Cytosolic Phospholipase A₂α Regulates Cell Growth in RET/PTC-Transformed Thyroid Cells

Modulation of cytosolic phospholipase A₂ (PLA₂) expression levels and production of its metabolites have been reported in several tumor types, indicating involvement of arachidonic acid and its derivatives in tumorigenesis. Following our demonstration that the PLA₂ group IV isoform α (PLA₂IVα) controls TSH-independent growth of normal thyroid (PCCl₃) cells, we have investigated the mitogenic role of PLA₂IVα in rat thyroid cells transformed by the RET/PTC oncogenes (PC-PTC cells). We now report that PLA₂IVα acts downstream of the RET/PTC oncogenes in a novel pathway controlling RET-dependent cell proliferation. In addition, we show that PLA₂IVα is in its phosphorylated/active form not only in RET/PTC-transformed cells and in cells derived from human papillary carcinomas but also in lysates from tumor tissues, thus relating constitutive activation of PLA₂IVα to RET/PTC-dependent tumorigenesis. Moreover, p38 stress-activated protein kinase is the downstream effector of RET/PTC that is responsible for PLA₂IVα phosphorylation and activity. In summary, our data elucidate a novel mechanism in the control of thyroid tumor cell growth that is induced by the RET/PTC oncogenes and which is distinguishable from that of other oncogenes, such as BRAF. This mechanism is mediated by PLA₂IVα and should be amenable to targeted pharmacologic intervention. [Cancer Res 2007;67(24):11769–78]

The different forms of phospholipase A₂ (PLA₂) thus far identified are all recognized regulators of cell function, whereby their phospholipid-hydrolyzing activities lead to formation of lipid metabolites that are directly involved in processes such as inflammation, apoptosis, proliferation, and carcinogenesis (1). Activation of the PLA₂s results in formation of two main products: free fatty acids (with arachidonic acid the most common in mammals; ref. 2) and lysolipids. These products can lead to similar activities in different cell systems. However, the localization of the different PLA₂ isoforms and their modes of activation, substrate specificities, and relative abundances, as well as the coordinated activation of arachidonic-acid–metabolizing enzymes, such as the cyclooxygenases (COX), lipooxygenases (LOX), and cytochrome P450, all determine the final, highly specific activities of the PLA₂s (2–4). Accordingly, different secretory and cytosolic isoforms of PLA₂ serve different functions, whereby some behave mainly as housekeeping enzymes devoted to phospholipid remodeling and others are rapid responders to receptor activation and cytosolic calcium modulation (4, 5).

We have recently shown that PLA₂ group IV isoform α (PLA₂IVα) controls TSH-independent cell proliferation in PCCl₃ rat thyroid epithelial cells (6). This pathway is regulated by adrenergic and purinergic receptors, and most importantly, it leads to preferential hydrolysis of phosphatidylinositol to form lysophosphatidylinositol, glycerophosphoinositol (GroPIns), and arachidonic acid. PLA₂IVα activation is downstream of the receptor-activated extracellular signal-regulated kinase (ERK) 1/2 mitogen-activated protein kinases (MAPK) and p38 stress-activated protein kinase (SAPK), and the phosphoinositide metabolites that promote cell proliferation are arachidonic acid and GroPIns [as shown by specific inhibitors, RNA interference (RNAi), and a dominant-negative PLA₂IVα_1-522; ref. 6]. Although present, the mitogenic role of PLA₂IVα in normal thyroid tissue is minor compared with the known modulation by physiologic concentrations of TSH, which is mediated via the adenylyl cyclase/cyclic AMP (cAMP) cascade (7). However, in a different context, such as oncogenic cell transformation that results in loss of the TSH/cAMP control in the thyroid, the role of PLA₂IVα might predominate. In support of this, a PLA₂ activity associated with Ras-induced transformation has previously been reported in thyroid cells (8); in addition, this activity is specific for the phosphoinositides and can be detected by increases in the glycerophosphoinositols (6, 8–12).

To directly investigate the role of PLA₂IVα in thyroid tumor growth, we have used thyroid cells transformed by the RET/PTC oncogenes: PC-PTC cells (13). The RET proto-oncogene encodes a tyrosine kinase receptor for growth factors of the glial-derived neurotrophic factor family (14). RET activation results in phosphorylation of 12 tyrosine residues in its cytoplasmic tail, which initiates multiple signaling cascades that involve phospholipase Cγ, c-Src, and adaptors such as the Shc proteins, IRS1/2, FRS2, and DOK1/4/5 (15). Germ-line point mutations convert RET into a dominantly transforming oncogene in multiple endocrine neoplasia 2A and 2B and in familial medullary thyroid carcinoma, whereas chromosomal aberrations cause recombination of the intracellular kinase-encoding domain of RET with heterologous genes, generating the constitutively active RET/PTC oncogene chimeras that are responsible for papillary thyroid carcinomas (PTC; refs. 16, 17).

