PDIA3: a versatile and pleiotropic protein disulfide isomerase

Introduction PDIA3, also known as ERp57 or GRP58, is a member of the protein disulfide isomerase family; its structure is characterized by four thioredoxin-like domains: a, b, b’ and a’; a and a’ domains contain the redox active site while b and b’ domains are redox inactive. PDIA3 is localized predominantly in the ER (Endoplasmic Reticulum), where it is involved in the correct folding of newly synthesized glycoproteins and in the assembly of the MHC class I complex, but it is also present in the cytosol, in the nucleus and on the cell surface. PDIA3 is a multifunctional protein disulfide isomerase with a wide range of functions. It has been shown that this protein is involved in the cellular response to stress as well as in several diseases such as cancer, prion disorders, Alzheimer’s and Parkinson’s diseases. Considering this and given that PDIA3 is able to interact with a number of macromolecules and small ligands, such as green tea catechins, in the first part of my research I focused my attention on finding molecules that can interact with and modulate PDIA3 activity. This study, previously started in our lab analyzing the major catechins present in the extracts of green tea, has been expanded to various classes of flavonoids to verify if their activity was in some way connected to the modulation of PDIA3 functions. Flavonoids are a large class of plant secondary metabolites of low molecular weight present in fruits, vegetables and in products such as tea and red wine. Their basic structure shows two benzene rings (A and B) linked by the heterocyclic pyran ring, and, according to type and position of substituents on the central structure, they can be divided into different classes. These molecules show antioxidant, anti-inflammatory, antithrombotic, antiviral and antitumor activity. However, for many of them the molecular and cellular basis of their activities are not well known. They can act on different targets affecting regulation of cell signalling and cell cycle, free radical scavenging, inhibition of angiogenesis, initiation of DNA repair mechanisms, apoptotic induction and inhibition of metastasis. For this reason, we undertook a screening study for assessing the interaction and impact on PDIA3 protein activity of several types of flavonoids. Alzheimer’s disease (AD) is a common neurodegenerative disorder in humans, characterized by deposition in the brain of β amyloid (Aβ) plaques and neurofibrillary tangles (NFTs). Aβ plaques are constituted by Aβ1-40 and Aβ1-42 peptides, which are the results of the APP (amyloid precursor protein) proteolytic cleavage and are thought to be the main cause of the Alzheimer development. Aβ is present in the plaques of Alzheimer’s patients only as a naked peptide, while it is complexed to PDIA3 and calreticulin in the cerebrospinal fluid (CSF) of healthy individuals. It has also been reported the beneficial effect of diosgenin on the memory deficit in an AD mice model and on retrieval of axonal and presynaptic degeneration in the cerebral cortex and hippocampus, detecting PDIA3 as a target of diosgenin. Interestingly, the diosgenin-induced axonal growth was significantly inhibited in primary cortical neurons after PDIA3 knockdown. Given the evidences of PDIA3 involvement in AD, in the second part of my PHD I decided to better investigate PDIA3 role in amyloid beta deposits and in Alzheimer’s disease. The last part of my research, started in the Children’s Hospital of Philadelphia in Yair Argon’s laboratory, has been focused on ER stress and UPR (Unfolded Protein Response). In particular I studied one of the sensors of UPR, IRE1. IRE1 is a transmembrane receptor kinase located on the surface of the ER. This protein consists of different domains: the luminal domain, the transmembrane domain (TM), the linker domain, the kinase domain and the endoribonuclease domain (KEN). When ER stress occurs, BIP dissociates from the luminal domain of IRE1; interactions of the stress-sensing luminal domains of two IRE1 monomers promotes trans-autophosphorylation of the kinase and RNase domains on the cytoplasmic side of the ER membrane Phosphorylation triggers conformational changes in IRE1, which stabilize the dimer. This rearrangement in the RNase domain places the residues necessary for the catalysis in the correct orientation. Moreover IRE1 can also arrange in higher-order oligomers. The result is the cleavage of an