Identification of a novel Protein Disulfide-isomerase A3 (PDIA3) transcript variant as a potential biomarker associated with late stage prostate cancer

Background Prostate cancer (PC) is a heterogeneous and unpredictable disease and becomes untreatable when the tumor progress to castrate-resistant (CR) or androgen independent (AI). A major clinical challenge in prostate cancer is the lack of diagnostic and prognostic tests that distinguish between benign and aggressive tumors. Isoforms of gene transcripts are emerging as suitable candidates to represent disease progression. Vitamin D receptor (VDR and PDIA3) transcript isoforms could be the target candidates of study since they have been related with anti-tumoral effects and carcinogenesis in several cancer types. Methods The current study investigates the role of vitamin D receptor transcript isoforms in prostate cancer progression by using Next Generation Sequencing (NGS), Droplet Digital PCR (ddPCR) and several functional prediction tools. Results The NGS analysis revealed a novel PDIA3 transcript isoform (PDIA3N) that is higher expressed than the PDIA3 isoform that codifies for the receptor protein, in prostate cells. The expression of PDIA3N was validated by droplet digital PCR (ddPCR) absolute quantification, which confirmed the findings from the NGS analyses. The PDIA3N isoform was present in higher levels than PDIA3, in the metastatic androgen dependent LNCaP cells. Furthermore, analysis of the novel PDIA3 isoform sequence indicate that the variations present in its sequence are altering the original protein function and structure as well as the predicted subcellular localization of the protein. Conclusions We conclude that, PDIA3N due to the high expression in LNCaP cells and its abnormality in predicted structure, localization and function, is a potential biomarker for prostate cancer disease that needs to be further investigated in prostate cancer samples. (NGS) progression. The NGS analysis identified a new PDIA3 transcript isoform in prostate cells that is higher expressed than the actual isoform that codifies for the PDIA3 receptor protein. We did an absolute quantification by droplet digital PCR (ddPCR), absolute quantification, which confirmed that this PDIA3N isoform was present in higher levels, compared to the normal PDIA3 isoform, in the metastatic and androgen dependent stage of the disease. These results also show that there is change in expression level of PDIA3 isoforms between normal and metastatic


Abstract
Background Prostate cancer (PC) is a heterogeneous and unpredictable disease and becomes untreatable when the tumor progress to castrate-resistant (CR) or androgen independent (AI). A major clinical challenge in prostate cancer is the lack of diagnostic and prognostic tests that distinguish between benign and aggressive tumors. Isoforms of gene transcripts are emerging as suitable candidates to represent disease progression. Vitamin D receptor (VDR and PDIA3) transcript isoforms could be the target candidates of study since they have been related with anti-tumoral effects and carcinogenesis in several cancer types. Methods The current study investigates the role of vitamin D receptor transcript isoforms in prostate cancer progression by using Next Generation Sequencing (NGS), Droplet Digital PCR (ddPCR) and several functional prediction tools. Results The NGS analysis revealed a novel PDIA3 transcript isoform (PDIA3N) that is higher expressed than the PDIA3 isoform that codifies for the receptor protein, in prostate cells. The expression of PDIA3N was validated by droplet digital PCR (ddPCR) absolute quantification, which confirmed the findings from the NGS analyses. The PDIA3N isoform was present in higher levels than PDIA3, in the metastatic androgen dependent LNCaP cells. Furthermore, analysis of the novel PDIA3 isoform sequence indicate that the variations present in its sequence are altering the original protein function and structure as well as the predicted subcellular localization of the protein. Conclusions We conclude that, PDIA3N due to the high expression in LNCaP cells and its abnormality in predicted structure, localization and function, is a potential biomarker for prostate cancer disease that needs to be further investigated in prostate cancer samples. 4 Background Prostate cancer (PCa) is one of the most common cancer types in men worldwide and is the second leading cause of death among men in the United States (1)(2)(3). This disease is very heterogeneous and unpredictable since prostate cancer cells can easily pass from indolent to rapidly progressive and fatal. Furthermore, when the tumor progress to castrate-resistant (CR) or androgen independent (AI) it does not respond to androgen deprivation therapy and it becomes untreatable (4).
