Painless photodynamic therapy triggers innate and adaptive immune responses in a murine model of UV-induced squamous skin pre-cancer

Abstract

Painless photodynamic therapy (p-PDT), which involves application of photosensitizer and immediate exposure to light to treat actinic keratosis (AK) in patients, causes negligible pain on the day of treatment but leads to delayed inflammation and effective lesion clearance [Kaw et al. (2020) J Am Acad Dermatol, 82, 862–868]. To better understand how p-PDT works, hairless mice with UV-induced AK were treated with p-PDT and monitored for 2 weeks. Lesion clearance after p-PDT was similar to clearance after conventional PDT (c-PDT). However, lesion biopsies showed minimal cell death and less production of reactive oxygen species (ROS) in p-PDT-treated than in c-PDT-treated lesions. Interestingly, p-PDT triggered vigorous recruitment of immune cells associated with innate immunity. Neutrophils (Ly6G+) and macrophages (F4/80+) appeared at 4 h and peaked at 24 h after p-PDT. Damage Associated Molecular Patterns (DAMPs) including calreticulin, HMGB1, and HSP70, were expressed at maximum levels around 24 h post p-PDT. Total T-cells (CD3+) were increased at 24 h, whereas large increases in cytotoxic (CD8+) and regulatory (Foxp3+) T-cells were observed at 1- and 2- weeks post p-PDT. In summary, the ability of p-PDT to eliminate AK lesions, despite very little overt cellular damage, appears to involve stimulation of a local immune response.

1. INTRODUCTION

Skin cancer incidence continues to rise in Caucasian population worldwide, as more people are diagnosed annually with skin cancer than all other cancers combined (1). Nonmelanoma skin cancers (NMSC), which include basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), typically arise in sun-exposed areas of the body after many decades of chronic sun exposure (2), with the UVB component of sunlight (280–320 nm) acting as both a tumor-initiating and tumor-promoting agent (3–5). Actinic keratosis (AK, or pre-SCC) are pre-cancerous lesions caused by prolonged sun exposure in Caucasian individuals, and are a very frequent reason for visits to the dermatology clinic (6–8). Since AK have a potential for malignant progression to SCC, it is important to treat these lesions at an early stage (6,9,10). Several therapeutic options exist, including surgery, cryosurgery, and curettage. However, a major limitation of these surgical approaches is that lesions must be large enough for the clinician to visualize and treat, when in fact most sun-damaged skin consists of broad regions of microscopic dysplasia (a phenomenon known as ‘field cancerization’); those invisible diseased areas are often missed (11).

Photodynamic therapy (PDT) is a cancer-targeting treatment that is well suited for treating broad areas of field cancerization, and has the additional advantage of being a non-scarring modality that can be safely repeated as many times as necessary (6,9–12). PDT requires a photosensitizer (PS), visible light (either noncoherent or laser source), and tissue oxygen to kill cancer cells (13–15). One advantage of PDT compared to other cancer treatment modalities is its dual selectivity. First, the PS accumulates and is selectively retained to a much higher extent in cancer and pre-cancer cells relative to normal cells; second, the light/laser source can be precisely focused on the tumor to avoid off-target damage to surrounding tissues (13). While pre-formed PS are typically utilized for PDT treatment of internal cancers, skin cancers are more often treated using a prodrug (5-aminolevulinic acid; ALA) that is applied topically. ALA is then metabolically converted to PS (protoporphyrin IX, PpIX) within the cancer cells (6,10,15,16). The fact that ALA is converted into PpIX preferentially within the mitochondria of neoplastic and pre-neoplastic cells (relative to normal cells) provides exquisite tumor selectivity. Exposure of PpIX to intense visible light generates reactive oxygen species (ROS) that kill the targeted cells through a cascade of lethal events such as apoptosis, necrosis, autophagy and paraptosis (17,18). In addition to its widespread use for AK, PDT is being increasingly used to treat BCC and thin SCC, particularly in Europe and other countries overseas (10,16,19,20).

When treating AK with ALA-based PDT, the most typical regimen involves topical ALA application and incubation for 1–4 h to allow PpIX to build up within the cancer cells (16,21,22). When excited by visible light, PpIX becomes activated, generates ROS, and triggers tumor cell death (14,15,23,24). While this conventional regimen of ALA-PDT is effective, an important adverse effect is burning and stinging pain that occurs during illumination; this can be a major reason for patients to refuse further treatments, thereby limiting overall efficacy (25–29). The underlying mechanisms of pain during PDT are not fully understood, but appear to involve activation of PpIX within peripheral nerve endings (pain fibers) (25,30–33). Other than application of ice and evaporative cooling with a fan, very few techniques can effectively ameliorate this pain (25).

