PDIA3 Reduce Glioma-associated Macrophage/microglia Pro-tumor Activation trough IL-6-STAT3-PDIA3 Pathway.

Background: Glioblastoma (GB - grade IV glioma) is the most aggressive and common cancer of central nervous system with an overall survival of 14-16 months. The GB tumor microenvironment includes cells of the innate immune system identied as glioma-associated microglia/macrophages (GAMs). It is known that between GAMs and GB cells there is a double interaction, but the role of GAMs is still poorly characterized. The endoplasmic reticulum (ER) protein ERp57, also known as PDIA3, is a thiol oxidoreductase with main function related on glycoprotein folding in endoplasmic reticulum. However, PDIA3 shows different functions. In fact, the various subcellular localizations and binding partners of PDIA3 affect numerous physiological processes and diseases: different regulation and modulation of PDIA3 has been reported in multiple pathologies including neurodegenerative diseases and cancer. Methods: In the present work, we evaluated in both GB cells and microglia-macrophage cells the expression of PDIA3 using specimens collected after surgical from 18 GB patients. In addition, we studied in vitro microglia-glioma interaction to determine the role of PDIA3 in viability and the activation of both GB and microglia cells. The study was carried using PDIA3-silenced T98G cells and/or using a pharmacological inhibitor of PDIA3 activity (Punicalagin-PUN). Results: We initially investigated the role of the PDIA3 in GB survival by inquiring The Cancer Genome Atlas dataset. The results indicated that 352 out of 690 patients reported over-expression of PDIA3, which signicantly correlated with a ~55% reduction of overall survival. Subsequently, for the rst time, we investigated the PDIA3 expression in the tumor and the nearby parenchyma of 18 GB patients and our data showed a signicant upregulation (15% vs 10%) of ERp57/PDIA3 in GAMs of tumor specimens respect the microglia present in parenchyma. In addition, we show that conditioned medium (CMs) obtained from both wild type T98G and PDIA3 silenced T98G induced primers. AGCCATGGCAGAAGTACCGT Reverse: TCCATGGCCACAACAACTGA; GAPDH

The GB tumor microenvironment includes cells of the innate immune system identi ed as gliomaassociated microglia/macrophages (GAMs) that represent the largest population in ltrating GB tumor [3].
GAMs possess different dynamic states of activation: the M1 state, which is tumor suppressive, and the M2 state, which is tumor supportive, contributing to tumor growth. These different 2 phenotypes are thought to re ect a spectrum of plastic functional conditions rather than a set of discrete activation states [4; 5]. However, although it is known that between GAMs and GB cells there is a double interaction where the rst support the latter and viceversa, the role of GAMs in the GB context is still poorly characterized.
The endoplasmic reticulum (ER) protein ERp57, also known as PDIA3, GRP58, ERp60 is a thiol oxidoreductase and is a member of the protein disul de isomerases (PDIs) family that counts members with different localization and functionality. PDIs protein family have a TRX-like domain and its active or inactive form restrict the PDIs mostly to the endoplasmic reticulum [6; 7], however different localizations have been reported [8]. The rst and main function of PDIA3 is related on glycoprotein folding in endoplasmic reticulum (ER) [9]. In fact, PDIA3 interacts with calnexin and calreticulin regulating the folding of newly glycoproteins catalysing the formation and disruption of disul de bonds between cysteine residues [10; 11]. However, PDIA3 plays different functions (depending on cellular localization) beyond its abilities in the ER. In the cell membrane, for example, PDIA3 acts as a membrane receptor for 1α, 25-dihydroxy-vitamin D3 [12; 13]. In the cytoplasm, PDIA3 co-localizes with NF-κB or mTOR, forming in the latter case a complex that has been implicated in various developmental processes [14]. In the nucleus, PDIA3 directly interacts with DNA or enhances the DNA-binding of the signal transducer and activator of transcription 3 (STAT3) complex, in uencing binding of the transcription factor to DNA, and facilitating nuclear import and export of transcription factor [15; 16; 17]. Therefore, the various subcellular localizations and binding partners of PDIA3 affect numerous physiological processes and diseases. In fact, different regulation and modulation of PDIA3 has been reported in multiple pathologies including neurodegenerative diseases [18; 19], and cancer [20; 21; 22]. Moreover, PDIA3 expression level has been evaluated as a biomarker in several conditions [23; 24] and could be a novel pharmacological target.
PDIA3 signalling has a pharmacological natural inhibitor, Punicalagin (PUN). Punicalagin is a natural compound deriving from the secondary metabolism of different plants and represents the known largest molecular weight polyphenol found in forms alpha and beta in pomegranates (Punica granatum), in Terminalia catappa and Terminalia myriocarpa and in Combretum molle. PUN shows several properties among which anti-in ammatory activity and recent studies shown the binding with PDIA3 and its redox activity inhibition [25].
In the last years, our group has investigated the role of GAMs in different murine and human models of GB microenvironment, looking at GAM polarization status [26], at the involvement of mTOR pathway in GAM activation [27; 28], as well as at the role of chemokine receptor CCR5 in GAM migration and activation [29]. In addition, recently, we have also investigated VEGFR-1 expression in GAMs [30]. In the present work, we evaluated both in GB cells and microglia-macrophage cells the expression of PDIA3 in GB specimens collected after surgical from 18 patients. In addition, we studied in vitro microglia-glioma interaction to determine the role of PDIA3 in viability and the activation of both GB and microglia cells.

