Induction of cPLA2 in Lung Epithelial Cells and Non-small Cell Lung Cancer Is Mediated by Sp1 and c-Jun*

Activating mutations in rasgenes are frequently associated with non-small cell lung cancer cells (NSCLC) and contribute to transformed growth in these cells. Expression of oncogenic forms of Ras in these cells is associated with increased expression and activity of cytosolic phospholipase A2(cPLA2) and cyclooxygenase-2 (COX-2), leading to constitutively elevated levels of prostaglandin production. Expression of oncogenic Ras is sufficient to induce these enzymes in normal lung epithelial cells. We have previously reported that the JNK and ERK pathways are necessary for induction of cPLA2 and have defined a minimal region of the cPLA2 promoter from −58 to −12 that is required for Ha-Ras-mediated induction. To further characterize the cis-regulatory elements within this region involved in this response, site-directed mutagenesis was used to make mutations at various sites. Three cis-regulatory elements were identified: regions −21/−18, −37/−30, and −55/−53. Mutations in any of these elements decreased basal and Ha-Ras-induced cPLA2 promoter activity in both normal lung epithelial cells, as well as steady state promoter activity in A549 cells, with a mutation in element −21/−18 completely eliminating all promoter activity. Overexpression studies and gel shift assays indicated that Sp1 may serve as a transcription factor functionally regulating promoter activity by directly interacting with two of thecis-regulatory elements, −21/−18 and −37/−30. Expression of Ha-Ras led to induction of c-Jun protein, which showed functional cooperation with Sp1 in driving promoter activity. Additional unidentified transcription factors bound to the regions from −55/−53 and −37/−34.

Activating mutations in ras genes are frequently associated with non-small cell lung cancer cells (NSCLC) and contribute to transformed growth in these cells. Expression of oncogenic forms of Ras in these cells is associated with increased expression and activity of cytosolic phospholipase A 2 (cPLA 2 ) and cyclooxygenase-2 (COX-2), leading to constitutively elevated levels of prostaglandin production. Expression of oncogenic Ras is sufficient to induce these enzymes in normal lung epithelial cells. We have previously reported that the JNK and ERK pathways are necessary for induction of cPLA 2 and have defined a minimal region of the cPLA 2 promoter from ؊58 to ؊12 that is required for Ha-Ras-mediated induction. To further characterize the cis-regulatory elements within this region involved in this response, site-directed mutagenesis was used to make mutations at various sites. Three cis-regulatory elements were identified: regions ؊21/؊18, ؊37/؊30, and ؊55/؊53. Mutations in any of these elements decreased basal and Ha-Ras-induced cPLA 2 promoter activity in both normal lung epithelial cells, as well as steady state promoter activity in A549 cells, with a mutation in element ؊21/؊18 completely eliminating all promoter activity. Overexpression studies and gel shift assays indicated that Sp1 may serve as a transcription factor functionally regulating promoter activity by directly interacting with two of the cis-regulatory elements, ؊21/؊18 and ؊37/؊30. Expression of Ha-Ras led to induction of c-Jun protein, which showed functional cooperation with Sp1 in driving promoter activity. Additional unidentified transcription factors bound to the regions from ؊55/؊53 and ؊37/؊34.