As a means to reduce pain during PDT, a number of clinical trials have explored alternative regimens that feature lower irradiance, lower PS concentrations, or both (30,31,34,35). “Daylight PDT”, in which low-irradiance sunlight is used to treat AK, with or without drug-light interval, is not only painless but also nearly as effective as conventional PDT using artificial light sources (30,31,34,35). The overall hypothesis behind daylight PDT is that a low steady-state level of PpIX, simultaneously produced and activated by light, may be sufficient for effective therapy. Less pain is a significant benefit here because the high levels of PpIX that normally accumulate during long ALA preincubations with conventional PDT (c-PDT), and which drive PpIX diffusion into nerve endings (25,28,31,34–36), do not occur with the daylight regimen (33). We recently completed a clinical trial at Cleveland Clinic that applied this principle in an indoor environment, using an artificial light source (BluU^™) to simulate daylight PDT for ALA-PDT of actinic keratosis in patients (37). We call this new regimen “painless PDT” (p-PDT). Our bilaterally-controlled clinical trial used a split-body design to compare the clearance rates of AK lesions on two halves of the body (one treated with p-PDT, the other with c-PDT). Pain levels reported by patients (0–10 on a visual analog scale) were much lower on the p-PDT side (0–1) relative to c-PDT side (4–6), as expected. More importantly, the treatment efficacy (AK clearance at 3 months) was equally effective whether using painless or conventional PDT approaches (37).

Because detailed mechanisms are very difficult to study in human patients, in this paper we employed an animal model (mice with UVB-induced AK lesions) to explore microscopic events by which p-PDT may be working to shrink and successfully eliminate squamous precancer lesions. In the prior clinical study, patients showed surprisingly little erythema immediately after p-PDT, yet they developed an intense inflammatory reaction at 24–48 post treatment (37). Such a delayed response, we reasoned, is a strong indicator of immune system activation. In the case of conventional PDT, much is already known about the mechanisms by which PDT damages tumor cells and tumor vasculature directly, as well as about how certain elements of anti-tumor immunity are activated after PDT exposure (24,38–40). The most well documented immunological effect after c-PDT is stimulation of innate immunity, including rapid tumor infiltration by neutrophils, macrophages and dendritic cells, and the release of inflammatory cytokines (38,41–43). There is also some evidence for involvement of T-cells in tumors treated with c-PDT (see Discussion). Here, however, our question was whether or not the p-PDT regimen, which causes very little overt damage to lesional skin, can nevertheless activate the immune system and therefore offer a plausible explanation for the observed reduction and elimination of pre-neoplastic lesions.

2. MATERIALS AND METHODS

2.1. Mouse model of skin pre-cancer by UV exposure:

A murine model of squamous pre-cancer of the skin (actinic keratosis; AK) was generated in SKH-1 hairless mice (Charles River Laboratories, Wilmington, MA; purchased at ~8 weeks of age) by exposure to UV-irradiation twice weekly, for approximately 12–15 weeks, using a set of UV lamps that provided 80% UVB and 20% UVA, respectively. The chronic UV exposure was incrementally increased from 90 mJ/cm²/sec to 180 mJ/cm²/sec, resulting in a gradual appearance of AK lesions that morphologically and histologically resembled actinic keratosis in humans (Fig. 1A and 1B) (44,45). Histological and pre-neoplastic characteristics of the AK lesions were confirmed by Hematoxylin/Eosin (H&E) staining (Fig. 1B) and by in vivo imaging of protoporphyrin IX (PpIX) fluorescence (Fig. 1C). For histological analysis, formalin-fixed, paraffin embedded sections of AK lesions were evaluated by H&E staining, following standard procedures. The pre-neoplastic nature of AK lesions was visualized by in vivo analysis of PpIX, which selectively accumulated in the lesions on the back of the mice after 1 h of incubation with topical aminolevulinic acid (ALA; Levulan Kerastick^™, Sun Pharmaceuticals, Princeton, NJ). The PpIX fluorescence was captured as an image cube (excitation 640 nm and emission 680 nm) using an In Vivo Imaging System (IVIS; Perkin Elmer, Inc. Waltham, MA) and was spectrally unmixed using Live Imaging software (Perkin Elmer, Inc. Waltham, MA). All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Cleveland Clinic, Cleveland, Ohio.

Figure 1. Measurement of AK lesions and quantification of PDT-induced lesion response.

