Contribution of the unfolded protein response to breast and prostate tissue homeostasis and its significance to cancer endocrine response

Resistant breast and prostate cancers remain a major clinical problem, new therapeutic approaches and better predictors of therapeutic response are clearly needed. Because of the involvement of the unfolded protein response (UPR) in cell proliferation and apoptosis evasion, an increasing number of publications support the hypothesis that impairments in this network trigger and/or exacerbate cancer. Moreover, UPR activation could contribute to the development of drug resistance phenotypes in both breast and prostate cancers. Therefore, targeting this pathway has recently emerged as a promising strategy in anticancer therapy. This review addresses the contribution of UPR to breast and prostate tissues homeostasis and its significance to cancer endocrine response with focus on the current progress on UPR research related to cancer biology, detection, prognosis and treatment.

A c c e p t e d M a n u s c r i p t 5 IRE1α branch: Upon dissociation from BiP, IRE1α dimerizes and autophosphorylates activating its RNAse domain, leading to the subsequent splicing of X-box binding protein 1 (XBP-1) mRNA. Spliced XBP-1 (XBP-1s) encodes a transcription factor that up-regulates genes involved in protein folding, lipid metabolism, quality control and endoplasmic-reticulum-associated degradation (ERAD). XBP-1 heterodimerizes with several other transcription factors, hence its target genes may vary according to the cell context and stimuli. Unspliced XBP-1 (XBP-1u) can function as a negative regulator of XBP-1s activity, therefore XBP-1s/XBP-1u balance can have significant consequences for UPR activation as well as XBP-1 function and transcription of its target genes (29,30). IRE1α also regulates the stability of multiple RNAs through its endonuclease activity in a process known as regulated IRE1-dependent decay (RIDD) which targets glucose metabolism, inflammation and apoptosis. IRE1α cleavage of XBP-1 or activation of RIDD follow different kinetics and may depend on IRE1α oligomeric state (31).
PERK branch: Release of PERK from BiP results in PERK oligomerization and transautophosphorylation, which activates its kinase function. p-PERK phosphorylates eukaryotic initiation factor 2α (eIF2α) in Ser51 to reduce the rate of formation of the eIF2 ternary complex (eIF2-GTP-tRNA Methionine) which is essential for ribosome binding to the start codon (32). Despite this translational inhibition, some mRNAs, such as activating transcription factor 4 (ATF4), escape and are translated (33). ATF4 induces the expression of anti-oxidative enzymes, promotes amino acid synthesis, autophagy, protein folding and differentiation, and downregulates genes involved in cellular senescence and inhibitors of angiogenesis (27,34). In response to chronic stress, sustained ATF4 expression induces C/EBP homologous protein (CHOP) gene transcription, which encodes a transcription factor that stimulates growth arrest and apoptosis (35). CHOP´s target gene, growth arrest and DNA damage-inducible protein Downloaded from https://academic.oup.com/carcin/advance-article-abstract/doi/10.1093/carcin/bgy182/5266691 by Mount Allison University user on 02 January 2019 A c c e p t e d M a n u s c r i p t 6 34 (GADD34) in association with phosphatase protein 1 (PP1) dephosphorylate eIF2α, which enables the recovery of protein translation (36). However, if CHOP accumulates, due to chronic stress, Bcl-2-like protein 11 (BIM) expression is induced, and cells commit to apoptosis (35). Thus, PERK arm integrates adaptive and chronic EnR stress responses.
ATF6α branch: Following its release from BiP, ATF6α translocates to the Golgi where it is cleaved by the proteases SP1 and SP2, releasing an N-terminal fragment (ATF6f) that acts as a transcription factor of XBP-1u, EnR chaperones including BiP and ERAD-associated proteins (26,27,37). ATF6f also forms heterodimers with XBP-1s which drives specific gene expression programs. Despite the functional overlap between ATF6α and IRE1α branch gene targets, it appears that ATF6αhas evolved to enhance the protective mechanisms of PERK and IRE1α signaling, contributing for cell survival during chronic stress (38). In fact, ATF6α deletion results in impaired function of the secretory pathway during EnR stress thus resulting in impaired long-term EnR function (39).