Immunostaining Analysis
For quantitative analysis, two blinded examiners counted the number of PDIA3+, IBA1+, or both PDIA3+ and IBA1+ cells in a number of 50 cells total in three randomly different areas of the slides. In particular, two blinded examiners have examined three different areas of the same slides and have counted 50 cells that included the number of positive cells for each antibody, the number of positive cells for both antibodies and the number of negative cells [31]. In total, the average of six counts was reported as percentage.

Cell cultures
The human microglia cell line (CHME-5; RRID: CVCL_5J53) was kindly provided by professor Pierre Talbot [32]. CHME5 cells were grown in DMEM media containing 10% FCS and antibiotics; experimental conditions were reached with DMEM at low concentration of FCS (1%) and cells were splitted at the 80% of the con uence.
Glioblastoma cell line T98G [T98-G] (ATCC® CRL-1690™) was kindly provided by professor Grazia Graziani (Tor Vergata University-Rome). T98G cells were grown in DMEM containing 10% FCS and antibiotics. Experimental conditions were reached with DMEM containing 1% FCS and antibiotics. Cells were splitted at the 80% of the con uence. All the experiments received institutional approval. Conditioned media from activated glioma cells was generated following a protocol described before [28]. Brie y, Basal Conditioned Media (B-CM) was prepared with 4 hours incubation in plain medium, followed by 3 washes with phosphate buffered saline (PBS) and addition of fresh plain medium for 24 hours. After that, CM was collected, centrifuged to remove cellular debris and stored at -80°C. Prestimulated Conditioned Media (PS-CM) was prepared with 4 hours incubation with a mixture of cytokines (10 ng/ml TNFα, 10 ng/ml IL1β, 10 UI/ml hIFNγ called TII), followed by 3 washes with PBS and addition of fresh plain medium for 24 hours. After that, the CM was collected, centrifuged and stored at -80°C.
RNA interference (siRNA) siRNA for PDIA3 was kindly provided by professor Fabio Altieri (University of Rome -La Sapienza) and Lipofectamine™ 2000 was purchased from Invitrogen. The day before transfection, 5 × 10 5 cells per well were seeded in a 6-well plate and were grown in normal conditions. The transfection was carried out according to the manufacturer's instructions and the siRNA was used at 1 µg per milliliter nal concentration. Cell lines were incubated 6 hours with the transfection complex under their normal conditions and after 48 hours incubation the selection with puromycin at 1µg/ml was conducted. Cells were grown with puromycin for at least two weeks and then PDIA3 gene and protein expression were carried out.
Nitrite assay iNOS activity was assessed indirectly by measuring nitrite accumulation in the incubation media. Brie y, an aliquot of the cell culture media (80 μL) was mixed with 40 μL Griess Reagent (Sigma-Aldrich, St Louis, MO, USA) and the absorbance measured at 550 nm in a spectrophotometric microplate reader (PerkinElmer Inc. Waltham, MA, USA). A standard curve was generated during each assay in the range of concentrations 0-100 μM using NaNO2 (Sigma-Aldrich) as standard. In this range, standard detection resulted linear and the minimum detectable concentration of NaNO2 was ‡ 3.12 μM. In the absence of stimuli, basal levels of nitrites were below the detection limit of the assay at all the time points studied. The levels of NO were normalized with the protein content determined by Bradford's method (Bio-Rad, Hercules, CA, USA) using BSA as standard.

Urea assay
Urea levels in CHME5 and T98G cells were detected by the QuantiChrom Urea Assay kit (BIOassay System, Hayward, CA, USA), used according to the manufacturer's instructions. Brie y after 48h of incubation with the B-CM and PS-CM, an aliquot of cell culture media (50 µl) was mixed with 200 µL Urea Reagent (Bioassay system) and the absorbance measured at 430 nm in a spectrophotometric microplate reader (PerkinElmer Inc., MA, USA). A standard curve was generated during each assay in the range of concentrations 0-100 µg/ml using Urea as standard. In this range, standard detection resulted linear and the minimum detectable concentration of Urea was 3.12 µg/ml. The protein content in each sample was determined by Bradford's method (Biorad, Hercules, CA, USA) using bovine serum albumin as standard.

