Tumor cell survival pathways activated by photodynamic therapy: a molecular basis for pharmacological inhibition strategies

The production of ROS occurs during irradiation of the photosensitizer. Although these primary ROS are short-lived, there is ample evidence that PDT induces prolonged oxidative stress in PDT-treated cells [25, 26]. The post-PDT oxidative stress stems from (per)oxidized reaction products such as lipids [26] and proteins [27] that have a longer lifetime and, in addition to acutely generated ROS, depletion of intracellular antioxidants [28] and, hence, further exacerbation of already perturbed intracellular redox homeostasis.

The generation of ROS and oxidative stress by PDT leads to the activation of three distinct tumoricidal mechanisms. The first mechanism is based on the direct toxicity of photoproduced ROS, which oxidizes and damages biomolecules and affects organelle and cell function. For example, 8-hydroxydeoxyguanosine is a reaction product of ROS with guanosine [29] and may contribute to the induction of DNA damage by PDT [30‐38]. Furthermore, 8-oxo-7,8-dihydro-2′-guanosine is a product of RNA oxidation reactions that leads to impaired RNA-protein translation [39, 40].

With respect to phospholipids, linoleic acids are prominent targets for ROS-mediated peroxidation [41], yielding 9-, 10-, 12-, and 13-hydroperoxyoctadecadienoic acids as specific products of ¹O₂-mediated linoleic acid oxidation [42]. Other membrane constituents such as cholesterol, α-tocopherol, aldehydes, prostanes, and prostaglandins are susceptible to oxidation by type I and type II photochemical reaction-derived ROS [41, 43‐46]. The (per)oxidative modifications of phospholipids and membrane-embedded molecules by ROS lead to changes in membrane fluidity, permeability, phase-transition properties, and membrane protein functionality [47‐60]. Since many photosensitizers are lipophilic, the oxidation of membrane constituents by PDT is likely a prominent cause of cell death.

In addition to nucleic acids and lipids, most protein residues are also susceptible to oxidation by type I and type II photochemical reaction-derived ROS, which can potentially lead to rupture of the polypeptide backbone as a result of peptide bond hydrolysis, main chain scission, or the formation of protein-protein cross-links [61]. Specific amino acids such as histidine, tryptophan, tyrosine, cysteine, and methionine that may be involved in the active sites of enzymes can be oxidized. Proteins that are most abundantly modified by PDT-generated ROS include proteins involved in energy metabolism (e.g., α-enolase, glyceraldehyde-3-phosphate dehydrogenase), chaperone proteins (e.g., heat shock proteins (HSP)70 and 90), and cytoskeletal proteins (e.g., cytoplasmic actin 1 and filamin A α) [62]. Besides detrimental effects on protein function, oxidative modification of these biologically essential substrates disrupts the normophysiological redox state of cells, leading to oxidative stress and, in case of excessive damage or stress, cell death via necrosis, apoptosis (reviewed in [63]), or necroptosis [64], depending on which intracellular substrates are most affected by ROS (reviewed in [65]).

Surviving cells may activate adaptation mechanisms in order to (1) restore the intracellular redox homeostasis (antioxidant response), (2) activate a stress response that aids in survival or stimulates apoptosis (immediate early stress response), and (3) facilitate in refolding or degradation of carbonylated proteins (proteotoxic stress response). Autophagy as a result of mitochondrial or ER stress may prevent apoptotic cell death and thereby constitutes a survival mechanism in sublethally damaged tumor cells following PDT [66].

The second tumoricidal mechanism of PDT involves the induction of local hypoxia in the irradiated tumor bulk. The acute induction of hypoxia is a result of O₂ depletion in consequence to the O₂ → ¹O₂ or O₂ ^•– conversion and subsequent oxidation of biomolecules during PDT [67] and the shutdown of tumor vasculature after PDT [68]. The majority of systemic first- and second-generation photosensitizers localize primarily in endothelial cells as well as tumor cells that line the tumor vasculature after short drug-light intervals [69, 70], defined as the time between photosensitizer administration and light delivery. Endothelial photosensitization in particular is associated with vasculature-damaging effects [71‐74] that translate to a favorable therapeutic outcome. Prolonged hypoxia due to the destruction of intratumoral vasculature was found to be crucial in the massive induction of cell death following PDT as a result of thrombosis, hemostasis, and cessation of oxygen and nutrient supply (reviewed in [68]). A state of hypoxia or even anoxia reduces the ability of cells to generate ATP by oxidative phosphorylation [75]. As will be reviewed here, hypoxia causes cells to resort to ATP production through anaerobic metabolism to sustain cell function and restore homeostasis and promote angiogenesis to resolve the hypoxic conditions. Cells that are incapable of sustaining ATP production anaerobically due to extensive oxidative stress undergo necrotic cell death (an ATP-independent mode of cell death), which is the strongest trigger for the third tumoricidal mechanism: the antitumor immune response.

The antitumor immune response, which is triggered by a form of sterile inflammation, constitutes an important process in the post-PDT removal of the treated malignancy. Various studies in mice have shown that activation of the immune system after PDT is necessary for complete eradication of the tumor [76, 77]. The tumor cell death that occurs directly from photochemical damage or as a result of vascular shutdown-mediated hypoxia/anoxia and hyponutrition is the key precursor event for the antitumor immune response.

The PDT-treated cancer cells die as a result of necrosis, apoptosis [78], necroptosis [64], and/or autophagy [79]. In all modes of cell death, intracellular molecules are released that, following their release, act as so-called damage-associated molecular patterns (DAMPs) [80]. The released molecules also comprise tumor-associated antigens (TAAs) that are otherwise shielded from recognition by immune cells and hence are nonimmunogenic until released [81]. Accordingly, the extracellular DAMPs and TAAs alert cells of the innate and adaptive immune system of impending cellular demise and the presence of malignant tissue, respectively, and consequently trigger a sterile immune response aimed at removing the PDT-treated tumor [82]. A major advantage of the PDT-triggered oncoimmunological pathways is that these pathways can trigger an antitumor immune response mediated by antigen-specific T-cells against distant tumor cells that were not subjected to PDT (referred to as abscopal effects) [83, 84].

NRF2 is a bZIP transcription factor that is constitutively expressed in most cells and tissue types [91‐93]. Under normoxic conditions, NRF2 associates with Kelch-like ECH-associated protein 1 (KEAP1) that is bound to the cytoplasmic cytoskeleton and therefore sequesters NRF2 in the cytosol [94, 95]. Moreover, KEAP1 binds Cullin-3 that forms a scaffold for E3 ubiquitin ligases to facilitate polyubiquitination and subsequent proteasomal degradation of NRF2. Thus, under normoxic conditions, the antioxidant stress response is inactivated by high levels of cytosolic retention and degradation of NRF2 (reviewed in [86]). During oxidative stress, the NRF2-binding domain of KEAP1 is oxidized at Cys273 and Cys288, resulting in impaired KEAP1 binding to NRF2 [96]. Consequently, free NRF2 accumulates in the cytoplasm where it is activated by oxidation at Cys183, after which it is able to translocate to the nucleus [86]. Additional phosphorylation of NRF2 at serine (Ser)40 by p38α/β and/or JNK1, which are also induced by PDT (Section 3.4), may also play a role in the dissociation of the NRF2-KEAP1 complex or the prevention of NRF2-KEAP1 binding [97‐99]. Once in the nucleus, NRF2 dimerizes with members of the AP-1 family, such as JUN and musculoaponeurotic fibrosarcoma oncogene homologue (MAF) subfamily proteins [100, 101], and binds to antioxidant response element (ARE) sequences to induce the transcription of antioxidant genes. An overview on the activation mechanisms of NRF2 and downstream effects is presented in Fig. 3. An elaborate review on the activation mechanisms of NRF2 is provided in [86].

