Proteomic Profiling of Iron-Treated Ovarian Cells Identifies AKT Activation, which Modulates the CLEAR Network

Abstract

Although iron is essential for cell survival, dysregulated levels can contribute to cancer development or even cell death. The underlying mechanisms mediating these events remain unclear. Herein, we assessed proteomic alterations in iron-treated ovarian cell lines using Reverse Phase Protein Array (RPPA) technology and potential functional responses via Ingenuity Pathway Analysis (IPA). Using these approaches, we noted upregulation of pathways modulating organismal death with alterations in mTOR, MAPK, and AKT signaling in HEY ovarian cancer cells in contrast to T80 non-malignant ovarian cells. Since modulation of cell death is mediated in part via microphthalmia-associated transcription factor (MiTF) family, which regulate lysosomal biogenesis and autophagosome formation by upregulating expression of coordinated lysosomal expression and regulation (CLEAR) network, we investigated expression changes in these factors in response to iron. We identified increased transcription factor EB (TFEB) in T80 (relative to HEY), accompanied by its nuclear translocation and increased CLEAR network gene expression with iron. We noted that inhibition of AKT altered these responses in contrast to mTOR inhibition, which had little effect. Collectively, these findings support use of RPPA/IPA technology to predict functional responses to iron and further implicate AKT pathway and MiTF members in iron-induced cellular responses in ovarian cells.

1. Introduction

Although iron is an essential nutrient needed to maintain survival, its ability to generate reactive oxygen species via Fenton reactions also classifies it as a mutagenic agent which may promote cancer pathogenesis ^[1–3] as well as cell death ^[4]. Specifically, iron overload disorders (i.e., haemochromatosis) are associated with increased cancer risk ^[1]. Furthermore, exogenous iron has been reported to induce ferroptosis ^[5] as well as other cell death-like responses ^[6]. However, well-defined signaling mechanisms involved in these iron-induced cellular responses have not yet been identified. Our earlier studies focused primarily on the Ras/MAPK signaling cascade and its contribution to the iron-induced biochemical changes and cell death responses in gynecological cell lines ^{[6, 7]}. Such transition metals including copper and iron have been shown to alter lysosome function and increase lysosomal biogenesis ^[8]. Moreover, we recently identified that ovarian epithelial cells harbored an abundance of lysosomes rich in iron particles, which may have contributed to enhanced clearing of excess intracellular iron ^[6]. Apart from the Ras/MAPK pathway, it is likely that numerous other signaling cascades may additionally contribute to these observed changes.

In the present study, we utilized the focused reverse phase protein array (RPPA) technology to identify changes in signaling pathways in response to iron treatment in a variety of ovarian cell lines. In contrast to untargeted mass spectrometric approaches, RPPA technology has limitations due to the requirement for validated antibodies although it has been used in prior studies for targeted analysis of specific pathways, including the PI3K, MAPK, LKB1, JAK/STAT, apoptosis, cell polarity, TGFβ, and tyrosine kinase signaling cascades (i.e., EGFR) ^{[9, 10]}. Nonetheless, RPPA techniques have been employed for predicting drug sensitivity in a wide array of cancers ^[11]; however, to the best of our knowledge, this is the first targeted proteomic study investigating signaling pathways which are altered in response to iron. In this regard, we noted significant alterations in the AKT, MAPK, and mTOR pathways between HEY ovarian cancer and T80 immortalized ovarian surface epithelial cells. We validated these changes via western analysis and investigated cellular responses to iron that may be mediated through these specific pathways. Herein, although we identified iron-induced changes in the AKT/mTOR pathways in our gynecological cell lines, further investigations are needed to delineate additional global proteomic alterations using an unbiased and untargeted mass spectrometric approach.

2. Materials and methods

2.1. Cell culture

T80 (human immortalized (with LTAg and hTERT) ovarian epithelial), T80 cells overexpressing H-Ras (T80+HRas), and HEY (human ovarian carcinoma) cell lines were kindly provided by Dr. Gordon Mills and Dr. Jinsong Liu (MD Anderson Cancer Center (Houston, TX, USA)). These cell lines were maintained in RPMI 1640 supplemented with 8% FBS and 1% penicillin/streptomycin as previously described ^[6]. TOV112D endometrioid ovarian carcinoma cells (obtained from ATCC, Manassas, VA, USA) and TOV21G clear cell ovarian cancer cells (kindly provided by Dr. Jonathan Lancaster, Moffitt Cancer Center, Tampa, FL, USA) were maintained in MCDB131:Medium 199 (1:1 ratio) supplemented with 8% FBS and 1% penicillin/streptomycin as previously described ^[6]. All cell lines were authenticated by STR profiling (Genetica Laboratories (Cincinnati, OH, USA)), and were confirmed to be negative for mycoplasma.

2.2. Cellular treatments

Ferric ammonium citrate (FAC) was obtained from Fisher Scientific (Pittsburgh, PA, USA). FAC was prepared as a 50mM stock in phosphate-buffered saline (PBS) and used at a final concentration of 250μM for the indicated times, as previously described ^[6]. The mTOR inhibitor Torin 1 (#14379, Cell Signaling Technology (Danvers, MA, USA)) was prepared in dimethyl sulfoxide (DMSO) and used at a final concentration of 10nM as reported in ^[12]; control cells were treated with an equivalent percentage of DMSO. The AKT inhibitor GDC0941 (Selleckchem (Houston, TX, USA)) was prepared in DMSO and used at a final concentration of 1μM as previously published in ^[13].

