Tumour cells are reported to display an imbalance in the levels of ROS (reactive oxygen species). Frequently, elevated ROS production goes along with compensatory up-regulation of antioxidant enzymes. Accordingly, we found in a previous study that protein levels of several peroxiredoxins, including PRDX6 (peroxiredoxin 6), are highly elevated in experimentally induced melanomas. In the present study, we investigated the functional role of PRDX6 in human melanoma cells. PRDX6 is a bifunctional enzyme, which harbours iPLA2 (Ca2+-independent phospholipase A2) activity in addition to its peroxidase function. Our results show that PRDX6 is strongly expressed in most melanoma cells and its expression levels are maintained in a post-transcriptional manner, particularly by EGFR (epidermal growth factor receptor)-dependent signalling. PRDX6 enhances cell viability mainly by enhancing proliferation, which goes along with activation of Src family kinases. Interestingly, we were able to show that the phospholipase activity of the enzyme mediates the pro-proliferative effect of PRDX6. We identified AA (arachidonic acid) as a crucial effector of PRDX6-dependent proliferation and inducer of Src family kinase activation. These results support further the biological importance of the emerging field of lipid signalling in melanoma and highlight the particular functional relevance of PRDX6-dependent phospholipase activity.
Compared with normal healthy tissue, tumour tissue displays many features which are common to different cancer types. Among these features is elevated ROS (reactive oxygen species) metabolism, which is the result of various cancer-associated alterations, e.g. enhanced mitochondrial activity, fatty acid metabolism, oncogene activity or carcinogen exposure [1–3]. Accordingly, tumour cells are far more sensitive to pro-oxidant agents than non-transformed cells . To maintain acceptable ROS levels, which allow tumour maintenance, tumour cells employ several antioxidative mechanisms to counteract the raised ROS production . In comparison with other solid tumours, melanomas are particularly prone to elevated ROS levels. This is presumably the result of their frequent exposure to UV radiation as well as their melanocytic origin, as both the brown eumelanin and the red pheomelanin have pro-oxidative features [5–8]. Although moderate ROS levels are important for tumorigenic signalling in melanoma, including angiogenesis [9,10], malignant melanocytic tumours often express increased levels of antioxidant enzymes [11–13].
The Xiphophorus melanoma model is the oldest described animal melanoma model, where the activation of Xmrk, an oncogenic version of the fish EGFR (epidermal growth factor receptor), can lead either to the development of benign pigmented lesions or to malignant melanomas . In a previous comparative proteome analysis of healthy skin, benign pigmented lesions and melanomas of this fish melanoma model, we found that the expression of several antioxidant enzymes is increased in melanomas . Most prominently, PRDX6 (peroxiredoxin 6) expression correlated with melanoma malignancy .
PRDX6 is a member of the family of non-selenium thiol peroxidases (peroxiredoxins 1–6), which are up-regulated in breast cancer , lung cancer [17,18] or squamous cell carcinoma [19,20]. Whereas PRDX1–PRDX5 are thioredoxin-dependent 2-Cys peroxidases, the 1-Cys peroxidase PRDX6, which utilizes glutathione (GSH) as a cofactor for its peroxidase activity, is an exceptional family member. Interestingly, PRDX6 is a bifunctional enzyme that contains, in addition to its peroxidase activity, iPLA2 (Ca2+-independent phospholipase A2) activity. The latter plays a major role in physiological lung function by contributing to the production of surfactant, but is also involved in lung cancer metastasis [21,22]. In human melanoma, the functional roles of peroxiredoxins have only been scarcely investigated to date.
Although the perspective for melanoma patients is now much better than only a few years ago, successful melanoma treatment is still hampered by the development of resistance mechanisms to targeted therapy or by the incomplete and unpredictable response to immunotherapy. Thus the evaluation of new targetable signalling pathways is of major importance for future drug development. As the Xiphophorus melanoma model displays in many respects molecular characteristics that are similar to human melanoma , we describe in the present paper the investigation of the functional role of PRDX6 in human melanoma cells.
We found that PRDX6 is elevated in human melanoma cells in an EGFR-dependent manner. Our results show that PRDX6 is an important driver of melanoma cell proliferation in an iPLA2-dependent manner. Mechanistically, PRDX6 exerts its effect via enhanced production of AA (arachidonic acid) and increased activation of SFKs (Src family kinases).