The products of NRF2 target genes are involved in the synthesis and redox cycling of antioxidants as well as the removal of potentially harmful oxidation products. The NRF2/AP-1 target genes include NAD(P)H:quinone oxidoreductase 1 (NQO1) and NQO2, heme oxygenase-1 (HO-1, HMOX1), glutamate-cysteine ligase (GCL), microsomal epoxide hydroxylase (EH-1), glutathione S-transferases (GSTs), sulfiredoxin 1 (SRXN1), and carboxylesterase 1A1 (CES1A1) [102]. EH-1 neutralizes epoxides, whereas NQO1 and NQO2 reduce oxidized quinones to prevent further cell damage by these reactive species [103, 104]. CES1A1 hydrolyzes esters and thioesters [105]. HO-1 neutralizes certain types of ROS directly as well as oxidized metabolites (lipid radicals) indirectly by producing the antioxidant molecule bilirubin from heme [106, 107]. Moreover, proteins involved in the reduction and reactivation of radical scavengers such as glutathione (GSH) and peroxiredoxins are upregulated by NRF2, including GCL (subunits GCLC and GCLM), GSTs, and SXRN1 [108, 109]. NRF2 further upregulates ATP-binding cassette subfamily C (ABCC) 2, 3, 4, and 6 (also known as multidrug resistance proteins (MRPs)) and the multidrug efflux pump ATP-binding cassette subfamily G member 2 (ABCG2), which mediate efflux of organic anions and glutathionylated, glucuronidated, and/or sulfated (bio)molecules (reviewed in [102, 110‐112]). Such biomolecules are the products of oxidative stress that are potentially harmful by themselves [102]. MRPs may therefore be crucial for the detoxification of tumor cells that have survived the initial PDT-induced ROS attack and can aid in restoring the intracellular redox balance.

HO-1 in particular has been linked to cancer cell survival following PDT. Besides being upregulated by NRF2, HO-1 (encoded by the HMOX1 gene) is upregulated by HIF-1 [113], which is also induced by PDT (Section 3.3). The function of HO-1 is to convert mitochondrially produced heme into carbon monoxide (CO) and biliverdin, of which the latter is reduced by biliverdin reductases to bilirubin [114]. Bilirubin scavenges peroxidized lipids [115, 116] and may significantly contribute to tumor cell survival following PDT by terminating lipid oxidation chain reactions. Furthermore, at low concentrations, CO possesses vasodilating, proangiogenic, anti-inflammatory, and antiapopotic properties, which can contribute to angiogenesis, tumor survival, and tumor regeneration in vivo [117, 118]. Although the degradation of heme to bilirubin also liberates Fe²⁺ that contributes to a prooxidant state, the release of Fe²⁺ by HO-1 was found to concomitantly increase the transcription of ferritin [119], which chelates and neutralizes free Fe²⁺ [120].

Another major pathway augmented by NRF2 is the GSH synthesis pathway, which yields an effective redox machinery aimed at scavenging ROS and neutralizing reactive intermediates such as oxidized protein residues (by glutathionylation) [121]. Synthesis of the GSH tripeptide occurs by ligation of l-glutamate and l-cysteine by GCL and addition of glycine by GSH synthetase. GSH can reduce ROS through oxidation of its thiol moiety (GSH → GS•), after which the reactive thiol is neutralized by GS-GS homodimerization (GSSG) with another GS• through disulfide bridge formation. Recycling of GSSG to GSH is catalyzed by GSSG reductase (reviewed in [121]). GSH can also react with oxidized cysteine residues, resulting in protein glutathionylation and subsequent cellular efflux via proteins of the MRP family [110]. Moreover, GSTs of different classes are upregulated by NRF2, which are responsible for the glutathionylation of oxidized proteins resulting in increased MRP transporter-mediated efflux of glutathionylated peptides [122]. Another role for GSTs is to inhibit molecular constituents in the ASK1 pathway, including ASK1 (by GSTM), JNK (by GSTP/GSTA), and tumor necrosis factor receptor associated factor 2 (TRAF2) (by GSTP), although the inhibitory efficacy decreases upon oxidative stress [122]. This may prevent prolonged activation of the ASK1 pathway and could stimulate cell survival as is discussed in Section 3.4. In sum, the activation of NRF2 is essential for the production of proteins involved in GSH synthesis and redox regulation, as well as the neutralization of oxidative compounds and their cellular efflux.

Although NRF2 activation by ROS is well-established, its activation by PDT has been sparsely investigated. Nuclear translocation and thus activation of NRF2 was observed by Kocanova et al. in human bladder cancer (T24) cells and human cervical cancer (HeLa) cells following hypericin-PDT [85]. Furthermore, NRF2 target genes were overexpressed in various cancer cells after PDT, which include HO-1 [123], GCLC and GCLM subunits of GCL [124], NQO1 [124, 125], and ABCG2 [111]. The inhibition of p38^MAPK (p38α and p38β, Section 3.4.2) with PD169316 reduced HO-1 messenger RNA (mRNA) levels and increased the susceptibility of T24 cells to PDT [85]. These findings indicate that NRF2 is activated following PDT, that p38^MAPK-mediated phosphorylation enhances the activity of NRF2 post-PDT, and that the expression of HO-1 by NRF2 is cytoprotective. Several reports have corroborated HO-1-mediated cytoprotection following PDT [123, 126, 127]. However, HO-1 was also found to be induced by aminolevulinic acid (ALA) prior to PDT [111], and targeted knockdown of HO-1 has been related to reduced intracellular protoporphyrin IX (PPIX) accumulation [128], indicating that HO-1 can both inhibit PS accumulation as well as reduce the PDT response. Interestingly, of the MDR proteins induced by NRF2, at least ABCG2 has been confirmed to facilitate cytoprotection against PDT by mediating the cellular efflux of photosensitizers PPIX, pyropheophorbide A, and benzoporphyrin deri-vative monoacid ring A [129] but not meso-tetrahydroxyphenylchloride and porfimer sodium [130].

Retinoic acid has been identified as an inhibitor of NRF2 in human mammary carcinoma (MCF-7) cells transfected with an ARE-luciferase reporter construct. Retinoic acid abolished the expression of genes with ARE sequences in their promoter regions [131] but did not affect the nuclear translocation or degradation of NRF2. Rather, retinoic acid inhibited the function of NRF2 by activating retinoic acid receptor α (RARα) in the nucleus. RARα sequesters NRF2 in the nucleus, thereby inhibiting the association between NRF2 and ARE sequences [131] (Table 1). Unfortunately, not much is known about the binding specificity of retinoic acid, nor has retinoic acid or any of its analogs been studied in the context of PDT. Nevertheless, retinoic acid and its analogs are also involved in the inhibition of AP-1 transcription factors (Section 3.4.2.2 Prolonged downstream effects of ASK1 activation), which constitute the main dimerization partners for NRF2. Thus, PDT with RARα activators may potentially enhance the cytotoxic effects of PDT by inhibiting both AP-1 and NRF2 survival signaling.

