DCs [13–15]. This suggests that this class of vaccine car-riers hold potential for intracellular delivery of vaccine protein anti-gen in view of enhancing cellular immune responses. In the presentwork we aim to explore the potential of the nanoporous polyelectro-lyte microcarriers for vaccine delivery in vitro and in vivo. First weoptimized the fabrication procedure to obtain microspheres thatare non-aggregated, with high particle recovery yield and size distri-bution below 10 μm to enhance cellular uptake while still assuringoptimum encapsulation efficiency. Second, we evaluated the poten-tial of the microparticles to enhance antigen cross-presentation toCD8 T cells in vitro. Third we evaluated the microcarriers in vivowith respect to their tissue response and antigen specific immune re-sponse in mouse models.

2. Materials and methods

2.1. Materials

Mannitol was obtained from Cargill. Dextran sulfate (DS; MW: 9–20 kDa), poly-L-arginine (PLARG; MW N70 kDa), ovalbumin (OVA)were obtained from Sigma-Aldrich. OVA-Alexa488, and Phosphatebuffered saline (PBS) were obtained from Invitrogen. All water used inthe experiments was of Milli-Q grade.

2.2. Formulation optimization

The influence of spray drying parameters on the yield and shape ofthe particles was investigated by varying different process parameters.Formulations prior to spray drying were prepared as earlier reported[15] and described in detail in Section 2.3.

Spray dryingwas performedwith a lab-scale Büchi B290 spray dryerequipped with a two-fluid nozzle (0.7mmdiameter). The setting of theinlet temperature was 120 °C and gas flow varies between 0.23–0.75bar. The mixtures were fed via a peristaltic feed pump at a feed flowof 1 ml/min to 10 ml/min. Dry powder was collected.

2.3. Preparation of the optimal microparticle formulation

Mannitol, DS, OVA and PLARG were mixed in water in a 40:4:1:5ratio at a total solid concentration of 1%. Two different adding sequenceswere compared. In the first case, 200 mg of mannitol, 20 mg of DS and5 mg of OVA were dissolved in 20 ml water. Subsequently 25 mg ofPLARG was dissolved in 5 ml water and was added in drops to the stir-ring mannitol/DS/OVA solution. Secondly, 200 mg mannitol and20mgDSwere dissolved in 19mlwater, further PLARGwas added anal-ogously and finally 5 mg OVAwas dissolved in 1 ml water and added indrops. Fluorescent particles were prepared using a mixture of OVAwithAlexa Fluor488 conjugated ovalbumin in a 50:1 ratio. Spray drying wasperformed with a lab-scale Büchi B290 spray dryer equipped with atwo-fluid nozzle (0.7 mm diameter). The setting of the inlet tempera-ture was 120 °C and gas flow was 0.75 bar. The mixtures were fed viaa peristaltic feed pump at a feed flow of 1 ml/min. Dry powder wascollected.

101S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

The encapsulation efficiency was determined by resuspending aknown amount of microparticles loaded with Alexa Fluor488 conjugat-ed ovalbumin (thus containing a known amount of protein (proteinconcdry microparticles) in phosphate buffered saline followed by centrifu-gation and measuring the OVA-Alexa488 concentration (proteinconcsupernatant)) in the supernatant using a PerkinElmer Envisionmultilabel plate reader. The encapsulation efficiency is then calculatedas follows:

Encapsulation efficiency ¼ 100� protein concdry micropartciles−protein concsupernatant

protein concdry microparticles

2.4. Microscopy

Scanning Electron Microscopy (SEM) was conducted on a Quanta200 FEG FEI scanning electron microscope. Samples were sputteredwith a palladium–gold layer prior to imaging. Transmission electronmi-croscopy (TEM) was performed on a JEOL 1010 instrument. Confocalmicroscopy was conducted on a Leica SP5 microscope equipped witha 63X oil immersion objective.

2.5. Cell lines and animals

C57BL/6 mice were obtained from Janvier. OT-I transgenic mice(C57BL/6) were purchased from Harlan. The mice were housed underspecific-pathogen-free conditions. All animal experiments were ap-proved by the Local Ethical Committee of Ghent University. The immor-talizedmouse dendritic cell line DC2.4was a kind gift from Prof. Dr. KenRock (Dana-Farber Cancer Institute, Boston, MA, USA). Bone-marrow-derived DCs were generated by flushing tibia and femurs of 2–4-month old C57BL/6 mice. After red blood cell lysis, the cells were cul-tured in complete RPMI (Roswell Park Memorial Institute) mediumcontaining 20 ng/ml GM-CSF (granulocyte macrophage colony-stimulating factor) for 6–8 days.

2.6. Cell toxicity assay

The cytotoxicity of the spray-dried particles was assessed accordingto De Koker et al. DC2.4 cells were grown and seeded in a 96-well plateat a density of 5 × 103 cells/well and incubated with different concentra-tions of the respective samples for 24 h. Afterwards, the medium wasrefreshed and the cells were cultured for another 48 h. Medium was re-moved and MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide) was added. MTT is reduced by mitochondrial dehydrogenasesof living cells into an insoluble purple formazan dye. After 4 h ofincubation at 37 °C, the cells are solubilized by dimethylsulfoxide(DMSO) and the released, solubilized formazan ismeasured spectropho-tometrically at 590 nm. The absorbance is a measure of the viability ofthe cells.