Although to date there is no evidence that a PLA₂ is part of RET signaling, the activation of Ras and Ras/ERK (by the Shc-Grb2-Sos complex; refs. 15, 16) is suggestive of such an involvement because it is known that, together with other kinases, such as c-Jun NH₂-terminal kinase (JNK) and p38 SAPK, ERK1/2 MAPKs can also promote phosphorylation and activation of PLA₂, which could initiate a mitogenic pathway parallel to that in normal thyroid cells (6).

We have analyzed this possibility and now show that PLA₂IVα and arachidonic acid are indeed major regulators of cell proliferation in RET/PTC-transformed cells. This pathway is also active in human papillary carcinoma cells and in thyroid tumor tissues, indicating an essential role for PLA₂IVα in promoting the mitogenic activity of this oncogene.

Reagents and antibodies. RPMI 1640 and Medium 199 were from Life Technologies. The anti-phosphorylated MAPK-activated protein (MAPKAP) kinase-2 (Thr²²²) antibody was from Cell Signaling Technology, the anti-MAPKAP kinase-2 antibody was from Upstate, and the anti-actin antibody was from Sigma. Pyrrophenone and ZD6474 were kindly provided by Dr. K. Seno (Shionogi Research Laboratories, Osaka, Japan; ref. 18) and AstraZeneca Pharmaceuticals (19), respectively. SP600125 was from Calbiochem. All other cell culture reagents, antibodies, and inhibitors were obtained as in ref. 6.

Cell lines and RNAi. PCCl₃ rat thyroid cells were grown in Coon's modified Ham's F12 medium, 5% calf serum, and a six-hormone mix (6H), including TSH (7, 20). The RET/PTC3- and RET/PTC1-transformed PCCl₃ cells (PC-PTC3 and PC-PTC1 cells, respectively) and the BRAF-transformed PCCl₃ cells (PC-BRAF cells) were grown under the same conditions but in the absence of hormones (13). The BHP10-3 and TPC1 cells were grown as described previously, 10% fetal bovine serum in RPMI 1640 and DMEM, respectively (21, 22). Although BHP10-3 and TPC1 cells might have originally derived from the same cell population,⁴

they display distinct phenotypes perhaps due to in vitro evolution (morphology, growth rate, RET/PTC1 protein levels, and PLA₂IVα phosphorylation levels; see [³H]thymidine incorporation and Results).

Sequence-specific PLA₂IVα silencing in rat PC-PTC3 cells was as described previously (6) using an electroporated small interfering RNA (siRNA; ref. 23), with a siCONTROL nontargeting duplex (Dharmacon, Inc.) as control. For sequence-specific silencing in human BHP10-3 cells, the human PLA₂IVα siGENOME SMARTpool siRNAs (Dharmacon) and the nontargeting siCONTROL duplex (Dharmacon) were used with Oligofectamine reagent (Invitrogen) following the manufacturer's instructions.

Analysis of [³H]inositol-containing metabolites. PC-PTC3, PCCl₃, and PC-BRAF cells were labeled for 72 h in Medium 199, 1% calf serum, containing [³H]myo-inositol (5 μCi/mL), with (PCCl₃ cells) or without (PC-PTC3 and PC-BRAF) 6H. Cell treatment, extraction, and high-performance liquid chromatography (HPLC) separation of [³H]inositol-labeled water-soluble metabolites were as described previously (24, 25). HPLC cpm are expressed as percentages of their individual controls for calculations of means in figures.

[³H]Thymidine incorporation. PC-PTC3 and PCCl₃ cells were starved for 48 h in Coon's modified Ham's F12 medium with 0.3% bovine serum albumin and 2 mmol/L l-glutamine, with treatments and subsequent [³H]thymidine incorporation in the same medium (for PCCl₃ cells, hormones were added, except for TSH) as described previously (6). In nonstimulated cells and across the full experimental data, the total [³H]thymidine incorporation levels under these conditions were 9,787 ± 1,522 cpm/well for PC-PTC3 cells and 4,258 ± 1,088 cpm/well for PCCl₃ cells.