In addition to inhibiting NRF2-mediated gene expression, the downstream gene products of NRF2, such as HO-1 and members of the GSH antioxidant machinery, may also be successfully inhibited by small molecular compounds (e.g., Zn-protoporphyrin IX (ZnPP) [132, 165], Table 1). Inasmuch as HO-1 catalyzes the degradation of heme into the antioxidants bilirubin and CO, inhibition of HO-1 with ZnPP during PDT is expected to enhance tumoricidal efficacy. Indeed, HO-1 inhibition with 2.5 μM ZnPP considerably reduced cell viability following porfimer sodium-PDT in both human (MDAH2774) and murine (C26) colon carcinoma cell lines. The addition of bilirubin or CO could not rescue cells from PDT-induced cell death upon HO-1 inhibition, suggesting a more elaborate role of HO-1 in the survival of tumor cells than merely the synthesis of antioxidants [123]. Similar results with HO-1 inhibition were obtained in WM541Lu human melanoma cells subjected to ALA-PDT, where the addition of anti-HO-1 siRNA (24 h prior to PDT) or tin-PPIX (SnPP [133]) increased the susceptibility of these cells to PDT [126]. SiRNA-mediated knockdown of HO-1 also increased the susceptibility of UM-UC-3 (but not T24, KU7, UM-CU-2, and UM-CU-4) human urothelial carcinoma cell lines to ALA-PDT [128]. With respect to other NRF2-upregulated antioxidants, Kimani et al. tested several inhibitors of the glutathione redox system and ROS scavenging enzymes (superoxide dismutases (SODs) and catalase) to increase the efficacy of disulfonated aluminum-phthalocyanine-PDT of MCF-7 breast cancer cells [166]. Diethyl-dithiocarbamate (DDC) and 2-methoxyestradiol (2-ME) were used as inhibitors of Cu-SOD/Zn-SOD [167] and Mn-SOD (although 2-ME has been shown not to inhibit Mn-SOD [134]), respectively (Table 1). l-Buthionine sulfoximine (BSO) was used as an inhibitor of GCL (glutathione synthesis [135]) and 3-amino-1,2,4-triazole (3-AT) as an inhibitor of catalase [136] (Table 1). All inhibitors were efficient in exacerbating PDT-induced apoptosis. The strongest effects were observed when a combination of BSO with either 3-AT or 2-ME was used, which achieved similar results as when all the inhibitors were combined. This suggests that the inhibition of H₂O₂ scavenging (by inhibition of catalase) and •OH scavenging (by inhibition of GCL and glutathione synthesis) and inhibition of the enzymatic dismutation of O₂ ^•– to H₂O₂ (by SODs) significantly increase the susceptibility of cells to PDT [166]. In conclusion, the preemptive inhibition of both the bilirubin and glutathione synthesis pathways revealed a protective effect of these pathways on the survival of tumor cells following PDT, altogether indicating that the NRF2 pathway counteracts the cytotoxicity of PDT.

NRF2 is the main trigger for the antioxidant stress response that restores the intracellular redox status toward normophysiological levels in PDT-surviving cells. The antioxidant stress response is activated by oxidative stress (Section 3.1.1) and culminates in the neutralization, modification, and cellular export of oxidized/oxidizing compounds and/or potentially hazardous products of oxidation reactions (Section 3.1.2). Given the experimental evidence that NRF2 is activated following PDT (Section 3.1.3) and that inhibition of NRF2-upregulated processes potentiates the efficacy of PDT (Section 3.1.4), NRF2 seems to be an important mediator of tumor cell survival following PDT.

It is important to realize that the short-lived ROS produced during PDT cannot be scavenged by antioxidants produced downstream of the NRF2 signaling pathway since these are produced long past the half-lives of these ROS, unless there is constitutive overexpression of this pathway. Rather, NRF2 may act as an essential factor for PDT-surviving tumor cells to restore the redox imbalance and promote prolonged survival in a post-PDT microenvironment. Moreover, since NRF2-upregulated proteins HO-1, MDR1, and ABCG2 are often upregulated in many cancer types, NRF2 is likely constitutively active in tumor cells, potentially desensitizing these cells to PDT and thereby playing an instrumental role in also neutralizing the first wave of ROS directly produced by PDT. Therefore, NRF2 inhibition strategies aimed at preventing NRF2 activity prior and/or post-PDT may prove to be beneficial for the enhancement of PDT efficacy as a result of impaired tumor cell adaptation to oxidative stress.

NF-κB comprises a family of proteins that include reticuloendotheliosis (REL) A, RELB, and c-Rel, as well as NF-κB1 and NF-κB2 [171, 172]. Two types of heterodimeric complexes can be formed from these proteins, each induced by different stimuli. NF-κB transcription factors composed of RELA, c-REL, and NF-κB1 are activated in the presence of proinflammatory cytokines and/or hypoxia. NF-κB complexes composed of RELB and NF-κB2 are induced solely by TNF-α. Both complexes mediate the transcription of similar target genes that contain κB elements in their promoter region and thus initiate an inflammatory response to, e.g., ROS and TNF-α [172]. Under normal conditions, NF-κB transcription factors are retained in the cytosol by inhibitors of κB (IκB) [168]. NF-κB is activated when IκB is phosphorylated by the IκB kinase (IKK) complex at Ser32 and Ser36, which results in the ubiquitination and degradation of IκB and corollary release and nuclear translocation of NF-κB [172]. Accordingly, the IKK complex plays a major role in the activation of NF-κB. The IKK complex is able to deactivate the IκB protein in response to three independent factors, namely in response to ROS, hypoxia, and TNF-α (Fig. 4).

The different NF-κB transcription factor complexes essentially share the same target genes that are associated with cell proliferation, inflammation, angiogenesis, and survival [172] (Fig. 4). NF-κB transcription factors induce cell proliferation (upregulation of cyclin D1 (encoded by CCND1) and VEGF); cause inflammatory cells to be recruited toward the tumor site (via the production and secretion of interleukin (IL)-1α/β, IL-2, IL-6, IL-8 (CXCL8), granulocyte-macrophage colony stimulating factor (GM-CSF, CSF2), TNF-α, cluster of differentiation 40 ligand (CD40LG), chemokine C-X-C motif ligand (CXCL) 2, and cyclooxygenase 2 (COX-2, encoded by prostaglandin synthase 2 (PTGS2)); trigger angiogenesis by upregulation of matrix metalloproteinases (MMPs), VEGF, and PTGS2; and facilitate inflammatory cell binding via selectin E (SELE), intercellular adhesion molecule (ICAMs), and vascular cell adhesion molecule (VCAMs) [168, 172, 191‐193]. The role of NF-κB target gene products ICAM and VCAM appears to be controversial insofar as PDT reduced gene and protein expression levels despite activation of NF-κB [194, 195]. Of the inflammation-associated proteins, IL-6 plays an important role in tumor cell survival following PDT, as discussed in Section 3.2.2.4 IL-6, whereas TNF-α is also directly responsible for inducing cell death via apoptosis and necrosis pathways, as discussed in Section 3.2.2.3 TNF-α. To ensure survival of immune cells in a hypoxic environment, NF-κB desensitizes cells to apoptosis through the upregulation of cIAP1 (baculoviral inhibitor of apoptosis repeat-containing 2, BIRC2), cIAP2 (BIRC3), and survivin (BIRC5) as well as CFLAR, COX-2, and antiapoptotic members of the BCL2 family (BCL2A1, BCL2L1) [192, 196]. Especially survivin and COX-2 have been implicated in cell survival following PDT (Sections 3.2.2.1 COX-2 and 3.2.2.2 Survivin). In addition to these antiapoptotic proteins, NF-κB triggers HIF1A transcription that promotes immune and tumor cell survival in a hypoxic environment as a result of the upregulated production of HIF-1 transcription factor [197] (Section 3.3). NF-κB further initiates a negative feedback loop toward its own activity by inducing the expression of IκB subunits and the NF-κB inhibitor A20 [172, 198].