(A), Bright field images showing normal (left) and examples of AK lesions on the back of a UVB-treated mouse (right panel, arrows). (B), Histological appearance of normal skin (left) and an AK lesion (right) by hematoxylin and eosin (H&E) staining. (C) Illustration of PpIX accumulation in AK lesions at 1 h after ALA application (right panel, red fluorescence) by in vivo imaging. (D) Schematic of conventional regimen of PDT and quantification of lesion reduction after conventional PDT, in 21 tumors from 3 mice at 1 week and 2 weeks post-PDT. (E) Schematic of the painless regimen and quantification of lesion reduction after painless PDT, in 34 tumors in 4 mice at 1 week and 2 weeks post-PDT. Mean differences were statistically significant by a non-parametric test (Mann-Whitney); *, P value < 0.0001.

2.2. Conventional and Painless photodynamic therapy:

Mice with AK lesions on their backs were anesthetized by inhalant isoflurane delivered using a vaporizer, and received PDT treatments using 20% aminolevulinic acid (Levulan Kerastick, Sun Pharmaceuticals, Princeton, NJ) and a 417 ±5 nm blue light source (Blu-U^™, Sun Pharmaceuticals) whose output was calibrated using a FieldMate laser power meter (Coherent Inc. Santa Clara, CA). The c-PDT was administered by topical application of ALA which was allowed to incubate for a 1-h period prior to the start of illumination (8.25 minutes, 5 Joules). For p-PDT, application of ALA was immediately followed by illumination (60 minutes, 36 Joules).

2.3. Treatment responses at 1 and 2 weeks after PDT of murine AK lesions:

The treatment efficacy of p-PDT and c-PDT was analyzed by measuring lesion size at 1- and 2-weeks post PDT. Since AK lesions are typically irregular and do not form a perfect hemisphere, the size of the lesions was estimated as follows. Diameter (D) was measured in millimeters (mm) using calipers; height (H) was estimated on a 3-point scale (1, 1.5, or 2 mm); then lesion size was calculated as D x H and expressed as a relative (-fold) change between baseline and post-PDT conditions. Twenty-one lesions from 3 mice and 34 lesions from 4 mice were included in the lesion clearance analysis for c-PDT and p-PDT regimens, respectively. Three AK lesions that showed an increase from the baseline, probably due to poor penetration of ALA limited by the presence of crust on these lesions, were not included in the analyses in Figure 1D and 1E.

2.4. Analysis of reactive oxygen species and cell death in murine AK samples:

Generation of reactive oxygen species (ROS) in murine AK lesions following p-PDT or c-PDT was analyzed in fresh-frozen sections using dichlorodihydrofluorescein di-acetate (CM-H2DCFDA) following the protocol described by Srivastava et al. (46). CM-H2DCFDA, which is a non-fluorescent dye, is converted into a fluorescent dichlorofluorescein (DCF) upon oxygenation with PDT-induced ROS produced in AK lesions. Briefly, cryosections (10 μm) from fresh-frozen AK lesions embedded in OCT compound (Tissue-Tek) were incubated with 10 μM CM-H2DCFDA dissolved in ACAS buffer (127 mM NaCl, 0.8 mM MgCl₂, 3.8 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM CaCl₂, 5 mM glucose & 10 mM HEPES PH-7.4) for 1 h at room temperature in the dark, followed by three washes with PBS and mounting in Vectashield (Thermo Fisher Scientific; Waltham, MA) with DAPI. Relative levels of ROS produced following p-PDT or c-PDT treatments were analyzed using a fluorescence microscope from Leica Biosystems (Buffalo Grove, IL) (46). The intensity of the DCF signal was captured using fluorescein isothiocyanate (FITC) fluorescence settings for fluorescein (excitation max 490 nm and emission max 525 nm) using a Leica fluorescence microscope and quantitated from digital images using IPLab image analysis software (Scanalytics Inc. Fairfax, VA). PDT-induced cell death in AK lesions was analyzed by DNA nick-end labeling (TUNEL assay) using the fluorescein-labeled in situ cell death detection kit by Millipore Sigma (St Louis, MO), following the manufacturer’s instructions exactly (47). As a second assay for cell death, the activation (cleavage) of pro-caspase-3 was measured in tissue sections using IHC and an antibody to cleaved caspase-3; see below.

2.5. Histological and Immunohistochemical staining of AK samples:

Murine AK lesions treated with conventional or painless PDT, along with no-PDT controls, were harvested at the times indicated in the figures. Formalin-fixed, paraffin embedded tissue sections were evaluated by Hematoxylin and Eosin (H&E) staining prior to immunohistochemistry (IHC). For immunofluorescent staining of protein markers, tissue samples were fixed in Histochoice (VWR Life Science, Radnor, PA). Antibodies from mouse immune cell phenotyping IHC antibody sampler kit (Cell Signaling Technology; Danvers, MA) were used to analyze macrophages (F4/80), dendritic cells (anti-CD11c) and T cells (CD3, CD8, FoxP3). Other primary antibodies used to analyze neutrophils (Ly6G; Thermo Fisher Scientific; Waltham, MA), calreticulin (CLR), high mobility group protein B1 (HMGB1) and heat shock protein 70 (HSP70) (all three from Cell Signaling), cleaved caspase-3 (BioVision Inc., Milpitas, CA); and secondary antibodies, Cy3-conjugated donkey anti-rabbit IgG and Cy3-conjugated donkey anti-rat IgG (Jackson ImmunoResearch (Westgrove, PA) were from sources indicated. Relative expression levels of marker proteins in tissue sections were analyzed using fluorescence microscopy and quantitated from digital images using IPLab as described above for ROS analysis.