Lung cancer is the leading cause of cancer death in the United States, but, despite a significant research effort, the critical pathways mediating transformed growth remain poorly understood. Non-small cell lung cancer (NSCLC) 1 constitutes the majority of lung cancers, and gain of function mutations in ras genes are frequently associated with NSCLC, occurring in 30% of adenocarcinomas, and just under 10% of other NSCLC types (1). Activating mutations in ras genes have also been detected in other types of human cancer, including colon, prostate, and pancreatic. It is believed that these mutated forms of Ras lacking intrinsic GTPase activity mediate transformation by constitutively activating downstream effector pathways. We recently reported that expression of oncogenic forms of Ras was associated with increased expression of cytosolic phospholipase A 2 (cPLA 2 ) and cyclooxygenase-2 (COX-2) in a panel of NSCLC cell lines (2). cPLA 2 is the major intracellular form of PLA 2 , which preferentially hydrolyzes membrane phospholipids at the sn-2 position to release arachidonic acid and represents the rate-limiting enzyme in eicosanoid production (3)(4)(5). Free arachidonic acid is metabolized through three major pathways to produce eicosanoids. COX converts arachidonic acid to prostaglandins and thromboxane, lipoxygenases produce leukotrienes and hydroxyeicosatetraenoic acids, and cytochrome P-450 epoxygenase produces epoxyeicosatrienoic acids. Two forms of cyclooxygenase have been identified (6). COX-1 is constitutively expressed in most cell types and involved in maintaining vascular tone, whereas COX-2 is an immediate early response gene (7) induced by mitogenic stimuli and associated with inflammation. Constitutively high levels of prostaglandin production are observed in NSCLC as a result of elevated levels of cPLA 2 and COX-2 (8,9). We (2) and others (10 -12) have shown that nonsteroidal anti-inflammatory agents, which inhibit eicosanoid production, block the transformed growth of NSCLC expressing Ras mutations. These drugs do not hinder the growth of most nontransformed cells, suggesting that this pathway plays a critical role in transformation.
Although studies in NSCLC cell lines have implicated a role for Ras in the induction of cPLA 2 , these cells contain a large number of mutations and aberrations in signaling pathways, making it difficult to define the critical molecular pathways regulating cPLA 2 expression. We therefore sought to examine the effects of constitutively active forms of Ras on cPLA 2 expression in untransformed lung epithelial cells. Ras has been shown to connect to numerous effector pathways, leading to the activation of diverse physiological responses (see Ref. 13 for review). These pathways regulate cell proliferation as well as changes in the cytoskeleton. We have recently demonstrated that expression of oncogenic forms of Ras directly increased cPLA 2 expression in normal lung epithelial cells through activation of the JNK and ERK pathways (14).
The regulatory elements within the cPLA 2 promoter critical for this induction by oncogenic Ras have not been defined. The cPLA 2 promoter has been isolated from both human (15) and rat (16) and contains a number of putative regulatory elements including AP-1 sites, NF-B sites, and glucocorticoid regulatory elements. The promoter region resembles that of a housekeeping gene in that it contains no TATA box and no CAAT box, but it is atypical in that it is not GC-rich (34.5%) and has no consensus Sp1 sites (15,17). We reported previously that the region of the promoter covering residues Ϫ58 to Ϫ12 is crucial for induction of promoter activity by oncogenic Ras (14) in RL-65 cells, a neonatal, untransformed rat epithelial cell line (18). In this study, we define three cis-regulatory elements within this region of the promoter critical for both basal and Ha-Ras-induced cPLA 2 promoter activity, and have begun to define transcription factors acting at these sites.
For transient transfections, cells were electroporated as described previously (2). Briefly, 2 million cells were electroporated in duplicate dishes using a geneZAPPER (IBI). Unless otherwise stated, cells were transfected with 2 g of the Ϫ58 truncation mutant of the cPLA 2 promoter, 2 g of CMV-␤-galactosidase, and 2 g of other DNA (e.g. CMV-Ha-Ras, CMV-Sp1). Total DNA concentration for each transfection was matched with pcDNA-3.1. Following electroporation, cells were incubated in standard media for 48 h. Cells were then harvested, and luciferase and ␤-galactosidase activity determined as described previously (2). Results are expressed as luciferase units normalized to ␤-galactosidase units.
