Immunoediting accompanies cancer progression
Tumors develop despite the ability of the immune system to recognize and eliminate tumor cells. This paradox can be explained by tumor immunoediting, a process associated to the pressure exerted by the immune system during tumorigenesis.
Immunoediting, first described by the team of Robert Schreiber, occurs in three sequential steps: elimination, equilibrium and escape. In the elimination phase, long before a tumor gets clinically detectable, emerging cancer cells are under immunosurveillance by the innate and adaptive immune system, including in particular CTLs, NK cells and proinflammatory cytokines, as previously described. This might lead to complete tumor destruction. However, some cancer variants may survive and enter the “equilibrium” phase, in which tumor outgrowth is maintained in a state of dormancy by mediators belonging to the adaptive immune system. NK cells in particular are dispensable in this phase. Under the continuous pressure exerted by the immune system on genetically unstable cancer cells, tumors undergo transformation (or immunoediting), until resistant clones may emerge, which (i) are not recognized by the immune system due to antigen variation or a defect in antigen presentation, (ii) are resistant to mechanisms of immune destruction and/or (iii) promote an immunosuppressive tumor micro-environment. In the escape phase, tumor proliferation cannot anymore be controlled by the immune system and a clinically detectable tumor forms (Dunn et al, 2004a; Dunn et al, 2004b; Schreiber et al, 2011) (Figure 3).
Figure 3. Cancer immunoediting (Schreiber et al, 2011). When tumor cells emerge, they are recognized and eliminated by the immune system, including in particular natural killer (NK), NKT cells, dendritic cells (DCs) primed CD8+ T cells, T cells, macrophages (Mᶲ) and CD4+ T cells, the pro-inflammatory cytokines interferon (IFN) , IFN , interleukin (IL)-12, tumor necrosis factor (TNF), natural killer group 2 member D (NKG2D) and the cytotoxic enzymes TNF related apoptosis inducing ligand (TRAIL) and perforin; this constitutes the elimination phase. Under the pressure of these immune mediators, tumor will undergo transformation to resist to the immune attacks; this corresponds to the equilibrium phase, a state of functional dormancy in which proliferation is controlled by mediators of the adaptive immune response. This may last for a while, until variants arise which have acquired the capacity not to be recognized and/or eliminated by the immune system; this constitutes the escape phase. CTLA-4, cytotoxic T-lymphocyte-associated protein 4; IDO, indoleamine 2,3-dioxygenase; MDSC, myeloid-derived suppressor cells; MHC, major histocompatibility complex; NKR, natural killer receptor; PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; Treg, regulatory T cell.
At this stage, the constitution of the tumor immune infiltrate indicates the outcome of the disease, with the presence of CTLs, tertiary lymphoid structures and M1 macrophages as good prognostic factors, whereas Tregs and M2 macrophages being worst prognosis factors (Fridman et al, 2017; Pages et al, 2018). It can be driven by the choice of anticancer therapy: ICB prevents from CTLs exhaustion, vaccination enhances tumor specific CD8+ T cells and immunogenic cell death inducers promote the emission of danger signals.
Adjuvanticity through immunogenic cell death
For a long time, cell death has been classified based on morphological features in a dichotomic manner. Apoptosis was described as a regulated, physiological and non-immunogenic type of cell death, whereas necrosis was considered as an uncontrolled pathological and proinflammatory mechanism. Since then, we learned that regulated cell death can exert morphological features of necrosis whereas apoptosis can trigger an immune response (Galluzzi et al, 2015a). The immunogenicity of cell death cannot be defined merely based on morphological traits, but need to be addressed with vaccination experiment which consists in the injection of dying cells into syngeneic host on their sequential rechallenge with live cells (Kepp et al, 2014). An absence of (or delayed) tumor growth in this model is an indication for anticancer vaccination efficacy and thus for the immunogenicity of cell death.