BHP10-3 and TPC1 cells were grown under normal growth conditions for 48 h and stimulated for a further 48 h in normal growth medium. [³H]Thymidine (0.5 μCi/well; Perkin-Elmer Life Sciences) was added 4 h before the end of this incubation. Incubations were ended as reported previously (6). As above, in nonstimulated cells, total [³H]thymidine incorporation levels under these conditions were 85,664 ± 9,927 cpm/well for BHP10-3 cells and 21,780 ± 1,563 cpm/well for TPC1 cells. Growth rates for BHP10-3 cells under RNAi were evaluated by blinded cell counting in a Neubauer chamber.

Sample preparation for LC-MS/MS. Cells were grown to near confluence in their respective medium and then detached, washed, and resuspended in 0.9% NaCl solution. Samples were used to measure relative cell volumes in comparison with PCCl₃ cells by flow cytometry. The PCCl₃ cell volume was taken as 0.97 ± 0.03 pL (12).

Parallel cell sample had 6 mL methanol/ 1 mol/L HCl (1:1; −20°C) added per 100-mm petri dish, with 5.5 mL chloroform added following cell scraping; this provided a two-phase acid extraction (24). The upper (aqueous) phase was lyophilized and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS; ref. 26). Each cell line was assessed in triplicate from at least three independent Petri dishes of each cell type.

Immunoblotting, intracellular cAMP, and [³H]arachidonic acid release. These were performed as reported previously (6, 27). Frozen thyroid tissue samples from patients affected by PTCs, assessed and processed as described (13), were obtained from Prof. F. Basolo (Department of Surgery, Division of Anatomic Pathology, University of Pisa, Pisa, Italy).

Statistical analysis. The data are presented as mean ± SE from at least three independent experiments. Statistical analysis was by the Mann-Whitney U test.

PLA₂IVα is involved in the control of cell proliferation in RET/PTC3-transformed cells. Following our recent demonstration that receptor-activated PLA₂IVα is involved in TSH-independent (cAMP-independent) proliferation of PCCl₃ thyroid cells (6), we investigated the role of PLA₂IVα in growth control of transformed thyroid cells (13, 17, 20).

[³H]Thymidine incorporation was evaluated in thyroid cells transformed by the chimeric RET/PTC3 oncogene (PC-PTC3 cells; Fig. 1A; ref. 13). The specific PLA₂IVα inhibitor pyrrophenone (0.1–0.5 μmol/L, 48 h) was used to assess involvement of PLA₂IVα in control of cell growth; pyrrophenone significantly inhibited basal PC-PTC3 cell proliferation at concentrations as low as 0.1 μmol/L. As shown in Fig. 1A, basal cell proliferation of normal PCCl₃ cells was also inhibited by pyrrophenone (0.5 μmol/L), although to a lesser extent. As previously shown (28), the RET kinase inhibitor ZD6474 (1 μmol/L, 16 h) also reduced basal [³H]thymidine incorporation in transformed PC-PTC3 cells (Fig. 1A). These data indicate that PLA₂IVα-dependent signaling is involved in growth stimulation induced by expression of the RET/PTC3 oncogene.

Next, the expression and activation levels of PLA₂IVα in PC-PTC3 cells were evaluated. Western blotting using a specific anti-PLA₂IVα antibody [anti-cytosolic PLA₂ (cPLA₂; N-216); ref. 6] revealed that PLA₂IVα expression levels in PC-PTC3 cells were lower than normal cells (∼10-fold; Fig. 1B). Surprisingly, however, in these PC-PTC3, RET/PTC3-transformed cells, PLA₂IVα was in its phosphorylated, and hence active, state (Fig. 1C). To analyze the relationships between PLA₂IVα phosphorylation and RET/PTC3 transformation, we again used the RET kinase inhibitor ZD6474 (28). At 1 μmol/L ZD6474, PLA₂IVα phosphorylation was reduced by 20% after 10 min (data not shown), by 40% to 50% after 30 min (Fig. 1C), and by >90% after 2 h of treatment (data not shown). This inhibition was still evident after 24 h (∼30%; 1 μmol/L ZD6474; Fig. 1C).

