Photosensitizer and Light Pave the Way for Cytosolic Targeting and Generation of Cytosolic CD8 T Cells Using PLGA Vaccine Particles

The CD8 CTLs play a central role in fighting diseases caused by tumor and intracellular pathogens, and huge research efforts have been made to stimulate and to engage such cells for disease management. However, the generation of CTLs is MHC class I (MHCI) restricted, whereas vaccine Ags typically enter into the MHC class II (MHCII) pathway of Ag presentation for stimulation of CD4 T cells and eventual production of Abs (1). Therefore, a great scientific challenge is to target vaccine to the MHCI pathway of Ag presentation, which would require Ag delivery into the cytosol of APCs, especially dendritic cells (DCs). Cytosolic delivery can principally be achieved by Ag translocation through the cell membrane or by Ag escape from endosome or phagosomes.

Several approaches have been proposed for the delivery of Ags and drugs to the cytosol, such as cell-penetrating peptides that carry their cargo across the plasma membrane (2, 3), pH-sensitive and fusogenic liposomes that break up phagosomes (4–6), micelle-based immune-stimulating complexes that may facilitate Ag cross-presentation (7, 8), and recombinantly modified viruses (9–11) and bacteria (12–15) that trigger CTLs. Also, microparticles or nanoparticles made from biodegradable polymers, such as poly(lactide-co-glycolide) (PLGA), have been investigated for the cytosolic delivery of vaccines. PLGA particles may enable controlled and sustained Ag release (16–18) and can, as particles in general, deliver large amounts of Ag to APCs at the single-cell level, thereby increasing the probability of reaching MHCI (19–23).

We recently suggested a photochemical approach for cytosolic targeting of Ags. By using a photosensitizer and light, soluble protein could be translocated to the cytosol of APCs in vitro (24) and in vivo (25, 26). This so-called “photochemical internalization” (PCI) represents a novel technology for controlled permeabilization of endosomes (27). PCI is based on photosensitizers that translocate from the plasma membrane to the endosomal membrane after endocytosis (Fig. 7). Upon subsequent light activation, the photosensitizer generates free radicals and reactive oxygen species, which rupture the endosomes and release the endocytosed material, such as Ags and adjuvants, into the cytosol. In the cytosol, Ags can enter the MHCI presentation for stimulation of CD8 T cell responses and CTLs (28). In this study, we combined PCI technology with PLGA-based vaccine particles. The microparticles, loaded with an Ag and a photosensitizer, were tested in a mouse model for their capacity to trigger Ag-specific CD8 T cell responses. The hypothesized benefit of vaccine particles over that of soluble vaccines, apart from their better recognition by the APCs, is that both Ag and photosensitizer are jointly targeted to the APCs. This is important because the effect of PCI-based vaccines assumes that photosensitizer and Ag are taken up by the same individual cell that is subsequently illuminated for cytosolic delivery of the Ag. To our knowledge, this is the first study to demonstrate that PLGA vaccine particles can be efficiently used to target Ag to the cytosol for MHCI-restricted and strong stimulation of cytotoxic CD8 T cells that prevent tumor growth.

PLGA 50:50 (Resomer RG503) was obtained from Evonik Industries. OVA, dichloromethane, DMSO, and chlorhexidine diacetate were obtained from Sigma-Aldrich (Buchs, Switzerland). Poly(vinyl alcohol) (PVA; Mowiol 8-88) was from Kuraray Europe (Hattersheim, Germany). The photosensitizer tetraphenyl chlorine disulfonate (TPCS2a; Amphinex) was kindly provided by PCI Biotech (Lysaker, Norway).