The Sp1 consensus binding site oligonucleotide (5Ј-CCCTTGGT-GGGGGCGGGGCCTAAGCTGCG-3Ј; Sp1 consensus site in bold italics) was purchased from Geneka Biotechnology (Toronto, Canada). Supershift assays were carried out using a specific monoclonal Sp1 antibody and a specific polyclonal c-Jun antibody (Santa Cruz Biotechnology, Santa Cruz, CA). Nuclear extracts (2 g) were incubated in binding reaction buffer with a final KCl concentration of 100 mM for 30 min on ice. For supershift or competition assays, the relevant antibodies or unlabeled oligonucleotides were incubated with nuclear extracts prior to adding the labeled probe. [ 32 P]dCTP-labeled probe (100,000 cpm) was added for an additional 30 min in a total volume of 20 l. For competition assays, a 50-or 100-fold excess of unlabeled double-stranded oligonucleotide was added to the binding reaction. Samples were resolved on a nondenaturing 5% acrylamide gel (29:1 acrylamide:bisacrylamide), in 1ϫ TGE (25 mM Tris, 1.0 mM EDTA, and 190 mM glycine) at 20 mA/gel for ϳ90 min. Gels were dried and exposed to film for autoradiography.

RESULTS
In earlier studies, we defined a minimal region of the rat cPLA 2 promoter spanning residues Ϫ58 to Ϫ12, which was necessary and sufficient for Ha-Ras-mediated induction of cPLA 2 in RL-65 cells and expression in NSCLC (14). An additional 5Јtruncation down to Ϫ37 reduced basal promoter activity, and inhibited but did not completely block induction by Ha-Ras. Ultimately, truncation down to Ϫ12 completely abolished both basal and Ha-Ras-induced promoter activity. These studies were paralleled in A549 cells, a human NSCLC cell line expressing an activating mutation in Ras. Promoter activity was slightly elevated upon truncation from Ϫ2.4 kilobase pairs to Ϫ58 bp. Further deletion to Ϫ37 decreased promoter activity by ϳ60%, and truncation to Ϫ12 entirely abolished promoter activity.
Examination of the sequence of the cPLA 2 promoter revealed that the 58-bp region contained no consensus sites for any known transcription factors. Therefore, to define regulatory elements, a series of mutations were introduced throughout the region with the original Ϫ58 truncation mutant as a template (Fig. 1). With the exception of mutation Ϫ37/Ϫ34, all mutations correspond to regions of the promoter that are conserved between the rat and human promoter (15). Plasmids encoding each mutation were transiently co-transfected into RL-65 cells along with an expression plasmid for Ha-Ras or pcDNA-3 as a control. Expression of Ha-Ras increased wild type promoter activity by 3-5-fold ( Fig. 2A), consistent with earlier studies. Mutation Ϫ21/Ϫ18 almost completely abolished both basal and Ha-Ras stimulated promoter activity. Mutations Ϫ37/Ϫ34 and Ϫ33/Ϫ30 inhibited both basal and Ha-Ras-mediated promoter activity by ϳ80 -90%. With the double mutant Ϫ37/Ϫ30, activity was further inhibited to ϳ10% of wild-type promoter. Mutation at Ϫ55/Ϫ53 decreased both basal promoter activity and Ha-Ras-mediated induction by 66%. There was no significant change in either basal promoter activity or Ha-Ras-mediated induction with mutation Ϫ49/Ϫ47.
Stimulation of RL-65 cells with EGF also increased cPLA 2 promoter activity (Fig. 2B). This induction was blocked by co-expression with N17-Ras (data not shown), indicating that the effects of EGF are mediated at least in part through the Ras signaling pathway. EGF stimulation was blocked with mutants Ϫ55/Ϫ53, Ϫ37/Ϫ34, Ϫ33/Ϫ30, or Ϫ21/Ϫ18 to an extent similar to that seen with expression of Ha-Ras (Fig. 2B). Finally, the effects of these mutations in the cPLA 2 promoter were also examined in A549 cells. Steady-state promoter activity was almost entirely eliminated with mutation Ϫ21/Ϫ18. Mutations Ϫ55/Ϫ53, Ϫ37/Ϫ34, and Ϫ33/Ϫ30 inhibited steady-state promoter activity by ϳ85-95% (Fig. 3). The double mutant Ϫ37/ Ϫ30 decreased promoter activity by about 95%. Mutation Ϫ49/ Ϫ47 reduced promoter activity by ϳ60%.
