Abrogation of group III phospholipase A2 inhibits tumor growth and metastasis in ovarian cancer and promotes chemo-sensitization

Aberrant lipogenicity and deregulated autophagy are common in most advanced human cancer and therapeutic strategies to exploit these pathways are currently under consideration. Group III Phospholipase A2 (sPLA2-III/PLA2G3), an atypical secretory PLA2, is recognized as a regulator of lipid metabolism associated with oncogenesis. Though recent studies reveal that high PLA2G3 expression signicantly correlates with poor prognosis in several cancers, however, role of PLA2G3 in ovarian cancer (OC) pathogenesis is still undetermined.

restoration of PC for future OC treatment and the critical role of PLA2G3 in regulating ciliary function by coordinating interface between lipogenesis and metastasis.

Background
Peritoneal dissemination at diagnosis contribute to poor prognosis in ovarian cancer (OC) [1,2]. Understanding the molecular alterations that promote the aggressive behavior of OC can lead to new therapeutic options. Transformed cells rewire their metabolism [3] that enables them to survive and adapt to prevailing stress [4]. In addition to an exacerbated glycolysis, transformed cells exhibit adaptations in lipid/cholesterol metabolism to support their high growth rate [5] by scavenging exogenous lipids or by activating endogenous lipogenesis [6,7]. Excessive lipids/cholesterol stored as lipid droplets (LDs) in malignant cells are considered as one of the hallmarks of tumor aggressiveness conferring chemoresistance [8,9]. Also stress-induced FAs released from stored LDs supplied energy to maintain survival and metastatic phenotype of OC cells [10,11]. Assessment of tumor LD content by Raman spectroscopy is an emerging tool to monitor therapeutic response in patients [12]. Thus adapting novel therapeutic strategies to exploit the lipid-related metabolic dependence in cancer may improve the overall survival.
Phospholipase A 2 (PLA 2 ), a group of enzymes that hydrolyze phospholipids to release fatty acids (FA) and lysophospholipids, are critical regulator of lipid metabolism of transformed cells and associated with cancer progression [13]. Amongst them, Group III sPLA 2 (PLA2G3) was identi ed as a candidate biomarker for colon cancer positively correlated with short survival and high lymph node metastasis [14].
The biologically active lipid mediators generated by these enzymes stimulate proliferation abrogate apoptosis, increase in ammation and angiogenesis. Recent studies implicate sPLA 2 s in the regulation of basal lipid metabolism, thus opening a new avenue to uncover the diverse role of this secretory enzyme in cancer [15].
Autophagy maintains the energy balance and cellular homeostasis through turnover of unwanted proteins/organelles in lysosomes [16]. Reports suggest that in addition to acting as substrates for lipases, LDs under stress conditions undergo lipophagy, an autophagy-mediated breakdown [17]. Furthermore, autophagy plays a critical role in regulating primary ciliogenesis depending on cancer types [18,19]. Primary cilia (PC), a conserved microtubular appendage originating from mother centrosome and extending to extracellular space, are signal hubs in maintaining development and tissue homeostasis [20,21]. Dysregulated ciliary function is associated with oncogenesis and the loss of ciliogenesis in preinvasive stages of cancer is considered an early oncogenic event [22][23][24][25]. Aberrant activation of lipogenic transcription factor SREBP1c mediates ciliary loss in well-ciliated non-malignant cellular models [26]. However, the connection between lipophagy and ciliogenesis is not well characterized. Our study revealed that PLA2G3 regulates lipogenesis and alters the ciliary process in the OC cells which potentially impacts the oncogenesis. We found that targeting lipogenesis with metabolic inhibitor PFK158 attenuates PLA2G3 expression in an autophagy-dependent manner and restores the PC, thereby counteracting OC progression.

Materials And Methods
Reagents PFK158 inhibitor is acquired on MTA from Gossamer Bio (San Diego, CA). Other reagents and antibodies used were listed in Table S1.

Cell culture
The maintenance and the list of cell lines used in this study are shown in Table S2. Patient-derived ascites were obtained through the Mayo Clinic Ovarian SPORE program (IRB-1288-03, Dr. Shridhar) and in collaboration with the University of Minnesota Cancer Center Tissue Procurement Facility with IRB approval and cultured as mentioned [27,28].