Human melanoma cell lines A375, MEL-HO, SK-MEL-28, UACC-62, SK-MEL-2, UACC-257, M14 and MDA-MB-435 and the human lung adenocarcinoma cell line A549 were cultured in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) FBS (PAA), 100 units/ml penicillin (PAA) and 100 μg/ml streptomycin (PAA). NHEMs (normal human epidermal melanocytes) were purchased from Promocell and were cultivated in Ham's F10 supplemented with 20% (v/v) FBS, 100 nM phorbol-12-myristat-13-acetat (Calbiochem), 200 pM cholera toxin (Calbiochem), 100 units/ml penicillin (PAA) and 100 μg/ml streptomycin (PAA), 100 μM IBMX (3-isobutyl-1-methylxanthine) and ITS™ Premix (1:1000 dilution; BD Bioscience).
Murine melanocytes (‘melan-a’) were originally isolated in the laboratory of Dorothy Bennett (St George's, University of London, London, U.K. ) and were obtained from the Wellcome Trust Functional Genomic Cell Bank (St George's, University of London). The generation and maintenance of transgenic murine melan-a melanocytes expressing either the HERmrk construct or human EGFR was described previously [25,26]. For stimulation assays, subconfluent transgenic melan-a cells were cultured in starving medium [DMEM supplemented with 2.5% (v/v) dialysed FBS (PAA), 100 units/ml penicillin (PAA) and 100 μg/ml streptomycin (PAA)] for 3 days before being treated with 100 ng/ml human EGF (epidermal growth factor) (PeproTech) for the indicated timespans. Inhibitors were generally administered 30 min before EGF stimulation.
The following inhibitors were applied (concentrations are indicated in the Figure legends): LY294002 (LC Laboratories), AG1478 (Calbiochem), erlotinib (Selleckchem), ERK (extracellular-signal-regulated kinase) inhibitor (Calbiochem), Gö6983 (Calbiochem), Src inhibitor (Calbiochem), BEL (bromoenol lactone) (Cayman Chemicals), COX2 (cyclo-oxygenase 2) inhibitor FK3311 (Calbiochem). Glucose oxidase, tiron, vitamin E, NAC (N-acetylcysteine) and H2O2 were obtained from Sigma–Aldrich.
RNA isolation and RT (reverse transcription)–PCR
RNA isolation was performed with pegGOLD TriFast Reagent (Peqlab) according to the supplier's instructions. Reverse transcription was carried out on 2 μg of total RNA with RevertAid First Strand cDNA Synthesis Kit (Fermentas) and random hexamer primers according to the manufacturer's protocol. Fluorescence-based quantitative real-time PCR was performed with the Mastercycler ep realplex (Eppendorf) using SYBR Green reagent. Gene expression was normalized to Actb (β-actin) or RPS14 (ribosomal protein S14) for murine and human samples respectively. The following oligonucleotides were used: murine Prdx6 forward primer, 5′-TCATGGGGCATTCTCTTTTC-3′; murine Prdx6 reverse primer, 5′-GTCCCTGCCCTTATCATCAA-3′; murine Actb forward primer, 5′-GCTACAGCTTCACCACCACA-3′; murine Actb reverse primer, 5′-AAGGAAGGCTGGAAAAGAGC-3′; human PRDX6 forward primer, 5′-CGTGTGGTGTTTGT-TTTTGG-3′; human PRDX6 reverse primer, 5′-CCATCA-CACTATCCCCATCC-3′; human RPS14 forward primer, 5′-CTCAGGTGGCTGAAGGAGAG-3′; human RPS14 reverse primer, 5′-GCAGCCAACATAGCAGCATA-3′. Relative expression levels were calculated using the ΔΔCT method.
Establishment of expression vectors and transgenic cell lines
To obtain inducible PRDX6-knockdown constructs, several shRNA constructs were cloned into the pLKO-Tet-On vector as described by the supplier (Addgene). The most efficient knockdown was obtained using the oligonucleotides 5′-AATT-AAAAACGCATCCGTTTCCACGACTTTCTCGAGAAAGTC-GTGGAAACGGATGCG-3′ and 5′-CCGGCGCATCCGTTTC-CACGACTTTCTCGAGAAAGTCGTGGAAACGGATGCGT-TTTT-3′ (PRDX6 target sequence underlined) and was used in the experiments. The oligonucleotides were annealed and ligated into pLKO-Tet-On pre-digested with AgeI and EcoRI.