2.7. In vitro microparticle uptake

For the confocal microscopy experiments, DC2.4 were seeded in 8-well microscopy chambers (Nunc) at a density of 50,000 cells per well(containing 200 μL culture medium) and allowed to adhere overnight.Subsequently, OVA-Alexa Fluor488 loaded-microparticles (5 μL of a1 mg/ml microparticle suspension in demi water) were added and thecells were further cultured for 24 h. Finally, the cells were washedwith PBS and fixated with 4% paraformaldehyde. Prior to imaging thecellswere stainedwithHoechst andAlexa Fluor 647-conjugated choleratoxin subunit B.

For flow cytometry, DC2.4 cells were seeded in 24-well plates at adensity of 105 cells per well and allowed to adhere overnight. Subse-quently, OVA-Alexa Fluor488 loaded microparticles (10 μL of a 10 mg/ml suspension) were added and the wells were cultured for another

24 h. Finally, the cells were washed, detached from the wells andmeasured on a BD Accuri C6 flow cytometer. Data processing wasdone in FlowJo. Statistical analysis was done in GraphPad viaStudent's t-test.

2.8. In vitro antigen-presentation assay

Cell suspensions of OVA-specific CD8 T cells were prepared fromspleen and lymph nodes from OT-I mice. Single cell suspensions wereprepared, and CD8 T cells were isolated from the suspensions usingDynal mouse CD8 negative isolation kit (Invitrogen) according tothe manufacturer's instructions and subsequently labeled with CFSE(carboxyfluorescein diacetate succinimidyl ester). DCs obtained fromthe bone marrow of C57BL/6 mice were pulsed with serial dilutions ofthe respective samples for 24 h, washed, counted and subsequentlyco-cultured with OT-I T cells at different DC:T cell ratios for 48 h. After48 h, the division of the OT-I T cells was measured by flow cytometryusing a BD LSR II.

2.9. Readout of in vivo antibody response (ELISA)

The mice were vaccinated twice with a 3 week interval by subcuta-neous injection of 100 μL containing 20 μg of either soluble or encapsu-lated OVA. For the detection of anti-OVA antibodies, blood sampleswere collected from the ventral tail vein. Maxisorp (Nunc) plates werecoated with OVA (10 mg/ml) and incubated with serial dilutions ofserum. Antibody titers were subsequently detected with goat anti-mouse IgG1-HRP (Southern Biotech; HRP = horseradish peroxidase).The data show antibody titers of individual mice. Statistical analysiswas done in GraphPad via Student's t-test. The data are representativefor at least independent experiments.

2.10. Readout of in vivo cellular response (ELISPOT and ELISA)

The mice were vaccinated twice with a 3 week interval by subcu-taneous injection of 100 μL containing 20 μg of either soluble or en-capsulated OVA. Splenocytes were harvested 3 weeks after thebooster immunization. Suspensions of 2 × 105 splenocytes were cul-tured onto IFN-γ ELISPOT plates (Diaclone) in triplicate and restim-ulated with 5 μg/ml of either the OVA MHCI epitope peptideSIINFEKL or the OVA MHCII epitope peptide ISQAVHAAHAEINEAGR(both Anaspec). ELISPOTs were developed after a 24 h incubationperiod. The amount of IFN-γ in the supernatant of splenocytes incu-bated with the MHCI or MHCII epitope of OVA was determined usingthe mouse IFN-γ Ready-Set-Go!® ELISA kit (eBioscience) accordingto the manufacturer's instructions. Splenocyte supernatants wereincubated on the coated ELISA plates at a 1/2 dilution at 4 °C over-night. Statistical analysis was done in GraphPad via the Student's t-test. The data are representative for at least independentexperiments.

2.11. In vivo tissue response

OVA-Alexa Fluor488 loadedmicroparticleswere injected subcutane-ously in the flanks of mice. At different time intervals mice weresacrificed and the injection site was dissected. Tissue samples werefixed in 4% paraformaldehyde in PBS, dehydrated in ethanol, embeddedin paraffin and 5 μm sections were cut with a microtome. Afterdeparaffinization, sections were either mounted with DAPI-containingVectashieldmountingmedium (VectorLabs) or stainedwith hematoxy-lin and eosin and then mounted with Vectashield. The data are repre-sentative for at least independent experiments.

Table 1Influence of the mannitol to antigen ratio1 on the quality of the spray driedmicroparticulate formulations.

Mannitol:antigen ratio

Particlerecovery

Particleintegrity

Encapsulationefficiency

0:5 − ++ ++50:5 + ++ ++200:5 ++ ++ ++500:5 ++ − −1 Formulationswere spray dried on a 25 ml scale composed of respectively 20/25/5 mg

of DS/PLARG/OVA. The ratio is expressed as wt.%.