intron from the XBP1 mRNA, leading to the translation of a potent transcription factor, which regulates the expression of several genes encoding for ER chaperones. Activated IRE1 can also cleave a subset of mRNAs encoding proteins targeted to the ER, leading to mRNA degradation; this process has been called regulated IRE1-dependent decay (RIDD). RIDD regulates many physiological processes, including degradation of mRNAs encoding a subset of ER or secretory proteins prone to misfolding, and regulation of lipid metabolism genes. RIDD is also responsible for the activation of apoptosis through degradation of several miRNAs’ precursors, such as miR-17. Moreover IRE1, through its kinase domain, activates the c-Jun N-terminal kinases (JNKs) via the formation of a complex with the E3 ubiquitin ligase TRAF2 (TNF receptor-associated factor 2) and the apoptosis signal regulating kinase 1 (ASK1), leading to apoptosis. It has been demonstrated in several studies that alteration in IRE1 function occurs in different diseases such as cancer, diabetes, inflammatory and neurodegenerative diseases. It still not entirely known how IRE1 regulates cell fate; the common thinking is that XBP1 splicing has mostly a prosurvival effect, whereas RIDD shows a proapoptotic output. Nevertheless, the precise mechanisms of these actions need to be further investigated. For this reason the main goal of this research was to understand how the different IRE1 activities are related to each other. Moreover we wanted to investigate if they are related to protein dimerization and clustering, and how phosphorylation affects those activities. METHODOLOGIES PDIA3-flavonoids interaction was investigated by quenching analysis of protein intrinsic fluorescence. This analysis was extended to the PDIA3 a’ domain. Disulfide reductase activity of PDIA3 was monitored by sensitive fluorescent assay using dieosin glutathione disulfide (DiE-GSSG) as fluorogenic probe. Flavonoid’s effect on the DNA binding properties of PDIA3 was also evaluated by EMSA analysis. To investigate PDIA3 involvement in AD, the neuroblastoma cell line SH-SY5Y, a model of neuronal cell, was treated with the 25-35 fragment of the amyloid β peptide. PDIA3 protein levels were analyzed through western blot analysis; mRNA levels through real time-PCR. Immunofluorescence studies were conducted to follow PDIA3 localization under Aβ treatment. To assess PDIA3 levels in the culture media proteins were precipitated with trichloroacetic acid (TCA) and analyzed through western blot. To study IRE1 activities we used an IRE1GFP construct in a TetOn inducible expression system. We established a stable IRE1-/- HAP1 cell line, in which the expression of WT-IRE1GFP and its mutants was regulated by adjusting the doxycycline concentration. IRE1-complemented IRE1 -/-HAP1s were generated by two consecutive transductions with lentiviral particles carrying pLVX-Tet-On™ or pLVX-Tight-Puro™ plasmids (Clontech). XBP1 splicing assay and cell living image were performed under ER stress. RESULTS AND CONCLUSIONS In the first part of this study, the interaction of different flavonoids with PDIA3 and their effect on protein reductase activity were evaluated. Two molecules, eupatorin and eupatorin-5-methyl ether, showed the highest affinity for PDIA3 with a Kd near to 1.0x10-5 M. They also showed a noticeable inhibitory effect on disulphide reductase activity of PDIA3, but they did not significantly affect its DNA binding activity. The backbone structure of these two flavones is characterized by a more stable conformation where B and A rings are almost parallel. This structure, associated with a definite degree of polarity, due to the presence of several methoxyl- groups, seems to be an important feature to determine a good affinity toward PDIA3. We can hypothesize that flavones interact with a region of the protein involving the tryptophan residues close to the redox site and given that PDIA3 does not contain any evident deep cavity or slot where this kind of ligands can bind, the binding of flavonoids may occur mainly via a flat interaction with the protein surface. Therefore, the planarity of the molecule as well as the number and specific position of its functional groups (hydroxyl-, methoxyl- and carbohydrates) will definitively play a major role to determine the affinity for the protein. In conclusion, eupatorin and eupatorin-5-methyl ether represent leading compounds for the binding to PDIA3 and for the inhibition