2.3. Lysate preparation for Reverse Phase Protein Arrays (RPPA) and data processing

T80, T80+HRas, HEY, TOV21G, and TOV112D cells were treated for 1h, 18h, and 48h with 250μM FAC (prepared in PBS, as detailed above) and the Untreated Control corresponded to 48h FAC. Protein lysates were collected as described below, and three biological replicates were assessed. Samples were analyzed at the RPPA Core Facility at MD Anderson Cancer Center (Houston, TX, USA) as previously described ^{[14, 15]}. Briefly, cell lysates were two-fold-serial diluted five times (1:2 to 1:16 dilution) in dilution buffer containing SDS with 2-mercaptoethanol followed by their application to nitrocellulose-coated slides (11 samples by 11 samples array format). In total, there were 197 slides to which 170 antibodies were applied (with 22 sets of replicated antibodies). Using an automated BioGenex autostainer (San Ramon, CA), slides were probed with antibodies and signals were developed using the Dako Catalyzed Signal Amplification system (DakoCytomation California Inc., Carpinteria, CA) in which visualization was enabled using 3,3’-diaminobenzine (DAB). Please see Supplementary Table 1 for a complete list of antibodies used in this study, including details of antibody-host organism, dilution, and manufacturer. Quality control (QC) testing was completed for each antibody staining as previously described in ^[16]. Using a flatbed scanner, slides were scanned to produce a 16-bit Tiff image from which spots were then identified and their intensity quantified for data processing and statistical analyses using MicroVigene (which employs a Supercurve method developed using an R package) as previously described ^[15] thus generating Supercurve Log₂ values. These values were then normalized for protein loading, transformed to linear and then to Log₂ values followed by median-centering for Hierarchical Cluster analysis (using Cluster 2.51) which utilizes a Pearson Correlation and a centered metric. Out of the 170 analytes, we identified 15 antibodies that were not fully validated from the entire RPPA dataset and thus, we eliminated them from our final analyses for a final of 155 targets.

2.4. Protein isolation, SDS-PAGE, and western analyses

Protein lysates from cultured cells were prepared as previously described ^[17]. Protein lysates were quantified using the Bicinchoninic acid assay (BCA Assay, ThermoFisher). We then normalized all samples for each independent experiment equivalently (i.e., 1500μg/ml). When sufficient protein was not obtained (at least 1mg/ml), we standardized all samples within the independent experiment to the lowest concentration. Normalized samples were loaded onto 10% SDS-PAGE gels and transferred to polyvinylidene fluoride (PVDF) membranes. As indicated in the Figure Legends, samples which were re-run to prevent interference between antigens of similar molecular weights are separated with a dotted line. For all experiments, three biological replicates were performed. Western blotting was completed as described previously ^[18] with the following antibodies from Cell Signaling Technology (Danvers, MA, USA): TFEB rabbit polyclonal (#4240S, 1:1000), LC3B rabbit polyclonal (#2775S, 1:1000), FTH1 rabbit polyclonal (#3998S, 1:500), TFE3 rabbit polyclonal (#14779S, 1:1000), MiTF rabbit polyclonal (#12590S,1:1000), pS473 AKT rabbit monoclonal (#4060S, 1:1000), Total AKT rabbit monoclonal (#4685, 1:1000), pS21/S9 GSK3 rabbit polyclonal (#9331S, 1:1000), Pan-Actin rabbit polyclonal (#4968S, 1:500), pS235/pS236 S6K rabbit polyclonal (#4858P, 1:1000), total S6K rabbit polyclonal (#2217S, 1:2500), Lamin A/C mouse monoclonal (#4777S, 1:1000), and GAPDH rabbit polyclonal (#5174P, 1:1000). CD71 mouse monoclonal (sc-51829, 1:250) was obtained from Santa Cruz Biotechnology (Dallas, TX, USA). SQSTM1 mouse monoclonal (#610832, 1:1000) was obtained from BD Biosciences (San Jose, CA, USA). Multiple exposures (short/long exposures) were captured to film to identify signals within the linear range. Densitometric analyses were completed using ImageJ (NIH).

2.5. siRNA transfection

T80 and HEY cells were seeded at 350,000 cells per well in 6-well plates and transfected with siRNA as described previously ^[18]. The following siRNAs were utilized: (1) non-targeting control (D-001810-10-20), (2) TFEB (L-009798-00-0005), (3) TFE3 (L-009363-00-0005), and (4) MiTF (L-008674-00-0005). Two rounds of siRNA transfection were completed for MiTF knockdown to obtain efficient knockdown, while one round was completed for TFEB and TFE3 knockdown. For all siRNA experiments, three biological replicates were performed.

2.6. Subcellular fractionation

The NE-PER Nuclear and Cytoplasmic Extraction Reagents (#78835, Fisher Scientific (Pittsburgh, PA, USA)) were utilized according to the manufacturer’s protocol. Cytoplasmic and nuclear fractions were collected and run on 10% SDS-PAGE gels for western blotting, and three biological replicates were performed.

2.7. Immunofluorescence staining

T80 and HEY cells were seeded onto glass coverslips (250,000 cells per well), allowed to adhere overnight, treated with 250μM FAC, and processed according to previously published methodology ^[6]. For colocalization studies, cells were fixed in 4% formaldehyde (in PBS), followed by blocking in 5% goat serum and 0.1% Triton-X-100 (in PBS) for 1h at room temperature. The cells were then stained with anti-TFEB mouse monoclonal (#H00007942, 20μg/mL, Abnova (Taipei City, Taiwan)) antibody overnight in a humidified chamber at 4°C followed by a 1h incubation with AlexaFluor-488 anti-mouse antibody (#A12379, 1:500, Fisher Scientific (Pittsburgh, PA, USA)). Next, the fixed cells were incubated with either anti-LAMP1 (#9091, 1:200, Cell Signaling Technology (Danvers, MA, USA)) or LC3B rabbit polyclonal (#2775, 1:2000, Cell Signaling Technology (Danvers, MA, USA)) antibody overnight in a humidified chamber at 4°C followed by 1h incubation with AlexaFluor-546 anti-rabbit secondary (#A11035, 1:500, Fisher Scientific (Pittsburgh, PA, USA)). Slides were viewed and imaged at 63X magnification using a PerkinElmer UltraVIEW Confocal Spinning Disc Microscope (PerkinElmer Incorporation). Three independent experiments were performed. Representative images were captured and colocalization was quantified via Pearson Correlation Coefficient using the Velocity Software (Version 6.1.1).