102 S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

3. Results and discussion

3.1. Formulation and process optimization

In a first series of experiments we aimed at exploring the effect offormulation and process conditions on themicroparticulate vaccine for-mulations. As depicted in Fig. 1, dextran sulfate (DS) was used as nega-tively charged polyelectrolyte (i.e. polyanion) and poly-L-arginine(PLARG)was used as positively charged polyelectrolyte (i.e. polycation).Ovalbumin (OVA) was used as model vaccine antigen, it has a Mw~43 kDa and an isoelectric point of 4.5 meaning that at neutral pH itwill bear an overall negative charge. The three major parameters thatreflect the quality of the microparticulate formulations are:

• Particle recovery: The amount of dry microparticles that is collected inthe recipient after the cyclone in the spray drier.

• Particle integrity: The ability of the obtained dry power to bereconstituted in aqueous medium into a monomodal suspension ofindividual non-aggregating and non-disintegrating microparticles.

• Encapsulation efficiency: The fraction of protein antigen that is, uponredispersion in aqueous medium, retained within the microparticlesand this is not released into the outer medium.

First, we varied the relative amount of mannitol that was added tothe solution prior to spray drying. Table 1 summarizes the ratios that

Fig. 2. (A)Microparticle recovery expressed as theweight percentage of themicroparticles collespray drying. (B) Scanning electronmicroscopy image of the spray dried microparticles before rmicroparticles redispersed in water. (B) and (C) were recorded from microparticles producedimages of epoxy-embedded and ultramicrotomed porous microparticles obtained after resusp

were used and the outcome of the spray drying experiments. Thelowermannitol to antigen ratios, or absence ofmannitol lead to poor re-covery due to sticking of the powder to the wall of the spray dryingequipment. This is in analogywith the commonuse ofmannitol in phar-maceutical technology to enhance the yield of spray drying [16], owingto its excellent flowing capability. However, too high mannitol to anti-gen ratios yield particles that do not retain their spherical morphologyupon redispersion in water and disassemble into smaller polyelectro-lyte coacervates and release a significant amount of antigen into the ex-ternal medium. These findings prompted us at using a mannitol/DS/PLARG/OVA ratio of 200/20/25/5 for further experiments. Interestingly,this ratio is very similar to the ratio of polyelectrolytes and antigen ap-plied to yield hollow Layer-by-Layer [17] capsules [18–21] where po-rous calcium carbonate microparticles are loaded with OVA andsubsequently coated with alternating layers of dextran sulfate andpoly-L-arginine followed by dissolution of the calcium carbonate coretemplates [22–27]. Evidently, in the present work the single calciumcarbonate pore former [28] (which creates capsules with a hollowvoid) is replaced by mannitol that creates a nanoporous internal struc-ture as earlier reported.

Second, we investigated the effect of various process conditions onparticle recovery. Using the above-mentioned composition of the for-mulation, spray drying was performed at different flow rates (i.e. theflow at which the liquid is fed to the nozzle) and at different pressuresof the atomizing air stream. The graph in Fig. 2A depicts particle recov-ery as function of the different processing parameters. Increasing theflow rate leads to a decrease in particle recovery whereas increasingthe air pressure affords a higher particle recovery. Based on these find-ings, experiments were continued using a flow rate of 1 ml/min and anair pressure of 0.75 bar. To verify that these formulation and processingconditions afforded the proper production of microparticles SEM wasused to characterize the microparticles after spray drying. As shown inFig. 2B, perfectly spherical shaped microparticles are obtained with asize below 10 μm. After redispersion in water, particle size (Fig. 2C),remained below 10 μm. TEM (Fig. 2D), recorded from epoxy embeddedand ultramicrotomed microparticles, revealed the highly porous

cted in the recipient after the cyclone to the original dry weight of the formulation prior toesuspension in aqueousmedium. (C) Size distributionmeasured by laser diffraction of theat optimal formulation and processing conditions. (D) Transmission electron microscopyension in aqueous medium.

Table 2Influence of the sequence of addition on the formulation properties.(1)

Formulation Sequence Ratio (wt.%) ζ-potential Encapsulation efficiency

1 DS/OVA/PLARG 20/5/25 42 ± 3 mV 110 ± 11%2 DS/PLARG/OVA 20/25/5 −30 ± 4 mV 110 ± 10%3 DS/PLARG (2) 20/25 40 ± 1 mV(3)

26 ± 1 mV(4)99 ± 0.1%

4 DS/PLARG(2) 25/20 −43 ± 1 mV(3)

−43 ± 1 mV(4)100 ± 0.2%

(1) Formulationswere spray dried on a 25 ml scale. The ratio is expressed aswt.%. All formulations containedmannitol at a ratio of 200.Measurementswere performed in three differentbatches.

(2) Soluble OVA was added after spray drying.(3) ζ-potential measured before soluble OVA was added.(4) ζ-potential measured after soluble OVA was added.

103S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

internal structure of the particles that is created upon dissolution of themannitol in aqueous medium.