For the establishment of PRDX6 overexpression constructs, murine Prdx6 was amplified by PCR using cDNA from melan-a cells as template (forward cloning primer, 5′-GCGCGAATTCAT-GGATTACAAGGATGACGACGATAAGCCCGGAGGGTTG-CTTCTCG-3′; reverse cloning primer, 5′-GCGCGTCGAC-TTAAGGCTGGGGTGTATAAC-3′). The PCR product was digested with EcoRI and SalI and subcloned into the pBabe-hygro vector. This vector was used as template for site-directed mutagenesis (S32A forward primer, 5′-TCCTGGGAGATGCA-TGGGGCATTC-3′; S32A reverse primer, 5′-GAATGCCCCAT-GCATCTCCCAGGA-3′; C47S forward primer, 5′-CTTTACC-CCAGTGTCCACCACAGAAC-3′; C47S reverse primer, 5′-GT-TCTGTGGTGGACACTGGGGTAAAG-3′). Murine FLAG-tagged wild-type PRDX6 and the C47S and S32A variants were then cloned into the lentiviral p201iEP vector. All constructs were verified by sequencing. Stable cell lines were generated via lentiviral transduction. After 48 h, infected cells were treated with 2 μg/ml puromycin (Calbiochem) for 2 weeks to generate stable transgenic cell lines.
Commercially available control siRNA (ON-TARGET plus Non-Targeting siRNA) and siRNA against human PRDX6 (ON-TARGET plus Human PRDX6-01 and PRDX6-02) were purchased from Thermo Scientific. Cells were transfected in six-well dishes using X-treme Gene siRNA transfection reagent (Roche), as recommended by the manufacturer. At 24 h after transfection, cells were transferred to new dishes for ensuing experiments.
Western blot analysis
Cells were harvested and lysed in lysis buffer (20 mM Hepes, pH 7.8, 500 mM NaCl, 5 mM MgCl2, 5 mM KCl, 0.1% sodium deoxycholate, 0.5% Nonidet P40, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 200 μM Na3VO4, 1 mM PMSF and 100 mM NaF). Then, 40–60 μg of protein was resolved by SDS/PAGE and was analysed by immunoblotting. Antibodies detecting p-Akt (Ser473), p-MAPK (mitogen-activated protein kinase) p42/44 (Thr202/Tyr204), p-SFK (Tyr416), p-Rb (retinoblastoma protein) (Ser780), EGFR and p-PKC (protein kinase C) (βII Ser660) were purchased from Cell Signaling Technology. The anti-phosphotyrosine antibody (PY-20) was acquired from Life Technologies. Anti-PRDX6 and anti-β-actin antibodies were purchased from Abcam and Santa Cruz Biotechnology respectively. Densitometric analyses of Western blots from three independent experiments (if not stated otherwise) were conducted using ImageJ software (NIH). Statistical analyses were carried out using an unpaired Student's t test.
The xCELLigence system (Roche) was used to indirectly investigate proliferation rates by monitoring impedance over time. siRNA-transfected cells were seeded into xCELLigence 96-well plates (Roche) with 2×103 cells per well and cellular impedance was monitored for several days. In parallel, cells were seeded in six-well plates and harvested at day 5 after siRNA transfection for monitoring successful PRDX6 knockdown by protein blotting. The relative impedance values measured at 80 h after the start of the experiment were used for calculation of the percentage of growth. Statistical analyses were carried out using an unpaired Student's t test.
Melanoma cells were seeded in triplicate into six-well dishes (MEL-HO: 6×104 cells per well; UACC-62 8×104 cells per well) and were treated with DMSO (as control) or the indicated inhibitors. After 4 days, cells were harvested and were counted manually, using the Trypan Blue dye-exclusion method. Of note, medium including inhibitors was refreshed after 2 days. The experiment was carried out three times independently, each time in triplicate. Statistical analyses were carried out using an unpaired Student's t test.
PGE2 (prostaglandin E2) sandwich ELISA
To measure cellular PGE2 content, subconfluent melanoma cells were transfected with control or PRDX6-specific siRNA and were seeded into six-well plates the next day. After 24 h, cells received fresh culture medium. Secretion of PGE2 into the medium was allowed for another 2 days, before the medium was harvested and was analysed with the PGE2 high-sensitivity ELISA Kit (Enzo Life Science), as recommended by the manufacturer. Analysis was carried out using the Tecan microplate reader system.