Overall, NF-κB stimulates tumor cell survival by inhibiting apoptosis and recruiting the immune system to facilitate angiogenesis and promote cell proliferation. The induction of NF-κB and the consequent production of cytokines may also be essential to the antitumor immune response (Section 2.2.3), which is essential for complete tumor eradication [76, 77] and long-term deterrence of tumor regrowth [199].

NF-κB is one of the major transcription factors induced by PDT [194, 195, 235‐239], although in some instances NF-κB was also found to be downregulated following PDT, such as in nasopharyngeal carcinoma (hypericin as photosensitizer) and breast cancer cell lines (C-phycocyanin as photosensitizer) [240, 241]. Despite the elusive NF-κB activation mechanism(s) in case of PDT, it is clear that NF-κB activation does occur after PDT on the basis of findings concerning at least two downstream targets of the NF-κB transcription factor, namely COX-2 and survivin. COX-2 mRNA and protein levels as well as COX-2 activity were increased after PDT in a multitude of studies [202, 239, 242‐246], albeit COX-2 activity was not necessarily attributed to NF-κB activation [247] but rather to IL-6 or p38^MAPK signaling [243, 244, 248]. Similarly, survivin (Section 3.2.2.2 Survivin) was upregulated and phosphorylated after PDT in a number of studies [249‐253]. This upregulation was most likely mediated by E2F and STAT3 transcription factors [254], which are indirectly activated by PDT through growth factors (e.g., epidermal growth factor (upregulated via the ASK1-AP-1 pathway, Section 3.4.2.2 Prolonged downstream effects of ASK1 activation) and VEGF) and cytokines (IL-6) downstream of the HIF-1 and NF-κB pathways (Section 3.2.2). IL-6 functions as a survival factor and also as a regulator of the antitumor immune response after PDT by activating STAT3 and COX-2. Presently, it is unclear whether inhibition of IL-6 signaling by for example blocking AP-1 and/or NF-κB is beneficial or detrimental to tumor response. Several studies have explored the function of IL-6 following PDT, but the investigations have yielded contradictory results. First, expression levels of IL-6 vary depending on the cell line, at least in case of nasopharyngeal cancer cell lines. Whereas CNE-2 cells showed a 13-fold increase in IL-6 mRNA levels compared to untreated cells, HK-1 cells exhibited only a 1.4-fold increase in IL-6 mRNA levels 6 h post-PDT. The effect of IL-6 overexpression on the response to PDT was not investigated [255]. Secondly, the outcomes regarding the prosurvival or prodeath role of IL-6 are conflicting. On the one hand, IL-6 stimulated tumor cell survival and negatively regulated the antitumor immune response in mice bearing Colo26 xenografts [256]. Similarly, IL-6 induction by PDT was associated with cell death inhibition and enhanced tumor growth in human basal cell carcinoma (BCC-1/KMC) cells [247] and mice bearing subdermal Co26 murine colon carcinomas or 4T1 mammary carcinomas [256]. On the other hand, a beneficial effect of IL-6 overexpression for PDT has been reported. Tumor growth in mice was reduced by IL-6 in human prostate cancer (LnCAP) xenografts [257] and human neuroblastoma (WAC2) xenografts [258]. Similarly, mice bearing Lewis lung carcinomas were more susceptible to PDT when the cells overexpressed IL-6 [259]. In a clinical setting, high levels of IL-6 following PDT of cholangiocarcinomas correlated positively with increased tumor mass, indicating that elevated IL-6 levels enhance tumor growth and/or recurrence following PDT [260]. With respect to transcriptional regulation of MMPs after PDT, the AP-1 transcription factors FOS and ATF2 that are activated in the ASK1 survival pathways (Section 3.4.2.1 Acute downstreameffects of ASK1 activation) as well as NF-κB are able to upregulate the expression of MMP1, MMP2, MMP3 [261, 262], and MMP9 [263]. However, the regulation of MMPs following PDT is ambivalent. For example, MMPs 1–3 were upregulated or activated in HK-1 nasopharyngeal carcinoma cells, MCF-7 mammary carcinoma cells and xenografts, in a Walker carcinosarcoma model, and in keratinocyte-associated fibroblasts that were subjected to PDT [240, 264‐266]. Furthermore, long-term upregulation of MMP9 but not MMP1, MMP3, MMP7, and MMP12 was observed in actinic keratosis patients treated by PDT [267]. Conversely, several studies have reported the downregulation of MMPs after PDT, including MMP2 and MMP9 in human nasopharyngeal carcinoma, oral cancer, medulloblastoma, and glioma cell lines (HK-1; UP and VB6; MED TE-671; and U87 and GBM6840 cells, respectively) [240, 268‐270]. The reduction in MMP2 and MMP9 levels was associated with retarded tumor cell migration [268, 269]. Furthermore, human glioma spheroids treated with PDT exhibited reduced MMP7 and MMP8 levels, a depolarized morphology, and a reduced migration and invasion capacity compared to untreated spheroids [271].

The contribution of NF-κB to the cell survival response appears to be well-established based on the studies that have demonstrated NF-κB activation following PDT (Section 3.2.3). At least three possible mechanisms are responsible for NF-κB activation after PDT (Section 3.2.1) and pharmacological interventions in the NF-κB survival pathway are possible to improve PDT outcomes (Section 3.2.4). However, such interventions may present a therapeutic quagmire. On the one hand, the downstream targets of NF-κB are instrumental for tumor cell survival following PDT, such as COX-2 and survivin, of which the inhibition results in increased tumor cell death and better tumor control (sections 3.2.4.2 Inhibition of COX-2 and 3.2.4.3 Inhibition of survivin). On the other hand, many proinflammatory cytokines are upregulated by NF-κB that can attract cells of the innate and adaptive immune system to mediate an antitumor immune response. Interfering with the capability of the treated cancer cells to produce a variety of cytokines and chemokines may therefore inhibit the antitumor immune response and reduce long-term therapeutic efficacy. In contrast to the postulations, it was recently shown that inhibition of NF-κB resulted in increased cytokine release and immunogenicity of PDT-treated tumor cells in vitro [274] and suggested that NF-κB may not be a suitable target for pharmacological inhibition in conjunction with PDT. These contrasting results demonstrate that further research on the in vitro and in vitro consequences is pivotal to understand the complex functions of NF-κB in a post-PDT tumor microenvironment.