2.8. Staining with LysoTracker Red

HEY cells were seeded onto glass coverslips at 250,000 cells per well in 6-well plates, allowed to adhere overnight, and then treated with 250μM FAC. LysoTracker Red staining was performed as previously reported ^[6]. Briefly, LysoTracker Red (#L7528, 75nM, Life Technologies (Carlsbad, CA)) was applied to the cells 1 hour prior to completion of FAC incubation. Cells were washed with PBS, fixed in 4% formaldehyde (in PBS) for 30 minutes, blocked with 5% goat serum and 0.1% Triton-X-100 (in PBS) for 1h at room temperature, and then incubated with TFEB mouse monoclonal antibody in a humidified chamber overnight at 4°C (see antibody details, above). Samples continued to be processed according to the methods described above.

2.9. RNA isolation and real-time PCR

The RNeasy Kit (QIAGEN (Valencia, CA, USA)) was utilized to isolate total RNA according to the manufacturer’s instructions. Real-time PCR was conducted using the One-step Master Mix from Applied Biosystems (#4392938 (Foster City, CA, USA)) with the following FAM-labeled probes/primers: (1) HO-1 (Hs01110250_m1), (2) SQSTM1 (Hs01061917_g1), (3) LAMP1 (Hs00174766_m1), (4) UVRAG (Hs01075434_m1), (5) Beclin1 (Hs00186838_m1), (5) CTSD (Hs00157205_m1), (6) MCOLN1 (Hs01100653_m1), and (7) ATP6AP1 (Hs00184593_m1). C_T values were normalized to β-actin (#401846, Applied Biosystems (Foster City, CA, USA)) and RNA-fold changes were derived using the formula 2^−ΔΔCT. For all analyses, three biological replicates were performed.

2.10. Cloning of TFEB and generation of TFEB retroviral HEY cell lines

Using the RNeasy Mini Kit (QIAGEN (Valencia, CA, USA)), RNA was isolated from T80 and HEY cells for reverse transcriptase (RT)-polymerase chain reaction (PCR) to amplify TFEB. The primers used were: (1) 5’-GGG-GAA-TTC-ACC-GCC-ATG-GCG-TCA-CGC-ATA-GG-3’ (forward primer) and (2) GGG-GAA-TTC-TCA-CAG-CAC-ATC-GCC-CTC-3’ (reverse primer). The following PCR conditions were used: 48°C for 30 min, 94°C for 2 min, 40 cycles of 94°C for 1 min, 55°C for 1 min, and 68°C for 5 min, with final extension at 72°C for 15 min. PCR products were analyzed on a 1% agarose gel followed by gel purification using the QIAquick Gel Extraction kit (QIAGEN (Valencia CA, USA)) following the manufacturer’s protocol. Products were next cloned into the pTOPO vector (Invitrogen (Carlsbad, CA, USA)). Positive clones were validated by sequencing (Molecular Genomics Core, Moffitt Cancer Center (Tampa, Florida, USA)) then subcloned into the pBABE-puro retroviral expression vector (Addgene plasmid #1764 ^[19]) at the EcoRI site.

HEK293T packaging cells were transfected with TFEB or empty pBABE-puro (as a control) vector along with pCGP and pVSVG vectors (1:1:1 ratio) using Fugene HD (Roche (Indianapolis, IN, USA)). Retroviral particles were collected at 48 and 72 h post-transfection, filtered (0.45μm), and utilized to infect HEY cells with 8μg/ml polybrene (first round of infection) or 16μg/ml polybrene (second round of infection) followed by antibiotic selection in 2μg/ml puromycin.

2.11. Statistical analyses

Graphpad Prism program, version 6.04 (GraphPad (La Jolla, CA, USA)), was used for data analysis. P-values were calculated using the non-parametric Student’s t-test. The error bars displayed reflect the mean ± standard deviation (SD). NS represents non-significant values; * represents p ≤ 0.05; ** represents p ≤ 0.01; *** represents p ≤ 0.001; and **** represents p ≤ 0.0001. For RPPA, normalized nonzero intensities of both phosphorylated and non-phosphorylated proteins were uploaded to Perseus processing suite (Perseus version 1.6.1.1) where the intensities were log₂ transformed. ANOVA and post hoc testing (FDR <0.05) were then performed to determine significant expression changes across all FAC treatment time points without a fold-change cutoff (to not restrict the bioinformatic analysis which could limit enrichment analysis based on the limited size of the dataset). Non-transformed ratio averages of differentially expressed proteins found significant in each of the following comparisons: 1h/0h, 18h/0h, and 48h/0h, were then uploaded to Ingenuity Pathway Analysis (IPA) to identify predicted activated or inhibited upstream regulators as well as over-represented canonical pathways and networks. For all IPA analyses, the Uniprot IDs were uploaded, along with phosphorylated protein ratios, phosphorylation sites, and non-phosphorylated protein expression ratios. Core analysis was performed separately for phosphorylated proteins and non-phosphorylated proteins via phosphorylation and expression analysis, respectively. The resulting predictions were considered for further analysis and validation based on a z-score in the range of approximately −2< z >2 indicating directional inhibition or activation, respectively. Further filtering criteria were overlap p-value of < 0.05 (Fisher’s exact test) in addition to consideration of biological relevance.