3.2. Exploring the influence of the sequence of addition

Here we aimed to investigate whether the sequence of addition ofthe respective components prior to spray drying can affect the overallcharge of the particles, the encapsulation efficiency and spatial distribu-tion of the antigen and the immuno-biological properties of antigenloaded particles. To do so, microparticles were produced as indicatedin Table 2. Besides OVA loaded microparticles, we also preparedempty microparticles where different dextran sulfate to poly-L-arginine ratios were applied in order to test whether the surface chargeof the microparticles can be modulated. We observed that changing theratio of dextran sulfate to poly-L-arginine allows to shift the ζ-potentialfrompositive to negative depending onwhether the polyanion (i.e. dex-tran sulfate) or the polycation (i.e. poly-L-arginine) is in excess. These

Fig. 3. Fluorescencemicroscopy images of the different spray driedmicroparticle formulation ei(i.e. sOVA) to empty microparticles.

latter two microparticles without OVA will further in this work beused for control experiments where they will be mixed with solubleOVA.

Interestingly, the ζ-potential of the OVA-loaded microparticlesstrongly depended on the sequence in which the respective compo-nents are added prior to spray drying. Indeed, when OVA is added last,a negative ζ-potential value of−30mVwasmeasuredwhereas an out-spoken positive value of 42 mV was measured when poly-L-argininewas added last. These findings suggest that the component that wasadded last would be distributed more at the surface of the particles. Toconfirm this hypothesis, microparticles were spray dried using greenfluorescently (i.e. Alexa Fluor 488) labeled OVA, and subsequently im-aged by fluorescence microscopy. These images, shown in Fig. 3, con-firm that when OVA is added before PLARG (i.e. DS/OVA/PLARG(+)),microparticles are obtainedwith OVA homogenously distributedwithinthe whole volume of the microparticles. This is in agreement with ourprevious findings [15]. However, when OVA is added last (i.e. DS/

ther encapsulating OVA-Alexa Fluor488 orwith OVA-Alexa Fluor488 added in solube form

Fig. 4. (A) Confocal microscopy images of DC2.4 cells incubated with OVA-Alexa Fluor488 loaded microparticles. A1: DS/OVA/PLARG and A2: DS/PLARG/OVA. The cell membrane was stainedwith AlexaFluor647-conjugated cholera toxin subunit B and cell nuclei were stained with Hoechst. The images show an overlay with the DIC channel. (B) Quantification of in vitro uptake ofmicroparticles by DC2.4 cells, measured by flow cytometry. (C) Cell toxicity, measured by MTT assay, of the microparticles and their respective components (n = 6, *: p b 0.05).

Fig. 5. (A) Flow cytometry gating strategy to assess OT-I cell proliferation. (B) Flow cytometryhistograms of OT-I proliferation in response to co-culturingwithDCs pulsedwith solubleOVAor encapsulated OVA at different OVA concentration. The OT-I cell to DC ratio was 1:20. (C) Quantitative representation of OT-I cell division as shown in the gating strategy in panel (A).(D) Cytokine secretion measured by ELISA in the supernatant of the DC – T cell co-cultures.

104 S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

Fig. 5 (continued).

105S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

PLARG/OVA(−)), it becomes preferentially located at the periphery ofthe particles. Additionally, the OVA distribution does not appear to behomogenous and bright spots of complexed OVA appear on the surfaceof the particles.We hypothesize that this is due to the formation of elec-trostatic coacervate complexes betweenDS andPLARGprior to the addi-tion of OVA. Subsequently, OVA will adsorb onto these complexes (byelectrostatic interaction) and thus form a punctuated pattern in thefinal microparticles, rather than being homogenously distributed. Inall cases the produced particles (irrespective of the sequence of addi-tion) remained stable in aqueous medium for over a week (data notshown).

As empty particleswill be used further on in thiswork for control ex-periments, we also wanted to investigate the interaction between theseempty microparticles and soluble OVA. Therefore, we incubated boththe positively and the negatively charged microparticles with solubleAlexa Fluor488 – conjugated OVA (i.e. sOVA). Subsequently, the micro-particles were imaged by fluorescence microscopy and the amount ofunbound OVA was measured in the supernatant after centrifugation ofthe microparticles. In addition, also the ζ-potential after incubationwith OVA was measured. As listed in Table 2 and confirmed in Fig. 3,both negatively and positively charged microparticles are capable ofbinding soluble OVA. However, in both cases aggregationwas observed.This was most severe in the case of positively charged microparticlesthat were likely subjected to bridging flocculation upon the additionof negatively charged OVA. Additionally, in both cases OVA was pre-dominantly located at the periphery of the microparticles.

3.3. In vitro interaction with DCs

In a first series of in vitro cell culture experimentswe assessedwheth-er the ζ-potential of the particles influences the uptake by DCs and possi-ble cytotoxic effects. For these experiments we used themouse dendriticcell line DC2.4 [29]. The cells were incubated overnight with microparti-cles and the uptake was measured by flow cytometry. As shown inFig. 4B, both positively and negatively charged microparticles were effi-ciently internalized byDCswithout any significant differences dependingon the surface charge of themicroparticles. Confocalmicroscopy (Fig. 4A)was used to confirm that indeed themicroparticles were taken up by thecells and not just adhered to the cell surface. These experiments alsoshowed no influence of particle size on cell uptake as irrespective oftheir size, particles were massively internalized by the DC2.4 cells.Note that, such behavior has been reported extensively in literature foractively phagocyting in vitro cultured macrophage and dendritic celllines [3].