Gas chromatography (GC)
Stable transgenic MEL-HO and UACC-62 cells containing a doxycycline-inducible PRDX6 shRNA were cultivated in 15-cm-diameter dishes and were treated with 100 ng/ml doxycycline for 6 days. Doxycycline-containing medium was replaced after 3 and 5 days. At day 6, cells were harvested and differences in the presence of AA were investigated by GC. Statistical analyses of the data were carried out using an unpaired Student's t test.
For quantification of the eicosanoic acids (AA, other unsaturated eicosanoic acids and saturated eicosanoic acid) in the harvested cells, 50 mg of cell pellet was mixed with 10 μl of standard [2.5 mM nonadecanoic acid methyl ester in methanol/chloroform (1:1, v/v)] and extracted with 3.55 ml of chloroform/methanol/water (10:60:1, by vol.). The extract was mixed with 1 ml of water and 4 ml of chloroform. After centrifugation, the lower phase was evaporated at 60°C under a stream of nitrogen gas. The residue was dissolved in 2 ml 1 M acetyl chloride in methanol and heated at 74°C for 3 h. The resulting fatty acid methyl esters were extracted twice with 2 ml of hexane each. The combined extracts were evaporated at 60°C under a stream of nitrogen gas. The residue was dissolved in 100 μl of hexane and applied to a silica gel TLC (thin layer chromatography) plate (Merck).
After evaluation of the plate in hexane/diethyl ether (7:3, v/v), the fatty acid methyl ester-containing region was scraped off. Fatty acid methyl esters were extracted from the silica gel by 1 ml diethyl ether. The extract was evaporated at 40°C under a stream of nitrogen gas. The residue was dissolved in 20 μl of dichloroethane. Then, 1 μl of the resulting solution was applied to the GC column (30 m Phenomenex Zebron ZB-1701 capillary GC column with 0.32 mm internal diameter). For separation of the fatty acid methyl esters, the following temperature programme was applied: 150–280°C with 4°C/min and 280–281°C with 0.1°C/min at a column head pressure of 100 kPa.
EGFR signalling strongly enhances PRDX6 protein levels
In our previous study on differentially expressed proteins in pigment cell lesions from the Xiphophorus fish melanoma model, we found that a group of ROS-associated proteins was strongly increased in malignant melanomas compared with benign lesions or healthy skin. The induction of PRDX6 was particularly striking, but PRDX2 was also up-regulated . To get an impression of the potential relevance of the six members of the peroxiredoxin family in human tumours, we used genomic data from The Cancer Genome Atlas (TCGA) and examined the frequency of genomic alterations in the PRDX1–PRDX6 genes in human melanoma using cBioPortal for Cancer Genomics [27,28]. The data from breast cancer and lung adenocarcinoma, two tumour types in which the pro-tumorigenic function of PRDX6 was reported previously [21,29,30], served as a reference. Altogether, the peroxiredoxins were altered in 12% of melanomas, 15% of breast cancers and 17% of lung adenocarcinomas (Figure 1). In all three cancer types, PRDX6 was the peroxiredoxin which was most frequently altered. Notably, it was mostly (melanoma, lung adenocarcinoma) or exclusively (breast cancer) amplified. These data suggest that PRDX6 is the major pro-tumorigenic peroxiredoxin family member in these cancer types.
Frequency of genomic PRDX alterations in cancer genomes
In the Xiphophorus melanoma model, the driving oncogene is the EGFR orthologue Xmrk. To find out whether PRDX6 is a direct target of the EGFR, we used melanocytes transgenic for the receptor HERmrk, a chimaeric construct bearing the extracellular domains of human EGFR and the intracellular domain of Xmrk, thus allowing directed receptor activation after addition of human EGF . HERmrk stimulation had no effect on RNA levels of Prdx6, as shown by real-time PCR, but strongly enhanced protein PRDX6 levels after 8 and 24 h of EGF treatment (Figures 2A and 2B). Similarly to HERmrk, human EGFR also induced PRDX6 protein levels, although the time of induction was delayed (Figure 2C). The same observation was made when we used the lung adenocarcinoma cell line A549 and stimulated the cells with EGF for 24 h (Figure 2D). Thus EGFRs are able to raise PRDX6 protein levels in a transcription-independent manner.