For immunization, female C57BL/6 mice were purchased from Harlan (Horst, The Netherlands) and used at 6–10 wk of age. Rag2-deficient OT-I mice, transgenic for the TCR that recognizes the MHCI H-2K^b epitope OVA257–264 (SIINFEKL), were originally purchased from Taconic Europe (Ry, Denmark) and bred in our facilities at the University of Zurich. All mice were kept under specific pathogen–free conditions, and the procedures performed were approved by the veterinary authorities of the canton of Zurich (license 69/2012). The light source used for the activation of TPCS2a was LumiSource (PCI Biotech). The lamp contains four 18-W Osram L18/67 light tubes with a fluence rate of 13.5 mW/cm² that peaks at 435 nm and with an energy of 0.81 J/cm². The octapeptide SIINFEKL was purchased from EMC Microcollections (Tübingen, Germany).

Microparticles were prepared by solvent evaporation using a water-in-oil-in-water (W/O/W) double-emulsion method (29). Briefly, 500 μl of an aqueous solution (W1) of OVA or OVA and TPCS2a was homogenized with a POLYTRON PT 6000 (Kinematica, Luzern, Switzerland), equipped with a 12-mm probe, at 6000 rpm for 1 min in 5 ml 6% (w/v) PLGA solution in dichloromethane. The obtained W/O emulsion was added to 20 ml a 1% (w/v) solution of PVA in purified water (W2), and the mixture was homogenized again (20 mm probe, 14000 rpm, 5 min) on ice. The resulting W/O/W emulsion was stirred with a magnetic stirrer (200 rpm) at room temperature for 5 h to evaporate dichloromethane. After solvent evaporation, the particles were washed twice with water to remove nonencapsulated material. The resulting mixture was supplemented with sucrose (1.5%), frozen in liquid nitrogen, and freeze-dried at 0.1 mbar for 48 h in an ALPHA 2-4 LSC lyophilizer (Martin Christ, Osterode am Harz, Germany).

For the encapsulation of OVA, W1 was composed of 0.8% (w/v) OVA and 1% (w/v) PVA, yielding a theoretical OVA loading of 1.4% (w/w) in the dried PLGA microparticles. For the encapsulation of TPCS2a, W1 was composed of 4% (w/v) TPCS2a and 2.5% (w/v) chlorhexidine diacetate, yielding a theoretical TPCS2a loading of 6% (w/w) in final microparticles. For the coencapsulation of OVA and TPCS2a, W1 was composed of 0.8% (w/v) OVA, 4% (w/v) TPCS2a, and 2.5% (w/v) chlorhexidine diacetate, yielding theoretical loadings of 1.4% (w/w) OVA and 6% (w/w) TPCS2a in the PLGA microparticles.

PLGA microparticles were diluted 1:100 in purified water and analyzed for size, size distribution, and ζ potential (Delsa Nano C; Beckman Coulter). For quantification of OVA and TPCS2a contents in the microparticles, 300 μl microparticle slurry (prior to freeze drying) was dissolved in DMSO/water (9:1, v/v) for OVA and TPCS2a extraction. Subsequently, OVA was assayed by Fluoraldehyde OPA Reagent Solution (Thermo Scientific Pierce, Reinach, Switzerland), according to the manufacturer’s specifications, and TPCS2a was assayed fluorimetrically using λ_ex of 424 nm and λ_em of 658 nm (Infinite M200 PRO; Tecan). When OVA and TPCS2a were coencapsulated, TPCS2a was assayed as described, whereas OVA was separated from TPCS2a, because of fluorescence interference, by precipitation with ProteoExtract Protein Precipitation Kit (Calbiochem, purchased from Impexron, Pfullingen, Germany). The protein was pellet suspended in 100 μl PBS (pH 7.4), sonicated for 1 min, and assayed by bicinchoninic acid assay (Micro BCA kit; Thermo Scientific Pierce). OVA and TPCS2a encapsulation efficiencies (%) were calculated as (measured compound concentration/theoretical compound concentration) × 100.