The activity of phosphorylated PLA₂IVα was then analyzed. Following the observation that, in normal PCCl₃ cells, PLA₂IVα preferentially hydrolyzes membrane phosphoinositides, thus releasing glycerophosphoinositols and arachidonic acid (see Introduction and Materials and Methods; ref. 6), formation of these phosphatidylinositol metabolites was measured in cells incubated with radiolabeled inositol and arachidonic acid, respectively (see Materials and Methods). First, basal levels of GroPIns and arachidonic acid release were analyzed: these were higher in PC-PTC3 cells compared with PCCl₃ cells (Fig. 2A). Then, stimulation of purinergic receptors (known to be present in PCCl₃ cells; ref. 6) was followed by addition of the agonist ATP (100 μmol/L): this significantly increased intracellular levels of GroPIns (Fig. 2B) and release of arachidonic acid (Fig. 2C) in both PCCl₃ and PC-PTC3 cells. However, whereas the increase in GroPIns was completely blocked by the specific PLA₂IVα inhibitor pyrrophenone (0.5 μmol/L; Fig. 2B; ref. 18), arachidonic acid release was inhibited only in part (Fig. 2C). Thus, this pyrrophenone-sensitive PLA₂IVα is active in both normal and RET/PTC3-transformed cells, and arachidonic acid, but not GroPIns, can also be released by other PLA₂ isoforms.

In summary, in PC-PTC3 cells, PLA₂IVα is phosphorylated and thus active under basal conditions; it can be further activated on agonist stimulation, probably as a consequence of increases in Ca²⁺, an essential cofactor of PLA₂IVα activity. In addition, GroPIns levels can be used to specifically monitor PLA₂IVα enzymatic activity.

In parallel experiments, intracellular GroPIns concentrations were also quantitatively determined by LC-MS/MS (see Materials and Methods). Under normal growth conditions, the GroPIns concentration was 420 ± 20 μmol/L in PC-PTC3–transformed cells and 270 ± 20 μmol/L in PCCl₃ cells. Therefore, RET/PTC3-induced cell transformation is associated with an increase in PLA₂ activity that leads to increased levels of GroPIns, as has also been reported for Ras-transformed cells (9, 12).

In addition, because activating point mutations of the BRAF serine-threonine kinase are frequent genetic events in PTCs (29), the same assays were performed in a PCCl₃ cell line transformed by the BRAF oncogene (BRAF V600E; ref. 13). In these PC-BRAF cells, both basal (Fig. 2A) and ATP-stimulated (Fig. 2B) GroPIns levels were lower than in normal PCCl₃ cells. ATP-stimulated release of arachidonic acid was also less pronounced than in PCCl₃ cells relative to their controls; however, similar to PCCl₃ cells, this was not fully inhibited by pyrrophenone in PC-BRAF cells (Fig. 2C), indicating that this specific PLA₂IVα isoform is not downstream of the BRAF oncogene.

To further analyze the relationships between RET/PTC3 expression and PLA₂IVα activity, we monitored the specific formation of GroPIns. After metabolic [³H] inositol labeling (see Materials and Methods), PC-PTC3 cells were exposed to the RET kinase inhibitor ZD6474 (1 μmol/L for 24 h). Here, basal intracellular GroPIns levels significantly decreased by ∼30% while remaining unchanged in normal PCCl₃ cells (Fig. 3A , left). This was also confirmed by mass spectrometry: GroPIns intracellular concentrations were reduced by ∼45% after a similar 24-h treatment with ZD6474 in PC-PTC3 cells. These data are consistent with increased activity of PLA₂IVα in PC-PTC3 cells that correlates with expression of the RET/PTC3 oncogene, indicating that PLA₂IVα is an important component of the signaling cascade downstream of this oncogene.

To further support this, we used both pharmacologic and molecular approaches. Treatment with pyrrophenone at a concentration specific for PLA₂IVα (0.1 μmol/L; ref. 18) inhibited basal, intracellular GroPIns levels in PC-PTC3 cells by up to 50% (and by up to 30% in PCCl₃ cells; Fig. 3A,, right). Inhibitors known to specifically act on secretory PLA₂ (50 μmol/L sPLA₂IIA inhibitor-I; 5 h; ref. 30) and on Ca²⁺-independent PLA₂ [1 μmol/L bromoenol lactone (BEL); 5 h; ref. 31] also had no effects on basal GroPIns levels in PC-PTC3 cells (Table 1).

As an alternative approach, PLA₂IVα was down-regulated by RNAi (by ∼70%, see Materials and Methods and Fig. 3B,, left). This led to a reduction in GroPIns levels of ∼40% in PC-PTC3 cells compared with PC-PTC3 cells treated with nontargeting duplexes (Fig. 3B,, right), in agreement with data using the specific inhibitor of PLA₂IVα (Table 1; Fig. 3A , right).

Altogether, these data identify PLA₂IVα as a component of the signaling pathway of the RET/PTC3 oncogene in the control of cell proliferation in PC-PTC3 thyroid cells.