The HIF-1 transcription factor is a basic helix-loop-helix (bHLH) heterodimeric protein composed of an α subunit (HIF-1α or HIF-2α) and a β subunit (HIF-1β) subunit [295]. HIF-1α is constantly transcribed but retained in the cytosol and rapidly degraded under normophysiological conditions. HIF-1β is constitutively expressed in the nucleus, where it is separated from its dimerization partner HIF-1α in the cytosol and thus kept inactive. Upon stabilization, HIF-1α translocates to the nucleus, dimerizes with HIF-1β, and binds DNA at hypoxia responsive elements (HREs) to initiate target gene expression [296, 297]. The effects of HIF-1 activation are profound, since over 500 genes are known to be a direct target of HIF-1. Moreover, HIF-1 is involved in chromatin remodeling complexes and microRNA expression that regulate gene expression at an epigenetic level [298‐301]. There are at least four different mechanisms by which HIF-1α may become activated after PDT, namely hypoxia, ROS, NF-κB, and COX-2. The pathways are summarized in Fig. 5.

When the growth of the tumor parenchyma is more extensive than the formation of new blood vessels, the deficiency in oxygen supply will trigger HIF-1 activation that in turn signals a metabolic switch to glycolysis and a consequent decrease in oxygen demand. The consequent decrease in O₂ consumption expands the area of O₂ availability within the tumor, allowing distally situated tumor cells (relative to the vasculature) to proliferate, which benefits tumor growth as a whole [297]. Inasmuch as persistent hypoxia can only be resolved by the formation of new blood vessels, HIF-1 signaling is programmed to stimulate angiogenesis [316] (Fig. 5). The vascularization of a tumor requires degradation of the extracellular matrix to enable vessel sprouting, migration, and maturation of mesenchymal cells into endothelial cells; tube formation; and pericyte recruitment to endothelialize the newly formed lumens (reviewed in [317]). Therefore, a hypoxic tumor microenvironment and the HIF-1 transcription factor are important mediators of cell survival and tumor regrowth following therapy.

With respect to glucose metabolism, tumor cells and tumor-associated cells become less dependent on oxygen during hypoxia by reducing oxidative phosphorylation and increasing anaerobic respiration (i.e., glycolysis; Warburg effect) [318]. HIF-1 is instrumental in this transformation by initiating the transcription of genes involved in glucose metabolism. The target gene products include glucose transferases 1 and 3 (GLUT1/3, SLC1A1/3), hexokinase (HK, HK1), lactate dehydrogenase A (LDHA), monocarboxylate transporters (MCTs, SLC16As), pyruvate dehydrogenase (PDH), pyruvate kinase (PKM), phosphofructokinase L (PFKL), and phosphoglycerate kinase I (PGK1) (reviewed in [297] and [296]) (Fig. 5). Despite the prevailing state of hyponutrition as a result of PDT-induced vascular shutdown, residual viable tumor cells may scavenge glucose from the tumor microenvironment to support anaerobic respiration. This glucose may have been released from tumor cells immediately killed by PDT to support anaerobic respiration. Intratumoral angiogenesis, endothelial cell proliferation, and matrix and vascular remodeling are modulated by HIF-1 via upregulation of VEGF, endothelin 1 (EDN1), plasminogen activator inhibitor 1 (PAI1, SERPINE1), (inducible) nitric oxide synthase 2 (NOS2), angiopoietin (ANGPT) 1 and 2, erythropoietin (EPO), and transforming growth factor (TGF)-β3 (TGFB3) [299, 319] (Fig. 5). Proliferation of tumor and tumor-associated cells is stimulated by HIF-1 through the induction of genes encoding insulin-like growth factor (IGF) 2 as well as IGF binding proteins 1, 2, and 3; TGF-α and TGF-β3; and VEGF [296, 297] (Fig. 5). In this process, COX-2, which is a target gene of HIF-1 (Section 3.3.1.4 HIF-1 activation by COX-2), orchestrates a positive feedback loop that reinforces the activity of both COX-2 and HIF-1 [201] (Fig. 5). PGE₂ is produced by COX-2 and enhances HIF1A transcription and induction of HIF-1, which subsequently binds the COX-2 promoter to upregulate its expression [201]. Taken altogether, HIF-1 potentiates numerous critical biological responses to PDT that revolve around tumor cell survival and enables cells to cope with and recover from the damage caused by PDT. Lastly, HIF-1 has been shown to have notable effects on cell death pathways. In addition to transcriptionally upregulating survivin (BIRC5) (Section 3.2.2.2 Survivin) and HO-1 (Section 3.1.2), HIF-1 regulates prosurvival proteins of the BCL2 family (BCL2 (BCL2A1), BCL-XL (BCL2L1), BID, and MCL-1 (MCL1)) (Fig. 5), although proapoptotic members of the same family have also been reported to be upregulated by HIF-1, including BCL2-homologous antagonist killer (BAK), BAX, BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (BNIP3), BNIP3 ligand (BNIP3L), and NOXA (phorbol-12-myristate-13-acetate-induced protein 1, PMAIP1) [320]. However, HIF-1-mediated induction of BNIP3 and BNIP3L may also be essential for hypoxia-driven cytoprotective autophagy and facilitate hypoxic survival, at least in human prostate cancer (PC-3) cells [321]. Furthermore, HIF-1 has been implicated in the stabilization of tumor suppressor protein p53 [322], which promotes apoptosis upon oncogenic stress and negatively regulates HIF-1α stability (Fig. 5) [323]. Although HIF-1 predominantly stimulates survival through various biological processes, proapoptotic signaling may also occur in the presence of DNA damage via p53-mediated activation of proapoptotic BCL2-family members that are upregulated by HIF-1.