2.6. Statistical analysis:

Lesion counts and lesion fluorescence staining intensities, were analyzed using GraphPad Prism (GraphPad Software, San Diego, CA). Data sets were first tested for Gaussian (normal) distribution, and then analyzed using the appropriate approach, i.e., by ANOVA and unpaired 2-sided t-tests if normally distributed, or by a nonparametric test (Mann-Whitney) if not normally distributed. P values < 0.05 were considered statistically significant.

3. RESULTS

To investigate mechanisms that might be responsible for similar rates of AK lesion clearance previously observed after conventional or painless PDT in humans (37), we designed experiments to simulate the two PDT regimens and performed a detailed time course analyses of underlying events in a murine AK model.

3.1. Conventional or painless PDT treatments elicit similar lesion clearance responses in a murine model of actinic keratosis:

Studies published by several groups including our own have shown that AK can be modeled in mice using a prolonged UVB-exposure regimen, and that the morphology and histology of the lesions in mice resemble those in humans (44,45). Based on these observations, SKH-1 mice were subjected to a 20-week UVB exposure regimen, as described in the methods section (44,48). Rough, red, and scaly AK lesions appeared on the backs of SKH-1 mice by ~Week 15 of the UVB exposure regimen (Fig. 1A; left vs. right image). Histological analysis of murine AK lesions by H&E staining showed epidermal hyperplasia and cellular atypia, as compared to the normal epidermis (Fig. 1B; left vs. right image). The pre-neoplastic nature of the AK lesions was visualized by application of ALA (for 1 h) and in vivo analysis of PpIX fluorescence by IVIS Spectrum imaging. Compared to the normal untreated skin, fluorescence from PpIX accumulated in discrete AK lesions (Fig. 1 C; left vs. right image). After the confirmation of murine AK lesions by morphology and histology, mice with AK lesions were treated with c-PDT or p-PDT and lesion clearance was analyzed by measuring the lesion size prior to, 1-week, and 2-weeks post PDT. Relative to pretreatment (100%), the average lesion size was reduced to 51% and 28% after c-PDT (Fig. 1D), and to 61% and 45% after p-PDT (Fig. 1E) at 1- and 2-weeks post PDT, respectively. This observation was similar to the one observed during the clinical trial reported by Kaw et al., showing that AK lesion clearance response between conventional and painless PDT regimens ranged between 40–60% at 3 months post PDT (37). Therefore, results shown in Figure 1 suggest that as observed in the clinical trial, p-PDT for the treatment of AK was as effective as c-PDT in this pre-clinical investigation using a murine model for AK (although the murine experiments were not statistically powered to test that assertion directly).

3.2. Reactive oxygen species and cell death are induced to different levels by conventional or painless PDT:

The results in Figure 1, showing similar lesion clearance rates after c-PDT and p-PDT, suggested that both treatments were possibly inducing similar cell death responses. To test that hypothesis, AK lesions were harvested at 24 h post PDT and cell death was analyzed in H&E-stained sections of AK (Fig. 2B–D). In the lesions harvested 24 h after c-PDT, the presence of apoptotic cells (characterized by shrunken, pyknotic nuclei), extravasated red cells, intracellular vacuoles, and large areas of cell loss (empty white spaces) were noted (Fig. 2C). These histological features of apoptotic cell death after c-PDT were very pronounced and visible throughout the depth of the epidermal lesion (Fig. 2C), as opposed to lesions treated with p-PDT in which cell death was only visible more superficially (Fig. 2D). Therefore, the results clearly indicated that cellular damage visible within 24 h post-treatment was much less after p-PDT than after c-PDT.

Figure 2. Histologic appearance of AK lesions, before and after conventional or painless treatment.

PDT-induced cell death was analyzed by comparing visual appearance and histomorphology of murine AK lesions at pre- and 24 h post-PDT. (A), Normal skin, untreated. (B), AK (pre-SCC) lesion, untreated. (C), AK biopsied at 24 h after conventional PDT. (D), AK biopsied at 24 h after painless PDT. Yellow arrows, erythrocytes in dilated blood vessels. Scale bars, 100μm. Dotted line, dermal-epidermal junction. sc, stratum corneum. epi, epidermis. hf, hair follicle.