Molecular cascade leading to PLA₂IVα activation. The mechanisms involved in the RET/PTC3-dependent activation of PLA₂IVα were then analyzed by specifically following hydrolysis of phosphoinositides and the consequent formation of GroPIns. RET/PTC3 is known to result in phosphorylation of several effectors (15). More specifically, activated RET has been shown to stimulate p38 SAPK activation (32). Under normal growth conditions, PC-PTC3 cells had an increased level of phosphorylated p38 SAPK compared with PCCl₃ cells (Fig. 3C , top left).

ERK1/2 MAPKs are downstream of RET/PTC3 (33), and ERK1/2 MAPK phosphorylation of PLA₂IVα at Ser⁵⁰⁵ produces an increase in enzymatic activity of this phospholipase (34). However, when PC-PTC3 cells were pretreated with the specific ERK1/2 MAPK inhibitor U0126 (30 μmol/L, 5 h), GroPIns levels were not affected (Table 1). Similarly, inhibitors of JNK (10 μmol/L SP600125) and protein kinase C (PKC; 5 μmol/L RO31-8220), two other potential regulators of PLA₂IVα downstream of RET/PTC3, were ineffective (Table 1; similar data were obtained also in normal PCCl₃ cells). In contrast, the phosphoinositide 3-kinase (PI3K) inhibitor wortmannin (1 μmol/L) inhibited GroPIns levels in PC-PTC3 cells by 20% (this effect was, however, not correlated to RET/PTC3 expression because similar inhibition was seen in normal PCCl₃ cells; Table 1). Finally, the inhibitor of p38 SAPK (25 μmol/L SB203580), a kinase known to regulate phosphorylation of PLA₂IVα (3), selectively decreased basal GroPIns levels in RET/PTC3-transformed cells (Table 1). This inhibition was evident from 5 to 48 h of SB203580 treatment, reaching a maximum of ∼30% (Table 1). In PCCl₃ cells, SB203580 treatment was either ineffective (5 h) or induced a significant increase in intracellular levels of GroPIns, again relating the effects reported above to RET/PTC oncogenic transformation (48 h; Table 1).

The role of p38 SAPK in this signaling cascade was confirmed by evaluating phosphorylation levels of p38 SAPK by Western blotting. As indicated above, these were higher (by 2.5-fold) in RET/PTC3-transformed cells compared with PCCl₃ cells (Fig. 3C,, top left). Interestingly, the specific p38 SAPK inhibitor SB203580 (25 μmol/L, 24 h) increased p38 SAPK phosphorylation (Fig. 3C,, top left). This is not surprising considering that SB203580 blocks the kinase activity of p38 SAPK and not the actual phosphorylation itself (see Discussion and ref. 35). In addition, SB203580 treatment decreased the phosphorylation of the p38 SAPK substrate MAPKAP kinase-2, as seen by Western blotting with an anti-phosphorylated MAPKAP kinase-2 antibody (Fig. 3C,, bottom left). This was paralleled by a decrease of up to 50% in PLA₂IVα phosphorylation in RET/PTC3-transformed cells (Fig. 3C,, right). In line with this, activation of PLA₂IVα was also seen in cells transformed by the RET/PTC1 oncogene (PC-PTC1; Fig. 3D,, top), where increased activation of p38 SAPK was seen (Fig. 3D,, middle), as with PC-PTC3 cells (Fig. 3C , top left).

These data are consistent with a signaling cascade that links active RET/PTC oncogenes to p38 SAPK phosphorylation, which in turn phosphorylates and activates PLA₂IVα, leading to release of phosphoinositide metabolites that have been previously shown to affect cell proliferation in normal thyroid cells (6, 32).

PLA₂IVα-dependent regulation of cell proliferation in RET/PTC3-transformed cells. As with other thyroid cancer cells and transformed cell lines, PC-PTC3 cells proliferate in a TSH- and cAMP-independent manner (13). Use of the inhibitor pyrrophenone, which inhibited basal cell proliferation (see above and Fig. 1A), indicated involvement of PLA₂IVα in control of cell growth in these RET/PTC3-transformed cells.