Bodipy staining
Cells were xed and stained with Bodipy as described [30] and examined under Zeiss-LSM 510 uorescence microscope.
Transmission electron microscopy (TEM) analysis SCG control-transfected and PLA2G3-KO OVCAR8 cells were xed in Trump's xative [31] and directed to Microscopy and Cell Analysis Core facility in Mayo Clinic, MN for further processing. Images were captured in JEM-1400 TEM (Jeol, USA).

Immunoblot assay
Cell lysates were subjected to immunoblot [29] with primary antibodies listed in Table S1. Target proteins were visualized by uorophore-conjugated secondary antibodies (LICOR) and using LI-COR OdysseyFc Imaging System (Nebraska, USA).
Immuno uorescence (IF) assay: PFK158-treated cells with/without BafA1 or the PLA2G3 KD cells were grown in 8-well chambered slide, xed and stained with tagged anti-acetylated α-tubulin-Alexa-Fluor561 and anti-PLA2G3 (1:100) as described [31] and visualized by Zeiss-LSM510 confocal microscope. Autophagic ux was measured by confocal microscopy upon transient transfection with RFP-LAMP2 and GFP-LC3B in the KO and SCG-control cells by cisplatin treatment and with GFP-RFP-LC3B followed by PFK158-treatment.
Cyto-ID autophagy detection by uorescence microscopy PLA2G3 KO and SCG-control cells were treated with Cisplatin and assessed for autophagic induction by Cyto-ID staining as per manufacturer's protocol.

Immunohistochemistry
IHC studies were performed on formalin xed de-para nized sections probed against Ki-67 as described [32].
Clonogenic assay PLA2G3 KD/KO and control-transfected cells were plated (500cells/well) and allowed to grow for ~ 14days until colonies became visible. Cells xed, stained with crystal violet and images provided.

In vivo xenograft study
Female athymic nude mice (nu/nu, 4-6weeks old; Jackson Laboratories, USA) were randomized into 4 groups (n = 7). OVCAR5 SCG control-transfected and PLA2G3 KO cells (5×10 6 cells; 2 groups each) were injected intraperitoneally (i.p). Seven days following injection, one group each from the SCG and KO cohorts was treated with CBP (50 mg/kg) once in a week for 4 consecutive weeks. All mice euthanized on day30 and tumor weight was determined. Tumors were preserved in formalin for IHC and fresh frozen for protein analysis. The experiments were carried out under the approved protocol and guidelines of Mayo Clinic Animal Care and Use Committee, MN.

Statistical analysis
All investigation was performed in triplicates for 3 independent experiments unless mentioned. The results were expressed as mean ± standard deviation. Signi cant changes (*p < 0.05, **p < 0.01) were determined using student's t-test unless otherwise noted.