Subsequently, MTT assay was used to investigate in vitro cytotoxiceffects of the microparticles and there components on DCs. As shownin Fig. 4C, DS, mannitol and ovalbumin are non-toxic. Contrarily,PLARG significantly reduces cell viability over a broad concentrationrange. However, the electrostatic complexed microparticles, onlyinduce cell toxicity at elevated (i.e. 1 mg/ml) concentrations. This is inaccordance with the recent observations by our research group on theeffect of complexation on the in vivo mucosal irritation potentialof polyelectrolytes. Indeed, whereas soluble polyelectrolytes, bothpolyanions and evenmore severely polycations, do inducemucosal irri-tation, this was fully suppressed in the case of polyelectrolyte com-plexes [30].

3.4. In vitro T cell presentation and T cell expansion

To assess the potential of the microparticles to enhance cross-presentation of antigenwe performed an in vitroCD8 T cell presentationassay. For these experiments, bone marrow derived DCs were pulsedwith soluble ovalbumin (OVA) or the equivalent amounts of encapsu-lated OVA. Both positively and negatively charged particles were evalu-ated. Subsequently, the DCs were co-cultured with OT I cells, which aretransgenic CD8 T cells having a T cell receptor selectively recognizing

Fig. 6. H&E staining (top panels) and confocal fluorescence (bottom panels) microscopy images recorded from tissue sections collected from the injection spot. Confocal images werestained with Hoechst to visualize the cell nuclei in blue. Green fluorescence originates from OVA-Alexa Fluor488 that was encapsulated within the microparticles. The confocal imagesalso include the overlay with the DIC channel. The dashed contours in the day 14 panel indicate cells that have internalized microparticles. Symbols: neutrophils (→), eosinophils( ), intact particles (▲), mononuclear cells ( ), Fibroblasts ( ).

106 S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

theMHC I epitope of OVA (SIINFEKL). Prior to the co-culturing with theantigen pulsed DCs, OT I cells were stained with CFSE enabling the as-sessment of T cell division by flow cytometry.

Besides different concentrations of soluble and encapsulated OVA,also multiple DC to T cell ratios were evaluated. The gating strategy forflow cytometry analysis of the CD8 T cell proliferation is shown inFig. 5A. As shown by the flow cytometry histograms in Fig. 5B, solubleOVAonlymarginally induces T cell proliferation. Contrarily, encapsulatedOVA strongly promotes antigen presentation. For OVA concentrations of2 and 5 μg/ml, DS/OVA/PLARG and DS/PLARG/OVA microparticles areequally potent in inducing T cell division. However, at the low OVAdose of 0.2 μg/ml, solely DEXS/PLARG/OVA microparticles are able topromote T cell proliferation.

In addition tomeasuring T cell expansion, cytokine (i.e. IFNg, IL2, IL13,IL17) secretion in the supernatant of the DC–T cell co-cultures was mea-sured. As shown in Fig. 5C, cytokine secretion follows a similar trend asobserved for T cell proliferation, with DEXS/PLARG/OVA microparticlesbeing clearly more potent than DEXS/OVA/PLARG microparticles at thelower antigen concentrations. Strongly enhanced T cell expansion uponnano- andmicroparticulate formulation of antigenhas beendemonstrat-ed by other groups aswell for different types of particles [7–9]. Our pres-ent findings confirm that, although OVA remains firmly encapsulatedwithin the polyelectrolyte matrix of the particles upon reconstitution inaqueous medium, it can be efficiently processed into peptide fragmentsand presented onto the DC surface as MHCI complex. This supports ourhypothesis that, upon cellular uptake, the fully hydrated polyelectrolytematrix allows inwards diffusion of phagosomal/endosomal proteasesthat can directly access the entrapped OVA and/or release the OVAby degrading first the PLARG. The observation that microparticleswith OVA situated at their periphery perform slightly better than

microparticles with OVA more embedded in their interior might bedue to a higher availability of the OVA for processing upon cellular up-take, at least within the timeframe of our experimental set-up.