Thus, the correlation between the oncogenic RET/p38 SAPK/PLA₂IVα signaling cascade that regulates GroPIns and arachidonic acid formation and cell proliferation was analyzed. The p38 SAPK inhibitor SB203580 (25 μmol/L, 48 h) significantly reduced basal [³H]thymidine incorporation in PC-PTC3 cells (Fig. 4A), in line with effects of ZD6474 and pyrrophenone shown above (Fig. 1A). Other inhibitors of PLA₂ activities, such as BEL (1 μmol/L), used at a concentration that specifically inhibits calcium-independent PLA₂ (PLA₂VI), did not affect [³H]thymidine incorporation, which was instead partially inhibited by specific inhibitor-I of secretory PLA₂IIA (Fig. 4A).

The results with this panel of inhibitors further support the involvement of the oncogenic RET/p38 SAPK/PLA₂IVα signaling cascade in the control of cell growth in RET/PTC3-transformed cells because inhibition of each of the components hampered the cascade and in parallel prevented cell growth (Fig. 4A). The effect of inhibitor-I for secretory PLA₂IIA (Fig. 4A) correlated with residual arachidonic acid release seen on pyrrophenone treatment (Fig. 2C), suggesting potential involvement of secretory PLA₂IIA either downstream of (3), or independently from, PLA₂IVα activity.

To further elucidate this mechanism, the products of PLA₂IVα catalytic activity (i.e., GroPIns, arachidonic acid, and some arachidonic acid metabolites produced by COX activities) were added directly to growth medium under basal conditions or after inhibiting PLA₂IVα with pyrrophenone (0.5 μmol/L). These data are summarized in Table 2 and indicate that, in PC-PTC3 cells, direct addition of prostaglandin E1 (PGE1; 30 nmol/L), PGE2 (30 nmol/L), and thromboxane B2 (TBXB2; 30 nmol/L) did not affect [³H]thymidine incorporation; similarly, the general COX inhibitor lysinate acetylsalicylic acid (300 μmol/L) was ineffective, excluding autocrine modulation of PC-PTC3 cell growth by this pathway (Table 2). Instead, stimulation of [³H]thymidine incorporation was seen with GroPIns (300 μmol/L) and arachidonic acid (0.1–1 μmol/L), although to different levels (115% and 132–212%, respectively). This was also evident on cell pretreatment with pyrrophenone (0.5 μmol/L), which significantly decreased [³H]thymidine incorporation (by 50%; Table 2); this decrease could be rescued partially by addition of arachidonic acid (1 μmol/L, by 44%; Table 2). Similar experiments were also performed in PCCl₃ cells and indicated that arachidonic acid and GroPIns stimulate [³H]thymidine incorporation, in agreement with previously published data (Table 2; ref. 6).

These data are consistent with a role for RET/PTC3-activated PLA₂IVα in the control of cell proliferation in PC-PTC3 cells and they identify PLA₂IVα as an important effector of the mitogenic activity of the RET/PTC3 oncogene.

Role of PLA₂IVα in human tumor cell proliferation. Several cell lines derived from thyroid papillary carcinomas are now available. Among these, some express activated versions of the RET/PTC1 oncogene (21, 36). In principle, in these human cancer cells, proliferation should be regulated by the oncogenic RET/p38 SAPK/PLA₂IVα signaling cascade reported above. We analyzed this possibility using the BHP10-3 and TPC1 thyroid papillary carcinoma cell lines, which have been characterized for expression of the RET/PTC1 oncogene, which was some 2-fold higher in BHP10-3 cells over TPC1 cells (Fig. 4B , bottom).

These cells were incubated with inhibitors of RET, p38 SAPK, and PLA₂IVα that were effective in PC-PTC3 cells to determine whether this cascade also controls tumor cell proliferation. [³H]Thymidine incorporation in both BHP10-3 and TPC1 cells was inhibited by ∼70% by ZD6474 (1 μmol/L, 16 h), in line with the reported action of the RET/PTC oncogene (37). Inhibition of p38 SAPK (25 μmol/L SB203580, 48 h) also reduced cell proliferation (∼40% and ∼60% in BHP10-3 and TPC1 cells, respectively), and inhibition of PLA₂IVα by pyrrophenone (0.5 μmol/L, 48 h) clearly reduced [³H]thymidine incorporation in BHP10-3 cells, whereas it had no significant effect in TPC1 cells (Fig. 4B,, bottom). Interestingly, oncogenic RET/p38 SAPK/PLA₂IVα modulation seemed more evident in BHP10-3 cells, which express higher levels of RET/PTC oncogene (∼2-fold) and of phosphorylated PLA₂IVα (100% versus 80% of total PLA₂IVα in BHP10-3 and TPC1 cells, respectively; Fig. 4B , top). Inhibitors of arachidonic acid metabolism were also analyzed, and although the general COX inhibitor lysinate acetylsalicylic acid (300 μmol/L; 48 h) was ineffective, the inhibitor of thromboxane synthase and 5-LOX (50 μmol/L ketoconazole; 48 h) inhibited BHP10-3 cell proliferation by up to 90%, suggesting that either a thromboxane or a leukotriene is the active mitogenic metabolite.