PLA2G3-de cient OC cells showed attenuated tumorigenesis.
Immunoblot analysis showed PLA2G3 is highly expressed in several OC cells compared to normal fallopian tube epithelial FTEs 190 and 194 and is not expressed either in FTE240 [34] or normal ovarian broblast NOF151hTERT cells (Fig. 1A). TCGA analysis of high-grade serous subtype showed 2.26% cases with altered PLA2G3 expression (@cbioportal, Fig. S1A). To better understand the role of PLA2G3, we generated PLA2G3 knockout (KO) clones in OVCAR8 cells and two shRNA-mediated stable PLA2G3-KD clones of OVCAR5 (sh33/sh35) along with scrambled RNA (SCG) and non-targeted control (NTC) transduced cells as controls respectively. E cient PLA2G3 downregulation in the KO and KD cells was veri ed by immunoblot analysis (Fig. 1B).
Clonogenic survival assays of OVCAR8-KO and OVCAR5-sh33/sh35 KD cells showed signi cantly reduced number of colonies compared to respective controls ( Fig. 1C-D). Additionally, wound-healing assay showed signi cant reduced migration in both KO and KD cells compared to respective controls ( Fig. 1E-F, S1B-C). We reported that group-IVA cytosolic phospholipase A2 is a critical regulator for LD biogenesis in cancer [35]. To determine whether PLA2G3 affects lipid metabolism in OC cells, we assessed LD formation by Bodipy staining and found a decrease in the number of LDs in KO and sh35/33 KD cells compared to respective controls (Fig. 1G, S1D). Consistent with these results, TEM analysis also showed a signi cant reduction in the numbers of LDs in the OVCAR8-KO cells compared to SCG-transfected cells (Fig. 1H). Collectively, these results suggest that PLA2G3 abrogation impairs the lipogenesis pathway and attenuates OC tumorigenesis.
PLA2G3 KD cells are sensitive to Platinum based-drug treatment.
Given that LD-rich cancer cells exhibit chemo-resistive properties, we investigated whether PLA2G3 KD sensitizes cancer cells to cisplatin/carboplatin-induced cytotoxicity. The OVCAR8 KO and SCG-control cells treated with increasing concentrations of cisplatin for 24hrs showed a signi cant reduction in IC50 value from 9.91 µM in SCG-transfected cells to 4.75 µM in the KO cells ( Fig. 2A). Similarly, stable HeyA8MDR PLA2G3-KD cells (Fig. S2A) treated with increasing CBP doses for 24hrs showed a decrease in IC50 value to 95.4 µM compared to 170 µM in the NTC cells ( Fig. S2B-C). Additionally, the KD cells showed an improved dose-dependent decrease in cell survival upon CBP treatment compared to NTC cells (Fig. S2D). Together, these results substantiate that PLA2G3 downregulation sensitizes cells to platinum-drug mediated cell death.
Likewise, to understand whether induction of autophagy is essential for sensitizing OC cells to cisplatininduced cytotoxicity, we pretreated OVCAR8 cells with BafA1 (inhibitor of autophagolysosome formation) for 2hr followed by cisplatin treatment for 24hrs. Results showed that inhibition of autophagy diminished the cytotoxic effect of cisplatin in OC cells (IC50: 13 µM to 28 µM, Fig. 2B), which suggests that autophagy-mediated cytotoxicity is critical for sensitizing cancer cells to the platinum drug-induced cell death. To validate the role of PLA2G3 in sensitizing cells to platinum drug-induced cytotoxicity, IF analysis for GFP-LC3B puncta expression and its co-localization with lysosomal associated membrane protein 2 (RFP-LAMP2) upon cisplatin treatment in both KO and SCG-transfected OVCAR8 cells was performed. Results showed increased expression and co-localization of RFP-LAMP2 and GFP-LC3B in the KO cells compared to control upon cisplatin treatment (Fig. 2C). Likewise, Cyto-ID staining used as a readout for autophagic induction, showed a signi cant increase in uorescent signal in KO cells compared to SCG-transfected cells upon treatment with 5 µM cisplatin (Fig. 2D). Immunoblot analysis also revealed an increased expression of LC3BII and cleaved PARP1 with reduced p62/SQSTM1 levels in KO cells compared to SCG-transfected cells upon cisplatin treatment (Fig. 2E). Pretreatment for 2hr with BafA1 inhibited the cisplatin-induced autophagy with increase of the LC3BII and rescue of the p62/SQSTM1 levels respectively and a decreased induction of the cleaved PARP1 in the cells (Fig. 2E). Collectively, these results suggest that PLA2G3 KD increased sensitivity of the cancer cells to autophagy-induced cytotoxicity upon treatment with platinum drugs.