The data from these studies suggest that three regions of the cPLA 2 promoter appear to be critical for expression: regions Ϫ37/Ϫ30, Ϫ21/Ϫ18, and Ϫ55/Ϫ53. To examine complexes formed at these sites, EMSAs were performed using 32 P-labeled probes spanning each of the three cis-regulatory elements (Figs. 4 -6).
Using an oligonucleotide encoding Ϫ42 to Ϫ13, a number of specific bands were detected using nuclear extracts prepared from RL-65 cells (Fig. 4A). At least three complexes were com-peted off by excess cold oligonucleotide. Extracts from EGFstimulated cells showed the same pattern of bands, with no significant change in intensity (data not shown). Sp1 is a ubiquitously expressed transcription factor, which binds to GC-rich elements frequently found in a number of housekeeping genes. The region of the cPLA 2 promoter from Ϫ58 to Ϫ1 is 55% GC-rich, and several areas within this region represent putative, but not consensus Sp1 binding sites. Sp1 has been demonstrated to activate numerous TATA-less promoters (20,21) and has recently been shown to control expression of tissue specific genes as well (22,23). To examine whether Sp1 can form complexes with this region of the cPLA 2 promoter, we employed a specific Sp1 antibody in a "supershift" EMSA. In the presence of Sp1 antibody and nuclear extracts from RL-65 cells, a major complex (a) was supershifted to a slower migrating complex (b) (Fig. 4A). Using a probe with mutations at Ϫ33/Ϫ30, the intensity of band a was greatly diminished, and a new band (c) was detected (Fig. 4B). This band was not supershifted with anti-Sp1 antibody. Using a probe with mutations at Ϫ37/Ϫ34, resulted in decreased intensity of band a, and a detectable, but faint supershifted band b (Fig. 4C). The double mutation Ϫ37/Ϫ30 eliminated all three bands (Fig. 4D). A similar pattern of bands was observed using nuclear extracts prepared from A549 cells (Fig. 4E), with the slowest migrating band supershifted with anti-Sp1 antibodies. Mutations at Ϫ33/ Ϫ30 caused a similar disappearance of band a and appearance of a new band (c).
With a labeled probe encoding the region from Ϫ28 to Ϫ8, three major bands were detected using extracts from either RL-65 cells or A549 cells (Fig. 5). All of these bands were competed off by excess cold oligonucleotide (data not shown). The slowest migrating band (a) was supershifted with anti-Sp1 anti-  2 and 4) or without Sp1 antibody (lanes 1 and 3). Complex a (a) was supershifted to a slower migrating species (b) by Sp1 antibody. With mutations at Ϫ33/Ϫ30 probe, a fast migrating complex was detected (c). bodies to a slower migrating complex (b). When the identical extracts were incubated with probe containing a mutation at Ϫ21/Ϫ18, no specific bands were detected with extracts from RL-65 (Fig. 5A, lane 5) or A549 cells (data not shown). Finally, EMSA analysis was performed using a labeled probe corresponding to the region from Ϫ65 to Ϫ40. A single major complex was detected, which was competed off by excess cold oligonucleotide (Fig. 6A). Extracts from EGF-treated cells showed the same pattern, with no significant difference in the intensity of the band. Antibodies against Sp1 failed to supershift this band (Fig. 6B). Once again, extracts from A549 cells gave a complex of the same mobility, which was competed off by cold oligonucleotide and not shifted by Sp1 antibody (Fig. 6C). Based on these data, it appears that Sp1 can bind to both the Ϫ37/Ϫ30 region and the Ϫ21/Ϫ18 region. Because the Ϫ21/Ϫ18 region is also encoded within the Ϫ42/Ϫ13 oligonucleotide (Fig. 4), we performed an additional EMSA, with the Ϫ42/Ϫ13 oligonucleotide mutated at Ϫ21/Ϫ18. The pattern of bands observed with this oligonucleotide (data not shown) was indistinguishable from that obtained with the wildtype Ϫ42/Ϫ13 probe (Fig. 4A).