This pharmacologic demonstration of the involvement of PLA₂IVα in BHP10-3 cell growth was also confirmed with a molecular approach. PLA₂IVα was knocked down by 90% by RNAi (Fig. 4C,, top) and cell growth was monitored by cell counting; this was significantly inhibited in the presence of siRNAs for PLA₂IVα (Fig. 4C , bottom).

The involvement of PLA₂IVα in thyroid cell transformation was at this point examined in thyroid tissues obtained from surgery. PLA₂IVα activation was analyzed by Western blotting of lysates from RET/PTC-positive thyroid tumors and their respective normal tissues (see Materials and Methods). In line with data reported for transformed cell lines, tumor samples also showed high levels of phosphorylated PLA₂IVα (compared with nonphosphorylated PLA₂IVα, which was virtually undetectable), whereas normal tissues showed comparable levels of these two PLA₂IVα forms (Fig. 4D).

These data indicate that the oncogenic RET/p38 SAPK/PLA₂IVα pathway described in PC-PTC3 cells is also active in these human thyroid carcinoma cells. In addition, they indicate that loss of thyroid differentiation (and associated loss of cAMP-dependent growth) can lead to an alternative pathway for cell proliferation that involves RET/PTC activation of PLA₂IVα, with release of active lipid metabolites of the arachidonic acid cascade.

In this study, we report novel components of the mitogenic signaling cascade initiated by the RET/PTC oncogenes: a specific isoform of PLA₂, PLA₂IVα, and arachidonic-acid–derived lipid metabolites, which are shown here to be essential mediators of RET/PTC-induced thyroid tumor cell growth. More importantly, this mitogenic pathway is also relevant in human thyroid tumors characterized by oncogenic RET/PTC expression.

In human thyroid papillary carcinomas, cell proliferation is often promoted by phosphorylation cascades such as those induced by products of chimeric genes, such as RET/PTC (17, 38), or NTRAK1-derived oncogenes (39) or by rearrangements or mutations that constitutively activate the proto-oncogene BRAF, a kinase acting upstream of MAPKs (40).

Our recent finding that PLA₂IVα is specifically involved in TSH-independent growth control in normal rat thyroid cells (6) led to our analysis of the role of this cascade in tumor cell growth. Quite surprisingly, PLA₂IVα expression in PC-PTC3 and PC-PTC1 cells was relatively low, although PLA₂IVα was fully phosphorylated (in a RET/PTC-dependent manner) and active (see Results). This is in line with a study reporting the gene expression profile in PC-PTC3 cells, which showed that, among other signaling enzymes, the PLA₂IVα gene is also down-regulated 2.5-fold (13). Thus, on the one hand, expression of the RET/PTC3 oncogene results in decreased PLA₂IVα gene transcription and hence lower expression of PLA₂IVα, and on the other hand, it causes direct activation of PLA₂IVα by stimulating its phosphorylation and thus increases production of its lipid products.

Although these two effects on protein levels and activation may seem contradictory, there is evidence that although tumorigenesis correlates with PLA₂IVα activity, it does not necessarily correlate with increased PLA₂ levels (41). A clear example of this is seen in the Apc^Min and Apc^Δ716 mouse models of colon tumor (42, 43), where cytosolic PLA₂ expression and function are completely different within specific regions of the gut and small intestine, and in colon tumors, indicating that the tumorigenic process is driven by different pathophysiologic processes in different cells (41). The role of cytosolic PLA₂ should thus be considered in a more general context that takes into account the levels and activities of the enzymes acting downstream of, and on the products of, PLA₂, such as the COXs (44). In addition, besides being a substrate of COX, PLA₂-derived arachidonic acid is also an active molecule per se, as it is involved in induction of apoptosis (45). For example, in mouse and human colon tumors, cPLA₂ is greatly decreased at the mRNA and protein levels compared with normal tissues, whereas levels of COX-2 are increased. This leads to decreased production of arachidonic acid (due to lower PLA₂ activity) coupled to increased metabolism (due to higher COX-2 activity). The final result is thus higher PGE2 levels, which enhance tumorigenesis and angiogenesis, and lower free arachidonic acid, which protects tumor cells from apoptosis (44).