Cytometer analysis
For intracellular analysis, cells were xed and permeabilized with Fix/Perm buffer (ThermoFisher Scienti c, MA, USA) and then incubated with primary monoclonal antibody anti-ARG1 (C-2) (Santa Cruz Biotechnology, Inc, TX, USA). Secondary monoclonal Antibody Goat anti-Mouse Alexa Fluor®-488 (ThermoFisher Scienti c, MA, USA) was used. The purity of cell preparations was assessed by cyto uorimetric staining. Unstained cells were used as a negative control.

Cell Viability and Toxicity
In order to discriminate viable, non-viable cells and apoptosis detection in ow cytometry, cells were incubated with Propidium Iodide and Annexin V-FITC (Novus Biological -NBP2-29373). The assay was conducted following the manufacturer's instructions. Compensation control cells were provided and unstained cells were used as negative control. Flow cytometry analysis was conducted with FC 500 (Beckman Coulter, Brea, CA) and the data were analyzed with Kaluza software (Beckman Coulter, Brea, CA). At least 50,000 events were acquired.
The cell viability was also measured using a speci c luminescence kit: CellTiter-Glo® Luminescent Cell Viability Assay (Promega, WI, USA). Cell mortality was detected using a speci c uorescence kit: RealTime-Glo™ MT Cell Viability Assay (Promega, WI, USA). The assays were carried out according to the manufacturer's instructions.
At the end of the incubation time H2DCF-DA 20 μM were added and cells were incubated for 45 minutes at 37°C.The uorescence signal was quanti ed using a microplate uorescence reader (VictorXTM4 microplate reader, PerkinElmer Inc, Waltham, Ma, USA), using 485 nm as excitation and 535 nm as emission wavelength. mRNA analysis in real time PCR Total cytoplasmic RNA from cell lines was extracted using the TRIzol reagent protocol and RNA from FFPE tissues was extracted using the Absolutely RNA FFPE Kit (Agilent, CA, USA) using the manufacturer's instructions. RNA concentration was measured using the Qubit™ RNA HS Assay Kit (Thermo Fisher Scienti c). Aliquots (1 µg) of RNA were converted to cDNA using random hexamer primers. Quantitative changes in mRNA levels were estimated by real time PCR using the following 12% bis-tris plus gel (Invitrogen, CA, USA). Apparent molecular weights were estimated by comparison to colored molecular weight markers (Sigma-Aldrich, MO, USA). After electrophoresis, proteins were transferred to nitrocellulose membranes by iBlot™ 2 Gel Transfer Device (Invitrogen, CA, USA). The membranes were incubated in the presence of the primary and secondary antibody in the iBind™ Flex Western Device (Cat. No.: SLF2000 -Invitrogen™, CA, USA) for βactin and PDIA3. Primary antibody for IκBα, phospho-STAT3 and total STAT3 were incubated overnight with gentle shaking at 4°C. Each primary antibody diluition was 1:1000. Primary antibody was removed, membranes washed 3 times in Flex Solution, and further incubated for 1h at room temperature in the presence of speci c secondary antibody diluted 1:15000 for anti-rabbit and 1:3000 for anti-mouse. Following three washes in Flex Solution, bands were visualized by incubation in ECL reagents (Thermo Scienti c™) and exposure to Hyper lm ECL (GE Healthcare NY, USA).

Statistical analyses
Data were described as median ± Standard Deviation (SD) or SEM as indicated in gure legends. Statistical analysis of the differences between pairs of groups was performed by Student's t test. For multiple comparisons ANOVA analysis, followed by Sidak's post-test, was used. Statistical signi cance was determined at α = 0.05 level. Differences were considered statistically signi cant when p < 0.05.
Statistical analysis was performed with GraphPad software Prism version 7.04 (GraphPad Software, San Diego, CA, USA).