3.5. In vivo evaluation — tissue response

In a first series of in vivo experiments, we aimed at screening thetissue response upon subcutaneous injection of the microparticles.The mice were subcutaneously injected with DEXS/OVA/PLARGmi-croparticles and at different time intervals animals were sacrificedand the injection spot was dissected and analyzed by hematoxylinand eosin (H&E) staining and confocal fluorescence microscopy.For the latter microparticles loaded with (green fluorescent)OVA-AlexaFluor488 were used. Fig. 6 shows the correspondingmicrographs. At day 1 post infection, many polymorphonuclearcells – most of them neutrophils (→) but also significant numbersof eosinophils ( ) – can be seen surrounding and infiltratingthe injection spot, where large, intact spherical particles (▲) areeasily discernible. Three days post injection, infiltration of the mi-croparticle mass has further proceeded. Intact particles (▲) canstill be observed outside the cells. However, at the border of theinjection spot, some deformed particles appear to be inside phago-cytic mononuclear cells ( ), likely macrophages or newlyrecruited monocytes. Particle uptake continues over time, andat 2 weeks post injection, most of the particles – or particle rem-nants – are inside mononuclear phagocytic cells ( ). Neverthe-less, large, intact particles (▲) can still be observed on the confocalimages in Fig. 6. Fibroblasts ( ) can be seen surroundingthe injection spot. These data are largely similar to our earlier ob-servations, describing the mild, localized inflammatory response

Fig. 7. IgG1 titers after prime and boost injection. IFN-γ secreting CD8 and CD4 T cells in the spleen. IFN-γ cytokine secretions by splenic CD4 and CD8 T cells (n = 5, *: p b 0.05).

107S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

after injection or instillation of layer-by-layer (LbL) produced hol-low dextran-sulfate/poly-L-arginine microcapsules. Nevertheless,spray dried DS/pARG particles appear more resilient to cellular

uptake and degradation when compared to their LbL generatedcounterparts, a feature likely attributable to their slightly bigger di-mensions and their more rigid structure [25,27].

108 S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

3.6. In vivo evaluation — antigen specific immune response

The previous paragraph confirmed that antigen encapsulated in po-rous polyelectrolyte microparticles is highly efficiently cross-presentedby DCs to CD8 T cells in vitro. Additionally, in vivo low inflammatory re-sponses were observed upon injection of the microparticles. These en-couraging findings prompted us at further investigating the potentialof the porous polyelectrolyte microparticles at enhancing antigen-specific immune responses in vivo.

The mice (in cohort of 5) were immunized with a 10 μg dose ofeither soluble OVA or encapsulated OVA following a prime-boostscheme with a 3 week interval. As control groups, we immunizedmice with soluble OVA in PBS or with OVA mixed with the FDA ap-proved adjuvant Alum. The microparticle formulations used forthese experiments were those listed in Table 2, and for the sakeof clarity denoted in the following paragraphs and figures as:

1) DS/OVA/PLARG: positively chargedmicroparticles OVA-encapsulatingmicroparticles having the OVA embedded within themicroparticles.

2) DS/PLARG/OVA: negatively chargedmicroparticles OVA-encapsulatingmicroparticles having the OVA situated at the periphery of themicroparticles.

3) (+) DS/PLARG + sOVA: empty positively charged microparticles towhich soluble OVA was added.

4) (−) DS/PLARG + sOVA: empty negatively charged microparticles towhich soluble OVA was added.

Two weeks following the prime and boost immunization, blood wastaken from themice and antibody levels were determined in the serumby ELISA to quantify the humoral immune response. As shown in Fig. 7,encapsulated antigen outperforms soluble antigen and performs equal-ly well (or in the case of DS/OVA/PLARG microparticles slightly better)than soluble antigen formulated with alum. Finally, 3 weeks followingthe booster immunization, the mice were euthanized and theirsplenocytes were restimulated with either the MHCI or MHCII epitopeof OVA to quantify the cellular immune response. Subsequently, wemeasured IFN-γ cytokine secretion by splenic CD4 and CD8 T cells byELISA and the number of these T cells by ELISPOT. The humoral immuneresponsewas quantified bymeasuring anti-OVA IgG1 titers in serum viaELISA. From Fig. 7 it is clear that relative to soluble OVA encapsulatedOVA dramatically enhances cellular immunity. Also a strong enhance-ment relative to soluble OVA formulated with alum is observed. Thesefindings are in accordance with literature on nano- and micro-particulate antigen formulation that also reports on increasing antibodytiters and T cell responses relative to soluble antigen and antigen formu-lated with alum [7,8].

In between the different microparticle formulations the differencesare less outspoken. Positively charged empty microparticles with OVAadded afterwards in soluble form appears to be less potent while posi-tively charged OVA-loaded microparticles with OVA embedded, appearto be the most potent through the different assays. At this point it israther speculative to explain these observations. However, a major dif-ference between the overall best performing formulation (i.e. DS/OVA/PLARG) and the others is that the OVA is embedded within the interiorof the microparticles while in the case of the other formulations, theOVA is more situated at the periphery of the microparticles. The lattermight be beneficial to enhance in vitro antigen presentation, as in thiscase the antigen is readily available upon phagocytosis of themicropar-ticles. However, it might be less beneficial in vivo as antigen situated atthe periphery of themicroparticlesmight bemore prone to leaching outbefore cellular uptake. Such phenomenon can especially be expected totake place in a complex physiological environment such as the extracel-lularmedium. Furthermore, as depicted in Fig. 3, empty particles mixedwith soluble OVA are also prone to aggregation, which might reducetheir uptake by antigen presenting cells in vivo and thereby decreasethe amplitude of the evoked antigen-specific immune-response.