of its redox activity. Further experiments are required to better characterize the effect of flavonoids on PDIA3 and to understand if some of the biological activities of these compounds are depending on the interaction with PDIA3. Since these flavones and PDIA3 are both involved in proliferative and carcinogenic processes, our in vitro findings on their interaction suggest that some of the biological effects of flavones could be mediated by modulation of PDIA3 activity. Additionally, this study will help to define and identify compounds to be used as selective inhibitors/modulators of PDIA3 biological activities. Regarding the implications of PDIA3 in β-amyloid deposits and Alzheimer’s disease, we observed that β-amyloid peptide fragment 25-35 induced a decrease in PDIA3 protein but not in its mRNA levels. We demonstrated that this decrease was not a consequence of ER stress, since we proved that this specific fragment of the amyloid β peptide did not cause activation of the unfolded protein response. We also proved that this decrease was not due to protein degradation through proteasome. Moreover we observed a delocalization of PDIA3 toward the plasma membrane following Aβ treatment. Considering our data and since in literature evidences about PDIA3 extracellular presence can be found, we hypothesized that PDIA3 can be secreted under Aβ treatment. In this study we indeed showed that PDIA3 is secreted by SH-SY5Y, with a significant increase after 1 hour of Aβ25-35 treatment. We also observed that PDIA3 secretion seemed not to be dependent on the classical secretion pathway Golgi-mediated. An explanation for this observation could be that PDIA3 is released from the cell within exosomes. This is not totally unlikely since PDIA3 is present in the ExoCarta Database as an exosome-associated protein. From data presented in this work, our hypothesis is that β-amyloid peptide induces a PDIA3 delocalization and secretion in the extracellular fluid as a defence mechanism carried out by the cell to counteract the toxic action of Aβ. Considering the work of Erickson et al., it could be that PDIA3 pursues this aim through direct binding to amyloid β peptide in order to prevent its aggregation and keep it in solution. In the study of IRE1 activities we observed that, in order to have activation of the RNase domain of IRE1 and clustering, a functional luminal domain is required. Interestingly we found out that if the RNase domain of IRE1 is somehow inhibited, with a mutation or using a chemical inhibitor, IRE1 persists in clusters in the ER membrane. This means that IRE1 clustering does not require the endoribonuclease activity of IRE1. Moreover this led us to hypothesize that IRE1 clustering can be related to other activities, such as RIDD or activation of the JNK pathway. Another interesting finding in this study was that a stimulus coming from the cytosol induces XBP1 splicing but not IRE1 clustering. Indeed, the flavonoid luteolin, through direct binding to the interphase between the kinase and endoribonuclease domains, triggers the splicing of XBP1 mRNA but it does not cause IRE1 redistribution in clusters in the ER membrane. This could be explained by the finding that luteolin induces a different degree in IRE1 phosphorylation if compared with the one induced by a common ER stress activator, such as thapsigargin. Our hypothesis is that luteolin induces phosphorylation only in the activation loop, which is sufficient to have XBP1 splicing, but in order to have IRE1 clustering, phosphorylation in additional residues is required. More studies are needed to confirm our thinking and, more important, the next step will be to investigate the other IRE1 activities, RIDD and JNK-pathway activation, in order to relate every single IRE1 activity to its dimeric or oligomeric state. Moreover we want to better understand how the phosphorylation state of IRE1 affects its activities. Last, during the course of this study on IRE1, we discovered by chance a new IRE1 mutant, L827P. This mutation, not present in any of the catalytic domains of IRE1, has proven to led to the complete suppression of its endoribonuclease activity. We are now interested in better characterizing the mechanism of action of this mutant with the intent to develop, in the future, small peptides that can inhibit IRE1. Since it has been proved that IRE1 is involved in pathologies, such as multiple myeloma, it can be a valid and promising pharmacological target.