To determine whether Sp1 plays a functional role in regulation of cPLA 2 promoter activity, RL-65 cells were transiently co-transfected with cPLA 2 promoter constructs along with a Sp1 expression vector. Overexpression of Sp1 increased wild type promoter activity ϳ1.6-fold in RL-65 cells (Fig. 7). However, with the promoter construct with a mutation at Ϫ33/Ϫ30, no transactivation of promoter activity was detected in the context of Sp1 overexpression. Mutations at Ϫ37/Ϫ34 or Ϫ21/ Ϫ18 resulted in lower basal promoter activity, with some residual stimulation in the presence of Sp1.
Sp1 is capable of forming multimeric complexes through homotypic interactions (24,25) as well as interacting heterotypically with different types of nuclear proteins such as NF-Y (26,27), E2F (28), and AP-2 (29). The 12-lipoxygenase promoter has many similarities with the cPLA 2 promoter in that it is also a TATA-less promoter whose activity is regulated by the ERK and JNK mitogen-activated protein kinase pathways and induced by Ha-Ras expression (30,31). Sp1 has been reported to bind to several consensus sites within this promoter (32). The transcription factor c-Jun has been implicated in induction of the 12-lipoxygenase promoter through direct binding to Sp1 (33). Others have reported that c-Jun superactivates Sp1 to drive the human p21(WAF/Cip1) promoter by directly binding to Sp1 but not to the promoter (34). To determine whether c-Jun is involved in the induction of the cPLA 2 promoter, RL-65 cells were co-transfected with the cPLA 2 promoter construct along with expression plasmids encoding c-Jun and Sp1 either individually or in combination. Overexpression of c-Jun alone increased cPLA 2 promoter activity by itself and, in the setting of Sp1 overexpression, resulted in a synergistic increase in promoter activity (Fig. 7A). Overexpression of either Sp1 or c-Jun individually or together had a blunted effect on promoter activity with the Ϫ37/Ϫ34 mutant, and failed to cause significant increases for mutants Ϫ33/Ϫ30, or Ϫ21/Ϫ18 (Fig. 7A). Direct binding of c-Jun to the promoter was assessed by EMSA with an anti-c-Jun antibody. However, no supershifted complex was detected with this antibody using either region Ϫ42/Ϫ13 (Fig. 4) or Ϫ28/Ϫ8 (Fig. 5). Incubation of extracts with either oligonucleotide in the presence of antibodies against Sp3, NF-Y, or gut Kruppel-like factor (GKLF) failed to result in a supershifted complex (data not shown). Expression of v-Jun, which lacks the JNK docking domain (35), also stimulated promoter activity, and synergized with Sp1. In fact, co-expression of v-Jun and Sp1 increased promoter activity to levels comparable with that seen with expression of Ha-Ras.
One possible mechanism to account for the role of c-Jun in regulating the cPLA 2 promoter is that activation of Ras leads to increased expression of c-Jun. We therefore examined levels of c-Jun expression in RL-65 cells transfected with Ha-Ras. Expression of Ha-Ras induced c-Jun expression 3-4-fold, as assessed by immunoblotting (Fig. 7B). Increased phosphorylation of c-Jun was also increased as assessed by immunoblotting with a phospho-Jun-specific antibody (data not shown).

DISCUSSION
Activating mutations in Ras are characteristic of a variety of cancer cells and play a critical role in the transformation of these cells. We have shown previously that expression of oncogenic Ras is necessary for elevated expression of cPLA 2 and COX-2 in NSCLC (2), and sufficient to induce expression of both enzymes in nontransformed lung epithelial cells (14). In both cell types, induction of cPLA 2 involved activation of the ERK and the JNK families of mitogen-activated protein kinases.