EGFR-dependent PRDX6 induction
EGFRs are often expressed at low or intermediate levels in human melanoma cell lines  and play a major role in autocrine activation and drug sensitivity of melanoma [25,33,34]. We examined the protein and mRNA levels of PRDX6 in a panel of different human melanoma cell lines under normal conditions, without adding EGF. Compared with NHEMs, two cell lines displayed similar protein levels (MEL-HO and SK-MEL-28) and six cell lines showed enhanced (SK-MEL-2, UACC-257, M14 and MDA-MB-435) or strongly enhanced (A375 and UACC-62) protein levels of PRDX6 (Figure 3B). PRDX6 mRNA levels did not correlate with protein abundance (Figures 3A and 3B).
EGFR-dependent PRDX6 induction in human melanoma cells
To confirm the regulation of PRDX6 by human EGFR, we chose two cell lines with strong PRDX6 expression (A375 and UACC-62) and two cell lines with low PRDX6 expression (MEL-HO and SK-MEL-28) and first tested the presence of EGFR by Western blotting (Figure 3C). Interestingly, the degree of EGFR expression went along with that of PRDX6. Long-term treatment with erlotinib, a selective tyrosine kinase inhibitor of EGFR, caused a clear decrease in the protein expression of PRDX6 in all cell lines except for A375, where the decrease was rather weak (Figure 3D). Taken together, these data show that EGFR affects PRDX6 expression in human melanoma cells.
The PI3K (phosphoinositide 3-kinase) pathway is involved in the regulation of PRDX6
For the identification of responsible downstream pathways involved in the regulation of PRDX6, HERmrk cells were treated with different inhibitors and were analysed by Western blotting (Supplementary Figure S1). Whereas successful pathway inhibition was monitored after 30 min of EGF stimulation, PRDX6 expression was investigated after 24 h because of its delayed expression in response to EGF. As expected, the up-regulation of PRDX6 protein expression after HERmrk activation was completely abolished by the EGFR inhibitor AG1478. Inhibition of PI3K impeded the expression of PRDX6, although to a lower extent compared with EGFR inhibition. In contrast, blocking ERK1/2 or PKC had no effect on PRDX6 levels. We also tested the PI3K-dependent PRDX6 regulation in human cell lines (Figure 4A). Here, PI3K inhibition did not endure for 24 h in all cell lines, so we applied the inhibitor between 4 and 24 h. Again, a moderate PRDX6 down-regulation was detected after PI3K inhibition (Figure 4A). EGFRs including HERmrk led to the generation of ROS [11,35,36], e.g. via activation of NADPH oxidases (, and own unpublished work). As ROS can increase the abundance of several PRDX isoforms , and also induce PRDX6 expression in A549 adenocarcinoma cells , we tested whether this is also true for the observed PRDX6 regulation in our cell system. HERmrk cells were treated with the ROS scavengers tiron and vitamin E as well as the antioxidant NAC. To reduce NADPH oxidase activity, we administered apocynin. None of these reagents could prevent EGF-dependent PRDX6 induction (Figure 4B). Along the same line, enhanced oxidative stress caused by glucose oxidase or H2O2 had no effect on PRDX6 protein levels in human melanoma cells or NHEM cells (Figures 4C and 4D).
ROS-independent and PI3K-dependent protein expression of PRDX6
In conclusion, the EGFR-mediated up-regulation of PRDX6 is independent of ROS, but at least partly attributed to active PI3K signalling.
PRDX6 stimulates proliferation and affects SFK activity
To investigate the functional role of PRDX6 in human melanoma, siRNA-mediated knockdown of PRDX6 was performed and cellular proliferation was examined through xCELLigence impedance analysis in four melanoma cell lines. The impedance is a measure of the attached cell surface on the microplate and increases with cell density. PRDX6 knockdown was visible in all cell lines, using two different siRNAs (Figure 5A). Cell impedance was decreased by PRDX6 knockdown in all cell lines apart from A375 (Figure 5B, upper images). We calculated the percentage of growth after 80 h of impedance measurement and observed the same outcome with two independent siRNAs (Figure 5B, lower images).