3. Results

3.1. RPPA analysis identifies differential responses to FAC across multiple ovarian cell lines

Although increased iron is well-established to modulate both cell survival and cell death, the mechanisms by which this occurs has not yet been delineated ^{[1, 2, 4]}. We previously reported that iron treatment promoted cell death in HRas overexpressing T80 (immortalized ovarian surface epithelial), HEY ovarian cancer (harboring KRas mutation), TOV21G clear cell ovarian carcinoma (harboring KRas mutation), and, to a lesser extent in TOV112D endometrioid ovarian cancer cells (lacking Ras mutation), with no change in parental T80 (lacking HRas overexpression or mutation) cells ^[6]. To identify signaling pathways altered in response to iron that could mediate these functional differences, we first completed reverse phase protein array (RPPA) to investigate proteomic changes in specific signaling pathways (155 total markers containing 40 phosphorylated markers which were assessed; see Supplementary Table 1 for antibody details) using protein lysates collected from untreated, 1, 18, and 48 hour ferric ammonium citrate (FAC)-treated T80, T80+HRas, HEY, TOV21G, and TOV112D cells. The results obtained were analyzed utilizing Cluster/Treeview software to generate heatmaps. Supplementary Figure 1A depicts the heatmap generated by median-centering all 155 markers amongst all 5 cell lines. It is not surprising that we observed diverse baseline expression of these markers based on the different origins of these cells (HEY cells were previously described as serous ovarian cancer, but this classification has recently come into question ^[20]). Using a focused analyses comprised of 45 markers (derived from the total 155 markers and including 21 phosphorylated markers), we identified iron-induced changes in pAKT (S473), pMAPK, and mTOR that diverged between the Ras-mutated cell lines and those lacking this alteration (Figure 1A). In assessing changes associated with Ras status of these cells, we identified AKT, pS6K, and pGSK3 as being divergent. We cautiously express the limitations of our RPPA/IPA approach as the number of phosphorylated proteins assessed were biased towards the AKT/mTOR pathway (21 of the 40 phosphorylated markers). Thus, we may have missed changes in the phosphorylation status of other proteins that were not included in this RPPA analysis.

Figure 1. Reverse phase protein analysis of T80 and HEY cells treated with 250μM FAC for 1, 18, 48 hours.

(A) Reverse phase protein array results displayed as heatmaps with T80, T80+HRas, HEY, TOV21G, and TOV112D. Heatmap depicts the analysis focusing on the AKT and mTOR signaling pathways in all five cell lines. Three independent experiments are shown. (B) Predicted activity of the mTOR signaling pathway was found significantly activated in HEY cells for 1h/0h comparison and significantly inhibited in T80 cells for 48h/0h comparison. For detailed information regarding the IPA Figure Legend, see Supplementary Figure 2. Red color indicates observed upregulation while green color indicates observed downregulation. Orange indicates predicted activation while blue indicates predicted inhibition. A glowing shape indicates that the observed pathway activation was opposite of that predicted. An orange line indicates a relationship predicted to lead to activation of the indicated pathway. A blue line indicates a relationship predicted to lead to inhibition of the indicated pathway while a yellow line indicates that an observed activation is inconsistent with the state of the downstream molecule.

We also utilized Ingenuity Pathway Analysis (IPA) software using colors, shapes, and directionality to indicate strength in association between differential phosphorylation and upstream regulator and pathway predictions. Interestingly, based on the significant list of differentiated phosphorylated proteins from each condition, we identified predicted activation of MAPK1 and AKT signaling pathways in HEY relative to T80 cells (Supplementary Figure 2). HEY cells showed predicted activation of AKT (z=1.944, p= 1.03E-07) and MAPK1 (z= 2.403, p= 1.92E-06) robustly at 1 hour followed by a general decrease in activation of these predicted upstream regulators at 18 hours for AKT (z= 0.653, p= 1.46E-08) and MAPK1 (z= 1.132, p= 5.85E-06) as well as 48 hours for AKT (z= 0.492, p= 2.76E-04) and MAPK1 (z= 0.561, p= 7.71E-06). The general decrease in activation was represented by downstream targets showing expression changes opposite to that expected for predicted activation, which is demonstrated by yellow lines indicating phosphorylation status inconsistent with the state of the downstream molecule (Supplementary Figure 2A). In contrast, T80 cells showed no predicted activation of AKT initially at 1 or 18 hours but was then predicted to be significantly inhibited by 48 hours (z= −2.164, p= 1.99E-06). Additionally, T80 cells had an initial predicted activation of MAPK1 (z= 1.961, p=2.68E-05) but no predicted activation or inhibition at 18 or 48 hours (Supplementary Figure 2B). IPA also displayed a significant difference in mTOR signaling between HEY and T80 cells with FAC treatment (Figure 1B). Specifically, we observed a significant increase in activation of mTOR at 1 hour in FAC-treated HEY cells (z= 2.236, p= 4E-08), while T80 cells had a predicted increase in inhibition by 48 hours for mTOR (z= −2, p= 2.09E-07). Altogether, these results show that iron promotes different signaling responses between ovarian epithelial (T80) and ovarian cancer (HEY) cells.

Since our analyses demonstrated that activation of AKT, GSK3, and mTOR were increased in cell lines associated with Ras mutations (i.e., HEY), we therefore validated phosphorylation of these molecules in both T80 and HEY cells treated with FAC. Indeed, although pAKT (S473) was elevated with FAC treatment in both T80 and HEY cells, this was more pronounced in HEY cells (~5-fold increase at 1 hour FAC in T80; ~16-fold increase at 1 hour FAC in HEY relative to T80 Untreated) (Figure 2A). A similar observation was noted for pGSK3 (S9/S21) (~3-fold increase at 1 hour FAC in T80; ~8-fold increase at 1 hour FAC in HEY); this was as expected since AKT activation phosphorylates GSK3 ^[21]. We also assessed phosphorylation of S6K (a downstream target of mTOR) [19]. We noted that S6K phosphorylation (at Ser235 and Ser236) was elevated and relatively unchanged in response to FAC treatment in T80 cells, while its phosphorylation pattern increased transiently in HEY cells between 1 to 6 hours following FAC treatment (~2-fold increase at 1 hour FAC, relative to T80 Untreated).

Figure 2. TFEB expression is altered in response to FAC in ovarian cells.