Unless otherwise described, 2 × 10⁶ purified and RBC-free spleen cells from female Rag2/OT-I mice were adoptively transferred by i.v. injection into recipient female C57BL/6 mice 1 d prior to immunization. Freeze-dried PLGA microparticles were carefully reconstituted in purified water on the day of immunization; formulations were kept light protected and used within 10 min of preparation. Mice were immunized intradermally in the abdominal region. Two injections (50 μl each) were given, and 25G needles were used. OVA was tested at 10–30 μg/dose, which corresponds to TPCS2a doses of 70–210 μg for the combined OVA and TPCS2a particles. Eighteen hours after immunization, the mice were anesthetized i.p. with ketamine (25 mg/kg body weight) and xylazine (4 mg/kg) and placed for 6 min on the light source (4.86 J/cm²). Blood was harvested at various time points for analysis of Ag-specific T cells by flow cytometry and for preparation of serum for later analysis of Abs. At the end of the experiments, mice were euthanized, and heart blood was isolated for analysis of anti-OVA Igs by ELISA, whereas the splenocytes were analyzed ex vivo for Ag-specific T cell responses by flow cytometry and ELISA.

The proliferation and activation status of OVA-specific CD8 T cells in blood or spleen cells was monitored by staining the cells with PE-labeled H-2K^b–SIINFEKL Pro5 pentamer (ProImmune, Oxford, U.K.) and anti-CD8 and anti-CD44 fluorescent Abs for analysis by flow cytometry, as previously described (25). Intracellular cytokine staining was performed after stimulation of the cells with 5 μg/ml SIINFEKL (6 h) and 5 μg/ml brefeldin A (last 4 h). The cells were fixed and permeabilized using BD Cytofix/Cytoperm (BD Biosciences, Basel, Switzerland), according to the manufacturer’s instructions, and stained with anti–IFN-γ and anti–TNF-α for 35 min. All Abs were from eBioscience (Vienna, Austria). The stained cells were acquired using a FACSCanto (BD Biosciences, San Jose, CA) and analyzed using FlowJo 8.5.2 software (TreeStar, Ashland, OR). H-2Kb/SIINFEKL Pro5 pentamer–staining cells were gated on lymphocytes based on forward and side scatter properties, CD8, and CD44. Cytokine-producing cells were gated on lymphocytes based on forward and side scatter properties and on CD8.

For analysis of cytokine secretion by ELISA, 2 × 10⁵ splenocytes were restimulated in round-bottom 96-well plates with 0.1 μg/ml SIINFEKL, and supernatants were collected and analyzed for IL-2 (24 h) or IFN-γ, TNF-α, and granzyme B (96 h) secretion by ELISA (Ready-Set-GO! kits; eBioscience).

For analysis of OVA-specific Abs in blood serum, MaxiSorp ELISA plates were coated with 2 μg/ml OVA. IgG1 and IgG2c were detected with biotin-conjugated rat anti-mouse Abs (BD Pharmingen, San Diego, CA). The plates were developed with streptavidin-conjugated HRP and TMB substrate (eBioscience). The OD at a given serum dilution was measured.

In vivo cytotoxicity was analyzed on CFSE-labeled and SIINFEKL-pulsed naive spleen cells, as described (25). Briefly, mice adoptively transferred with 2 × 10⁶ OT-I cells were immunized with PLGA-OVA or PLGA-OVA-TPCS2a or were left untreated; light treatment was done as described above. After 6 wk, 5 × 10⁶ CFSE^hi and SIINFEKL-pulsed and 5 × 10⁶ CFSE^lo and nonpulsed spleen cells were injected i.v. After 18 h, the mice were bled, and the cells were analyzed by flow cytometry for CFSE. Specific lysis or cytotoxicity was calculated for the cells using the following formula: percentage specific cytotoxicity = 100 − [100 × (CFSE^hi/CFSE^lo)].

Three weeks after the injection of CFSE-labeled cells, the same mice were challenged with 500,000 B16-OVA melanoma cells by intradermal injection into one of the flanks. A caliper was used to measure the size of the tumor, and the tumor volume was calculated as (length × width²)/2.

The two-sided Mann–Whitney U test was applied for nonparametric data analysis from two test groups. When three or more groups were compared, the Kruskal–Wallis test with the Dunn post hoc test was used. Differences in tumor size were analyzed using the Wilcoxon signed-rank test. The significance level was set to 95%. All calculations were done using the GraphPad Prism 5.04 software (Prism, La Jolla, CA).