Influence of PRDX6 on melanoma cell proliferation
As the PRDX6 knockdown affected the slope of increase in electron impedance without preventing the general impedance upward trend, we assumed that proliferation, but not apoptosis, was affected. This hypothesis was tested in the two cell lines MEL-HO and UACC-62, the latter displaying the most profound effect after PRDX6 knockdown. In both cases, we saw no signs of apoptosis, as both the apoptosis indicator cleaved PARP [poly(ADP-ribose) polymerase] and the percentage of cells in sub-G1-phase were unaltered after PRDX6 knockdown (Supplementary Figure S2).
To obtain further insight into the signalling processes which are involved in PRDX6-dependent proliferation, we analysed the activity of selected signalling pathways associated with the regulation of cell growth and cell cycle control. We found that PRDX6 knockdown went along with a reduction in p-Rb (Ser780), which is a sign of impaired CDK4/6 (cyclin-dependent kinase 4/6) activity and reduced cell cycle progression. In addition, we detected a decreased activity of Src family kinases (SFKs), as indicated by reduced phosphorylation of Tyr416 in the activation loop of SFK proteins (Figure 5C).
The iPLA2 activity is involved in the proliferative function of PRDX6
As PRDX6 harbours the enzymatic functions of a peroxidase and a phospholipase, we were interested in the contribution of both enzymatic activities to the proliferative effect. To this end, we designed a complementation assay based on the knockdown of human endogenous PRDX6 in melanoma cells in combination with the expression of several mPRDX6 (murine PRDX6) constructs. We used murine PRDX6 expression constructs because they are not affected by the knockdown of endogenous human PRDX6. In addition to wild-type mPRDX6, we generated S32A and C47S mutants, which were reported to abolish the iPLA2 and peroxidase activity respectively without affecting the other catalytic activity . The active sites of peroxidase and iPLA2 are visualized in a structural model in Supplementary Figure S3A.
Transgenic UACC-62 cells stably expressing the FLAG-tagged mPRDX6 constructs were transfected with control and human PRDX6-specific siRNA, and cell proliferation was monitored by xCELLigence analysis (Figures 6A and 6B). Interestingly, the growth-suppressive effect of endogenous PRDX6 knockdown was largely rescued by wild-type mPRDX6 and the mPRDX6-C47S mutant, but not by the mPRDX6-S32A mutant, as we observed with two different siRNAs (Figures 6B and 6C). A similar experiment was performed with MEL-HO cells (Supplementary Figures S3B and S3C). Here, both mutants were unable to rescue PRDX6 knockdown, suggesting that the iPLA2 activity is required for proliferation in both cell lines, whereas only MEL-HO cells also depend on the peroxidase function for proliferation.
The impact of PRDX6 enzyme activities on proliferation
To independently test the importance of iPLA2 activity for melanoma proliferation, MEL-HO and UACC-62 cells were treated with BEL, a general inhibitor of iPLA2-type phospholipases. Like PRDX6 knockdown, BEL also caused a decrease in the phosphorylation of SFK proteins in both cell lines (Figure 6D). Importantly, the application of either BEL or Src inhibitor led to a significant reduction in cell proliferation compared with DMSO-treated cells in MEL-HO and UACC-62 cells (Figure 6E).
Cyclo-oxygenase activity is not involved in PRDX6-mediated growth promotion
PLA2 (phospholipase A2) enzymes are responsible for releasing AA from position 2 of phospholipids, which can be processed further by COXs to the tumour-relevant PGE2. As PGE2 was reported to be involved in proliferation, e.g. in colon cancer , we were interested in the effect of PRDX6 on PGE2 levels in MEL-HO and UACC-62 cells. MEL-HO cells had almost undetectable PGE2 levels which were not affected further by PRDX6 knockdown. In UACC-62 cells, however, PGE2 levels were generally higher and were strongly decreased by PRDX6 knockdown (Figure 7A). When we treated melanoma cell lines with a COX2 inhibitor, cell growth was unexpectedly enhanced (Figure 7B) and p-SFK levels were not affected (Figure 7C). Thus we concluded that COX2-catalysed effectors are not responsible for PRDX6-dependent proliferation in melanoma.