Although HIF-1 is considered an important transcription factor in the context of PDT [17], very few studies have investigated HIF-1 activity following PDT. Chemical induction of HIF-1 by preincubating human Het-1 esophageal cells with 500 μM CoCl₂ desensitized cells to ALA-PDT [324]. Mitra et al. demonstrated that HIF-1 is activated by porfimer sodium-PDT in murine breast cancer (EMT-6) cells transfected with a gene encoding green fluorescent protein (GFP) under the control of a promoter sequence with five HREs [293]. The expression of GFP after PDT occurred under normoxic conditions, underscoring the relevance of ROS-mediated activation of HIF-1 in the absence of hypoxia (Sections 3.3.1.2 HIF-1 activation by ROS and 3.3.1.3 HIF-1 activation by NF-κB). The authors argued that PGE₂ synthesized by COX-2 (Section 3.3.1.4 HIF-1 activation by COX-2) may be an important mediator of HIF-1 activity, although no corroborative evidence was obtained in COX-2 inhibition experiments [293]. The technical difficulties in studying HIF-1 in an in vitro PDT setting result from the requirement for hypoxic culture conditions and the short half-life of HIF-1α under normoxic conditions (5–8 min) [325]. Despite these difficulties, Krieg et al. showed increased HIF-1α protein expression following ALA-PDT in UROtsa, RT112, and J82 (but not RT4) human bladder cancer cells under normoxic conditions using reversed phase protein arrays [292]. Stabilization and activation of HIF-1 under hypoxic conditions was recently demonstrated in human epidermoid carcinoma (A431) and human extrahepatic cholangiocarcinoma (Sk-Cha1) cells after PDT with liposomal zinc phthalocyanine. In line with HIF-1α stabilization, VEGF, PTGS2, and HMOX-1 mRNA were upregulated to a greater extent after PDT than in untreated hypoxic cells (Broekgaarden, M. et al., Nano Research, in resubmission; Weijer, R. et al., Oncotarget, in resubmission). Additional evidence for the prominent role of HIF-1 in PDT was provided in a mouse model of Kaposi’s sarcoma using porfimer sodium-PDT. Tumors collected 1 h after PDT exhibited increased HIF-1α protein levels compared to untreated tumors. The HIF-1α protein levels in PDT-treated tumors were comparable to those in tumors of which the blood supply had been clamped for 30 min [326]. Similar results regarding HIF-1 activation were obtained in human nasopharyngeal carcinoma (CNE-2) xenografts in mice that had been subjected to hypericin-PDT [246] and in rat chorioretinal tissue treated with verteporfin-PDT [294]. The increased mRNA expression and protein levels of HIF-1α were associated with increased protein levels of VEGF, as was demonstrated in a murine model of mammary (BA) carcinoma treated with porfimer sodium-PDT [291], indicating that post-PDT HIF-1 signaling induces angiogenic remodeling of the affected tissue (Section 3.3.2 and Fig. 5). Moreover, a clinical study by Koukourakis et al. revealed that esophageal tumors with high intratumoral protein levels of HIF-1 were more resistant to PDT compared to tumors with low HIF-1 protein levels [327], attesting to the involvement of HIF-1-mediated survival pathways following PDT (Section 3.3.2 and Fig. 5). Increased HIF-1α protein levels were also observed in mouse porfimer sodium-PDT-treated murine BA mammary carcinoma tumors, but this was not reported for porfimer sodium-PDT-treated BA cells in vitro [250].

Due to the importance of HIF-1 in tumor survival, therapeutic interventions for cancer encompass the inhibition of HIF-1 [290]. However, most HIF-1 inhibitors are rather unspecific and also target the upstream modulators of HIF-1α protein synthesis, of which imatinib (an inhibitor of breakpoint cluster region protein (BCR)-ABL [328]), gefitinib, erlotinib, and cetuximab (an inhibitor of EGFR [329]), and everolimus (an inhibitor of mTOR [330]) are well-known examples [290] (Table 1). Another combination strategy is to interfere with the stabilization of HIF-1 by inhibition of chaperone binding using geldanamycin (an inhibitor of HSP90 [331]) or increasing the affinity for natural inhibitors of HIF-1 (e.g., amphothericin B [148]) (Table 1). Interfering with HIF-1 DNA binding is another approach to reduce HIF-1 signaling. For example, echinomycin competes with HIF-1 to bind to HREs and can therefore be used to reduce transcriptional activity of HIF-1 [149] (Table 1). As mentioned previously, these inhibitors are rather unspecific, which may be valuable in the development of a combinatorial cancer therapy. However, a more specific inhibitor of HIF-1 would be desirable when investigating the mechanism of HIF-1 on tumor cell survival following PDT.

α-Ketoglutarate may be a useful drug as a specific inhibitor of HIF-1 (Table 1). Under normophysiological conditions, PHDs are the major inhibitors of HIF-1 activity during normoxia but are rendered dysfunctional during hypoxia [332] (Section 3.3.1 and Fig. 5). The endogenous molecule α-ketoglutarate is a selective PHD substrate and agonist [312], and it is able to reactivate PHDs to inhibit HIF-1 regardless of intracellular oxygen tension [141]. Under normoxic conditions, PHDs facilitate the conversion of α-ketoglutarate and oxygen to succinate and carbon dioxide, respectively, but also transfer oxygen to prolyl residues in the HIF-1α oxygen-dependent degradation domain (ODD) [312]. Increasing the activity of PHDs after PDT with α-ketoglutarate may therefore render cells less susceptible to HIF-1-mediated survival. Studies by Mackenzie et al. have shown that, despite hypoxia, the activity of PHD2 and 3 and the concurrent destabilization of HIF-1 in various tumor cell lines and murine xenografts could be induced by the administration of α-ketoglutarate esters (esterification allows passage through the membrane into the cell) [141]. The inhibition of HIF-1 by α-ketoglutarate was associated with decreased tumor growth and increased apoptosis [277, 333]. Based on these investigations, HIF-1 inhibition by α-ketoglutarate may be a valuable strategy in potentiating the effects of PDT. However, recent studies by our group revealed that α-ketoglutarate did not increase the efficacy of PDT, but rather reduced PDT-induced oxidative stress as measured 4 h post-PDT in A431 cells. It was hypothesized that α-ketoglutarate was used as an energy source to fuel antioxidant responses or that it functions as an antioxidant itself and thus is not a suitable agent to enhance the PDT response (Broekgaarden, M. et al., Nano Research, in resubmission; Weijer, R. et al., Oncotarget, in resubmission).

Another nonspecific inhibitor of HIF-1 is the naturally occurring diphenolic compound curcumin (Table 1), which was found to promote proteasomal degradation of HIF-1α and HIF-1β (also known as aryl hydrocarbon receptor nuclear translocator, ARNT) [150, 334, 335]. Moreover, Choi et al. [150] showed that HIF-1 activity was reduced as a result of curcumin-stimulated oxidation and proteasomal degradation of HIF-1β. The authors further showed that the inhibition of HIF-1β was dependent on oxidation, as addition of the antioxidant N-acetylcysteine prevented HIF-1β inactivation by curcumin. However, it should be noted that curcumin exerts many other mainly cytostatic/toxic effects on tumor cells, including the inhibition of EGFR tyrosine kinase activity and downstream signaling; inhibition of protein kinase C, COX-2, and NF-κB; and induction of DNA damage [336]. These effects tend to increase PDT efficacy, as exemplified by the finding that curcumin increases ROS production, mitochondrial membrane permeabilization, mitochondrial cytochrome c release, and caspase activation after porfimer sodium-PDT, which exacerbated cell death in human head and neck cancer (AMC-NH3) cells [337]. In contrast, curcumin is also a potent antioxidant [336], a property that was found to reduce the extent of ROS production and thereby the degree of cell death in A431 human epidermoid carcinoma cells treated with rose bengal-PDT [338].

A recently discovered and rather specific inhibitor of HIF-1 activity is acriflavine (Table 1), which inhibits HIF-1α/HIF-1β and HIF-2α/HIF-1β dimerization by binding the Per-Arnt-Sim B domain of HIF-1α and HIF-2α [151]. The binding of acriflavine to HIF-1 blocks DNA binding and reduces transcriptional activity, tumorigenicity, and angiogenic signaling, as was demonstrated in xenotransplanted human prostate (PC-3) and hepatoma (Hep3B) tumors in mice [151]. Results from our lab confirm the feasibility of employing acriflavine to improve PDT efficacy, as the sensitivity of A431 cells to PDT increased as a result of reduced glycolytic activity; downregulation of the HIF-1 target genes VEGF, PTGS2, and PAI1; and upregulation of EDN1. Taken together, these results illustrate the importance of HIF-1 in the protection of tumor cells from PDT (Broekgaarden, M. et al., Nano Research, in resubmission; Weijer, R. et al., Oncotarget, in resubmission).