Since the extent of tissue damage appeared to differ between samples treated with c-PDT versus p-PDT (Figs. 2C vs. 2D), we analyzed the relative levels of ROS in frozen sections from AK lesions harvested at 2 h post PDT following the protocol by Srivastava et al. (46). During PDT, excitation of photosensitizer (PpIX) by visible light results in generation of reactive oxygen species (ROS) and triggers a downstream cascade of events, resulting in apoptotic cell death (15,24,49). Following conventional PDT treatment, a significant increase (~8-fold over no PDT control; Figs. 3A and 3B) in the levels of ROS was observed. However, the increase in ROS level in lesions treated with p-PDT was significantly smaller (~3-fold over no PDT control) compared to the lesions treated with the c-PDT regimen (Fig. 3B). We have previously reported the activation and cleavage of caspase-3, a hallmark of apoptotic cell death, during PDT-induced cell death in a murine model of human SCC (A431) (47). Caspase-3 activation following c-PDT or p-PDT was analyzed in AK lesions by immunofluorescence using an antibody specific to cleaved caspase-3 (Fig. 3C). Activated and cleaved caspase-3 staining in the basal layer of hyperplastic regions corresponding to dysplastic epidermis was observed exclusively in AK lesions treated with the c-PDT regimen (Fig. 3C; middle vs. bottom image and Fig. 3D) suggesting that unlike p-PDT, c-PDT invokes a classic caspase activation pathway for induction of apoptosis in murine AK lesions. Next, to quantitate the levels of cell death induced by the two PDT regimens, TUNEL staining with an in-situ cell death detection kit was utilized (Figs. 3E and 3F). TUNEL staining showed a robust staining of apoptotic nuclei in epidermal and dermal regions in c-PDT treated lesions, but only slightly increased staining in lesions treated with p-PDT (Fig. 3E). Quantitation of apoptotic nuclei per field showed ~21-fold and 4-fold increases in conventional and painless PDT treated lesions, respectively (Fig. 3F). In summary, relative to c-PDT, the reduced ROS production and apoptotic cell death observed in AK lesions after p-PDT treatment suggest that apoptotic cell death cannot be the primary mechanism to explain the similarity in lesion clearances after p-PDT and c-PDT treatments.

Figure 3. ROS production and cell death after painless PDT are much less than after conventional PDT.

(A) Fresh frozen cryosections of AK lesions were analyzed for; reactive oxygen species (ROS); see Methods. The ROS assay showed lack of ROS production in untreated control samples (A; top panel), low levels of ROS in painless PDT samples (A; bottom panel) and the highest ROS production in hyperplastic epidermal region of conventional PDT samples (A; middle panel). (B) Quantitation of pixels representing ROS signal in no treatment control, conventional or painless PDT treated samples. (C) Immunofluorescence staining of formalin-fixed tissues for cleaved Caspase-3 showed lack of Caspase-3 cleavage in non-treated control sample (C; top panel) and in the painless PDT treated AK lesion (C; bottom panel), but robust cleavage of Caspase-3 in the basal layer of AK lesions treated with conventional PDT (C; middle panel). (D) Quantitation of cleaved caspase-3 expressing cells in the basal layer of no treatment control, conventional or painless PDT treated samples. (E) TUNEL staining of formalin-fixed tissue samples showed lack of cell death in untreated control sample (E; top) and dying TUNEL-labeled nuclei in PDT-treated AK samples (E; middle and bottom). Note the lower amounts of TUNEL labeling observed in painless PDT (bottom panel) than in conventional PDT samples (middle panel), reflecting a much higher levels of cell death in the latter. (F) Quantitation of apoptotic nuclei in no treatment control, conventional or painless PDT treated samples. Fluorescence-positive cells or ROS signals were quantified from three images each from five lesions per treatment group and expressed per high power field (20X). Differences evaluated by Mann-Whitney test; *, P < 0.0001. A.U., arbitrary units.

3.3. Induction of innate immunity by infiltration of neutrophils and macrophages in actinic keratosis lesions treated with painless PDT:

The therapeutic effect of PDT is usually assumed to result from photosensitizer-mediated tumor destruction mediated by reactive oxygen species (ROS) produced during PDT illumination (14,15,24), yet the data in Figures 2 and 3 suggest that immediate PDT-induced cell death is not the primary mechanism responsible for the similar lesion clearance rates after c-PDT and p-PDT. Because the literature has clearly shown that tumor cell damage caused by c-PDT can trigger an acute inflammatory response, with activation of the innate immune system (24,50–52) as described in several recent studies of murine and human pre-SCC and SCC of the skin (53–55), we focused our attention here on the immune responses induced by p-PDT only. AK lesions were harvested at immediate (1 and 4 h), intermediate (24, 48 and 72 h) and late (1 and 2 weeks) times after p-PDT, and tissue sections stained using antibodies specific for neutrophils, macrophages and dendritic cells. Infiltration/recruitment of neutrophils (stained with Ly6G antibody) was observed in AK lesions as early as 1 h, peaking at 24 h and totally disappearing by 1-week post PDT (Fig. 4A and 4B). Macrophages stained for F4/80 (a marker of mature macrophages) were detected at low levels in untreated AK lesions, but then increased significantly by 1 h post-PDT, with maximum levels reached at 24 h and slowly declining at 48 h and 72 h post PDT (Fig. 4C and 4D). Interestingly, a biphasic response in macrophage recruitment was observed at ~1 week post PDT, a phenomenon that could have several possible explanations including an M1/M2 phenotypic switch (see Discussion) (56). Dendritic cells (probed with an anti-CD11c antibody specific for DCs), were not detectable in our p-PDT treated AK lesions.

Figure 4. Quantification of innate immune cells in AK lesions at various times after painless PDT.

(A) Immunofluorescence images showing painless PDT-induced infiltration/recruitment of neutrophils (Ly6G in red) in murine AK lesions following painless PDT (left vs. right panels). (B) Quantitation of a time course response of neutrophil infiltration in AK lesions after painless PDT. (C) Immunofluorescence staining for macrophages (F4/80 in red) showing their presence in painless PDT treated samples (left vs. right panels). (D) Quantitation of a time course response of macrophage recruitment in AK lesions after painless PDT. Nuclei were stained with DAPI (blue) for reference. Enumerated cells were counted from three images each from five lesions per treatment group and reported per high-power field (20X). Statistical analyses by ANOVA and 2-tailed Student T-test. *, P < 0.0001.

3.4. Induction of immunogenic cell death (ICD) and expression of damage associated molecular patterns (DAMPs) in murine AK lesions by painless PDT:

Others have shown that oxidative stress generated during PDT-induced cell death can result in expression and release of DAMPs from damaged cells at the site of tumor destruction (24,39,43). DAMPs bind to cellular receptors and activate innate immune cells like macrophages, dendritic cells and some types of T cells which are involved in the processing and presenting cellular debris as tumor associated antigens (TAA) (24,39,43). We were interested in the expression of three commonly reported DAMPs associated with ICD, namely calreticulin (CLR), high mobility group box 1 (HMGB1) and heat shock protein 70 (HSP70) (39). A time course of expression pattern of DAMPs was examined by immunofluorescence analysis of p-PDT treated murine AK lesions. Compared to no-treatment controls, a robust upregulation of calreticulin (Fig. 5A) with highest expression at 24 and 48 h post PDT (Fig. 5B) was observed (28-fold and 32-fold, respectively). Similarly, highly induced expression of HMGB1 (Fig. 5C) was observed at 24 h (30-fold), with gradual decrease in expression by 1 week (Fig. 5D). However, compared to CLR and HMGB1, expression of HSP70 was only slightly increased (~3-fold) at ~24 h post PDT relative to the no treatment controls (Fig. 5E and 5F).

Figure 5. Expression (activation) of Damage Associated Molecular Patterns (DAMPs) in pre-SCC lesions after painless PDT.

Histological sections of AK lesions were stained with antibodies specific for (A) Calreticulin (CLR; in red); (C) high mobility group protein B1 (HMGB1; in red) and (E) heat shock protein 70 (HSP70; in red). Compared to no PDT controls, a robust expression of CLR and HMGB1 in painless PDT-treated AK lesions was induced around 24 h post painless PDT, top vs. bottom images in (A) and (C). Note that HSP70 was slightly induced in painless PDT-treated AK lesions in our study [top vs. bottom images in (E)]. Intensity of the antigen specific fluorescence signals were quantified from three images each from five lesions per treatment group from high power field (20X) and expressed per unit area (A.U., arbitrary units) in a time course study for CLR (B), HMGB1 (D) and HSP70 (F). Nuclei were stained with DAPI (blue) for reference. Statistical analyses of differences were by ANOVA and 2-tailed Student T-test. *, P < 0.05; **, P < 0.005; ***, P < 0.0001.