Through truncational analysis, we have previously defined the minimal region of the promoter required for induction of cPLA 2 from Ϫ58 to Ϫ12 (14). Further truncation of this region indicated two smaller regions, which contribute to regulated expression. Promoter activity decreased when truncating from Ϫ58 to Ϫ37, but some basal activity was retained and the -fold induction of the promoter achieved by expression of Ha-Ras was similar. An additional truncation to Ϫ12 abolished both basal and Ha-Ras stimulated activity. We sought to further characterize the cis- FIG. 5. EMSA analysis in region ؊28/؊8 of the cPLA 2 promoter. Left panel, extracts from RL-65 cells were incubated with 32 P-labeled probe corresponding to region Ϫ28 to Ϫ8 of the cPLA 2 promoter (lanes 1-4) or with probe containing mutations at Ϫ21/Ϫ18 (lane 5). Extracts were incubated with Sp1 antibody, c-Jun antibody, or no addition as described in Fig. 4. A specific complex (a) was supershifted to a slower migrating form (b) in the presence of Sp1 antibody. Right panel, extracts from A549 cells were incubated with WT probe in the presence or absence of Sp1 or c-Jun antibody. regulatory elements in this region responsible for these effects. From the present study, we have identified three cis-regulatory elements from Ϫ55/Ϫ53, from Ϫ37/Ϫ30, and from Ϫ21/Ϫ18, which appear to be critical for regulated expression of cPLA 2 . In normal lung epithelial cells, mutations in each of these regions significantly decreased both basal promoter activity and stimulated promoter activity in response to expression of Ha-Ras or exposure to EGF. Similarly, each mutation decreased steadystate promoter activity in an NSCLC line expressing oncogenic Ras (A549). Overall, region Ϫ21/Ϫ18 appears to be the most crucial element in regulating promoter activity because mutation of this region abolishes all promoter activity.
By EMSA analysis, Sp1 bound to both the Ϫ37/Ϫ30 region and the Ϫ21/Ϫ18 region. Although these regions do not contain classical Sp1 sites, they are GC-rich. Our data suggest that at least two factors bind to the Ϫ37/Ϫ30 region. Mutating the Ϫ37/Ϫ34 region decreased the binding of Sp1, but still resulted in a fainter band, which was shifted by anti-Sp1 antibody (Fig.   4C). However, mutation of the Ϫ33/Ϫ30 region resulted in the formation of a complex that was not supershifted (Fig. 4B). We would therefore propose that a factor binding to the Ϫ37/Ϫ34 cooperates to promote Sp1 binding to the Ϫ33/Ϫ30 region. The identity of this second factor is unknown, but, based on supershift assays, it is not likely to be another member of the Sp1 family. Sp1 binding correlated with the ability of Ha-Ras to induce the promoter, because mutations that inhibited induction of promoter activity blunted or abolished Sp1 binding. Furthermore, overexpression of Sp1 in RL-65 cells was sufficient to induce promoter activity in the absence of Ha-Ras expression. Overexpressing Sp1 alone did not stimulate promoter activity to the extent seen with Ha-Ras. This may be a result of the fact that Sp1 is already highly expressed in most cells and is difficult to overexpress. Alternatively, Sp1 itself may not be sufficient to drive the promoter and an additional factor(s) is required (see above). Mutations in the Ϫ33/Ϫ30 region, which resulted in disappearance of complexes contain- ing Sp1, completely abolished transactivation of promoter activity by Sp1 overexpression. With a construct containing mutations in the Ϫ21/Ϫ18 region, overexpression of Sp1 still resulted in a small increase in promoter activity. Thus, we would suggest that Sp1 binding to both regions (Ϫ33/Ϫ30 and Ϫ21/Ϫ18) is required for promoter induction.