(A) Western analysis of T80 and HEY cells treated at the indicated time points with 250μM FAC. The dotted line separates samples which were re-run to prevent interference between antigens of similar molecular weights. (B) Western analysis of T80 cells (left panels) and HEY cells (right panels) treated at the indicated time points with 250μM FAC in the absence or presence of 10nM Torin 1. (C) T80 cells (left panels) and HEY cells (right panels) were treated at the indicated time points with 250μM FAC in the absence or presence of 1μM GDC0941. For all panels, data shown are representative of three independent experiments.

3.2. Altered expression of TFEB and TFE3 in response to FAC in ovarian cells

AKT and mTOR are well-known negative regulators of autophagy ^{[22, 23]}. MiTF family of transcription factors (including TFEB, TFE3, and MiTF) have been shown to regulate lysosome formation and alter autophagic response via transcriptional upregulation of the CLEAR network ^{[24, 25]}. Since MiTF transcription factors are regulated via mTOR kinase activity ^[25–27] and since activation of AKT pathway negatively regulates TFEB activity independently from mTOR ^{[28, 29]}, we next investigated whether TFEB protein levels differ between T80 and HEY cells following iron treatment. We observed that TFEB baseline expression was markedly lower in HEY relative to T80 cells (Figure 2A). Relative to T80 cells, TFEB expression was even further reduced between 18–24 hours of FAC treatment (relative to untreated) in HEY cells (Figure 2A). We also assessed other MiTF family members including TFE3 and MiTF ^{[24, 30]}; while MiTF was not detectable in our cells (data not shown), we clearly observed TFE3 protein expression in both cell lines which remained unchanged in T80 and reduced in HEY (~50%) following 24 hours of FAC treatment (relative to T80 untreated). Since we previously reported that FAC can modulate autophagic response ^[6], we assessed changes in LC3B (an autophagosomal marker) protein levels to allow comparison between the T80 and HEY cell lines. We observed increased lipidation (LC3B-II form) in both T80 (~5-fold) and HEY (~13-fold) cells following 18 hours FAC treatment relative to T80 Untreated (Figure 2A). In addition, we noted increased SQSTM1 following long-term FAC treatments in both T80 and HEY; specifically, basal SQSTM1 protein was markedly lower in HEY cells relative to T80. As expected, we identified that baseline transferrin receptor (CD71) protein was elevated in HEY cells (~3-fold, relative to T80) and was reduced in both cell lines following 24 hours of FAC (~76% (T80) and ~33% (HEY)) relative to T80 Untreated. In contrast, the ferritin heavy chain (FTH1, which is part of the iron storage complex) was increased in T80 cells beginning at 3 hours of FAC (~3-fold) but was delayed in appearance in the HEY cells, in which it reached detectable levels at only 18 hours of FAC treatment (~3-fold increase) relative to T80 Untreated.

3.3. Contribution of mTOR and AKT to the FAC-mediated changes in TFEB expression

Since MiTF family members are substrates of mTOR kinases ^[25–27] which hinders their movement to the nuclear compartment where they are required to transcriptionally regulate the CLEAR network, we assessed whether inhibition of mTOR kinase could lead to altered cellular responses to iron in ovarian cells. Therefore, we utilized the mTOR inhibitor Torin 1 in the absence or presence of FAC in T80 and HEY cells. Indeed, as shown in Figure 2B, the inhibitor ablated the phosphorylation of an mTOR substrate, S6K ^[31], in both cell lines. We observed that TFEB protein was reduced at an earlier timeframe (1 hour) upon iron treatment in combination with Torin 1 in both T80 and HEY cells (relative to FAC alone). Further, we noted an increase in LC3B lipidation in T80 cells following 1 hour of FAC treatment with Torin 1.

It was recently reported that the AKT signaling pathway regulates TFEB activity independently from mTORC ^{[28, 29]}. Since our analyses showed activation of the AKT pathway following iron treatment in T80 and HEY cells (see Figure 1), we next investigated whether AKT contributes to modulating downstream effectors of TFEB. To inhibit AKT, we utilized the PI3K inhibitor GDC0941. As shown in Figure 2C, pAKT was not detected up to 24 hours in either T80 or HEY cells with the FAC/GDC0941 combinatorial treatment while total AKT remained unchanged. While phosphorylation of S6K was weak/undetectable in HEY cells following the combinatorial treatment, pS6K was reduced in T80 cells relative to FAC treatment alone. We found no marked difference in TFEB expression with the combined FAC with GDC0941 treatment in T80 cells relative to FAC alone, whereas TFEB expression reduced at an earlier timepoint with combined treatment (6 hours) relative to FAC alone (18 hours). Neither FTH1 protein nor LC3B lipidation were altered with combined FAC and GDC0941 treatment (Figure 2C). Together, these data suggest that mTOR may regulate TFEB expression in both T80 and HEY cells, while AKT may regulate TFEB expression in HEY cells.

3.4. Altered TFEB localization patterns in response to FAC in ovarian cells

Since lysosome biogenesis and autophagy can be altered following translocation of MiTF family members to the nuclear compartment (where they transcriptionally regulate the CLEAR network ^{[32, 33]}), we next determined whether FAC treatment could modulate TFEB localization by assessing its expression in the cytoplasmic and nuclear compartments. Indeed, we identified reduced cytoplasmic levels of TFEB following FAC treatment which coincided with increased nuclear TFEB expression after 24 hours treatment in both T80 and HEY cells (Supplementary Figure 3A). Furthermore, immunofluorescence staining of TFEB showed increased cytoplasmic puncta formation at 24 hours iron treatment (relative to untreated) in both T80 and HEY cells (Supplementary Figure 3B and 3C) which was more pronounced in HEY cells. Collectively, these data show differing protein expression of the MiTF family transcription factors (TFEB and TFE3) and SQSTM1 between immortalized and malignant ovarian cells that may contribute to our previously reported biochemical and functional responses to iron in these cells ^{[6, 7]}.