3.5. Activation of adaptive immunity and infiltration of different T-cell populations in murine AK lesions by painless PDT:

PDT-induced activation of adaptive immunity involves interactions between antigen presenting cells (APCs) and T lymphocytes (T cells), resulting in proliferation of T-cell clones that recognize specific epitopes on the target tumor (24,39,41–43). To investigate the involvement of anti-tumor adaptive immunity as a possible mechanism for lesion clearance in p-PDT treated mice, AK lesions were harvested at different times after PDT treatment, along with no PDT controls, and analyzed by immunofluorescence using antibodies specific for individual T cell subtypes. While few CD3+ T cells were present in untreated AK lesions, p-PDT was found to stimulate the influx of CD3+ T cells as early as 1 h post PDT. CD3+ T-cell levels remained elevated (~3- to 5-fold) for the first 3 days then rose to 7-fold by 2 weeks post PDT (Fig. 6A and 6B). CD8+ cytotoxic T cells (CTCs, known for tumoricidal activities) showed a different recruitment pattern; while a 3- to 5-fold increase was observed between 4 h and 72 h, much larger inductions in CD8+ T-cells were observed at 1-week (29-fold) and 2-weeks (44-fold) post PDT (Fig. 6C and 6D). Similarly, Foxp3+ (forkhead box P3) regulatory T cells (Tregs; also referred to as suppressor T cells) were induced in p-PDT treated AK lesions. Foxp3+ T-cells were first increased at 48–72 h post PDT (a moderate 2–3-fold induction), but then became highly abundant at 1 week (19-fold) and 2-weeks (25-fold) (Fig 6E and 6F). Although involvement of CD4+ T helper cells in anti-tumor immunity by conventional PDT has been reported in other models/systems (42,43), we did not observe any CD4+ T cells in our study.

Figure 6. Adaptive immune response and quantification of T-cells in pre-SCC lesions at various times after painless PDT.

Immunofluorescence images showing painless PDT-induced infiltration/recruitment of T cells in murine AK lesions. (A) Immunofluorescence staining for CD3+ active T cells (CD3ε; in red), (C) CD8+ cytotoxic T cells or cytotoxic T lymphocytes (CTL; CD8α; in red) and (E) Foxp3+ regulatory T cells (Tregs) or suppressor T cells (Foxp3; in red) are shown. Compared to no PDT controls, note the presence of different types of T cells in painless PDT treated lesions, top vs. bottom images in (A), (C) and (E). Nuclei were stained with DAPI for reference. CD3 positive (B); CD8 positive (D) and FoxP3 positive (F) cells were counted from three images each from five lesions per treatment group in a time course experiment and reported per high-power field (20X). Statistical analyses for (B) were by ANOVA and 2-tailed Student T-test. ***, P < 0.0001. Statistical analyses for (D) and (F) were by Mann-Whitney test. *, P < 0.05; **, P < 0.005; ***, P < 0.0001.

4. DISCUSSION

In a prior clinical study, we had shown that p-PDT, which is similar to daylight PDT, can be successfully used for the treatment of actinic keratosis (AK) as a painless alternative to c-PDT. Here, to investigate a detailed time course of histological changes in AK lesional tissue following p-PDT (something difficult to do in human patients), we employed a murine model of AK. The results show that AK lesion clearance following p-PDT is very similar to lesion clearance after c-PDT. Furthermore, immunogenic cell death (ICD) rather than classical apoptosis-mediated cell death is the predominant mechanism, and therefore may be central to the therapeutic effects of p-PDT.

We find it interesting that in our murine model, the changes in lesion size observed after treatment (Fig. 1D and 1E) are very similar to changes observed during the clinical trial reported by Kaw et al. (37). This suggests that murine AK lesions are an excellent model of human AK for studying PDT responses. It also reinforces the conclusion of Kaw et al. that painless PDT may be just as effective for the treatment of AK as c-PDT. While the lesion clearance rates observed in our mice at 1- and 2-weeks after either PDT regimen are well below 100%, this is actually quite typical of clinical responses seen in patients after a single blue light PDT session, unless additional preparation of the lesions is performed. Steps that can be taken in clinical practice to optimize PDT responses include: physical abrasion of the lesions (curettage), use of an occlusive dressing to enhance ALA penetration, and use of repeated PDT treatment sessions (57,58).

With respect to tumor destruction, the therapeutic effects of PDT depend upon three well characterized events that occur sequentially and are mechanistically linked (59). First, direct PDT-mediated destruction of cancer cells triggers cell death, mainly by apoptosis, necrosis and autophagy. Second, targeting of the tumor vasculature can occur, which limits the blood supply and induces hypoxia. These two mechanisms are directly responsible for initial tumor destruction and production of tissue debris. Cellular debris primes the third event, inflammation and activation of the immune system, a response that lasts for days to weeks and can eventually exert systemic (abscopal) effects in some situations (59). Contrary to our initial hypothesis that the similar clearance rates after p-PDT and c-PDT implied similar rates of cell death, we found instead that cell death and ROS production were very low in p-PDT treated lesions, relative to c-PDT (Figs. 2 and 3). This clearly suggested that the classically-described mechanisms of tumor cell death after c-PDT may not be the primary mechanism leading to lesion clearance after p-PDT. Therefore, we considered the alternative hypothesis that immune-mediated anti-tumor immunity might be important. In support of this notion, our observation that p-PDT only causes minimal damage to tumor vasculature (unlike c-PDT) favors anti-tumor immunity, because an intact vasculature is necessary to recruit the circulating leukocytes required to mount an effective and long-lasting anti-tumor immune response (39).