PDIA3 in Glioblastoma specimens
Relevance of PDIA3 expression in GB patients' survival Difference in PDIA3 expression between tumour and brain parenchyma in human GB specimens In order to analyze the distribution of PDIA3 expression in GB, we examined the tumour and surrounding parenchyma of the same patients in tissue specimens collected after surgical removal of the tumour from 18 patients diagnosed with GB. Unexpectedly, the tumour tissue presented a signi cantly lower number of PDIA3 stained cells in comparison with the surrounding parenchyma (Figure 2A and B). In fact, in the parenchyma about 50% of the cells were positive for PDIA3, while in the tumour the percentage of cells expressing PDIA3 was about 40% ( Figure 2C). Interestingly, the staining of PDIA3 is not the same for all types of cells: there are cells that presented nuclear membrane strengthening and cells in which the cytoplasm resulted also positive for PDIA3. These data suggest that PDIA3 is expressed by GB cells as well as by cells of the tumour-associated microenvironment. Therefore, since microglia represent the cell type that mostly in ltrates the tumor we analyzed the expression of PDIA3 in GAMs. By double staining for PDIA3 and Iba1 (a macrophage-microglia marker) in these samples, we found that the percentage of microglia-macrophages expressing PDIA3 present in the tumour was signi cantly higher than in the parenchyma ( Figure 2D). In fact, taking into account only the microglia-macrophage cell population, in the tumour 15% of the Iba1 positive cells also expressed PDIA3, whereas in the parenchyma the percentage of double positive cells was 10%. The number of cells positive for Iba1 in the 18 patients studied was similar between the parenchyma and the tumour.

Effects of PDIA3 on T98G glioblastoma cell line activation and vitality
In order to study the effects of PDIA3 on the release of chemokines and cytokines and on viability of glioma cells, the PDIA3 gene was silenced in T98G cells or a pharmacological inhibitor of PDIA3 activity (punicalagin-PUN) was used. In particular after PDIA3 gene silencing on T98G cell line (about 35% of PDIA3 in both mRNA and protein level was silenced- Figure S1) the T98G showed a reduction of 40% in IL6 and COX2 gene expression and an increment of more than 50% in IL1β expression comparing to basal conditions of wild type T98G (Figure 3). In addition, we evaluated a panel of 105 cytokines and chemokines released from both PDIA3 silenced-T98G and wild type T98G under basal condition (the media collected was called B-CM). When PDIA3 is silenced a signi cant increase of Fibroblast Growth Factor (FGF-19), Platelet-derived growth factor (PDFG-AA) and Osteopontin occurred, while a signi cant decrease of insulin-like growth factor-binding protein 2 (IGFBP2) and IL8 is proved ( Figure 4C). In addition, PDIA3-silenced T98G released cytokines and chemokines never produced from the wild type T98G such as Angiopoietin 2, Complement factor D, DKK-1, MCSF, resitin and uPAR ( Figure 4D).
Moreover, in order to study the in ammatory activation of glioma cells we studied also the effects of PDIA3 on TII-stimulated both wild type and PDIA3-silenced T98G. Therefore, when wild typeT98G were treated with TII for 8 hours, data shown an increment of IL6, COX2 and IL1β compared to calibrator of more than 600-fold, 115-fold and 800-fold respectively. When PDIA3 was silenced and under the same experimental conditions, PDIA3-silenced T98G showed still an increase in the same gene expression, but the PDIA3 silencing limited the increment of about 50% for COX2 (115-fold vs 57-fold) and about 25% for IL1β (800-fold vs 590-fold) ( Figure 5). In addition, after 4 hours of TII stimulation, 3 washes and 24 hours of incubation with fresh media, we collected the media (called PS-CM) and we evaluated the same panel of 105 cytokines and chemokines released from both PDIA3 silenced-and wild type T98G. In PDIA3silenced PS-CM Chitinase 3-like, GM-CSF, Monocyte Chemoattractant protein 2 (MCP3 or CCL7) and C-C Motif Chemokine Ligand 20 (CCL20 or Macrophage In ammatory Protein 3 -MIP3A) are increased and Cystatin C, Emmprin, IGFBP2, Monocyte Chemoattractant protein 1 (MCP1 or CCL2), C-X-C motif chemokine 11 (CXCL11), Osteopontin and Thrombospondin-1 are decreased ( Figure 6C). In addition, PDIA3-silenced T98G released cytokines and chemokines never produced from the non-silenced counterpart such as Complement factor D, Granulocyte colony-stimulating factor (G-CSF) and Chemokine (C-C motif) ligand 3 and 4 (CCL3/CCL4) ( Figure 6D).
In ow cytometry analysis with Annexin V-Propidium Iodide assay, the effect of the block of PDIA3 activity induced by PUN on viability of wild type T98G cell was tested. We tested two different concentrations of PUN, 5µM and 50µM. After 24 hours of treatment, viable T98G cells were more than 70% in 5µM PUN with 15% of cells in early apoptosis (Figure 7). Punicalagin 50µM showed high toxicity with 80% of non viable cells, but through a necrotic way.
Effects of PDIA3 gene silencing in microglia-glioma interactions.
Effects of CMs from both PDIA3-silenced and wild type T98G on CHME-5 viability and activation.
Based on previous data showing the toxicity of CMs from glioma cells, CHME-5 were exposed to a challenge of B-CM and PS-CM from both wild-type and PDIA3-silenced T98G. The viability and the status of microglia activation were evaluated. Under a phase contrast electronic microscope, CHME-5 treated 24 hours with B-CM and PS-CM from PDIA3-silenced T98G preserve the same morphology as when CHME-5 were treated with CMs obtained by wild type T98G. Conversely, when microglia cells are treated with PDIA3-silenced CMs the number of cells is increased compared to treatments with CMs from wild type T98G ( Figure 8). In particular, the number of CHME-5 cells was between 30-40% more (the same as control) when treated with PDIA3 silenced T98G respect when treated with wild type T98G CMs. However, when we analyze the 24h-viability of CHME-5 measuring a direct inhibition of PDIA3 elicited by 5µM PUN, we found a signi cant reduction of CHME-5 viable cells when treated with PUN respect to un-treated cells.
In particular, we found more than 70% of viable cells and 20% in early apoptosis stage ( Figure 9).
These data indicate that the PDIA3 pathway is important for microglial cell survival (tested using PUN on CHME-5) but that the absence of the same pathway on glioma cells does not affect microglial viability as well (tested using CMs obtained from PDIA3-silenced T98G on CHME-5).
As parameter of M2 (or anti-tumor) activation, we measured urea release in the medium and the activity of ARG1 of CHME-5 in ow cytometry. Urea release, after 48 hours was reduced of 40% in CHME-5 treated with PDIA3-silenced T98G B-CM compared to wild type T98G B-CM. Similarly, CHME-5 treated with PS-CM obtained from PDIA3-silenced T98G shown 40% of urea decrease compared to cells treated PS-CM of T98G ( Figure 10A). To con rm the reduction of the M2 phenotype we investigated the ARG1 expression of CHME-5 on ow cytometry when cells are treated with CMs. In particular, measures of uorescence intensity from CHME-5 treated with B-CM and PS-CM from wild type T98G and B-CM and PS-CM from PDIA3-silenced T98G were collected after 24 hours of stimulation; accordingly with urea data, the expression of ARG1 when CHME-5 were treated with B-CM from wild type T98G was increased compared to B-CM from PDIA3-silenced T98G ( Figure 10B). Similarly, PS-CM treatments shown enhanced expression of ARG1 when compared to PS-CM from PDIA3-silenced T98G ( Figure 10C).
As parameter of M1 (pro-tumor) activation, we measured IL6 release from CHME-5 treated with CMs. No change was reported expect that IL6 release was reduced by half after 24 hours when cells were treated with PDIA3-silenced T98G PS-CM compared when the cells were treated with PS-CM obtained from wildtype T98G ( Figure 11A). Moreover, we investigated a transcription factor sensitive to cytokine levels: signal transducer and activator of transcription 3 (STAT3). In 2 hours western blot analysis, CHME-5 treated with CMs from wild type T98G and PDIA3-silenced T98G shown increasing levels of phospho-STAT3 in treatment with PS-CM from PDIA3-silenced T98G compared to PS-CM from wild type T98G ( Figure 11B). On the contrary, total STAT3 is increased in the B-CM from PDIA3-silenced T98G compared to B-CM from T98G while it is reduced in the PS-CM from PDIA3-silenced T98G compared to PS-CM from T98G ( Figure 11B). Similar changes in STAT3 are also reported after 24 hours of treatment ( Figure S2).
Therefore, all these data suggest that a partial block of PDIA3 on T98G could be bene cial on microglia activation tend to reduce the M2 phenotype without exacerbate a pro-in ammatory activation.