Role of the unfolded protein response in tissue homeostasis: focus on the mammary gland
During lactation, normal breast cells must balance the increased production of milk proteins with the risk that an excessive protein load accumulation could impair basic cell survival functions (28,37,40). Similarly, to observations in other secretory cells such as plasmocytes and pancreatic cells, mammary cells have a well-coordinated and active UPR to adapt to the high EnR activity required during their functional differentiation during pregnancy and lactation (41)(42)(43)(44)(45). Notably, to the date, there is no information available regarding UPR involvement in normal prostate physiology and development. A c c e p t e d M a n u s c r i p t 7 IRE1-α/XBP1 branch: A thorough characterization of mammary tissue from virgin, early pregnant and lactating mice lacking XBP-1 disclosed that its deletion correlates to poor branching morphogenesis and impaired terminal end bud formation at the virgin stage, possibly due to a stromal effect. XBP-1 deletion was sufficient to impair lobuloalveolar development during early lactation. This is due to reduced epithelial cell proliferation which prevents lobuloalveolar compartment expansion as shown using tissue transplantation techniques (41). The same authors also showed that XBP-1 was detectable only during lactation and was nearly absent in virgin and pregnant mice.
However, other have shown that XBP-1mRNA and protein gradually increase from pregnancy, reaching highest levels during lactation. Moreover, in agreement with a role for XBP-1 in functional differentiation, blocking its expression in HC11 mammary epithelial cells reduced lactogenic protein mRNA levels in response to dexamethasoneprolactin-insulin stimulation. The authors explained these as resulting from reduced mRNA of prolactin and insulin receptors in cells with XBP-1 knock-down (42). Therefore, the increased protein and lipid synthesis demands during lactogenic differentiation induce EnR stress to activate the IRE-1α/XBP-1 branch and consequently increase the EnR capacity needed to support alveolar expansion and the secretory phenotype.
PERK/ATF4 branch: PERK activation inhibited MCF10A acini formation under normal growth conditions and its inhibition with dominant-negative PERK mutants resulted in hyperproliferation and in vivo tumorigenicity (43). In line with this, overexpression of the PERK downstream effector ATF4 in mice decreased proliferation and differentiation of mammary alveolar epithelium and accelerated involution (44). The effect of this branch on involution appears to result from an interplay with autophagy regulation where elevated BiP and p-eIF2a expression (as well as XBPu and autophagy genes) occurred expression was increased in the irreversible involution phase (72-168 h) and stimulated CHOP expression, which coincided with the expression of apoptosis markers such as active caspases and cleaved PARP (46). This study showed a sequential contribution of UPR and autophagy pathways in the involution process, promoting pro-survival or death signaling during the reversible and irreversible involution phases, respectively (46). During mid-lactation, PERK physiological activation appears to be necessary for the lipogenic maturation of mammary epithelial cells, as demonstrated by the increased levels of p-eIF2αbetween lactation days 7 and 12 (45), as well as activation of the lipogenic phenotype which characterizes lactogenic differentiation (47). Tissue-specific PERK deletion in the mouse mammary epithelium reduced levels of the lipogenic genes (45). Regarding ATF4, its expression was found differentially regulated during the development of the mammary gland although results are not consistent. One study found that in total tissue lysates ATF4 expression was highest during virgin and pregnancy stages and lowest during lactation (44). The second study reported that ATF4 protein levels gradually increased during pregnancy and reached a significantly higher level on lactation days 4-7 (42). Supporting the latter study, in the HC11 cell line, ATF4 knock-down reduced insulin and glucocorticoid receptor mRNA levels which suggests that ATF4 is necessary to allow lactogenic protein synthesis. Moreover, in the same cell line insulin and prolactin increased CHOP mRNA and transcriptional activity on the STAT5a-driven beta-casein gene (48).