Aberrant PLA2G3 expression impairs PC formation in OC cells.
Since aberrant lipogenic signaling is associated with distortion of PC [26] we assessed whether LD deregulation due to aberrant PLA2G3 expression is involved in regulation of ciliogenesis in OC cells. Immunoblot analysis of OVCAR5-KD and OVCAR8-KO cells showed a signi cant increase in expression of acetylated α-tubulin (a marker for PC) compared to respective controls ( Fig. 3A-B). Likewise, IF study using uorescently tagged-acetylated α-tubulin revealed an increase in percent ciliation in OVCAR5 sh35-KD and OVCAR8-KO cells compared to controls (Fig. 3C-D respectively) with e cient downregulation of PLA2G3 under similar conditions. Together these data validate the importance of PLA2G3 in the regulation of lipid metabolic pathway and PC in OC. To determine the role of PC in OC progression, we transiently knockdown IFT88, a key factor regulating ciliogenesis, and as shown in g. S3A, a decrease in acetylated α-tubulin in KD cells was con rmed by immunoblot. IFT88 KD cells showed a signi cant increase in colony forming ability and increased migration of OVCAR5 cells (Fig. S3B-C).
Knockdown of PLA2G3 inhibits in vivo tumorigenesis and metastatic spread in OVCAR5 xenograft model.
To support our in vitro ndings, the effect of PLA2G3-KO alone and in combination with CBP treatment on tumor growth and metastatic spread was assessed in vivo as described (Fig. 4A). No signi cant alteration in health condition was observed (data not shown); however, two mice died in the control group due to unknown reasons very early on and therefore had to be excluded in the analysis. PLA2G3-KO tumor-bearing mice showed a signi cant reduction in tumor growth and metastatic spread compared to the SCG-control group (Fig. 4B). Interestingly, the KO tumor-bearing mice showed almost no tumor burden upon CBP treatment compared to the SCG-control cohort (Fig. 4B). Comparative statistical analysis of the tumor weight and Ki67 staining of tumor tissue sections showed a similar signi cant reduction in the KO model both with and without CBP treatment (Fig. 4C-D). Immunoblot analysis con rmed downregulated PLA2G3 expression and increased acetylated α-tubulin in the KO-cohort compared to SCG-derived xenografts (Fig. 4E) and an increased expression of LC3BII with p62 downregulation in CBP treated SCGcontrol and KO cohort of mice (Fig. 4F). Hence, our in vivo data supports the role of PLA2G3 in metastatic spread and its downregulation sensitizes cells to chemotherapy.
Targeting by PFK158 inhibitor restores PC by reducing PLA2G3 in an autophagy-dependent manner.
Although we highlighted the role of PFK158-induced autophagy in regulating lipophagy [35], it did not address if PFK158 regulated ciliation. Driven by our observations, we wondered if inhibition of lipogenic signaling by PFK158 can restore ciliation in OC cells. IF analysis with uorescently tagged-acetylated αtubulin, showed that PFK158 treatment signi cantly restored PC in both OVCAR8 and OVCAR5 cells (Fig. 5A, S4A). Quantitation of increase in percent cilia is shown in Figs. 5B-C. Immunoblot analysis of acetylated α-tubulin also showed increased expression in OVCAR8 and OVCAR5 cells upon treatment with PFK158 (Fig. 5D). To understand if PFK158-induced autophagy plays a role in induction of PC, we monitored induction of autophagic ux by GFP-RFP-LC3B transfection in PFK158 treated OVCAR5 cells.