Immunogenic cell death (ICD) following pathogens infection constitutes an ancient mechanism of defense. Indeed, when cells are infected by viruses or intracellular bacteria, they sense microbe-associated molecular patterns (MAMPs) via their specific PRRs which trigger an intracellular and micro-environmental danger response and consequent activation of immune cells. In 2005, it has been demonstrated that anthracyclines can also drive a kind of cell death which induces an anticancer adaptive immune response (Casares et al, 2005). In the following years, many other chemotherapeutics, like oxaliplatin (OXA), as well as physical modalities, like irradiation, photodynamic therapy (PDT) or high hydrostatic pressure (Fucikova et al, 2014; Garg et al, 2012a; Obeid et al, 2007a), were shown to be able to trigger ICD. Similarly to their microbial counterparts, these anticancer agents induce pre-mortem stress mechanisms leading to the emission of DAMPs, recognized by PRRs expressed on DCs (Figure 4). Calreticulin (CALR), adenosine triphosphate (ATP) and high-mobility group box 1 (HMGB1) are DAMPs common to almost all instances of ICD discovered until now (Galluzzi et al, 2017). Type I IFN and annexin A1 (ANXA1), discovered in the context of chemotherapeutics-induced ICD, have been further added to the list (Sistigu et al, 2014; Vacchelli et al, 2015a), but their general involvement in ICD remains to be investigated. Even though these five DAMPs may be general features of ICD, the pathways underlying their emission is particular to each form of ICD, as exemplified by autophagy, which is involved in anthracyclines-induced ICD but displays an immunosuppressive action in the context of hypericin-based PDT (Garg et al, 2013).
Figure 4 Activation of immunogenic cell death (Galluzzi et al, 2017). Chemotherapeutics induce calreticulin (CALR) exposure at the plasma membrane, as well as adenosine triphosphate (ATP), annexin A1 (ANXA1) and high-mobility box 1 (HMGB1) release from stressed and dying cancer cells, which are recognized by low density liproprotein receptor related protein 1 (LRP1), purinergic receptors P2Y2 (P2RY2) and P2X7 (P2RX7), formyl peptide receptor 1 (FPR1) and toll-like receptor 4 (TLR4) expressed by dendritic cells (DCs) respectively. In addition, they induce the autocrine/paracrine secretion of type I interferon (IFN) leading to C-X-C motif chemokine ligand 10 (CXCL10) release. These mechanisms allow DCs recruitment, activation and maturation and elicit an adaptive immune response involving IL-17 producing T cells and type I IFN secreting T cells. In addition to the establishment of an immune memory, this has the potential to eliminate cancer cells through cytotoxic T lymphocytes (CTLs)-mediated cytotoxicity.
In the context of ICD, inflammatory DCs (CD11b+ CD11c+ Ly6Chigh) sense the emission of DAMPs and in turn secrete proinflammatory cytokines and cross-present antigens to naïve T cells, most of which differentiate in IFN producing CD8+ T cells. IL-17 producing T cells are also activated and sustain the proliferation of CD8+ T cells (Figure 4) (Ma et al, 2013; Ma et al, 2011). In addition, neutrophils are recruited through the release of DAMPs, but also through the secretion of certain chemokines, such as C-X-C motif chemokine ligand (CXCL)1, C-C motif chemokine ligand (CCL)2 and CXCL10, by dying cancer cells (Garg et al, 2017b).
Altogether, these mechanisms participate in elimination of residual cancer cells.
Some signaling events in dying cells involving in particular RIPK1 and nuclear factor B (NF- B)-induced transcriptional programs have also been reported to be required for DCs cross-presentation and consequent CD8+ T cells activation, but the precise mechanism by which they activate DCs is not well understood (Giampazolias et al, 2017; Yatim et al, 2015).