Effects of PUN on microglia pro-in ammatory activation
In parallel, we also tested the effects of a pharmacological inhibition of PDIA3 on CHME-5 proin ammatory activation. Therefore, CHME-5 cells were treated for 24 or 48 hours with TII alone or in combination to serial dilutions of PUN [1nM-100µM]. Nitrites levels were signi cantly reduced when TII is combined with 5µM PUN going from 18 µM/mg proteins (TII alone) to 1.2 µM/mg proteins (TII + 5µM PUN) on average ( Figure 12A). The NO data of dose range were reported in Figure S3. In addition to NO levels, an evaluation of the intracellular production of total reactive oxygen species, was carried out measuring the species with the uorescent dye, 2',7'-dichlorodihydro uorescein diacetate (H2DCF-DA) and treating CHME-5 for 48 hours with TII alone and in association with 5µM PUN. Data shown a signi cant reduction of uorescence in TII combined with PUN and the compound given alone ( Figure   12B) reinforcing the idea of multi-level anti-in ammatory activity. In addition, IL6 release in the culture media of CHME-5 was evaluated after 24 hours of treatment with 5µM PUN and data were compared to IL6 release under basal conditions. Accordingly with the reduction of IL 6 release on PDIA3-silenced T98G, on CHME-5 treated with PDIA-silenced CMs, IL6 release is also reduced when cells are treated with 5µM PUN (78.5 pg/ml on average) compared to control (125 pg/ml on average) ( Figure 12C). These data clearly indicated a strong correlation between IL6 and PDIA3.
To explain the anti-in ammatory effects of PUN on CHME-5 we tested the involvement of the NFκB pathway and in particular, we measured IκBα levels (a NFκB inhibitor). Brie y, 2 hours experiments were carried out and we reported an increment of IκBα in the treatment with PUN with an increment of 93-fold were given alone and of 43,5-fold and 41,8-fold in TII combined with PUN and B-CM combined with PUN respectively (Figure 13). Taken together, data con rm an involvement of PDIA3 on microglia activation and suggests multi-level anti-in ammatory activity of PUN in microglia cells.