PLGA microparticles were prepared with a mean size of 1 μm and a size range of 0.4 –1.3 μm (data not shown). Upon lyophilization and reconstitution, the particle size range increased to 2–3.5 μm. The ζ-potential of the PLGA microparticles was in the range of −20 to −12 mV. The resulting microparticles contained 9.87 μg OVA/mg dry particles in the TPCS2a-free vaccine formulation and 8.30 μg/mg in the TPCS2a-containing vaccine; the loading efficiencies relative to the targeted loading were 69 and 58%, respectively. SDS-PAGE analysis of OVA extracted from microparticles demonstrated that the integrity of the protein was preserved during encapsulation and lyophilization (data not shown). An emulsion stabilizer was required for efficient encapsulation of TPCS2a. The addition of chlorhexidine diacetate in the inner aqueous phase of the W/O/W emulsion resulted in significantly higher TPCS2a encapsulation compared with other stabilizers tested (data not shown).

To facilitate analysis of MHCI Ag presentation, 2 × 10⁶ Rag2/OT-I lymphocytes were adoptively transferred to wild-type C57BL/6 mice 1 d prior to immunization with PLGA-based vaccines. The particles were loaded with OVA (PLGA-OVA) or with OVA and TPCS2a (PLGA-OVA-TPCS2a). Eighteen hours later, the mice were anesthetized and illuminated. Two weeks later, the effect of immunization on the proliferation of splenocytes was measured ex vivo by pentamer staining and flow cytometry and expressed as the percentage of pentamer-staining cells among the total CD8 lymphocyte population (Fig. 1A). Significantly higher numbers of SIINFEKL-specific CD8 T cells were observed in mice immunized with PLGA-OVA-TPCS2a than with PLGA-OVA (Fig. 1B, p = 0.0006, Mann–Whitney test), with the average frequencies being 12.6 and 3.4%, respectively, and 0.9% for untreated mice. Seven weeks postimmunization, a reduction in the frequency of SIINFEKL-specific CD8 T cells was observed with the TPCS2a-free vaccine, whereas the frequency increased further after immunization with PLGA particles containing both OVA and TPCS2a. The frequencies of SIINFEKL-specific CD8 T cells were ∼20.0% for PLGA-OVA-TPCS2a, 2.1% for PLGA-OVA, and 0.3% for untreated mice (Fig. 1C).

The splenocytes were restimulated in vitro with SIINFEKL and analyzed by flow cytometry for surface expression of CD8 and the effector/memory marker CD44, as well as for intracellular production of IFN-γ and TNF-α (Fig. 1D). Immunization with both PLGA-OVA and PLGA-OVA-TPCS2a resulted in activation of CD8 T cells with a CD44 phenotype, which produced both IFN-γ and TNF-α. However, the frequencies of IFN-γ– and TNF-α–producing CD8⁺ CD44⁺ cells were significantly higher after immunization with the TPCS2a-containing particles than with the TPCS2a-free particles at 2 wk (Fig. 1E, 5.3 versus 1.6%, p = 0.0033) and 7 wk (Fig. 1F, 9.9 versus 1.1%, p = 0.0006) after immunization. Hence, increases in both proliferation (Fig. 1B, 1C) and cytokine production (Fig. 1E, 1F) from weeks 2 to 7 indicate a persistent Ag presentation and immune stimulation by the TPCS2a-containing PLGA microparticles. Also, when no adoptive transfer with OT-I cells was done, a strong benefit of TPCS2a-containing PLGA-OVA vaccines was observed. The frequencies of pentamer-binding (Fig. 1G) and IFN-γ–producing (Fig. 1H) cells were much lower than after immunization of OT-I transferred mice (Fig. 1B, 1E), but they were significantly higher in mice immunized with the photosensitizer-containing particle vaccine TPCS2a than in mice that were not immunized with TPCS2a.