A major clinical challenge in prostate cancer is the lack of diagnostic tests that distinguish between benign and aggressive tumors (1,5). Standard approaches used in clinical decision making are tissue pathology and Gleason score, imaging and prostate specific antigen (PSA) serum levels (1)(2)(3)(4). Pathology analysis and Gleason score classification is more reliable but require invasive techniques to extract biopsies from the tumor. Imaging is not used if there is not any suspicion of tumor growth and usually misses small tumors. PSA screening has been used for more than 20 years as the most effective non-invasive method. However, this method lacks the sensitivity to detect early and late stage tumors and lacks specificity since very often gives false positives associated with other uropathies (5). Considering this, there is a need of finding and validating novel biomarkers for the prostate cancer disease progression and aggressiveness.
New gene transcript isoforms are emerging as suitable candidates to represent disease progression since high expression levels of specific isoforms may be associated with specific tumor stages or with the level of the disease progression (6)(7)(8)(9). An example is the discovery of an alternative splicing isoform signature associated with overall survival for hepatocellular carcinoma (10). Another example is the isoform switches that are highly predictive for cancer survival and 5 aggressiveness, such as the one for the DNA excision repair 1 (ERCC1) gene that results in a protein lacking the HHH domain which is associated with lower cancer survival rates (11).
Vitamin D and its metabolites have been suggested as potential candidates for the prevention and therapy of several cancer forms, including prostate cancer (12,13).
Numerous reports demonstrate that Vitamin D has an antitumor effect since it stimulates differentiation, increases apoptosis and inhibits proliferation, invasiveness and metastasis of cancer cells (13)(14)(15). Two receptors are involved in the activation of the Vitamin D signaling pathway and its mediated effects: The classic nuclear vitamin D receptor (VDR) and the protein disulfide isomerase family A, member 3 (PDIA3) receptor (16). VDR is a receptor localized in the cytosol that gets activated upon binding with vitamin D which initiates its heterodimerization with the RXR receptor. This complex migrates to the nucleus where it modulates gene transcription after binding with the vitamin D response element (VDRE) in the genome (17,18). Thus, VDR is responsible of the "genomic or long-term actions" which are the activation of gene transcription by chromatin remodeling and the regulation of vitamin D biosynthesis (18). The PDIA3 receptor is a chaperone localized mainly in the endoplasmic reticulum and it is responsible for "non-genomic or rapid actions" of vitamin D in the plasma membrane (16). These actions include the regulation of intracellular, extranuclear pathways and signaling cascades, such as the activation of the protein kinase C (PKC) pathway and the calcium transport (16). The role of PDIA3 in cancer regulation remain controversial since some studies suggest that it is responsible of the activation of proapoptotic pathways (19,20) and other studies that it is associated with cancer proliferation, inhibition of the apoptosis and poor prognosis (21,22). VDR was found to be lower expressed in 6 prostate tissue while PDIA3 was higher expressed at protein and transcript level in both normal and cancer prostate tissue (23).
The current study investigates the role of vitamin D receptor isoforms in prostate cancer progression. Furthermore, we searched for novel vitamin D receptor transcript isoforms, related to prostate cancer progression. Therefore, next generation sequence (NGS) was performed on prostate cancer RNA from stages of progression. The NGS analysis identified a new PDIA3 transcript isoform in prostate cells that is higher expressed than the actual isoform that codifies for the PDIA3 receptor protein. We did an absolute quantification by droplet digital PCR (ddPCR), absolute quantification, which confirmed that this PDIA3N isoform was present in higher levels, compared to the normal PDIA3 isoform, in the metastatic and androgen dependent stage of the disease. These results also show that there is change in expression level of PDIA3 isoforms between normal and metastatic cancer cells that could be an indicator of normal and aggressive stage of prostate cancer.
The novel PDIA3 isoform has a second and different translation initiation site (TIS) than the one observed in the normal PDIA3 isoform. This novel TIS lead to a different protein sequence in the N-terminus and it is an indication that this isoform is delivered to a different subcellular compartment in the cell to accomplish another function. Furthermore, this novel PDIA3 isoform shows a truncated N-terminus sequence that changes the active thioredoxin sites of the protein. Moreover, most of the amino acid substitutions as well as deletions in PDIA3N sequence, are predicted to be pathogenic or damaging in the PDIA3N sequence compared to the PDIA3 sequence.
Altogether, these data suggest that this novel transcript isoform of PDIA3 novel transcript isoform is a potential biomarker for the aggressiveness of the prostate cancer disease, that truly needs to be further investigated in patient samples in correlation with the Gleason score.