Discussion
For the rst time we evaluated the expression of ERp57/PDIA3 in glioma-associated microglia/macrophages of human glioblastoma specimens. In particular, we investigated the ERp57/PDIA3 expression in the tumor and the nearby parenchyma of 18 GB patients and our data shown an upregulation of ERp57/PDIA3 in GAMs of tumor specimens (Figure 2) supporting the idea of its potential involvement in cellular and molecular processes of GB as well as different cancers. In fact, ERp57/PDIA3 has been evaluated as therapeutic target in cancer progression for i.e. in renal cancer and in hepatocellular carcinoma [21; 33]. In addition, in several cancer cell lines the knockdown of ERp57 affects different pathways involved in its physiological functions [34; 14]. In this scenario, using a siRNA we made a knockdown of ERp57/PDIA3 in T98G glioblastoma cell lines (Figure 1 suppl) and we investigated the effects on mRNA expression of IL6, IL1β and COX2 genes ( Figure 3). Interestingly, ERp57/PDIA3 knockdown reduced the basal expression of IL6 and COX2 genes but increased IL1β expression (Figure 3), suggesting a pivotal role of PDIA3 on in ammatory activation of T98G. Accordingly, when we challenged for 4 hours with TII both PDIA3-silenced and wild type T98G we obtained a limited increase of IL1β and COX2 when we measured the gene expression on PDIA3-silenced T98G in comparison to wild type T98G, whereas IL6 gene expression was not modi ed or tend to be up regulated ( Figure 5). Accordingly, IL6 regulates the in ammatory activation by decreasing proin ammatory cytokines and upregulating anti-in ammatory cytokines [35]. Due to its dual role in in ammation, IL6 upregulation can be linked to the total reduction of in ammatory parameters when PDIA3 is silenced [36]. In addition, COX2 overexpression in cancer has been extensively reported and contributes to tumor development and progression [37; 38; 39; 40]. COX2 is also overexpressed in many gliomas and it is correlated with tumor grade and shorter survival [41; 42]. PDIA3-silenced T98G showed a downregulation of gene expression of COX2 when treated with activating stimuli reinforcing the idea of bene cial role of ERp57/PDIA3 inhibition in glioma patients.
Using the microglia-glioma interaction paradigm in functional experiments with T98G cells used as human glioma model and CHME-5 cells as human microglia model, we investigated the involvement of ERp57/PDIA3 in the bi-directional interaction between glioma and microglia cells. In this context, we produced conditioned media from both PDIA3-silenced and wild-type T98G cells and evaluated the urea release of CHME-5 when treated with CMs. In our studies, we used the urea release as parameter of M2 activation. Arginase-1 (ARG1) converts arginine to ornithine and urea, competing with nitric oxide synthase (NOS), usually used as M1 marker which utilizes arginine to produce nitric oxide [43; 44]. Our previous data shown an increase of urea production when CHME-5 were treated with CMs from wild type T98G, reinforcing the M2 phenotype [28]. Here, we interestingly reveal a phenotypic modi cation of CHME-5 by detaching the cells from the glioma-induced M2 phenotype with changing in urea production and in morphology. In fact, cyto uorimetric analysis showed less expression of ARG1 ( Figure 10) without an upregulation of NOS (data not shown) and in vitro assays shown a decrease of urea release when CHME-5 were treated with CMs from PDIA3-silenced T98G. Moreover, we investigated the IL6 release in such experimental conditions. In our work, we found that when microglia cells where treated with PDIA3silenced T98G PS-CM we assist to a decrease in IL6 levels. These data are linked to the decrease of CCL2 and IL6 from GAMs of glioma (mimicked by PDIA3-silenced T98G PS-CM) implicated in glioma growth and invasiveness [45]. This, associated with the anti-in ammatory role of IL6 and the urea decrease reported, supports our hypothesis of non-M2 phenotype of microglia. Interleuchin-6 also trigger the JAK/STAT3 pathway and in the GB pathology high levels of IL6 are related to poor outcome and overall survival [36; 46]. ERp57/PDIA3 modulates STAT3 signalling [15], is present in STAT3-DNA complexes [16] and ERp57/PDIA3 up or downregulation is respectively related to cancer progression and inhibition of proliferation through STAT3 [21; 47]. IL-10, IL6 and FGF are known STAT3 activators and STAT3 is constitutively activated in gliomas supporting tumorigenesis [48]. Moreover, STAT3 suppresses antitumor immunity in GB [49]. In this scenario, we investigated STAT3 expression when microglia cells interact with wild type T98G glioma cell and PDIA3-silenced T98G. Interestingly, comparing the CMs effects we are faced with opposite effects: on the one hand there is an up regulation of total STAT3 when we treat the cells with PDIA3-silenced