The effect of photosensitization on the immunogenicity of protein-containing PLGA microparticles also was studied by measuring cytokine secretion in vitro after restimulation of splenocytes with SIINFEKL for CD8 T cell–specific immune responses. SIINFEKL-specific cytokine secretion was enhanced in mice immunized with PLGA-OVA-TPCS2a compared with mice that received PLGA-OVA (p < 0.01). Seven weeks after immunization, the secretion of IL-2 (Fig. 2A, p = 0.0006), IFN-γ (Fig. 2B, p = 0.0006), and TNF-α (Fig. 2C, p = 0.0012) was significantly higher in splenocytes from mice immunized with PLGA-OVA-TPCS2a than from mice immunized with PLGA-OVA. Nonetheless, PLGA-OVA microparticles stimulated significantly stronger secretion of IFN-γ and TNF-α (both p = 0.0167), but not IL-2 (p = 0.116), compared with nonimmunized mice.

To further test the effector function of the stimulated CD8 T cells, cytotoxicity was analyzed. First, the release of the cytotoxic mediator granzyme B was analyzed in the supernatants of cultured and SIINFEKL-restimulated splenocytes from immunized mice. Although immunization with PLGA-OVA caused a small increase in granzyme B production (Fig. 2D, p = 0.0167 compared with untreated mice), a very strong secretion of granzyme B was observed after immunization with concomitant photosensitization. The effect was significantly stronger than after immunization without TPCS2a (p = 0.0021). This effector function was further confirmed in vivo using an in vivo cytotoxicity assay with CFSE-labeled targets (Fig. 3A, 3B). None of the nonimmunized mice showed SIINFEKL-specific cytotoxicity (mean 6.2%), whereas all seven mice immunized with PLGA-OVA-TPCS2a exhibited specific killing of target cells (mean, 98.0%; range: 93.5–99.5%). Five of seven mice immunized with PLGA-OVA showed specific killing (mean, 60.7%; range: 5.1–92.1%), but the effect was significantly weaker than after immunization with photosensitization (p = 0.0021).

The mice used in the in vivo cytotoxicity assay (Fig. 3A) were challenged 3 wk later with OVA-expressing B16 mouse melanoma cells and monitored for growth of tumor. The earliest onset of tumor was observed after 11 d for nonimmunized mice (data not shown). On day 16, six of eight nonimmunized and five of seven mice immunized with PLGA-OVA had solid tumors (Fig. 3C). None of the mice immunized with particulate Ag and photosensitizer had tumors.

Exogenous Ags, such as vaccines, are primarily presented via MHCII molecules and stimulate CD4 T cell–dependent Ab responses. To investigate the potential of PCI to deliver the Ag to cell cytosol, thus redirecting the Ag from the default MHCII pathway, mice were immunized with PLGA-OVA or PLGA-OVA-TPCS2a, and blood was collected for analysis of OVA-specific Abs. Immunization with PLGA-OVA particles stimulated high titers of anti-OVA IgG1, but an Ab response was not detected in mice that had received TPCS2a-containing PLGA microparticles (Fig. 4). Comparable results were obtained when IgG2 subclasses were measured (data not shown). Indirectly, this result suggests that the PCI-mediated cytosolic and MHCI targeting were so effective that the Ag did not reach the MHCII pathway of Ag presentation.