Even though ICD was first described as an apoptotic kind of cell death (Casares et al, 2005), it has been recently reported that necroptosis, a form of regulated necrosis mediated by receptor-interacting serine-threonine kinase 1 (RIPK1), RIPK3 and mixed lineage kinase domain-like (MLKL), is also able to drive an anticancer immune response (Aaes et al, 2016; Yang et al, 2016). Necroptosis can be induced in cells engineered for RIP3K3 to be inducible by doxycycline or by the combination of TNF- with a synthetic second mitochondria derived activator of caspase (SMAC) mimetic and the caspase inhibitor z-VAD-FMK. Similarly to apoptotic ICD, it triggers the emission of DAMPs (CALR, ATP, HMGB1) that confer long term protection in mice against subsequent challenge with living cancer cells of the same type. Interestingly, anthracyclines and OXA’s immunogenic properties are attributable to necroptosis in addition to apoptosis (Aaes et al, 2016; Yang et al, 2016).
The success of a specific anticancer immune response relies both on tumor antigenicity, which is related to its mutational load (Samstein et al, 2019), and on the choice of the therapy driving the antigenicity through DAMPs emission.
Mechanisms of chemotherapeutics-driven immunogenic cell death
Most ICD mechanisms discovered so far rely on the exposure of CALR, the secretion of ATP and the release of HMGB1, but the mechanisms underlying the emission of these DAMPs depend on the type of ICD. Moreover, more recently discovered hallmarks of ICD, i.e. type I IFN and ANXA1 release have been much less studied and we don’t know yet if they are specific to chemotherapeutics-induced ICD or shared with other modalities. The present work aims at finding new ICD inducers among chemotherapeutics agents. Thus, here, we will present the mechanisms of chemotherapeutics-driven ICD (Figure 5), which have been most described so far.
The canonical response to ER stress is called unfolded protein response (UPR) and involves three pathways, which are mediated by the transmembrane proteins eukaryotic translation initiation factor 2-alpha kinase 3 (eIF2AK3, also called PERK for protein kinase R (PKR)-like endoplasmic reticulum kinase), inositol-requiring enzyme 1 (IRE1α) and the activating transcription factor (ATF) 6. In normal conditions, these proteins are maintained in an inactive state by the binding immunoglobulin protein (BiP) chaperone. During ER stress, BiP dissociates from them to bind the misfolded/unfolded proteins, resulting in the activation of the three arms of UPR. EIF2AK3 is a type I ER transmembrane kinase that contains a PEK-like catalytic domain in its cytosolic C-terminal region. When dissociated from BiP, EIF2AK3 oligomerizes, autophosphorylates and promotes phosphorylation of the serine 51 on eukaryotic initiation factor 2 (eIF2α). EIF2α phosphorylation leads to messenger RNA (mRNA) translational attenuation by preventing cap-dependent ribosomal initiation complexes formation. In addition, phosphorylated eIF2α selectively activates the translation of the mRNA encoding for ATF4. ATF4 is a basic leucine zipper (b-ZIP) transcription factor that induces growth arrest and upregulates genes coding for chaperones, antioxidants, XBP1 as well as DNA damage-inducible transcript 3 (DDIT3, also known as CHOP for CCAAT/enhancer-binding homologous protein). IRE1 is also a type I ER transmembrane endoribonuclease/kinase that has a kinase domain and an endoribonuclease domain in its cytosolic N-terminal luminal domain. IRE1 exists as two isoforms: IRE1α and IRE1β. IRE1α is present in all cell types and has been widely studied. Upon ER stress conditions, it becomes active by dimerization and autophosphorylation. Activated IRE1α catalyzes the splicing of a 26 nucleotides intron from the X-box binding protein 1 (XBP1) mRNA. Spliced XBP1 encodes a b-ZIP transcription factor that upregulates UPR genes, including genes involved in ER-associated protein degradation (ERAD) as well as genes involved in protein folding. The third pathway involves ATF6, a type II ER transmembrane protein that has a transcriptional activation domain in its cytosolic region. ATF6 has two isoforms, ATF6α and ATF6β; when dissociated from BiP, ATF6α transits from the ER membrane to the Golgi where it is cleaved by the Golgi resident site 1 and 2 proteases (respectively S1P and S2P), leading to the generation of an activated b-ZIP factor. The N-terminal fragment of ATF6α is transported to the nucleus where it activates UPR genes, like chaperones, XBP1 and BiP overexpression, which promote protein folding (Figure 6-A) (Kato & Nishitoh, 2015; Oslowski & Urano, 2011).