B-CM, on the other hand we reported a downregulation when we are looking at the PDIA3-silenced PS-CM. It could be speculated that these opposite ways are led by FGF-19 augmented release in PDIA3-silenced B-CM and IL6 and FGF-19 decreased release in PDIA3-silenced PS-CM. At this stage, it is not possible to directly correlate the effect of PDIA3 silencing to the STAT3 activity but since PDIA3 is related to STAT3 at different levels, taken together these data are reinforcing the idea of a hybrid phenotype neither M1 nor M2 and the hybrid phenotype could be orchestrated by the STAT3 activation pathway. In addition, the different effect on STAT-3 by B-CM and PS-CM may be because to obtain PS-CM, the cells were subjected to stimulation with cytokines that use STAT3 dependent pathways. So the stimulation of these cells can be affected by the inhibition of PDIA3 (being linked to STAT3).
It is known that GAMs interact with glioma cells and the crosstalk is supported by cytokines and chemokines released from both tumor cells and GAMs [50]. In particular, glioma cells recruit GAMs through the monocyte chemoattractant protein MCP-1, also known as C-C motif ligand 2 (CCL2) [51].
In this sense, we evaluated the expression of 105 cytokines and chemokines released from both wild type and PDIA3-silenced T98G cells. Several are the cytokines and chemokines released in B-CM and some of them are pro-angiogenic such as Angiogenin, Angiopoietin 1 and VEGF. Additionally, we reported the release of Macrophage migration inhibitory factor (MIF) but probably such inhibitory effect is mitigated by the release of MCP-1. More generally, T98G glioma cells release chemoattractant factors and activation factors of the innate immune response. In PDIA3-silenced B-CM we reported mostly an enhancement of such responses but on the other hand we shown a signi cant reduction of IFGBP-2 and IL8 ( Figure 4). In concert with releasing factors belonging only to PDIA3-silenced B-CM as Complement factor D and M-CSF, the decrease of IL8 and IGFBP-2, and the increase of FGF-19, Osteopontin, that is involved in enhancing production of IFNγ and reducing production of IL-10, and PDGF-AA, that is a mitogen factor, is linked to the different phenotype experienced in vitro [45].
More complex is the situation in PS-CM with 43 cytokines and chemokines released from glioma cells. In fact, PS-CM is a condition that mimics late stage of pathology [27] and it's characterized by the release of in ammatory cytokines such as IL6, IL8, C-X-C motif chemokine 10 (CXCL10) also known as Interferon gamma-induced protein 10 (IP-10) as well as GM-CSF and MCP-1 to name a few. Interestingly, when we evaluate the cytokines released by the PDIA3-silenced PS-CM we are in front of a mitigation of the response triggered by activating stimuli with decreased secretion of most of them ( Figure 6). Some exceptions are chitinase 3-like, secreted by activated macrophages, GM-CSF, which is an activator of microglia, MCP-3, that is a chemotactic and activating factor of cells of in ammatory response, and CCL20, that is chemotactic for lymphocytes. In fact, such chemokines and cytokines are increased and are probably linked to the different number of cells reported in our in vitro assays. Moreover, PDIA3silenced T98G PS-CM releases also IL-24, G-CSF, LIF and CCL3/CCL4 that are not expressed before in PS-CM. It could be speculated that when PDIA3 is silenced and glioma cells are exposed to an activating stimulus, glioma cells are not able anymore to release high amount of IL6 and IL8 to induce the M2 phenotype in GAMs in concert with the IL-24 release, which is a tumor suppressing protein, and LIF, that inhibits the cell differentiation. Thus, glioma cells continue to attract microglia/macrophages in tumor microenvironment but with other differentiation that could be speculate as tumor resolving. It is clear and interesting that ERp57/PDIA3 knockdown in fact provokes different responses in GAMs and its inhibition could be bene cial and obtained with a natural compound: Punicalagin.
PUN has been demonstrated to affect ERp57/PDIA3 reductase activity as a non-competitive inhibitor [25] and neuroin ammation in microglia cells [55; 56; 57] but PUN activity in gliomas is poorly characterized and seems to be related to autophagic cell death and apoptosis [58]. In this context, we evaluated the anti-in ammatory effect of PUN in microglia cells and we investigated the viability of such cells in ow cytometry and in in vitro assays. It might be hypothesized that PUN effects are correlated to the cell type involved because the anti-in ammatory activity was well established in our experiments but not the proapoptotic activity; our dose of 5 micro molar was not enough to trigger a cytotoxic effect while with a ten time higher concentration ow cytometry analysis reveals a toxic effect of PUN.