Since TFEB regulates lysosome biogenesis and lysosomes are required for autophagosome turnover ^{[34, 35]}, we next assessed whether the FAC-induced increase in TFEB puncta observed in HEY cells via immunofluorescence (see Supplementary Figure 3C) was associated with lysosomes and/or autophagosomes. We completed colocalization studies (as determined by a yellow fluorescence signal) via immunofluorescence for TFEB with lysosomal-associated membrane protein 1 (LAMP1), LysoTracker Red (which detects acidic organelles), and/or with autophagosomes (assessed via LC3B protein). We identified increased colocalization between TFEB and LAMP1 (~2-fold, p=0.0175) (Figure 3A, left panel) as well as with LysoTracker Red (~2-fold, p=0.1959) (Figure 3A, middle panel); however, we did not observe any significant colocalization between TFEB and LC3B upon 24 hours FAC relative to untreated HEY cells (Figure 3A, right panel). These results suggest that TFEB may participate in lysosomal functions.

Figure 3. TFEB localization is mediated in an mTOR-independent yet AKT-dependent manner.

(A) HEY cells were treated with 250μM FAC for 0, 6, and 24 hours. Colocalization (yellow arrowhead) of TFEB with LAMP1 (left panels), with Lysotracker Red (middle panels), with LC3B (right panels) along with DAPI are shown. (B) Western analysis of cytoplasmic and nuclear fractions from T80 and HEY cells treated with 250μM FAC in the absence or presence of 10nM Torin 1. (C) Western analysis of cytoplasmic and nuclear fractions from T80 and HEY cells treated with 250μM FAC in the absence or presence of 1μM GDC0941. For all panels, data shown are representative of three independent experiments.

Since our analyses (see Figure 1 and Figure 2A) showed altered activation of the mTOR and AKT pathways, which are known to regulate TFEB ^{[24, 26–29]}, we next assessed the localization of TFEB following combined treatment (Torin 1 or GDC0941 with FAC) via subcellular fractionation. While we did not see an increase in TFEB translocation to the nuclear compartment with the Torin 1 and FAC co-treatment in either T80 or HEY cells (Figure 3B), we did observe an increased proportion of TFEB that was localized to the nuclear compartment relative to that in the cytoplasmic compartment with the GDC0941 and FAC co-treatment in both T80 and HEY cells (Figure 3C).

3.5. Induction of HO-1 and the CLEAR network via iron and the contribution of mTOR and AKT pathways in ovarian cells

It has been previously reported that MiTF family members can upregulate the CLEAR network ^[35] which contributes to increased lysosome and autophagosome numbers. Therefore, to assess whether this CLEAR network is upregulated in response to iron, we completed real-time PCR using RNA isolated from FAC-treated T80 and HEY cells for selected markers of this network (for a review of these markers, see ^[36]). As shown in Figure 4A, we observed increased expression of SQSTM1 (~3-fold, p<0.0001), LAMP1 (~2-fold, p=0.0005), MCOLN1 (~2-fold, p=0.0091), CTSD1 (~2-fold, p=0.0003), ATP6AP1 (~2-fold, p=0.0002), and UVRAG (~2-fold, p=0.0051) with 24 hours FAC in T80 cells. In contrast, at 3 hours FAC treatment in HEY cells, we initially observed reduced expression of LAMP1 (~45%, p<0.0001), MCOLN1 (~31%, p=0.0035), CTSD1 (~43%, p<0.0001), ATP6AP1 (~39%, p<0.0001), UVRAG (~29%, p=0.0084), and BECN1 (~38%, p<0.0001) while SQSTM1 was dramatically increased at 24 hours treatment (~8-fold, p=0.0487). We also assessed HO-1 mRNA levels and show that it was markedly increased in HEY cells (~89-fold, p=0.0159) relative to T80 cells (~4-fold, p=0.0010) following 24 hours FAC treatment. Together, these results suggest that TFEB alters transcription of the CLEAR network in ovarian cells and could therefore protect cells from the damaging effects of iron overload conditions.

Figure 4. FAC upregulates the CLEAR network in T80 and HEY cells in an AKT-dependent manner.

RNA was isolated and utilized for real-time PCR to assess transcript expression for heme oxygenase 1 (HO-1) and CLEAR network genes (SQSTM1, LAMP1, MCOLN1, CTSD1, ATP6AP1, UVRAG, and BECN1). (A) T80 (left panels) and HEY (right panels) cells were treated at the indicated time points with 250μM FAC. (B) T80 (left panels) and HEY (right panels) cells were treated in the absence or presence of 250μM FAC for 24 hours with or without 1μM GDC0941. For all panels, data shown are representative of three independent experiments.

We next investigated whether the FAC-mediated upregulation of HO-1 and the CLEAR network was regulated by mTOR by completing real-time PCR using RNA isolated from FAC and Torin 1-treated T80 and HEY cells. We observed little to no change in the majority of the transcripts assessed within the CLEAR network in both cell lines, though we noted in T80 cells that HO-1 was significantly reduced with FAC and Torin 1 co-treatment at 24 hours (~41%, p=0.0138) relative to FAC only (Supplementary Figure 4). This suggests that the majority of the CLEAR network genes assessed appear to be regulated in an mTOR-independent manner in these ovarian cells, yet upregulation of HO-1 mRNA in T80 cells in response to iron occurs in an mTOR-dependent manner.

Since TFEB localization was altered with co-treatment of GDC0941 and FAC relative to FAC alone (see Figure 3C), we next assessed whether the combinatorial treatment of FAC with AKT inhibitor affected the expression of HO-1 or the CLEAR network. Interestingly, co-treatment appeared to further promote the expression of CLEAR network genes relative to FAC only (Figure 4B). Indeed, in T80 cells, we noted significant increases for mRNA transcripts of SQSTM1 (p=0.0028), HO-1 (p=0.0073), LAMP1 (p=0.0025), MCOLN1 (p<0.0001), CTSD1 (p=0.0003), ATP6AP1 (p=0.0018), UVRAG (p=0.0175), and BECN1 (p=0.0094) relative to FAC treatment alone. In HEY cells, while we observed elevated expression of SQSTM1 (p=0.0051), LAMP1 (p=0.0012), MCOLN1 (p=0.0057), CTSD1 (p=0.0008), and ATP6AP1 (p=0.0025) mRNA, we noted reduced levels of HO-1 mRNA following combinatorial treatment (FAC with GDC0941) relative to FAC alone (p=0.0031). Altogether, these data suggest that AKT could mediate the activity of TFEB in response to iron in these ovarian cells and furthermore, AKT could potentially regulate the expression of HO-1 in HEY cells.