Furthermore, whereas a relatively large number of CD4 cells secretingIFNg were measured, the amount of IFNγ secreted was relatively lowin the DS/OVA/PLARG group. Exactly the opposite was observed forthe (−)DS/PLARG + sOVA group. Apparently, in the case of DS/OVA/PLARG a relatively high number of CD4 T cells is activated but they pro-duce little IFNgwhile in the case of (−)DS/PLARG+sOVA fewT cells getactivated, but those that do, produce lots of IFNg. The reason for this isnot yet fully understood, but these findings might reflect differencesin the transport of particles to the lymph nodes by DCs or in the amountof antigen that is available for processing. In future studies, we aim tounravel the reason behind this and to address the functional relevanceof these differences.

4. Conclusions

Summarizing, we have demonstrated in this work that porous poly-electrolyte microparticles are efficient in delivering antigen to DCs andthereby promoting antigen cross presentation to CD8 T cells in vitro. Invivo in mice we have demonstrated that upon subcutaneous injectiona mild tissue response is observed. Finally we analyzed the antigen spe-cific cellular and humoral immune response against amodel vaccine an-tigen. These experiments demonstrated that encapsulating antigen intomicroporous microparticles strongly enhanced serum antibody titersand splenic T cell responses. Taken together our findings have demon-strated the potential of this formulation technology for vaccine delivery.This technology is a simple one-step method that yields a dry powderthat hold potential for application where long-term storage undernon-refrigerated conditions is important, e.g. for pandemic vaccinesand vaccines intended for the developing world.

Acknowledgement

SDK, KF and BDG acknowledge the FWO-Flanders for funding.

References

[1] S. De Koker, B.N. Lambrecht, M.A. Willart, Y. van Kooyk, J. Grooten, C. Vervaet, J.P.Remon, B.G. De Geest, Designing polymeric particles for antigen delivery, Chem.Soc. Rev. 40 (2011) 320–339.

[2] J.A. Hubbell, S.N. Thomas, M.A. Swartz, Materials engineering for immunomodulation,Nature 462 (2009) 449–460.

[3] J.J. Moon, B. Huang, D.J. Irvine, Engineering nano- and microparticles to tune immu-nity, Adv. Mater. 24 (2012) 3724–3746.

[4] B. Pulendran, S. Kasturi, I. Skountzou, R. Albrecht, D. Koutsonanos, T. Hua, H. Nakaya,R. Ravindran, S. Stewart, M. Alam, M. Kwissa, F. Villinger, N. Murthy, J. Steel, J. Jacob,R. Hogan, A. Garcia-Sastre, R. Compans, Programming the magnitude and persis-tence of antibody responses with innate immunity, Nature 470 (2011) 543–547.

[5] J. Banchereau, F. Briere, C. Caux, J. Davoust, S. Lebecque, Y.T. Liu, B. Pulendran, K.Palucka, Immunobiology of dendritic cells, Annu. Rev. Immunol. 18 (2000)767–811.

[6] M. Houde, S. Bertholet, E. Gagnon, S. Brunet, G. Goyette, A. Laplante, M. Princiotta, P.Thibault, D. Sacks, M. Desjardins, Phagosomes are competent organelles for antigencross-presentation, Nature 425 (2003) 402–406.

[7] M. Amidi, E. Mastrobattista, W. Jiskoot, W.E. Hennink, Chitosan-based delivery sys-tems for protein therapeutics and antigens, Adv. Drug Deliv. Rev. 62 (2010) 59–82.

[8] H. Tamber, P. Johansen, H.P. Merkle, B. Gander, Formulation aspects of biodegrad-able polymeric microspheres for antigen delivery, Adv. Drug Deliv. Rev. 57 (2005)357–376.

[9] J. Frechet, K. Broaders, J. Cohen, T. Beaudette, E. Bachelder, Acetalated dextran is achemically and biologically tunable material for particulate immunotherapy, Proc.Natl. Acad. Sci. U. S. A. 106 (2009) 5497–5502.

[10] R. Rappuoli, Bridging the knowledge gaps in vaccine design, Nat. Biotechnol. 25(2007) 1361–1366.

[11] G. Parmiani, C. Castelli, P. Dalerba, R. Mortarini, L. Rivoltini, F. Marincola, A. Anichini,Cancer immunotherapy with peptide-based vaccines: what have we achieved?Where are we going? J. Natl. Cancer Inst. 94 (2002) 805–818.

[12] S. Setia, H. Mainzer, M.L. Washington, G. Coil, R. Synder, B.G. Weniger, Frequencyand causes of vaccine wastage, Vaccine 20 (2002) 1148–1156.

[13] (a) B. Devriendt, K. Baert, M. Dierendonck, H. Favoreel, S. De Koker, J.P. Remon, B.G.De Geest, E. Cox, One-step spray-dried polyelectrolyte microparticles enhancethe antigen cross-presentation capacity of porcine dendritic cells, Eur. J.Pharm. Biopharm. 84 (2013) 421–429;

(b) R. De Smet, S. Verschuere, L. Allais, G. Leclercq, M. Dierendonck, B.G. De Geest, I.Vqann Driessche, T. De Moor, C.A. Cuvelier, Spray-dried polyelectrolyte

109S. De Koker et al. / Journal of Controlled Release 195 (2014) 99–109

microparticles in oral antigen delivery: stability, biocompatibility and cellularuptake, Biomacromolecules 15 (2014) 2301–2309.