However, CHOP´s function in lactogenic differentiation remains to be clarified since in whole mouse mammary tissue CHOP mRNA significantly decreased from day 15 of pregnancy to day 4 of lactation (42), while in another study it transiently peaked at lactation day 5 (48).
In summary, PERK activation may induce a positive feedback loop where ATF4 reduces proliferation and increases the response to lactogenic hormones, which in turn ATF6 branch: there was no significant change in the expression level of ATF6 during the mammary gland developmental stages (42). However, since ATF6 branch enhances the response to IRE-1α and PERK activation, more studies are needed to asses ATF6 activation.

4.The Unfolded Protein Response in Breast and Prostate Cancer
Cancer cells have increased metabolic demands to sustain biomolecule biosynthesis, survive chronic hypoxia, acidosis and nutrient depletion (50). In addition, accumulation of gene mutations alters protein folding which could increase formation of toxic protein aggregates (51). Cancer cells have evolved PQC mechanisms that allow their survival in normally deleterious conditions, with the overall result of chronically disturbed proteostasis along with enhanced survival (52). One explanation for this phenomenon is that cancer cells display sub-functional death pathways, thus chronic proteostasis loss does not result in cell death, but may be a selective advantage contributing to cancer cell survival (27).
Due to the importance of UPR in morphogenesis and differentiation of the secretory epithelium, the exploitation of UPR by neoplastic breast and prostate cells in order to deal with metabolic and oxidative stress seems predictable. This evolutionary strategy could induce cancer cell dormancy (34) and may be involved in the acquisition of a therapy resistant cancer phenotype. Amongst PQC effectors the ubiquitinproteasome-system (UPS) is frequently up-regulated in cancer cells but its activity per seis not sufficient to maintain cancer cell proteostasis (53). Curiously, BC and PC A number of studies have shown alterations in BC and PC that support the hypothesis that impairments in UPR appear to trigger and/or exacerbate the disease as well as influence response to anti-cancer therapy (15,18,24,28,(59)(60)(61)(62)(63)(64)(65)(66)(67). However, the current knowledge on UPR function in PC is still limited and controversial: So, et al found that all the 3 UPR branches are selectively down-regulated in mouse models of prostate tumorigenesis (68); on the other hand, Liu et al found that IRE1α, PERK and ATF6 were increased and significantly associated with Gleason grade, T and M stages, PSA level ad shorter survival (62). The following sections focus on receptor-positive BC and PC and the significance of UPR on endocrine response.

Regulation of UPR branch activation by ER and AR agonists and antagonists
Estrogens and androgens exert their effects through distinct molecular mechanisms ( Figure 2). In the classical nuclear receptor (NR) pathway, ER and AR bind their agonist, dimerize and bind to DNA consensus sequences which results in direct gene regulation. ER and AR may also interact with other transcription factors and regulate gene expression by indirect DNA binding (69)(70)(71). In the absence of agonist, these NRs can be activated by phosphorylation through cross-talk with receptor tyrosine kinase signaling which results in ligand-independent regulation of gene expression. On the other hand, a ligand activates a membrane-associated receptor or a receptor located in the cytoplasm, such as phospholipase C (PLC)/protein kinase C Downloaded from https://academic.oup.com/carcin/advance-article-abstract/doi/10.1093/carcin/bgy182/5266691 by Mount Allison University user on 02 January 2019 A c c e p t e d M a n u s c r i p t 11 (PKCs), Ras/Raf/MAPK, phosphatidyl inositol 3 kinase (PI3K)/Akt and cAMP/ protein kinase A (PKA), to induce a rapid physiological response without direct gene regulation by ER or AR (69)(70)(71). Increasing evidence shows that ER or AR activate or inhibit the UPR branches in a cell context-dependent manner and that this could be associated to endocrine therapy resistance.