3.6. TFEB overexpression upregulates HO-1 and the CLEAR network in malignant ovarian cancer cells

We next set out to determine whether MiTF transcription factors contribute to modulating the CLEAR network and HO-1 expression in ovarian cells. We first completed siRNA-mediated knockdown of MiTF transcription factors (TFEB and TFE3). TFEB knockdown in T80 cells (~81% reduction) was accompanied by reduced FTH1 (~21%) protein at 24 hours FAC relative to control (non-targeting) siRNA (Supplemental Figure 5A). In HEY cells, neither FTH1 nor LC3B-II proteins were altered upon TFEB knockdown (~75% reduction). We then used real-time PCR to assess whether SQSTM1 and HO-1 mRNA (the most dramatically increased markers upon iron treatment, see Figure 4A) were altered upon TFEB knockdown with FAC but did not observe any changes in expression (relative to control siRNA at 24 hours FAC) in either cell line (Supplemental Figure 5B); this suggests that other MiTF family members may contribute to their regulation. TFE3 knockdown in T80 cells (~80% reduction) correlated with ~29% reduction of LC3B-II protein with 24 hours of FAC treatment (Supplemental Figure 3C). In HEY cells, TFE3 knockdown (~94% reduction) resulted in reduced protein for both FTH1 and LC3B-II (~25% and ~22%, respectively) at 24 hours FAC treatment (Supplemental Figure 5C). Although TFE3 knockdown did not significantly alter SQSTM1 or HO-1 mRNA with 24 hours FAC (relative to control) in T80 cells, we observed significant reductions in SQSTM1 without (~65%, p<0.0001) and with (~47%, p=0.0003) FAC treatment (Supplemental Figure 5D) in HEY cells. These results suggest that TFE3 contributes to SQSTM1 regulation in HEY cells and that there may be redundancy across the MiTF transcription factors in their ability to regulate the CLEAR network.

We next overexpressed TFEB in HEY cells (which express lower levels of this molecule, see Figure 2A). As shown in Figure 5A, TFEB overexpression increased FTH1 protein as well as SQSTM1 relative to control cell lines and also increased TFEB translocation to the nuclear compartment following FAC treatment in the overexpressing cell lines (relative to controls) (Figure 5B). We also observed significant increases in an array of the genes assessed in the TFEB overexpressing HEY cell lines (Figure 5C): SQSTM1 (~7-fold without FAC (p<0.0001) and ~2-fold with FAC (p=0.0223)), LAMP1 (~1.5-fold without FAC (p=0.0168) and ~2-fold with FAC (p=0.0127)), MCOLN1 (~2-fold without FAC (p=0.0043) and ~4-fold with FAC (p=0.0002)), CTSD1 (~2-fold without FAC (p=0.0036) and ~3-fold with FAC (p=0.00831)), and UVRAG (~1.6-fold without FAC (p=0.0059) and ~2-fold with FAC (p=0.0003)) mRNA, relative to controls. We also identified a significant increase in HO-1 mRNA levels in the TFEB overexpressing cells in the absence of iron (~8-fold (p<0.0001)). Altogether, these data show that TFEB overexpression in the HEY cell line can further elevate the transcriptional levels of genes involved in the CLEAR network as well as increase HO-1 expression.

Figure 5. Overexpression of TFEB in HEY cells upregulates HO-1 and the CLEAR network.

(A) Western analysis of Control and TFEB overexpressing HEY cells were untreated or treated with 250μM FAC for 24 hours. All three independent cell lines are shown. (B) Western analysis of cytoplasmic and nuclear fractions from Control and TFEB overexpressing HEY cells treated with 250μM FAC for 24 hours. Data shown are representative of three independent experiments. (C) Real-time PCR analysis of Control and TFEB overexpressing HEY cells treated with 250μM FAC for 24 hours was completed as detailed in Figure 4. Data shown are representative of three independent experiments.

4. Discussion

Although the role of iron in cancer pathogenesis is still not fully understood, it has been suggested that iron (as found in iron overload conditions) induces reactive oxygen species (via its role in Fenton reactions) and thus promotes mutagenesis ^[3]. On the other hand, increased iron can also promote cell death via nonapoptotic means including necroptosis (activated by tumor necrosis factor) and ferroptosis (characterized by increased lipid peroxidation and reduced glutathione levels). However, our knowledge of how iron elicits this effect remains unknown ^[4]. The RPPA/IPA work presented herein sets the foundation for the next step of conducting a more global, untargeted mass spectrometric approach, which would be beneficial towards more comprehensively identifying signaling mechanisms that are altered in response to iron which mediate cellular death responses. Indeed, we previously reported that iron overload conditions promote cell death in an ovarian carcinoma cell line (HEY) with minimal changes in survival function in immortalized ovarian epithelial cells (T80) ^[6]. This differential functional response appeared to be dependent on the status of the Ras/MAPK signaling cascade ^[6], but this study presented herein applies a proteomic technology (though focused to a subset of phosphorylated proteins, including AKT/mTOR) for elucidating the mechanism by which iron promotes cell death in HEY cells. Limitations of RPPA include assessment of a limited set of targets and the availability of validated antibodies for assessing phosphorylation status. Nonetheless, RPPA methodology has been used to predict drug sensitivity in a wide array of cancers ^[11]. However, to our knowledge, this approach has not been used to assess status of signaling pathways in response to iron and thus, we used this technology to identify differences amongst multiple ovarian cell lines in response to treatment with FAC. We demonstrate that activation of specific signaling networks occur following iron treatment, specifically the mTOR/AKT pathway (identified as two of the more significantly altered pathways in T80 and HEY cells in response to iron treatment) which is differentially regulated between the HEY and T80 cell lines. Importantly, the results from this analysis were validated in functional studies, which enabled us to identify that TFEB (a regulator of autophagy and lysosome biogenesis ^[27]) protein was elevated in T80 relative to HEY cells and translocated to the nuclear compartment in response to iron to upregulate the CLEAR network as well as HO-1.