The adjuvant effect of the photosensitizer on the microencapsulated Ag is assumed to be light dependent. To test this hypothesis or to test any intrinsic and light-independent adjuvant effects of TPCS2a, mice were immunized with TPCS2a-containing PLGA microparticles loaded with 10 or 30 μg OVA and tested again for stimulation of SIINFEKL-specific CD8 T cell responses as a function of light activation. In mice that received 10 μg OVA and 70 μg TPCS2a in PLGA microparticles, but were not subsequently light treated, a baseline level of CD8 T cell proliferation was detected, with an average frequency of 0.3% SIINFEKL-specific CD8 T cells in the spleen and as assessed 2 wk after immunization (Fig. 5A). When light was applied to the same vaccine, a significantly increased proliferation was observed (mean, 2.1%, p = 0.0079, Mann–Whitney test, n = 5). The adjuvant effect of light-activated TPCS2a also was evident at the higher doses of 30 μg OVA and 210 μg TPCS2a, with the average frequency of specific CD8 T cells increasing from ∼2.0 to 9.0% upon exposure to light (p = 0.0079, n = 5). Intracellular staining for secretion of IFN-γ and TNF-α revealed that immunization with TPCS2a- and OVA-containing PLGA microparticles elicited double-producing cells and that the degree of elicitation was dependent on light activation of TPCS2a, with the light causing a 5–10-fold increase in the frequencies of cytokine-producing CD8 T cells (Fig. 5B). Restimulation of splenocytes with SIINFEKL and analysis of granzyme B secretion by ELISA confirmed that light activation of the photosensitizer was needed to obtain adjuvant effects (Fig. 5C). Also, the increase in Ag-specific secretion of IL-2, IFN-γ, and TNF-α was dependent upon light irradiation of the photosensitizer (Fig. 5D–F).

Because the hypothesized basis of PCI-mediated immunization is that light activation of photosensitizers in the endosomes causes these to get leaky and to release their content of Ag, it is important that both photosensitizer and Ags are taken up by the same cells. To test this, PLGA-OVA was compared with the same vaccine mixed with soluble TPCS2a, with TPCS2a and OVA contained in separate particles (PLGA-OVA mixed with PLGA-TPCS2a), and, finally, with TPCS2a and OVA contained in the same particles (PLGA-OVA-TPCS2a). The OVA and TPCS2a doses were kept constant. Only when OVA and TPCS2a were coencapsulated were significant Ag-specific proliferation and cytokine secretion observed (Fig. 6).

The majority of vaccines fail to generate strong CD8 T cell responses because exogenous proteins are mainly processed in lysosomes and presented by the MHCII molecules to stimulate proliferation and differentiation of CD4 Th cells, which again can trigger B cells for production of Abs. However, malignancies and infections by intracellular pathogens are better controlled by cytotoxic CD8 T cell responses (30–33). Therefore, strategies to stimulate CTLs are central in modern vaccine development.

For more than two decades, the induction of CTLs by biodegradable PLGA microparticles has been investigated (17, 21, 34–37). However, despite the fact that the particles are easily recognized by APCs (38, 39), that PLGA itself is immunologically inert (i.e., not recognized by specific Abs and eliminated by phagocytes), and that the particles can contain a high load of Ag that would facilitate CTLs and Th1-like immune responses, PLGA microparticles have not lived up to the initial expectations with regard to CD8 T cell activation. The problem is that particles typically remain contained in the APC phagosomes and are consequentially shuffled to the lysosomes for loading on MHCII molecules. If phagosomal maturation could be blocked and the Ag released into cytosol, it could enable correct processing of antigenic peptides by proteasomes and loading of peptides on MHCI for CTL induction. In this regard, it was shown that photosensitizers can mediate cytosolic delivery of chemotherapeutic drugs (28) and, more recently, of soluble Ags (25, 26). Briefly, the photosensitizer binds to the plasma membrane of APCs, and upon endocytosis of extracellular photosensitizer and Ag, the photosensitizer is translocated to the endosomal or phagosomal membranes. Upon subsequent light activation of the photosensitizer, the endosomal membranes are disrupted, and the endosomal content is released into the cytosol as illustrated in Fig. 7.