Investigation of the involvement of COX in PRDX6-dependent proliferation
The proliferative effect of PRDX6-iPLA2 is caused by AA
To identify lipid mediators which are involved in the PRDX6-mediated effect, we analysed the impact of PRDX6 on cellular lipid abundance. We first generated a doxycycline-inducible PRDX6-specific shRNA construct, which allowed us to generate a large number of cells with consistently repressed PRDX6 levels. After 5 days of doxycycline treatment, reduction in PRDX6 and p-SFK levels were well visible (Figure 8A). Cells were harvested and analysed for the content of total long-chain desaturated and saturated fatty acids by gas chromatography (GC). This analysis revealed a clear decrease in the abundance of AA in response to PRDX6 reduction in both cell lines (Figure 8B).
Effect of PRDX6-dependent AA levels on melanoma cells
Next, we wanted to find out whether exogenously added AA is able to rescue the growth-suppressive effect of PRDX6 reduction. Interestingly, AA rescued SFK phosphorylation and largely rescued growth suppression in both cell lines (Figures 8C and 8D). In summary, our data indicate that AA-mediated lipid signalling is a crucial mediator of PRDX6-dependent proliferation in melanoma.
In the present study, we investigated the functional role of PRDX6 in melanoma cell proliferation. We show that EGFR signalling triggers PRDX6 protein expression, resulting in enhanced AA release catalysed by the iPLA2 activity of the enzyme. This, in turn, activates SFK signalling and is followed by enhanced proliferation.
The expression of PRDX6 in melanocytes and melanoma is regulated by EGFR signalling in a manner which depends, at least partially, on the PI3K pathway. Interestingly, our results revealed that this regulation is transcription-independent. The underlying mechanism is currently not known. However, it can be speculated that PI3K activity affects the protein stability of PRDX6. The search for interaction partners of PRDX6 by STRING analyses (http://string-db.org) and BioGRID (http://thebiogrid.org) revealed UBC (ubiquitin C) as the best physical interaction hit, as identified within the scope of ten different high-throughput analyses (e.g. [41–43]). This is a common phenomenon observed in proteins with high turnover rates. The related peroxiredoxin family member PRDX3 was identified previously as substrate for the CUL4B (cullin 4B) ubiquitin ligase , which polyubiquitinates PRDX3 and thereby targets it for degradation. Interestingly, PRDX6 was identified in a search for cullin RING ubiquitin ligase substrates, but its interaction was not tested in independent validations . It is, however, possible that PRDX6 is regulated in a manner which is comparable with that of PRDX3. Furthermore, the PI3K pathway regulates both the expression of UBC and the transcription of ubiquitin protein ligase components , and might thereby contribute to PRDX6 stability.
The functional involvement of PRDX6 in cancer is contro-versial. In breast cancer, PRDX6 was previously described as a tumour marker. Here, both tumour tissue and serum of patients display elevated PRDX6 levels compared with healthy subjects , and PRDX6 expression is associated with proliferation, invasion and development of cisplatin resistance in vitro [29,47]. In papillary thyroid carcinomas, however, tumour size and lymph node metastases are inversely correlated with PRDX6 expression . Overall, the pro-tumorigenic role of PRDX6 seems to depend on the tumour type as well as the stage of tumour development. When Prdx6 was depleted from keratinocytes, HPV-8- (human papillomavirus 8) or chemically induced skin carcinogenesis was found to be enhanced in mice . Accordingly, Prdx6 overexpression in keratinocytes protected Prdx6-transgenic mice against HPV-8-induced skin cancer. This effect was accompanied by a reduction in oxidized lipids in these animals, suggesting that PRDX6 prevents the accumulation of oxidative damage and thereby blocks the early steps of tumorigenesis. However, PRDX6 overexpression in existing tumours promotes malignant conversion .
The tumour-promoting function of PRDX6 is best investigated in lung cancer. In a mouse model using allografted lung carcinoma cells, tumour growth was enhanced in mice with constitutive PRDX6 overexpression. This went along with enhanced proliferation markers such as Ki67, cyclin-dependent kinases, cyclins and AP-1 (activator protein 1) transcription factor activity . However, PRDX6 is also a strong promoter of invasion and metastasis in lung cancer [21,30]. Ho et al.  described that this feature is mediated by the iPLA2 activity of the enzyme. Interestingly, the growth-promoting function of PRDX6 in lung cancer was attributed to the peroxidase activity . This is in contrast with our results, where iPLA2 activity was most important for melanoma cell proliferation. This might be attributed to different expression of the downstream effectors of ROS- and lipid-dependent pathways in lung cancer and melanoma cells.