Importantly, eIF2 phosphorylation is induced by eIFAK3 during ER stress, but can also be activated by eIF2AK1 (also called HRI for heme regulated inhibitor) upon heme deprivation, eIF2AK2 (also named PKR for protein kinase R) in response to viral infection or eIF2AK4 (also called GCN2 for general control nonderepressible 2) in the context of nutrient deprivation.
In ICD induced by chemotherapeutics, CALR exposure is preceded by the phosphorylation of eIF2 , which has been showed to be mediated by eIF2AK3 in the murine methylcholanthrene-induced fibrosarcoma MCA205, the human osteosarcoma U2OS and colon carcinoma CT26 (Bezu et al, 2018a; Obeid et al, 2007b; Panaretakis et al, 2009). However, it was recently demonstrated that in melanoma human cells, anthracyclines induce eIF2 phosphorylation via the activation of other kinases: eIF2AK2 or eIF2AK4 (Giglio et al, 2018). Importantly, during ER stress elicited by chemotherapeutics, solely eIF2 phosphorylation is involved, without the downstream expression of ATF4 and DDIT3 and without activation of the two other arms of ER stress, i.e. the translocation of ATF6 and the alternative splicing of XBP1 (Figure 6-B) (Bezu et al, 2018a). This split ER stress response was also observed for crizotinib-induced ICD in various cell lines (Liu et al, 2019b) and for anthracyclines-treated human melanoma cells (Giglio et al, 2018). In addition, following administration of the EGFR specific antibody cetuximab, ICD was abolished by overexpression of XBP1 (Pozzi et al, 2016).
Figure 6. Split ER stress response involved in chemotherapeutics driven ICD (Bezu et al, 2018b). Canonical endoplasmic reticulum (ER) stress activates three pathways: (i) protein kinase R-like endoplasmic reticulum kinase (PERK)-mediated eukaryotic initiation factor 2 (eIF2 ) phosphorylation leading to activating transcription factor (ATF) 4 translation due to particular open reading frame (ORF) arrangement, (ii) inositol-requiring enzyme 1 (IRE1) dimerization leading to X-box protein 1 alternative splicing (XBP1s) and (iii) ATF6 translocation to the Golgi where it undergoes proteolytic cleavage followed by translocation to the nucleus of the N-terminal part; all aiming at eliminating and/or repairing misfolded proteins (A). Immunogenic cell death (ICD) inducers mediate PERK-dependent eIF2 phosphorylation but none of the other protein activation (B).
CALR needs to be engaged in a complex with PDIA3 to be able to translocate from the ER to the membrane (Liu et al, 2019a; Panaretakis et al, 2008). Following eIF2 phosphorylation, B-cell receptor-associated protein 31 (BAP31) is proteolyzed in a caspase 8-dependent manner, leading to the activation of the pro-apoptotic Bcl-2–associated factors X et K (BAX and BAK) and the anterograde translocation of the CALR/PDIA3 complex from the ER lumen to the plasma membrane via the Golgi apparatus. CALR is finally secreted by soluble N-ethylmaleimide-sensitive-factor attachment protein receptor (SNARE)-dependent exocytosis (Panaretakis et al, 2009). Inhibition of any step of this pathway abolished CALR exposure. In addition, the inhibition of (i) eIF2AK3 phosphorylation with a siRNA, (ii) caspase 8 activation with a pharmacologic inhibitor (Z-VAD-fmk) or siRNA, (iii) BAX with a siRNA, (iv) the ER-Golgi traffic with brefeldin A or (iv) SNARE-dependent exocytosis with a synaptosomal-associated protein 23 (SNAP23) siRNA also abolished vaccination capacity of anthracyclines (Panaretakis et al, 2009). Of note, even though eIF2AK3-dependent CALR exposure is also required for the immunogenicity of cells treated with hypericin based PDT, this does not involve eIF2 phosphorylation, caspase 8 and PDIA3 (Garg et al, 2012b).