Our data suggest that, similar to regulation of the 12-lipoxygenase promoter by Ha-Ras (30,31), c-Jun cooperates with Sp1 to induce cPLA 2 expression. Overexpression of c-Jun alone increased cPLA 2 promoter activity; importantly, a synergistic induction of promoter activity occurred when both c-Jun and Sp1 were overexpressed. With oligonucleotides spanning the entire 58-bp region, no complex was detected in EMSA assays that was supershifted with anti-Jun antibodies. This is consistent with the lack of an AP-1 site in this region. These findings suggest that c-Jun does not bind directly to the promoter, but may form a complex with Sp1, leading to activation. Although we have not been able to detect such a complex using an EMSA assay, the affinity of binding may be too low. Others workers have reported that c-Jun is involved in driving expression of the p21(WAF1/Cip1) promoter by binding directly to Sp1 but not to the promoter (34), but also failed to detect a c-Jun/Sp1 complex in an EMSA assay. Our previous studies have implicated a role for JNK activation in Ha-Ras-mediated induction of the cPLA 2 promoter. Phosphorylation of c-Jun by JNKs on serines 63 and 73 is critical for activation of c-Jun-mediated transcription at AP-1 sites (36). It is therefore possible that activation of JNKs in response to expression of Ha-Ras mediates induction of cPLA 2 promoter activity by phosphorylating c-Jun. However, the ability of v-Jun, which lacks the JNK docking domain, to synergize with Sp1 in inducing promoter activity, suggests that increased expression of c-Jun may mediate induction of the promoter. The role of the JNK pathways may therefore be more important in regulating expression of c-Jun, rather than direct phosphorylation of the protein.
Sp1 belongs to a family of 16 different transcription factors containing a highly conserved zinc finger DNA-binding domain (22). All these proteins bind to similar DNA sites, which are specifically G-rich elements such as GC and GT/CACC boxes (22,37). Although the Sp subgroup of proteins prefer the GCrich box and the Kruppel-like factors favor the GT/CACC element, some cis elements have affinities for members of both subfamilies (37). GKLF binds to a GC-rich consensus sequence, but can bind with lower affinity to a classical GC-rich Sp1 site (38), whereas GKLF and Sp1 can bind with strong affinity to the basic transcription element found in the promoters of a family of cytochrome P450 genes (37,39). Another member of this family is lung Kruppel-like factor (LKLF). The major site of expression of LKLF is the lung (22), and it has been shown to play a role in normal lung development (40). LKLF can bind to the CACC element of the ␤-globin promoter known to be a site for erythroid Kruppel-like factor (41). We have preliminary data, using a yeast one-hybrid screen, that LKLF can also bind to the cPLA 2 promoter. 2 As the minimal region of the cPLA 2 promoter (Ϫ58 to Ϫ12) does not contain any classical Sp1 sites, it may be that multiple family members have an affinity for the cis-regulatory elements within this region. Sp1 may be able to bind to the promoter with some affinity and drive the promoter when it is highly expressed, but multiple Kruppel-like factors may share affinity for these cis-regulatory elements and regulate the promoter in vivo.
In summary, we propose a hypothetical model for induction of cPLA 2 expression by oncogenic Ras. Constitutively active Ras signals through the ERK and JNK mitogen-activated protein kinase pathways as well as some additional pathway to activate downstream transcription factors that work to regulate cPLA 2 expression through three cis-regulatory elements within its promoter, from Ϫ21/Ϫ18, Ϫ37/Ϫ30, and Ϫ55/Ϫ53. A candidate transcription factor includes Sp1 which may bind to two of the elements, Ϫ21/Ϫ18 and Ϫ33/Ϫ30. Additional factors bind to the region from Ϫ37/Ϫ34 and the third element from Ϫ55/Ϫ53. c-Jun expression is induced by Ha-Ras and may play a functional role by interacting with Sp1 without directly binding to the promoter. The similarities in the pathways mediating induction of cPLA 2 and 12-lipoxygenase by Ha-Ras also suggest that the coordinated induction of these enzymes is required for enhanced eicosanoid production.