The HIF-1 transcription factor is essential for the survival of cells in a hypoxic environment and under conditions of acute stress, such as oxidative stress induced by PDT. The survival signaling manifests itself through direct intracellular events (e.g., by the production of prosurvival proteins and the shift to oxygen-independent metabolism) as well as through tissue processes (e.g., angiogenesis and proliferation) (Section 3.3.2). In case of PDT, it has been shown that HIF-1 activation causes the release of proangiogenic factors (Section 3.3.3).

Since malignant tumors proliferate faster than the rate of neovascularization, most tumors are hypoxic in nature and constitutively activate HIF-1 [316]. Poorly vascularized tumors may therefore be more resistant against PDT due to hypoxic preconditioning (in addition to the suboptimal accumulation of systemically administered photosensitizer molecules as a result of the poor blood supply). Tumors that overexpress HIF-1 are less sensitive to therapy and are associated with poor survival in patients. Accordingly, the coadministration of HIF-1 inhibitors as neoadjuvants increases the efficacy of PDT, as has been demonstrated in several studies (Section 3.3.4).

Direct ASK1 activation following PDT has never been demonstrated, so the involvement of this pathway in response to PDT can only be deduced from the effects on downstream kinases and transcription factors. Considerable increases in c-FOS and c-JUN mRNA levels were found after porfimer sodium-PDT of RIF-1 cells. Levels of mRNA peaked 90 min post-PDT, after which mRNA levels gradually dropped to baseline during the subsequent 8 h [406]. Furthermore, protein kinase inhibitors such as staurosporine effectively blocked the synthesis of c-FOS mRNA, hinting toward the involvement of upstream kinases p38^MAPK and JNK [406]. Activation of JNK and p38^MAPK, but not other MAPKs such as ERK1 or ERK2, in benzoporphyrin derivative-PDT-subjected murine PAM212 keratinocytes confirmed the involvement of the AP-1 response. The activation of JNK and p38^MAPK was abrogated in the presence of antioxidants [407]. Activation of the AP-1 response in the PAM212 keratinocytes was further confirmed by ATF2 and JUN phosphorylation following PDT [407].

Since the discovery of their involvement in the response to PDT, the effects of the MAPKs and AP-1 proteins on the biological fate of PDT-affected cancer cells have been elusive or inconsistent. Two types of approaches can be used to inhibit the ASK1 pathway. The first approach is to inhibit the function of AP-1 transcription factors, while the other method prevents AP-1 activation through inhibition of the upstream kinases JNK and p38^MAPK. Direct chemical inhibition of AP-1 transcription factors is possible with retinoic acid and its analog SR11302 (Table 1). More specifically, retinoic acid or retinoic acid analogs are able to inhibit ATF2 [408] and c-Jun [409]. It was shown that these inhibitors were efficient in blocking AP-1 activity, which prevented tetradecanoyl phorbol acetate-induced papilloma formation in mice [152, 410]. This illustrates the potential for a treatment modality in which retinoic acid is used to sensitize tumor cells to PDT. However, to date no studies have employed direct inhibition of AP-1 in conjunction with PDT.

Indirect inhibition of AP-1 activation has been achieved with inhibitors of JNK and p38α/β. JNK1 can be inhibited by the anthrapyrazole SP600125 (Table 1), even though this compound also mildly impairs the function of p38β and ERK2 [153]. The pyridinyl imidazole derivatives SB202190, SB203580, and PD169316 have been used for the selective inhibition of p38α/β, but not the δ or γ isozymes (Table 1). The JNK and p38^MAPK inhibitors all compete with ATP for the ATP-binding domain of the kinases, resulting in the reversible inhibition of kinase activity [154]. However, the roles of these kinases are rather unpredictable and the effects of inactivation are dependent on the cell type and the extent of damage. The contradictory effects that these MAPKs can have on tumor cell survival following PDT are shown in Table 2. The data reveal that p38^MAPK and JNK induce both survival and apoptosis but also suggest that the inhibition of p38α/β augments tumor cell death, whereas inhibition of JNK either has no effect or impairs PDT-induced cell death. This is in line with the dichotomous effects of JNK1 activation and the cytoprotective function of p38α/β as described in Section 3.4.2.2 Prolonged downstream effects of ASK1 activation, in which p38α/β mainly functions as activator of AP-1 and facilitators of transcription. Therefore, it can be deduced that JUN and ATF2 (both downstream of JNK) do not play a significant role in cell survival, but FOS (only activated by p38^MAPK) may in fact be a decisive factor for cell survival. Alternatively, in a recent study by Rubio et al., p38^MAPK was implicated in the removal of ubiquitin aggregates via autophagy and activation of NRF2 after hypericin-PDT, which resulted in increased survival of fibroblasts [386].

The ASK1 pathway is one of the most complicated pathways activated by PDT. Although ASK1 itself has, to our knowledge, never been investigated in the context of PDT, many of its downstream targets have often been implicated in cellular responses to PDT. However, the effects of this pathway are highly divergent, ranging from stimulation of survival to causing inflammation and stimulation of cell death. It is arguable that JNK1 has a particularly important role in the stimulation of apoptosis in cells exposed to severe and prolonged oxidative stress or concomitant TNF-α signaling. Experimental evidence regarding the inhibition of JNK-1 and p38α/β is in agreement with this hypothesis. Survival signaling as a result of transient JNK and p38^MAPK activity may arise from the phosphorylation of AP-1 transcription factors and the triggering of the immediate early survival response. Therefore, given the dichotomous nature of the ASK1-MAPK pathway, it is hypothesized that pharmacological inhibition of AP-1 transcription factor activation could improve PDT efficacy, while the function of the MAPKs JNK and p38^MAPK should remain intact.

Of the four HSFs isoforms 1–4, HSF1 is the most important [421] and will be discussed in the context of the proteotoxic stress response. Under nonstressed conditions, HSF1 forms inactive monomers in the cytoplasm that are acetylated by p300/CREB binding protein (CBP). HSP40, 70, and 90 form complexes with HSF1 and negatively regulate either its transactivation mechanism (HSP90) or its DNA-binding capacity (HSP40 and HSP70) [421, 425]. Whenever HSPs need to be engaged to combat proteotoxic stress by ROS or hypoxia, HSPs dissociate from the HSF complex and bind to misfolded proteins. This binding relieves the negative regulation of HSF, leading to HSF homotrimerization and activation [421]. Moreover, stress-induced sirtuin (SIR2) deacetylates HSF and maintains the complex in a state capable of binding DNA [425] (Fig. 9). Additionally, JNK2 has been implicated in phosphorylating the transactivation domain of HSF1, contributing to its activation and subsequent binding to genomic heat shock elements [426]. Other sensors of the ER stress response include IRE1, ATF6, and PERK. These proteins are embedded in ER membranes and are activated when their chaperones (mainly HSP70A5, also known as glucose-regulated protein 78 (GRP78) and binding immunoglobulin protein (BIP)) are recruited away to bind to unfolded proteins (Fig. 10).