BiP: BC and PC cells usually overexpress molecular chaperones such as HSP27, HSP60, HSPA1A, GRP94 and BiP (24,72,73) which aid to restore proteostasis by facilitating protein folding and the pro-survival and cytoprotective response of cancer cells to environmental stress (74). ERα, which initiates a fruitless cycle of EnR Ca 2+ depletion and ATP consumption and converts UPR from cytoprotective to cytotoxic (76). ERα abundantly interacts with BiP in the EnR, which probably reflects the need of this chaperone for ERα mediated nongenomic signaling (77). In T47-D cells, pre-exposure to E2 for 8 days elicits a 10-fold increase in the concentration of tunicamycin necessary to induce apoptosis (66).
Therefore, cells can exploit the UPR anticipatory response produced by ERα activation as a protective mechanism against chemotherapy Supporting this findings, BiP silencing in these cells induced antiestrogen sensitivity (16). On the contrary, BiP over-expression in antiestrogen sensitive MCF7 and LCC1 cells lead to a reduced proliferation rate even in the absence of antiestrogen treatment.
The authors suggested that BiP may be involved in the growth-inhibition process and that this function is lost in resistant cells (16). Therefore, BiP appears to be necessary for the maintenance of a resistance phenotype since its over-expression was related with resistance to E2 starvation-induced apoptosis in MCF7-5C, mimicking hormonal therapy resistance (80) and its down-regulation with shRNAs re-sensitizes antiestrogen resistant cells to anti-estrogen treatment (16).
Novel studies have recently shown that in LCC9 anti-estrogen resistant cells and tumors, reducing BiP with an antisense morpholino diminishes de novo fatty acid synthesis and mitochondrial fatty acid transport through down-regulation of SRFBP1 A c c e p t e d M a n u s c r i p t 14 BiP has also been detected in the cell surface of BC cells (89)(90)(91) and was associated with early stages of the disease, with progesterone receptor expression, highest p53 levels and with good prognosis in ERα-positive tumors (90). On the other hand, in triple negative BC, BiP cell surface expression was related with growth inhibition, apoptosis and reduced anti-BiP antibodies in mouse serum (89). Presently, the functional significance of BiP in the cell surface is unknown but may be related to non-genomic effects as shown in PC cells (see below).
 BiP in prostate cancer: in PC tissue samples BiP is over-expressed as disease progresses from early to metastatic androgen-independent state (64,65,92). BiP seems to be essential for PC cell survival, to allow cells to resolve EnR overload in response to the AR anabolic signaling in nutrient-deprived conditions (93). In fact, BiP temporal up-regulation can occur independently of EnR stress, and promotes acute adaptation to nutrient starvation by blocking autophagy (93). However, BiP-mediated autophagy was shown to be critical for the development of androgen-resistant PC (16,93).
Therefore, it is not surprising that BiP is also associated with endocrine therapy resistance and its inhibition restored endocrine response in C42B castration resistant cells (15  XBP-1s protein interaction with hypoxia-inducible factor 1α (HIF1α) increases cell tolerance to hypoxia, facilitating tumor growth by a mechanism independent of angiogenesis (104,105). HIF-1α is an ERα direct transcriptional target and both proteins share many target genes. HIF-1α is able to confer anti-estrogen resistance to MCF-7 cells (106). Therefore, XBP-1s co-activator function can enhance HIF-1α/ERα cross-talk to facilitate endocrine resistance.
The cross-talk between ERα and NF-κβ in endocrine resistance is well documented and has been reviewed elsewhere (107). These interdependencies have a significant impact on cell survival, especially in cells with elevated IKK/ NF-κβ activity such as breast (108)  A c c e p t e d M a n u s c r i p t 17 cell fate decisions by affecting the balance between apoptosis and pro-survival autophagy, XBP-1s is more potent in activating NF-κβ signaling (103). In addition, in combination with translation repression by PERK, IRE1α was able to maintain IKK basal activity, which is critical for maximal NF-κβ activation during UPR activation (109).