The experiments described in this study were designed to exploit the potential of Ag-containing PLGA microparticles combined with a photosensitizer (TPCS2a) to trigger cytosolic Ag delivery and generation of CTLs. Immunization of mice with PLGA microparticles loaded with both Ag and photosensitizer strongly facilitated CD8 T cell responses compared with immunization with PLGA particles containing Ag only. Proliferation of Ag-specific CD8 T cells in vivo and their function in vivo, ex vivo, and in vitro were strengthened by using the photosensitizer TPCS2a. Compared with immunization with soluble OVA and TPCS2a (25) or with in vitro–generated DCs loaded with OVA and TPCS2a (26), immunization with the PLGA-based vaccine produced much stronger proliferation of Ag-specific CD8 T cells; the beneficial immunological effect of particulate Ag–delivery systems over soluble Ags was shown earlier for formulations without photosensitizers (40, 41). In the current study, as much as 20–30% of the CD8 T cells in the spleen were specific for the immunodominant SIINFEKL epitope, whereas ∼2–10% were obtained after immunization with soluble OVA (25, 26) or after autologous immunization with OVA- and PCI-treated DCs (24). Moreover, although the soluble Ag and photosensitizer produced a proliferation peak ∼10–12 d after immunization, after which the frequency of specific CD8 T cells decreased to ∼10–20% of the peak response within 2–4 wk, the PLGA-based and TPCS2a-containing vaccine, but not the TPCS2a-free vaccine, provided a sustained immune stimulation. The frequency of Ag-specific CD8 T cells increased over ≥7 wk reaching as much as 30–35% of all CD8 T cells.

Compared with our previous reports on soluble OVA and TPCS2a (25, 26), no additive effect was observed when combining soluble TPCS2a with particulate OVA in PLGA particles or when combining TPCS2a-containing particles with OVA-containing particles. At first glance, this may appear to be a contradiction. However, when both Ag and photosensitizer are administered as soluble compounds, their local biodistribution in the skin are expected to be similar or at least comparable, whereas when the Ag is contained in PLGA particles and the TPCS2a is given as a solute or contained in different particles, the biodistribution and migratory properties are expected to be less similar and comparable. Hence, the probability that the two compounds reach the same APC is lower when both are given as solutes than when both are contained in the same particles. Moreover, we observed that soluble OVA and TPCS2a can form colloidal complexes in vitro (B. Gander, unpublished observations). TPCS2a has two anionic sulfonate substituents, and it can undergo several protonation steps due to its imino and pyrrole nitrogens (42). As a consequence, such functional groups can react with amino acid side chains and end groups, thereby allowing TPCS2a–protein complexes that may function as particles and, thereby, facilitate concerted targeting to individual APCs.

Part of the increased proliferation of Ag-specific cells may be explained by the very strong effect on IL-2 secretion, a growth factor of T cells. In vitro IL-2 secretion of spleen cells from mice immunized with TPCS2a-containing PLGA particles produced ∼10-fold higher culture concentrations than did spleen cells from mice immunized with TPCS2a-free particles. Moreover, although immunization with soluble OVA and TPCS2a resulted in cells producing IFN-γ, a small fraction of which also were producing TNF-α (25), all Ag-specific CD8 T cells were IFN-γ and TNF-α double producers upon immunization with the PLGA-based vaccine. Finally, in contrast to soluble OVA and TPCS2a, particulate OVA-TPCS2a stimulated granzyme B production in CD8 T cells; when infected target cells or Ag-expressing tumor cells are recognized by CTLs, granzyme B is secreted into the target cells along with the pore-forming perforin to mediate cell death. The potential effector function of this granzyme B was tested in an in vivo cytotoxicity test: all mice immunized with PLGA-OVA-TPCS2a produced strong positive killing of Ag-loaded target cells, whereas this function was impaired in mice immunized with particles that did not contain the photosensitizer. Hence, the quantity of potential effector cells, as well as their effector quality, was found to be improved after immunization with photosensitizer-containing PLGA particles.

Interestingly, medium-control cultures of splenocytes from TPCS2a-immunized mice also showed significant production of IFN-γ (data not shown). The secretion was orders of magnitude lower than that after restimulation of the cells with Ag, but it suggests that 7 wk after the immunization, the spleen still contained APCs with Ag-containing PLGA particles that caused in vitro restimulation of the in vivo–primed T cells.