Our data show that AA is a crucial mediator of the observed PRDX6-mediated growth-promoting effect, as it rescues reduced SFK activity and growth inhibition caused by PRDX6 reduction. AA can be metabolized further to bioactive lipids via COX, LOX (lipoxygenase) and cytochrome P450 epoxygenase pathways. These lipids can act as inflammatory mediators or participate in cellular signal transduction.
The inducible COX isoform COX2 catalyses the conversion of AA into PGE2, thereby increasing melanoma migration and epithelial–mesenchymal transition [51–53]. Although PRDX6 was involved in PGE2 production in the melanoma cell line UACC-62, we observed that COX2 inhibition did not suppress melanoma cell proliferation, but instead increased it in both UACC-62 and MEL-HO cells. It was described previously that some pro-migratory cues can inhibit proliferation . Thus PGE2 or other products of COX2 might have pro-tumorigenic effects in melanoma even if they suppress proliferation.
The AA metabolites 12(S)-HETE (hydroxyeicosatetraenoic acid) and 15(S)-HETE are produced via 12/15-LOX activity. 12(S)-HETE binds to a G-protein-coupled receptor and stimulates various signalling molecules including SFKs [55–57]. It is likely that a similar mechanism is responsible for PRDX6-dependent SFK activation in melanoma. Src activity is detected in most melanoma cell lines, and SFKs play an important role in melanoma biology. In addition to the described effect on proliferation, Src is involved in invasion and transendothelial migration . SFKs also enhance ERK1/2 activation, which is highly relevant for melanoma growth, and are required for the expression of several transcriptional targets downstream of ERK1/2 [59,60]. Several Src inhibitors are already used in the clinic. Although they are not yet used for melanoma treatment, pre-clinical studies show promising anti-melanoma effects .
The pro-proliferative effect of AA which we detected in melanoma cells is supported by data from other groups. Using the B16 melanoma model, He et al.  described recently that inhibition of Δ6-desaturase, the rate-limiting enzyme for AA synthesis, suppresses melanoma growth as well as angiogenesis and inflammation in mice . Interestingly, AA can modulate the activity of potassium channels, as in the case of the hEAG1 (human ether-a-go-go 1) channels, where AA binding lowers the activation threshold and thereby increases proliferation of hEAG1-expressing melanoma cells .
The fact that the phospholipase activity of PRDX6 can be blocked by iPLA2 inhibitors poses an interesting therapeutic option. To date, inhibitors targeting other PLA2 family members such as sPLA2 (soluble PLA2) and lpPLA2 (lipoprotein-associated PLA2) have been developed and are being tested in clinical trials for the treatment of atherosclerosis . However, less specific PLA2 inhibitors such as the antimalarial agent quinacrine are also available and show activity in melanoma as well as in other tumour entities in vitro [65–67]. In future studies, we will investigate their effect on tumour maintenance of EGFR-expressing melanomas in vivo.
In the present study, we show that melanoma cell growth is strongly affected by lipid signalling. The bifunctional protein PRDX6 affects this pro-proliferative lipid metabolism to a large extent. Via its iPLA2 activity, PRDX6 raises AA levels, which ultimately results in enhanced SFK activity and proliferation. PRDX6 expression is linked to EGFR and PI3K activity, thereby providing a connection between oncogenic signalling and lipid metabolism in melanoma. The fact that inhibition of iPLA2 activity strongly blocks melanoma growth reveals an interesting therapeutic promise of this type of enzyme, which will be analysed further by us in the future.
Alexandra Schmitt performed and analysed the majority of the experiments and prepared the Figures. Werner Schmitz performed and analysed the chromatography experiments. Anita Hufnagel performed some western blots and X-celligence analyses. Svenja Meierjohann designed the project and wrote the paper. Alexandra Schmitt and Manfred Schartl helped with writing the paper.
We thank Jochen Kuper for help with the PRDX6 structural model.
This work was supported by the Melanoma Research Network of the Deutsche Krebshilfe e.V. (German Cancer Aid) and by the research unit FOR2314 (German Research Foundation).
Dulbecco's modified Eagle's medium
epidermal growth factor
epidermal growth factor receptor
human ether-a-go-go 1
human papillomavirus 8
normal human epidermal melanocyte
protein kinase C
reactive oxygen species
Src family kinase