Collectively, we acknowledge that RPPA analysis is a targeted approach for assessing signaling changes, as the investigated markers are dependent on the availability of validated antibodies ^[10] and hence its discovering capabilities are low. Future investigations would assess global proteomic changes between iron-treated and control cells using mass spectrometry. Overall, our results, via the use of RPPA and IPA technology, identify specific signaling pathways dysregulated upon exposure to iron, and provides justification for further work to identify other pathways through which this transition metal may promote the pathogenesis of cancers.

Supplementary Material

Supplementary FIgures

Supplementary Figure 1. Reverse phase protein analysis of T80 and HEY cells treated with 250μM FAC for 1, 18, 48 hours.

Reverse phase protein array results with T80, T80+HRas, HEY, TOV21G, and TOV112D displayed as a heatmap which depicts the analysis for all 155 analyzed antibodies, clustered by experiment and genes.

Supplementary Figure 2. Ingenuity Pathway Analysis of T80 and HEY cells treated with 250μM FAC for 1, 18, 48 hours.

Network analysis of significantly phosphorylated proteins and phosphorylation sites in all FAC-treated time points for HEY (A) and T80 (B) cells. Differences are shown between both AKT and MAPK1 as predicted upstream regulators with the activation z-scores and overlap p-values as described in the manuscript text. Legends for Ingenuity Pathway Analysis (IPA) of T80 and HEY cells treated with 250μM FAC depict orange and blue dotted lines to indicate strong direct association between quantified proteins and the activation or inhibition of a predicted upstream regulator, respectively. The outer nodes represent differentially phosphorylated proteins where red represents upregulation and green represents downregulation. Depth of color in the nodes denotes the magnitude of differential phosphorylation. Network shapes include representation of the type of proteins expressed and predicted including complexes, kinases, and transcription regulators.

Supplementary Figure 3. TFEB localization in T80 and HEY cells following cell treatment with 250μM FAC for 1, 6, 24 hours

(A) Western analysis of cytoplasmic and nuclear fractions from T80 and HEY cells treated with 250μM FAC for 0, 6, and 24 hours. (B) T80 cells were treated with 250μM FAC for 0, 6, and 24 hours followed by staining with antibodies for TFEB. Images were captured at a magnification of 63X. (C) HEY cells were treated with 250μM FAC for 0, 6, and 24 hours followed by staining with antibodies for TFEB. Images were captured at a magnification of 63X.

Supplementary Figure 4. FAC upregulates the CLEAR network in T80 and HEY cells in an mTOR-independent manner.

RNA was isolated and utilized for real-time PCR to assess transcript expression for heme oxygenase 1 (HO-1) and CLEAR network genes (SQSTM1, LAMP1, MCOLN1, CTSD1, ATP6AP1, UVRAG, and BECN1). T80 (left panels) and HEY (right panels) cells were treated in the absence or presence of 250μM FAC for 24 hours with or without 10nM Torin 1.

Supplementary Figure 5. TFE3 contributes to SQSTM1, a CLEAR network gene, in HEY cells.

(A) Western analysis of T80 (left panels) and HEY (right panels) cells were treated at the indicated time points with 250μM FAC following knockdown of TFEB. (B) Real-time PCR analysis of HO-1 and SQSTM1 in T80 and HEY cells treated at the indicated time points with 250μM FAC following knockdown of TFEB. (C) Western analysis of T80 (left panels) and HEY (right panels) cells treated at the indicated time points with 250μM FAC following knockdown of TFE3. (D) Real-time PCR analysis of HO-1 and SQSTM1 in T80 and HEY cells treated at the indicated time points with 250μM FAC following knockdown of TFE3. For all panels, data shown are representative of three independent experiments.

Significance of the study.

Since patients with iron overload diseases are at increased risk of developing cancer and iron can also mediate cell death, there is a dire need to understand the mechanisms by which iron may regulate these events. Herein, we utilize RPPA and IPA technologies to identify signaling pathways that are altered in response to exogenous iron treatment in multiple ovarian cell lines. To our knowledge, this is the first study to uncover signaling pathways that are altered in iron overload conditions using a focused proteomics approach. We identify differential activation of AKT, MAPK, and mTOR pathways as well as their potential functional response in ovarian cells. Our study demonstrates that a focused proteomic approach can be used to identify mechanisms through which iron may regulate cell survival.

Abbreviations

ATP6AP1

ATPase H+ transporting accessory protein 1

BECN1

beclin 1

CD71

transferrin receptor

CLEAR

coordinated lysosomal expression and regulation

CTSD1

cathepsin D1

FAC

ferric ammonium citrate

FTH1

ferritin heavy chain 1

GSK3

Glycogen synthase kinase-3

HO-1

heme oxygenase 1

hTERT

human telomerase reverse transcriptase

LAMP1

lysosomal associated membrane protein 1

LC3B

microtubule associated protein 1 light chain 3 beta

LTAg

large t antigen

MiTF

microphthalmia-associated transcription factor

MCOLN1

mucolipin 1

mTOR

mammalian target of rapamycin

RPPA

reverse phase protein array

S6K

ribosomal protein S6 kinase

SQSTM1

sequestosome 1

TFEB

transcription factor EB

TFE3

transcription factor binding to IGHM enhancer 3

UVRAG

UV radiation resistance associated