PDT was found to activate HSF [442, 443] and stimulate the production of HSP70, HSP47, HSP60, and HSP27 [442, 444‐450]. Furthermore, high levels of HSP27, HSP60, HSP70, and HSP90 were linked to reduced susceptibility of tumor cells to PDT in vitro and in vivo [250, 444, 448, 450, 451]. The cytoprotective properties of HSPs after PDT likely arise from the alleviation of proteotoxic stress that ensues protein oxidation. The induction of ER stress by PDT was studied by Szokalska et al., who showed that porfimer sodium-PDT leads to extensive protein carbonylation, polyubiquitination, and widening of the ER lumen [27]. Moreover, PDT induced XBP1 activation and upregulation of HSP70A5 and calnexin, whereby ERAD was crucial for the survival of EMT-6 and HeLa cells after PDT in vitro and in vivo [27]. A microarray looking at the transcriptional response of T24 bladder cancer cells to hypericin-PDT showed significant mRNA upregulation of HSP70A5, 40, 47, 60, 90, 110, CHOP, GADD34, XBP1, PERK, ATF3, and ATF4, providing compelling evidence for the involvement of HSF1 and the UPR in response to PDT [124]. A follow-up study by the same group suggested that PERK-upregulated NOXA was the main instigator of tumor cell death after hypericin-PDT-induced ER stress [452].

With respect to the antitumor immune response, HSP70 in particular has been implicated in immune cell modulation after PDT [199, 422]. Apoptotic cells expressed HSP70 on their plasma membrane after PDT [445], most likely in an attempt to stabilize the plasma membrane or chaperone integral membrane proteins [422]. Moreover, HSP70 can bind protein fragments derived from tumor-specific antigens such as mutated, truncated, or misfolded proteins. When expressed on the plasma membrane or released from necrotic cells, these HSP70/tumor antigen complexes can be taken up by dendritic cells to induce their maturation, activation, and migration to lymph nodes, where they can initiate cross-presentation to naive T-cells and stimulate the formation of tumor-specific CD8⁺ cytotoxic T cells (reviewed in [199]). A similar mechanism in favor of dendritic cell activation has been ascribed to CRT (a downstream gene product of ATF6), which aids in ER protein folding. When CRT is expressed on the outer leaflet of the plasma membrane (ecto-CRT), it is associated with the induction of immunogenic cell death that stimulates antigen presentation and the activation of immune cells. Following hypericin-PDT (hypericin localizes to the ER), T24 cells expressed ecto-CRT in a PERK-dependent manner. Moreover, PERK was essential for the immunogenicity of CT26 cells treated with hypericin-PDT in vivo [453]. Thus, the proteotoxic stress response seems to play a key role in mediating an antitumor immune response.

It may be postulated that inhibition of HSPs would be beneficial when beneficial for PDT outcome given the role of HSPs in tumor cell survival [422]. However, their role in promoting the antitumor immune response also suggests that HSP inhibition may be detrimental [199]. Ferrario and Gomer used the geldanamycin derivative 17-AAG [155] to inhibit the function of HSP90 (Table 1) during PDT and found a reduction in protein levels of survivin, VEGF, phospho-AKT, and BCL2. A higher cure rate and long-term survival were observed in BA mammary tumor-bearing mice treated with PDT combined with 17-AAG [250, 252]. HSP70 inhibition with the bacterial cytotoxin SubA fused to EGF [160], (Table 1) was recently shown to augment the efficacy of porfimer sodium-PDT in human SW-900 lung cancer cells and DU-145 prostate cancer cells as a result of increased ER stress [454]. Taken together, these results point toward the beneficial effect of HSP inhibition in the enhancement of PDT efficacy. Besides 17-AAG, other HSP90 inhibitors are available and include different geldanamycin derivatives, although these may be associated with liver toxicity [455], as well as the synthetic small molecules CNF-2024/BIIB-021, NVP-AUY922, SN-X5422, and STA-9090 (Table 1), which are undergoing clinical trials [156‐159, 456]. However, inhibition of HSP90 typically exacerbates proteotoxic stress that induces HSP70 proteins [457] and may therefore alleviate any beneficial effects of these agents in terms of tumor cell death.

Alternatively or in addition to HSP90 inhibition, HSP70 inhibitors are also available. Schlecht et al. recently demonstrated the inhibition of HSP70 and HSC70 (a constantly expressed isozyme of HSP70) using VER-155008, a compound that binds the nucleotide binding domain of these proteins and reduces their ATPase activity (Table 1). In RNAi knockdown experiments, it was shown that concomitant inhibition of HSP70 and HSC70 was necessary to induce tumor cell death [161]. A more effective approach to completely abolish the heat shock response is to block HSF1 activity. KRIBB11 (N ²-(1H-indazole-5-yl)-N ⁶-methyl-3-nitropyridine-2,6-diamine) is an HSF1 inhibitor that blocks the association between HSF1 and positive elongation factor b, which is required for HSF1 transcriptional activity (Table 1). Accordingly, KRIBB11 was very effective in preventing HCT-116 tumor growth in nude mice [458]. Based on these results, inhibitors of the HSF pathway could be used to elucidate the role of this pathway in PDT and may provide promising approaches to improve PDT efficacy.

During ER stress, cells deal with the accumulation and aggregation of carbonylated proteins by polyubiquitination and proteasomal degradation. Therefore, Szokalska et al. investigated whether inhibition of the proteasome could exacerbate ER stress and increase the extent of cell death after PDT. Indeed, porfimer sodium-PDT on EMT-6 and HeLa cells pretreated with 4 ng/mL bortezomib (binds and inhibits the catalytic center of the 26S proteasome [162], Table 1) for 24 h increased the accumulation of carbonylated proteins and disrupted ERAD, leading to an increased sensitivity of cells to PDT [27]. Similar results were obtained for verteporfin-PDT in combination with bortezomib (2 mg/kg) in a PC-3 mouse xenograft model [459]. Thus, these results attest to the utility of pharmacological interventions in proteasome function as a means to augment ER stress and improve the therapeutic efficacy of PDT. Pharmacological inhibition of IRE1 and ATF6 (but not PERK) is possible with 4-phenylbutyric acid analogs (Table 1), although the exact mechanism has not been elucidated [163]. With respect to PERK, inhibition is possible with the synthetic compound GSK2656157 (Table 1), which competes with ATP to bind PERK specifically, and thus inhibits its kinase activity [164]. However, none of these UPR-inhibiting compounds have been investigated in combination with PDT.

Proteotoxic stress appears to be a primary response to PDT regardless of cell type and PDT strategy. The degree to which this response is triggered depends somewhat on the photosensitizer localization insofar as ER-localizing photosensitizers such as hypericin are more effective in inducing the UPR than photosensitizers that accumulate in other intracellular venues. While the functional outcome of this pathway may be both protective and destructive in tumor cells, the protective effects of the proteotoxic stress response can be pharmacologically blocked to promote tumor cell death. Inhibition of HSP70 and HSP90 was shown to increase the efficacy of PDT, as did inhibition of the proteasome by exacerbating ER stress. The HSF pathway is an essential component of the UPR in response after PDT. Given its reported induction by hypoxia and its constitutive activation in tumor cells [460], the UPR may protect tumors against anticancer therapies [424] such as PDT. Disrupting the cytoprotective effects of the UPR or interfering with the function of chaperones has been shown to enhance proteotoxic stress and stimulate cellular demise after PDT. Thus, the proteotoxic stress pathway is an important and feasible target for pharmacological interventions to enhance the therapeutic efficacy of PDT.