We recently suggested that the mechanism of action by which photosensitization mediates stimulation of CD8 T cell responses is an endosome-to-cytosol translocation of Ag and a subsequent TAP1-dependent and proteasome (trypsin- and caspase-like)–dependent MHCI Ag presentation (25). If a large fraction of Ag is funneled to MHCI, by consequence, less Ag should be available for MHCII. We tested this indirectly by measuring MHCII- and CD4-dependent Ab production. Immunization without photosensitizer led to strong Ab production with anti-OVA IgG1 titers ∼1:10⁴–1:10⁵ after a single injection with 30 μg OVA in PLGA. In contrast, anti-OVA Abs were not detectable in mice immunized with TPCS2a-containing vaccines. This finding supports the hypothesis that photosensitizers can inhibit default phagosomal maturation toward lysosome fusion and drive the localization of Ag toward MHCI by cytosolic delivery. Although not aimed for explicitly, the consequence of highly effective cytosolic targeting is suppression of the humoral response.

The adjuvant potential of TPCS2a was dependent on light activation and not on some other unknown intrinsic adjuvant effect of the compound, because enhanced immune responses were measured in mice only after exposure of mice to light. This observation is in agreement with previous studies in which cytosolic drug release from photosensitive liposomes was triggered by light activation (43, 44). Moreover, it could be argued that photosensitizer and light are solely producing damage and, thereby, providing danger signals that adjuvate cross-presentation and stimulation of CD8 T cell responses. This was described as one of the immunological effects of photodynamic therapy (45). However, in vitro experiments with Ag- and TPCS2a-loaded DCs did not support this, because DCs treated with high TPCS2a doses were not able to present Ag and stimulate CD8 T cell responses (data not shown). Therefore, we conclude that the adjuvant effect of TPCS2a is based on its light activation, with the consequence being the release of free radicals causing phagosome eruption and leaking of Ag or Ag-containing particles into the cytosol of viable APCs that are able to proteasomically degrade Ag and to present peptides in a TAP-dependent and MHCI-restricted manner to CD8 T cells.

Compared with soluble Ags, PLGA microparticles offer the chance to deliver high loads of entrapped Ag and adjuvant to APCs at a single-cell level. The polymer type, the surface decoration of the particles, as well as further excipients and adjuvants all represent factors that enable the modulation of the immunological properties of a particle-based Ag delivery system (35, 46–52). In studies in which PLGA microparticles were shown to stimulate APCs associated with the release of Ags into the cytosol as consequence of destabilization of the endosomal membrane (53–55). Although this destabilization was produced by the polymer itself, we show that a photosensitizer can be included in the particles for the purpose of destabilizing the endosomes, targeting of Ag to cytosol and MHCI, and generation of high numbers of CTLs. To achieve highly effective antitumor effects, reports indicate that vaccines should generate a certain threshold frequency of Ag-specific CD8 T cells (56, 57). Moreover, the polyfunctionality of Ag-specific CD8 T cells (e.g., their capacity to secrete multiple cytokines and release cytotoxic mediators), was reported to represent a predictive factor for the efficacy of immunotherapeutic vaccines (58, 59) and for disease progression in HIV patients (60). The current work demonstrated that the combination of PLGA microparticle–based Ag delivery and photosensitization was an effective immunization strategy for stimulation of high levels of polyfunctional Ag-specific CD8 T cells in a mouse model. These findings encourage further evaluation and development of PLGA microparticles as an Ag-delivery system to promote the treatment of or protection against CD8 T cell–dependent diseases and cancer, which remain among the most urgent unmet medical and societal needs.

We thank Jennifer Sand for technical assistance. The photosensitizer TPCS2a (Amphinex) and the light source (LumiSource) were gifts from PCI Biotech.

M.H. is an employee of PCI Biotech, which has field patents on the use of photosensitizers in vaccination. M.H. and P.J. received financial support from PCI Biotech. P.J. is listed as an inventor on patents describing the use of photosensitizers in vaccination. The other authors have no financial conflicts of interest.