As with normal PCCl₃ cells, we also see that, in the PC-PTC3 cells, PLA₂IVα preferentially hydrolyzes membrane phosphoinositides (6),⁵

which made possible this detailed study of the cascade leading from the RET/PTC3 oncogene to the specific PLA₂IVα product GroPIns, through use of a panel of inhibitors. Thus, although oncogenic RET/PTC3 is clearly associated with increased phosphorylation of ERK1/2 MAPKs, p38 SAPK, MAPKAP kinase-2, and finally PLA₂IVα, only when p38 SAPK phosphorylation and activation was inhibited did basal intracellular levels of GroPIns decrease, indicating that PLA₂IVα is downstream of p38 SAPK phosphorylation. Consequently, thymidine incorporation by PC-PTC3 cells is also inhibited by p38 SAPK and PLA₂IVα inhibitors, indicating that these enzymes act downstream of oncogenic RET/PTC in cell growth regulation (see Results).

Increased p38 SAPK phosphorylation in the presence of the inhibitor SB203580 has also been seen in other cell lines (Raw264.7 and CHO cells; ref. 45),⁵ and it probably arises from a block of p38 SAPK-dependent phosphatase activity, which results in increased p38 SAPK phosphorylation. Nevertheless, SB203580 inhibited p38 SAPK activity, as indicated by reduced phosphorylation of MAPKAP kinase-2.

According to the data herein reported, cell proliferation in RET/PTC3-transformed cells should be regulated by arachidonic acid or an arachidonic acid metabolite formed by COX, LOX, or cytochrome P450 activities. As the only effective inhibitor was ketoconazole, thromboxanes or leukotrienes are thus far the best candidates in this tumor cell growth (see Results).

Work is in progress to better define the specific mitogenic compound of this arachidonic acid cascade. This aspect is of particular interest because the possibility of inhibiting a specific intracellular mitogen could help to delineate a targeted therapy for thyroid and other tumors that involve PLA₂IVα. In support of this proposal, we show here that the same regulatory cascade operating in rat thyroid cells regulates human thyroid tumor cell growth. Thus, oncogenic RET/PTC expression promotes thyroid cell growth (15), which can be inhibited by preventing the PLA₂IVα cascade, either by specific inhibitors or by knocking down of the PLA₂IVα protein by RNAi (see Results). Interestingly, although the human tumor cell lines examined respond equally well to the RET kinase inhibitor (37), they have a differential response to p38 SAPK and PLA₂IVα inhibitors. This could be related to different levels of expression of RET/PTC and phosphorylated PLA₂IVα in these cells (see Results). It can thus be hypothesized that RET/PTC-induced cell proliferation is activated through different pathways, the choice of which is determined by the level of expression of specific downstream effectors. Thus, Rap1 (46), PI3K/Akt (47), Ras-MAPK (48), p27Kip1 (37), phospholipase Cγ (49), or PLA₂IVα (this study) can all be crucial mitogenic agents depending on levels of their expression/activation or concomitant expression of cofactors. It is of note, however, that the active PLA₂IVα pathway in tumor tissues (see Results) further strengthens the involvement of PLA₂IVα in RET/PTC-mediated tumorigenesis in vivo.

Altogether, these data further increase the complexity of signaling pathways present in thyroid tumors that can be promoted by different oncogenes linked to distinct signaling cascades. This is indeed the case for the RET/PTC and BRAF oncogenes; while activating common signaling pathways, such as the MAPK/ERK kinase–dependent cascade (40), they differ in their ability to initiate lipid signaling driven by PLA₂IVα, which is peculiar of RET/PTC.

These data also lead to the proposal that thyroid and other types of tumors that are regulated by PLA₂IVα signaling can be treated with already available agents that are targeted to elements of the PLA₂IVα cascade. A more specific therapy could emerge from the identification of the arachidonic acid metabolite that regulates tumor growth (e.g., in thyroid or colon cancers), which would then allow the development and use of highly specific, targeted drugs.

Grant support: Italian Association for Cancer Research (Milan, Italy), Telethon Italia (Italy), and Ministero dell'Università e della Ricerca (Italy). B.M. Filippi was a fellow of the Italian Foundation for Cancer Research (Milan, Italy).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank A. Luini, M. Di Girolamo, R.M. Melillo, F. Carlomagno, and C.P. Berrie for useful discussions; Dr. K. Seno for pyrrophenone; A.J. Ryan for ZD6474; J.M. Hershman for BHP10-3 cells; Prof. F. Basolo for the tumor samples; and E. Fontana for preparation of the figures.