Our demonstration that various cellular components of innate immunity are recruited to tumors after p-PDT is consistent with previous reports in the literature. Krosl et al. described significant influxes of immune cell populations in SCCVII tumors following PDT, including rapid infiltration by neutrophils (within minutes after PDT) and more gradual recruitment of macrophages (at ~8 h) (60). Others have demonstrated induction of local acute inflammatory responses after PDT, with the sequential arrival of neutrophils, mast cells and macrophages that appear to contribute to tumor regression via their tumoricidal properties (38,39,52,60). In our study, the recruitment of neutrophils to AK lesions between 1 h and 72 h post mPDT (Fig. 4B) and macrophages by 1 h and lasting through 1-week (Fig. 4D) indicates innate immune system involvement. The biphasic response of macrophage activation and infiltration seen in p-PDT treated lesions (Fig. 4D) has several plausible explanations, including (1) switching between classically activated (M1) macrophages to alternatively activated (M2) macrophages; (2) recruitment of regulatory macrophages with a primary role in limiting inflammation. The latter are often seen as regulatory macrophages in tumor microenvironments (56). We did not observe any DC activation or recruitment in our experiments using a CD11c antibody. Although another group had reported activation of dendritic cells (DCs) following ALA-PDT of UVB-induced SCC in hairless mice (61,62), that group used c-PDT (3 h of ALA, followed by red light) and different DC markers (CD1a, CD80, CD86, and MHC-II); therefore their results are not directly comparable with our blue light painless regimen.

Our finding that p-PDT causes robust induction of DAMPs is also in accord with the PDT literature. In recent work by Wang et al. in a murine SCC tumor model, induction of CLR, HMGB1 and HSP70 expression was reported at 1 h after conventional red-light PDT (63). In our study, the induction of CLR and HMBG1 (maximum at around 24 h) occurred more slowly, and the increase in HSP70 expression was smaller. Differences between these two studies could easily be the result of differences in light wavelengths and intensities. However, the fact that expression of CLR, HMGB1 and even HSP70 was quite vigorously induced by p-PDT strongly suggests that immunogenic cell death is an important part of the mechanism leading to lesion resolution after p-PDT.

Our study demonstrated clear-cut changes in the recruitment of T cells (critical elements of the adaptive immune response) following treatment with p-PDT. The involvement of adaptive immunity in PDT-induced responses was first described by Korbelik et al. who showed that PDT resulted in tumor cell killing in both immunocompromised and immunocompetent mice; importantly, the long-term anti-tumor effects were seen only in the latter (64). A quick reminder of the implications of this finding is warranted here. T cell receptor (TCR)-mediated activation of T cells results in activation of CD3, a multiprotein complex that directly associates with TCR. Activation of CD3 results in activation of both CD8+ cytotoxic T lymphocytes (CTL) and CD4+ T helper cells (39,42,43). In p-PDT-treated AK specimens, we observed infiltration of both CD3+ and CD8+ T cells, suggesting a tumoricidal role of CD3-activated, CD8+ CTLs in a PDT-mediated anti-tumor effect. The study by Korbelik et al. showed the involvement of CTLs using EMT6 mammary tumor model, in which CD8+ CTLs were depleted, resulted in a 50% reduction in cure rate when compared to control mice (64). In our study, the robust recruitment in regulatory T cells (Tregs; Foxp3+ T cells) in parallel with the recruitment of the CD8+ was an interesting finding. Tregs, also referred as suppressor T cells, carry inhibitory activities towards other effector T cells and thereby compromise the anti-tumor activities of the latter (42,43). Tregs have been suggested as a possible mechanism of tumor resistance to PDT (41,65). To address the functional importance of T-regs, CD8+ T-cells, or other leukocytes in mediating the clearance of p-PDT treated AK lesions, further experiments involving depletion of specific immune cell populations will be required. Planning for such experiments is currently underway.

In conclusion, this report provides concrete evidence that p-PDT induces an immune response which may represent an important part of its mechanism of action. This response involves elements of innate immunity (DAMPs, neutrophils and macrophages) as well as adaptive immunity (T-cells). These findings should lay the groundwork for further detailed investigations on p-PDT as an inducer of anti-tumor immunity, both in skin precancers and perhaps in other neoplastic conditions as well.