Functional analysis
Protein sequences were retrieved from UniprotKB (P30101) for PDIA3 and from UniParc (UPI000066D935) for PDIA3N (24). Both protein sequences were analysed by the I-TASSER (Iterative Threading ASSEmbly Refinement server) for generating the protein structure model (25). Differences in the protein sequences were highlighted with UCSF Chimera 1.0 (26). A prediction of the damage or pathogenicity of the 56 variations (amino acid substitutions and deletions) that PDIA3N present in comparison with PDIA3 was carried out with PolyPhen-2 (27) and PROVEAN (Protein Variation Effect Analyzer) v1.1 (28) software tools. The threshold score used to determine that the variant was pathogenic in PolyPhen-2 (HumDiv and HumVar algorithms) was 0.403 and in PROVEAN was -2.5. In PolyPhen-2 an amino acid substitution was considered "Possibly Damaging" if the score was higher than 0.403 and "Probably Damaging if it was higher than 0.975. In PROVEAN an amino acid substitution/deletion was considered "Deleterious" in PROVEAN if the score was under -2.5. A prediction of the subcellular localization of PDIA3 and PDIA3N was performed by DeepLoc-1.0 (29).   The model generated by I-TASSER for PDIA3N had higher confidence (C-Score) than the one generated for PDIA3 ( Figure 4B). At the secondary structure level the PDIA3N lacks an α-helix and one of the thioredoxin active sites is truncated to CGH from WCGH in PDIA3 ( Figure 4A, 4B). The total 56 variations that PDIA3N shows with respect to PDIA3 were analyzed with Polyphen-2 (27) and PROVEAN (28). From these 56 variations, 22 were confirmed as variants affecting the structure and function of the protein, by the two predictors PolyPhen-2 and PROVEAN ( Table 1). The rest of variants were either benign or neutral (Polyphen-2 Score < 0.432, PROVEAN Score >2.5) or not confirmed by both methods as damaging variants. The prediction of subcellular localization by DeepLoc-1.0 shows that PDIA3N is a soluble protein present in the cytoplasm of the cell while PDIA3 is present in the endoplasmic reticulum (Additional File 2).

Discussion
Previous studies pinpoint PDIA3 as a plausible candidate for studying cancer prognosis and as a target for cancer treatment (19)(20)(21)(22)31). The role of PDIA3 in cancer treatment remains controversial and the implication of different PDIA3 transcript isoforms has not yet been studied in connection with prostate cancer progression.
By performing NGS we could detect a novel PDIA3 transcript isoform in prostate cells. This novel isoform (PDIA3N) was previously reported by the Ensembl gene build (ENST00000538521.1, GRCh37.p13 Ensembl 2018) and is not included in the new genome assembly, GRCh38 (hg19) (30). The information reported by Ensembl GRCh37 revealed that the PDIA3N could be associated with the progression of kidney and colon cancer. However, there are not further studies evaluating the existence of this isoform and/or the level of expression in prostate cancer.
Furthermore, the NGS analysis showed that this isoform was higher expressed than the normal isoform in prostate cells and that the level of expression was especially high in LNCaP cells. Since the number of samples was not enough to assess statistical significance, these results were further validated by ddPCR. The results from ddPCR confirmed that PDIA3N is expressed in prostate cells and that the concentration is significantly higher in the metastatic androgen dependent LNCaP cells in comparison to the PNT2 control prostate cells. In the metastatic androgen independent cell lines DU145 and PC3, we could not detect any differences between the PDIA3N and PDIA3 isoform concentrations compared to PNT2 was not significant.
These results suggest that the overexpression of PDIA3N and PDIA3 may be somehow related with cells in advanced stages that have androgen dependent growth.
Several studies associate deregulation of PDIA3 with multiple pathologies including cancer and neurodegenerative disease (31)(32)(33). Aberrant expression of PDIA3 is shown to be correlated with poor prognosis in several cancer types as well as to the increase of cell proliferation mediated by vitamin D binding and subsequent activation of the epidermal growth factor receptor (EGFR) (31).
Until now, there have been only one study evaluating the expression of PDIA3 in prostate cancer tissue (19). In this study, they showed that tumor samples with a higher Gleason score (GS 8-10) had significant increased levels of the PDIA3 transcript abundance levels compared to GS 6 tumors. However, the difference in PDIA3 expression between GS 8-10 and benign tumor tissue was not significant (19).