In opposition to ERα, which promotes BC cell proliferation, ERβ can, in certain conditions, counterbalance ERα effects and inhibit proliferation and survival (110,111). NF-ĸβ over-expression can confer estrogen-independence and antiestrogen resistance because of an overlap in their target genes (132) and trans-repressive interaction between these two proteins (107). NF-κβ is a major stress-inducible antiapoptotic transcription factor and eIF2αinactivation by PERK inhibits the synthesis of NF-κβ inhibitor IkB, thereby enhancing NF-κβanti-apoptotic activity in stressed BC cells (133,134). Moreover, NF-κβ subunit p65 can repress CHOP expression in BC cells, thus protecting cells against EnR stress-induced death (135,136) through the expression of pro-survival genes like BCL-2s, TRAF1/TRAF2 and SOD (137).
Therefore, PERK activation could select BC cells for dependence on NF-ĸβ signaling and thus promote endocrine resistance.
Protein-tyrosine phosphatase 1B (PTP1B) reverts PERK phosphorylation in response to EnR stress (138). PTP1B specifically de-phosphorylates Tyr616, resulting in PERK inactivation and attenuation of this UPR branch (138). PTP1B is commonly over-expressed in BC (139,140), being correlated with ER (141). Therefore, it would be interesting to know if PTP1B is implicated in BC endocrine resistance through downregulation of the pro-apoptotic PERK-CHOP pathway.
PERK and elF2α were found overexpressed in BC samples and were significantly associated with high histological grade and with tumor-infiltrating lymphocytes (142). However, it is important to consider that most of the performed studies lack information about therapy and the phosphorylation state of these two proteins, being of interest to verify their expression pattern before and after endocrine therapy.
 PERK/eIF2α in prostate cancer: Androgen treatment in LNCaP cells had no short-term effect on total or phosphorylated PERK and eIF2α protein, while it reduced Interestingly, ATF4 and CHOP protein levels increased in a time-dependent manner from 24h until 72h (61). The authors suggest that upon androgen treatment dephosphorylation of PERK and eIF2αresults in a general increase in protein synthesis, compensating the effects observed for mRNA levels. However, in this work CHOP increase wasn´t sufficient to trigger apoptosis as it was counterbalanced by strong activation of IRE1α/XBP-1 pathway (61). On the other hand, when LNCaP cells were cultured in serum-starvation, a rapid PERK/p-eIF2α induction occurred after 2h androgen treatment and was maintained above basal levels for up to 24h, while CHOP protein levels were also found to be upregulated after 24h of starvation (93). Taken together, these data are somewhat contradictory as to whether androgens activate or inhibit PERK/eIF2α branch; yet coincide with the observation that CHOP activation does not lead to increased apoptosis. Currently, there is no information regarding ARmediated anticipatory/non-genomic activation of this arm, neither its regulation by AR antagonists, nor its function on castration-resistance phenotype. However, the fact that BiP subcellular distribution could impact non-genomic signaling directing cells toward apoptosis or survival needs to be explored.
ATF6 remains the less explored branch of UPR and its interaction with IRE1α/XBP-1s and PERK arms remain to be disclosed. It will also be interesting to study ATF6 threshold levels involved in differential UPR activation in high or mild stress conditions. These threshold levels may be important in the context of endocrine therapy resistance. M a n u s c r i p t  Upregulation in red; downregulation in green. *Without information about patients' treatment. ATF4, activating transcription factor 4; ATF6, activating transcription factor 6; BC, breast cancer; BiP, sensor-binding immunoglobulin protein; CHOP, C/EBP homologous protein; eIF2α, eukaryotic initiation factor 2α; IRE1α, inositol-requiring enzyme 1α; PC, prostate cancer; PERK, protein kinase RNA-like endoplasmic reticulum kinase; TAM, hydroxytamoxifen; XBP-1, X-box binding protein 1; XBP-1s, spliced XBP-1; XBP-1u, unspliced XBP-1.