Confocal analysis showed induction of autophagic ux through the formation of increased red puncta in PFK158-treated cells compared to untreated cells (Fig. 5E). Interestingly, pretreatment with BafA1 for 2hr inhibited PFK158-induced increase of acetylated α-tubulin in OVCAR8 cells (Fig. 5F, top-panel). Also, immunoblot analysis showed PFK158-treatment attenuated PLA2G3 expression which was restored when cells were pretreated with BafA1 (Fig. 5F, middle-panel). To understand whether inhibition of autophagy regulates PC levels, we treated OVCAR8 cells with 3MA and BafA1 individually and observed that both early and late stage inhibition of autophagy downregulated acetylated α-tubulin levels (Fig.  S4B). Effect of BafA1 as an inhibitor of PFK158-induced autophagy was con rmed by the resulting increases in LC3BII expression and the rescue of p62/SQSTM1 both OC cells (Fig. 5G-H; panels2-3). Under similar conditions, the PFK158-induced reduction of PLA2G3 was restored in both cells in presence of BafA1 (Fig. 5G-H; panel-1).
Consistent with these results, PFK158-induced ciliation was inhibited by BafA1 treatment as determined by levels of acetylated α-tubulin with confocal microscopy (Fig. 6A-B). Under parallel conditions, BafA1 treatment rescued LD formation that was reduced by PFK158 in OVCAR8 cells (Fig. 6C). Taken together, these results mechanistically support that PFK158-induced autophagy-mediated downregulation of PLA2G3 regulates PC in OC. By analyzing percent ciliated cells, we determined that BafA1 treatment downregulated acetylated α-tubulin in OVCAR5-sh35KD cells con rming the role of autophagy (Fig. 6D-E), which was also validated by western blot analysis, that showed a rescue of both LC3BII and p62 levels (Fig. 6F). To validate the role of autophagy in regulating ciliogenesis, we determined acetylated α-tubulin levels in WT and in autophagy compromised Atg5 −/− MEFs, by IF. Atg5 −/− MEFs showed reduced percent ciliated cells compared to their WT counterpart (Fig. 6G), which was also corroborated by western analysis (Fig. 6H, panel-2). Further, PFK158 treatment did not show a signi cant change in the expression of acetylated α-tubulin in Atg5 −/− MEFs (Fig. 6I, lower panel-1). In contrast there was a signi cant up-regulation in WT cells (Fig. 6I, top panel-1). Together, these results show PFK158-induced autophagy that leads to PLA2G3 degradation regulates ciliary maintenance.
PFK158-mediates autophagic degradation of PLA2G3 and reduces viability in patient-derived ascites.
To understand the clinical relevance, we determined PLA2G3 expression in 9 patient-derived ascites cells [27,28]. Immunoblot analysis with human epithelial speci c antigen marker (EpCAM) and broblast activated protein marker (FAP) showed that the ascitic cells are predominantly epithelial in nature (Fig.  S5A) and 5 out of 9 samples expressed PLA2G3 (Fig. 7A-B). A reduction in percent viability was observed in PLA2G3-expressing ascitic cells following PFK158 (0-20 µM) treatment at 24hr with IC50 values ranging between 4.0-9.0 µM (Fig. 7C-D). Immunoblot analysis showed PFK158-induced autophagy, as determined by an increase in LC3BII, decreased p62 levels, and downregulation of PLA2G3 respectively ( Fig. 7E-F). Further, IF analysis showed signi cant increase in percent ciliated cells upon PFK158treatment in the A7683, KP263 and A4832 ascitic cells model compared to untreated control ( Fig. 7G-I,  S5B). When cisplatin was combined with 1/2IC50 of PFK158, a substantial reduction in IC50 ranging from 29 µM to 7 µM (AM812), 37.5 µM to 8.5 µM (KP263) and 33 µM to 21 µM (JM076; Fig. 7J-L) was observed, suggestive of the ability of PFK158 to sensitize the cells to chemotherapy. Together PFK158treatment sensitizes the patient-derived OC ascites to chemotherapy at least in part through the degradation of PLA2G3.