Furthermore, the study did not focused on assessing potential new PDIA3 isoforms playing a role in the expression between benign and tumor tissues (19). The novel PDIA3 isoform detected in this study was significantly higher expressed in comparison with the benign prostate cells.
The protein predicted sequence reveals that, PDIA3N is shorter in length and contains a different N-terminus sequence than the normal isoform due to an alternative 5´-proximal TIS. It has been shown that several TIS can be recognized by the ribosomes in the coding sequence (CDS) at the same time. Thus, the 40S ribosomal subunit, during the scanning of an open-reading sequence, may recognize a second TIS and start the protein translation from that point. This mechanism is called "leaky scanning" and is the responsible of the translation of functional different isoforms (34)(35)(36). Most of these isoforms are N-truncated proteins that contain secretory signals in order to be delivered to different cell compartments (35). We have confirmed by prediction analysis that PDIA3N has a different subcellular location since it is mainly located in the cytoplasm while PDIA3 is in the endoplasmic reticulum. Considering all this information and our results, we suggest that PDIA3N is an N-truncated protein isoform of PDIA3 produced by alternative translation or "leaky scanning".
One of the predicted functional differences, compared to the normal PDIA3, (See results section, Figure 4B) is the lack of one α-helix in its secondary structure and the truncation in one active site of the thioredoxin (See results section, Figure 4A).
These differences are likely to affect, first of all the conformation of the protein and secondly the activity as a thiol oxidoreductase.
The analyses of the sequence variations in this novel isoform PDIA3N present in comparison with PDIA3 a confirmation that a big proportion of the variations are damaging or harmful for the function and/or structure of the protein.
Isoforms of gene transcripts have been suggested as plausible biomarkers for different diseases, such as cancer. The expression pattern of different isoforms could change depending on the stage or severity of the disease and, also the relation between the expression of two gene isoforms could be correlated with a specific disease pattern (7,11,37). In this study we confirm PDIA3 as a plausible biomarker to test in prostate cancer human samples. Furthermore, the detected novel isoform of PDIA3 (PDIA3N) is even higher expressed than the normal PDIA3 transcript in the metastatic androgen dependent cell line LNCaP. The relation of expression between PDIA3N and PDIA3 (Ratio PDIA3 Novel/PDIA3 Normal) could also be a predictor of the disease aggressiveness since changes, from almost two times in control samples to nine times higher expression of PDIA3N, are seen in the metastatic androgen dependent samples. Functionality analysis of the variations in the PDIA3N sequence also confirm the abnormality of the protein with respect to PDIA3.
Considering that VDR is lower expressed in prostate cancer samples and the increased expression of PDIA3, according to the cancer stage, we propose that PDIA3 could be inhibiting the competitively antitumorigenic actions of vitamin D in cancer cells. Thus, Vitamin D would to a greater extend bind the PDIA3 receptor instead of with VDR receptor due to the abundance of the PDIA3 receptor in pathogenic conditions. The role of PDIA3N remains unclear but what is evident is that it also plays a relevant role in pathogenesis due to its abundant expression in the metastatic androgen-dependent stage, compared to PDIA3, and due to its predicted aberrant structure and function. One explanation that supports our results is that PDIA3N is novel and not studied before is that the seen contradictory effects could have been wrongly attributed to the PDIA3 normal isoform/receptor, when they instead are caused by the PDIA3N isoform. Thus, there could be different PDIA3 isoforms playing a contrary role in cancer progression. The switch in expression between these isoforms could indicate a different stage or form of the disease.
PDIA3N could be the pathogenic form of PDIA3 which inhibit cell apoptosis and stimulates the pathways leading to cancer cell proliferation. This aberrant isoform/protein would only be overexpressed under aggressive stages of cancer.
We conclude that, PDIA3N due to the high expression in LNCaP cells and its abnormality in predicted structure, localization and function, is a potential biomarker for prostate cancer disease that needs to be further investigated in

Availability of data and materials
The analysis supporting the conclusions of this article are included within the article and its additional files. NGS and ddPCR Raw data are available from the corresponding author on reasonable request.  Graph showing the differences in PDIA3 and PDIA3N target concentration (copies/ µl) betwee 30 Figure 4 Predicted Functional analysis of PDIA3 normal isoform and PDIA3 novel isoform with I-TASSER