(c) US 2013/0149357.[14] M. Dierendonck, S. De Koker, C. Cuvelier, J. Grooten, C. Vervaet, J.P. Remon, B.G. De

Geest, Facile two-step synthesis of porous antigen-loaded degradable polyelectro-lyte microspheres, Angew. Chem. Int. Ed. 49 (2010) 8620–8624.

[15] M. Dierendonck, S. De Koker, R. De Rycke, P. Bogaert, J. Grooten, C. Vervaet, J.P.Remon, B.G. De Geest, Single-step formation of degradable intracellular biomoleculemicroreactors, ACS Nano 5 (2011) 6886–6893.

[16] Y. Gonnissen, J.P. Remon, C. Vervaet, Development of directly compressible powdersvia co-spray drying, Eur. J. Pharm. Biopharm. 67 (2007) 220–226.

[17] G. Decher, Fuzzy nanoassemblies: toward layered polymeric multicomposites, Sci-ence 277 (1997) 1232–1237.

[18] F. Caruso, R.A. Caruso, H.Mohwald, Nanoengineering of inorganic and hybrid hollowspheres by colloidal templating, Science 282 (1998) 1111–1114.

[19] S. De Koker, R. Hoogenboom, B.G. De Geest, Polymeric multilayer capsules for drugdelivery, Chem. Soc. Rev. 41 (2012) 2867–2884.

[20] E. Donath, G.B. Sukhorukov, F. Caruso, S.A. Davis, H. Mohwald, Novel hollow poly-mer shells by colloid-templated assembly of polyelectrolytes, Angew. Chem. Int.Ed. 37 (1998) 2202–2205.

[21] Y. Yan, G.K. Such, A.P.R. Johnston, H. Lomas, F. Caruso, Toward therapeutic deliverywith layer-by-layer engineered particles, ACS Nano 5 (2011) 4252–4257.

[22] B.G. De Geest, R.E. Vandenbroucke, A.M. Guenther, G.B. Sukhorukov, W.E. Hennink,N.N. Sanders, J. Demeester, S.C. De Smedt, Intracellularly degradable polyelectrolytemicrocapsules, Adv. Mater. 18 (2006) 1005–1009.

[23] B.G. De Geest, M.A.Willart, H. Hammad, B.N. Lambrecht, C. Pollard, P. Bogaert, M. DeFilette, X. Saelens, C. Vervaet, J.P. Remon, J. Grooten, S. De Koker, Polymeric

multilayer capsule-mediated vaccination induces protective immunity against can-cer and viral infection, ACS Nano 6 (2012) 2136–2149.

[24] B.G. De Geest, M.A. Willart, B.N. Lambrecht, C. Pollard, C. Vervaet, J.P. Remon, J.Grooten, S. De Koker, Surface engineered polyelectrolyte multilayer capsules— syn-thetic vaccines mimicking microbial structure and function, Angew. Chem. Int. Ed.51 (2012) 3862–3866.

[25] S. De Koker, B.G. De Geest, C. Cuvelier, L. Ferdinande, W. Deckers, W.E. Hennink, S.De Smedt, N. Mertens, In vivo cellular uptake, degradation, and biocompatibility ofpolyelectrolyte microcapsules, Adv. Funct. Mater. 17 (2007) 3754–3763.

[26] S. De Koker, B.G. De Geest, S.K. Singh, R. De Rycke, T. Naessens, Y. Van Kooyk, J.Demeester, S.C. De Smedt, J. Grooten, Polyelectrolyte microcapsules as antigen de-livery vehicles to dendritic cells: uptake, processing, and cross-presentation of en-capsulated antigens, Angew. Chem. Int. Ed. 48 (2009) 8485–8489.

[27] S. De Koker, T. Naessens, B.G. De Geest, P. Bogaert, J. Demeester, S. De Smedt, J.Grooten, Biodegradable polyelectrolyte microcapsules: antigen delivery tools withTh17 skewing activity after pulmonary delivery, J. Immunol. 184 (2010) 203–211.

[28] D.V. Volodkin, N.I. Larionova, G.B. Sukhorukov, Protein encapsulation via porousCaCO3 microparticles templating, Biomacromolecules 5 (2004) 1962–1972.

[29] Z. Shen, G. Reznikoff, G. Dranoff, K.L. Rock, Cloned dendritic cells can present exog-enous antigens on both MHC class I and class II molecules, J. Immunol. 158 (1997)2723–2730.

[30] L.J. De Cock, J. Lenoir, S. De Koker, V. Vermeersch, A.G. Skirtach, P. Dubruel, E.Adriaens, C. Vervaet, J.P. Remon, B.G. De Geest, Mucosal irritation potential of poly-electrolyte multilayer capsules, Biomaterials 32 (2011) 1967–1977.