Discussion
Although aberrant expression of several human sPLA2s is reported in the pathogenesis of different cancers [13], the role of sPLA2s is controversial, since it can function either as a positive or negative regulator of tumorigenesis depending on the isoform, tissue/cancer types [15]. Current evidence suggests that high expression of PLA2G3 signi cantly correlates with metastasis and poor prognosis [36]. Therefore, a better understanding on the role and regulation of PLA2G3 in OC can open new therapeutic opportunities for targeting these enzymes.
Once secreted the sPLA2s can function either in an autocrine or paracrine manner, or as enzymes acting on extracellular/cellular phospholipid substrates to change the nature of FAs and lysophospholipids in tumor microenvironment [15,37]. Current studies highlight the role of various sPLA2s in modulation of lipid metabolism, which add-ons to their functional complexities [38]. Recent development is focused on therapeutic peptides and small molecule inhibitors to inhibit the sPLA2 activity in different cancers [39]. Their potential as cancer biomarkers is well recognized [14], and due to their secretion into tumor microenvironment, can lead to the development of new target opportunities.
Increased de novo lipogenesis in transformed cells acts as an additional energy source for a high proliferative rate [38,[40][41] and is associated with chemoresistance [10]. Reports suggest that LDmediated resistance to chemotherapy is multifactorial and associated with poor prognosis [42,43] and support the concept that targeting LDs alone or in combination with standard chemotherapy may lead to new clinical outcomes in cancers with lipogenic phenotype. We found a signi cant attenuation in LD biogenesis in PLA2G3-de cient OC cells. Since PC is associated with altered lipogenesis [26], and based on the role of PLA2G3 in ciliogenesis, we further explored whether PLA2G3 downregulation can restore PC in OC and can sensitize them to chemotherapy. PC is the regulatory signaling hub [44] and its loss in pre-invasive stages act as an early-oncogenic event in namely pancreatic adenocarcinoma [25]. Deng et.al., reported the role of PC in sensitizing cells to transformation through mevalonate pathway activation indicating the regulation of metabolic plasticity in cancer and ciliogenesis [45]. In support, Gijs et.al., reported that aberrant activation of SREBP1c suppresses cilia formation [26]. To this end, our present study suggests that PLA2G3 downregulation results in the PC restoration and sensitizes OC cells to platinum-drugs. The outcome from our in vivo analysis of OVCAR5 xenografts showing substantial inhibition of tumor growth/metastasis was accompanied by an increase in acetylated α-tubulin in KOderived xenograft compared to control and suggests the role of PLA2G3 towards OC metastasis partially through regulation of ciliogenesis.
Our ndings provide novel insights into the abrogation of PLA2G3 towards sensitizing the OC cells to chemotherapy in part by downregulating LD biogenesis and restoring ciliogenesis. Our data add to current understanding on controversial interplay between PC and autophagy in other cancers. Tang  Importantly, PFK158 treatment of patient-derived ascites cells also reveals that PLA2G3 is degraded in an autophagy-dependent manner to sensitize cells to cisplatin and reduce cell viability, thus providing crucial in vivo support. Of relevance, our data showed that PFK158-induced acetylated α-tubulin levels are signi cantly higher in ascitic cells. We analyzed PLA2G3 expression between recurrent ovarian tumors (secondary debulking) and their autologous primary tumors (primary debulking) from 19 patients with triplicate cores on a tissue microarray (TMA) by IHC, and did not nd statistically signi cant differences between the groups (data not shown). One limitation is small number of patient tumors analyzed which limits statistical power. Given our data that PLA2G3-KD inhibits metastasis in vivo, analyzing primary ovarian tumors vs their autologous metastatic tumors (bowel/omental Mets) in patients, may be more informative to determine the role of PLA2G3 in the OC prognosis.
Taken together our ndings become signi cant as many recent studies aim at identifying drugs that can be repurposed and used to restore ciliogenesis in cancer cells [49]. Thus, elucidation of pathways and molecular inhibitors towards ciliogenesis will be of interest to forge forward to be tested in a novel clinical setting.

Conclusion
Our study in ovarian cancer provides signi cant novel insights into the abrogation of PLA2G3 towards sensitizing the cancer cells to chemotherapy. Since ciliation is found to be regulated in an autophagy dependent process, this nding also paves the path for future analysis of the small molecule autophagy modulators that can have a promising impact on inhibiting OC progression as well. The authors declare that there are no con icts of interest. Funding: This work was supported by grants from the Department of Experimental Pathology and Laboratory Medicine Discretionary Funds, and Mayo Clinic (VS), National Institutes of Health P50CA136393 for providing the ovarian TMA and ascites cells.
Authors' contributions: UR conceptualized the work, acquired and interpreted the data and drafted the manuscript. DR acquired and analyzed data. LJ acquired data. PT, JS, YX and EK reviewed and edited the manuscript. AWH, GC and KG analyzed and interpreted the TMA data. AO analyzed and interpreted the TMA data, reviewed and edited the manuscript. VS supervised the work, acquired the funding and substantively revised and edited the manuscript. All authors read and approved the nal manuscript.   and cleaved PARP1 following 5μM cisplatin treatment for 24hr with or without 2hr pretreatment with BafA1 (50nM). PCNA was probed for endogenous control.