Conclusion
In conclusion, we con rmed the ERp57/PDIA3 upregulation in glioma specimens with an upregulation in GAMs as well. We also demonstrated a dissimilar activation of GAMs when ERp57/PDIA3 is downregulated in glioma cells with differences both in the products downstream of ARG1 activation and at phenotypic level. We can hypothesize that PDIA3 correlates with STAT3 in glioblastoma cells in a feedback loop with IL6 as protagonist. In fact, IL6 is less secreted if PDIA3 was silenced and IL6 itself had less effect on the glioblastoma cell because IL6-dependent activation passes through STAT3, its activation may in turn require PDIA3. A block of PDIA3 makes glioblastoma cells less able to condition the surrounding medium in their favor (less M2 parameters released from microglia cells and the survival is inversely linked to the levels of PDIA3 in glioblastoma cells). In glial cells, however, the role of PDIA3 would be more protective. In fact, if PDIA3 is directly inhibited the apoptosis was favorable. Therefore, the ideal would be to inhibit PDIA3 in cancer cells but not in other cells. In this context, glioma cells are still able to retrieve GAMs from the surrounding tissue but with a different activation phenotype. Further studies are necessary to con rm the diverse phenotype caused as well as the use of Punicalagin needs to be con rmed in vivo.

Declarations
Ethics approval and consent to participate All patients signed an informed consent form, and the experimental protocol was approved by the Ethics Committee of Fondazione Policlinico A. Gemelli (Rome).

Consens for publication Not Applicable
Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.       Viability of CHME-5 after Punicalagin challenge: CHME-5 were treated with Punicalagin 5μM for 24 hours. *** p<0.001.

Figure 10
Analysis of M2 phenotype of CHME-5: A) CHME-5 were treated with CMs for 48h; B) Flow cytometry analysis of ARGINASE1: CHME-5 were treated for 24 hours with both CMs from T98G and PDIA3-silenced T98G. Data are shown as percentage of control. C) Flow cytometry: i) Dot plot Forward Scatter (FS) versus Side Scatter (SS). Each dot represents a single cell analyzed by the ow cytometer. The characteristic position of different cell population is determined by differences in cell size and granularity. ii) Fluorescence intensity measurements for Basal conditioned medium. Basal conditioned medium results increased in uorescence intensity to the right. iii) Fluorescence intensity measurements for Prestimulated conditioned medium. Prestimulated conditioned medium results increased in uorescence intensity to the right. * p<0.05; ° p<0.05; ** p<0.002.

Figure 10
Analysis of M2 phenotype of CHME-5: A) CHME-5 were treated with CMs for 48h; B) Flow cytometry analysis of ARGINASE1: CHME-5 were treated for 24 hours with both CMs from T98G and PDIA3-silenced T98G. Data are shown as percentage of control. C) Flow cytometry: i) Dot plot Forward Scatter (FS) versus Side Scatter (SS). Each dot represents a single cell analyzed by the ow cytometer. The characteristic position of different cell population is determined by differences in cell size and granularity. ii) Fluorescence intensity measurements for Basal conditioned medium. Basal conditioned medium results increased in uorescence intensity to the right. iii) Fluorescence intensity measurements for Prestimulated conditioned medium. Prestimulated conditioned medium results increased in uorescence intensity to the right. * p<0.05; ° p<0.05; ** p<0.002.

Figure 12
Anti-in ammatory properties of Punicalagin on CHME-5. A) CHME-5 treated for 48 hours with proin ammatory cytokines mix (TII) alone and in association with Punicalagin; B) IL6 release after 24 hours of treatment with Punicalagin; C) DCF production after 48 hours of treatment with TII alone and in association with Punicalagin. **** p<0.0001.

Figure 12
Anti-in ammatory properties of Punicalagin on CHME-5. A) CHME-5 treated for 48 hours with proin ammatory cytokines mix (TII) alone and in association with Punicalagin; B) IL6 release after 24 hours of treatment with Punicalagin; C) DCF production after 48 hours of treatment with TII alone and in association with Punicalagin. **** p<0.0001.