Additionally, CALR exposure following anthracyclines chemotherapy is not only dictated by cell intrinsic pathways, but also relies on autocrine/paracrine signalization pathways in stressed cancer cells, mediated by the CXCL8 (also known as IL-8) and its receptor CXCL2 (Sukkurwala et al, 2014b).
Eventually, it has been recently shown that CALR exposure also relies on ANXA1. Indeed, following treatment with anthracyclines, CALR exposure and the consequent tumor growth control is abolished in Anxa1-/- cells, effect which is restored by supplementation with rCALR. The underlying mechanism remains to be elucidated, but it may be related to the involvement of ANXA1 in unconventional protein secretion pathways or in intracellular trafficking (Baracco et al, 2019).
Strategies to modulate CALR exposure
Chemotherapeutics like CDDP or mitomycin C, which induce all ICD hallmarks apart from CALR exposure, are not able to induce ICD (Casares et al, 2005; Tesniere et al, 2010), which could be compensated by the co-incubation with CALR recombinant protein (rCALR) prior to vaccination (Obeid et al, 2007b; Tesniere et al, 2010). In line with this, coating with rCALR enhances phagocytosis and vaccination with apoptotic melanoma cells (Dudek-Peric et al, 2015; Qin et al, 2011). CALR trafficking can also be targeted: -integrins and PDIA3 interact with CALR in the ER and coordinate CALR translocation. Incubation of cells with 9EG7, a 1 integrin activating antibody, reduces CALR exposure; conversely, knocking down integrins leads to enhanced CALR exposure, even though this strategy can obviously not be transferred to the clinics (Liu et al, 2019a). The immunogenicity of CDDP or mitomycin C could also be restored in combination with activators of ER stress like the inhibitor of sarco/endoplasmic reticulum Ca2+ ATPase (SERCA) thapsigargin, or the inhibitor of N-glycosylation tunicamycin (Martins et al, 2011). Similarly, pyridoxine, a precursor of the bioactive vitamin B6 as well as zinc supplementations were able to restore immunogenicity to CDDP-induced cell death (Aranda et al, 2014a; Aranda et al, 2015; Cirone et al, 2013).
Table of contents :
1.2. Mechanisms of oncogenesis
2. Role of the immune system in antitumor therapy
2.1. From the discovery of tumor antigens to the development of immunotherapies
2.2. Mechanisms of antitumor immunity
2.3. Immunoediting accompanies cancer progression
2.4. Adjuvanticity through immunogenic cell death
2.4.1. Mechanisms of chemotherapeutics-driven immunogenic cell death
188.8.131.52. eIF2 phosphorylation-dependent calreticulin exposure
184.108.40.206. Autophagy-mediated ATP secretion
220.127.116.11. HMGB1 release and TLR4 mimicry
18.104.22.168. Autocrine signaling of type I interferon
22.214.171.124. The ANXA1 FPR1 axis
2.4.2. Methods to assess immunogenic cell death
2.4.1. Immunogenic cell death inducers
3. The anticancer agents dactinomycin inhibits transcription
3.1. Transcription and translation in eukaryotic cells
3.1.1. Mechanisms of transcription
3.1.2. Mechanisms of translation
126.96.36.199. From mRNA to protein
188.8.131.52. ER stress inhibits cap-dependent translation
3.1.3. The inhibition of transcription and translation to prevent neoplastic cells proliferation
3.2. Dactinomycin intercalates into the DNA and inhibits transcription
3.3. Other anticancer mechanisms of dactinomycin
3.3.1. Dactinomycin is a topoisomerase inhibitor
3.3.2. Dactinomycin inhibits protein synthesis
3.3.3. Dactinomycin induces apoptosis
3.3.4. Dactinomycin induces photosensitization
3.3.5. Dactinomycin inhibits respiration and glycolysis
3.3.6. The effect of dactinomycin on the immune system
3.4. Pharmacokinetics of dactinomycin
3.5. Dactinomycin in the clinic
3.5.1. Cancers treated with dactinomycin-based chemotherapy
3.5.2. Clinical trials involving dactinomycin