Functional Regulation of Phospholipase D Expression in Cancer and Inflammation*

Phospholipase D (PLD) regulates downstream effectors by generating phosphatidic acid. Growing links of dysregulation of PLD to human disease have spurred interest in therapeutics that target its function. Aberrant PLD expression has been identified in multiple facets of complex pathological states, including cancer and inflammatory diseases. Thus, it is important to understand how the signaling network of PLD expression is regulated and contributes to progression of these diseases. Interestingly, small molecule PLD inhibitors can suppress PLD expression as well as enzymatic activity of PLD and have been shown to be effective in pathological mice models, suggesting the potential for use of PLD inhibitors as therapeutics against cancer and inflammation. Here, we summarize recent scientific developments regarding the regulation of PLD expression and its role in cancer and inflammatory processes.

Phospholipase D produces phosphatidic acid (PA) 2 via hydrolysis of phospholipids such as phosphatidylcholine and cardiolipin. As a lipid second messenger, PA has been implicated in a wide range of pathophysiological processes including proliferation, oncogenesis, inflammation, phagocytosis, membrane fusion, and spermatogenesis. Phosphatidylcholine-specific PLD1 and PLD2 are the classic mammalian isoforms of PLD (1,2). Growth factor, mitogen, and inflammatory cytokines up-regulate expression and activity of PLD; however, aberrant dysregulation of PLD occurs in various cancers and inflammation-related diseases. Accordingly, it is important to understand how expression of PLD is regulated and contributes to these diseases.
An association between inflammation and cancer has been observed (3) and supported by recent epidemiological data that indicated ϳ20% of cancer deaths are linked to chronic infections and persistent inflammation (4). The PLD signaling pathway provides a possible molecular link between inflammation and cancer; however, systematic investigation of PLD as a therapeutic target has only begun within the last few years with the development of small molecule PLD inhibitors and PLD knock-out mice (5)(6)(7)(8)(9)(10)(11)(12)(13). This review covers recent advances in the regulation of PLD expression and its role in cancer and inflammation.
PLD isozymes have been linked to protumorigenic and prometastatic phenotypes, although they play different roles in this context (23). Overexpression of PLD1 and its activity are required for mutant H-Ras-induced transformation and tumorigenesis (24). Moreover, PEA-15, a binding partner of PLD1, promotes H-Ras-mediated cell transformation via enhanced expression and activation of PLD1 (23). PLD2-generated PA in response to EGF activates Ras by recruiting its immediate activator, SOS, to translocate to the plasma membrane (16). Interestingly, PLD2 binds to Ras and acts as a guanine nucleotide exchange factor (GEF) of Ras.
The GEF action of PLD2 is due to the PLD2 protein itself and occurs independent of the lipase activity. Moreover, the GEF activity of PLD2 is greatly elevated in rapidly growing and highly metastatic cancer cells (25). Thus, PLD1 might be a critical downstream mediator of H-Ras-induced tumorigenesis, and PLD2-generated PA and/or PLD2 protein itself play crucial roles in the activation of Ras. PLD also activates mTOR and Wnt signaling and is linked to oncogenesis (18 -22). In addition, PLD enhances secretion of matrix metalloproteinases (MMPs), which degrade surrounding extracellular matrix to facilitate cellular movement (26,27). Thus, it seems that aberrant expression and activation of PLD in cancers accelerates these signaling pathways and contributes to cancerous phenotypes.
Simpler organisms, especially those containing a single gene, offer a useful setting to examine the role of PLD in cellular physiology. Deficiency of PLD in yeast is tolerated during vegetable growth, but the enzyme plays a critical role during sporulation (31). Inactivation of PLD in Caenorhabditis elegans results in viable progeny with no overt phenotype (32). Drosophila also has a single gene, but deficiency again results in a benign phenotype (33). Zebrafish have two PLD genes, and inhibition of PLD1 expression impairs blood vessel development in this organism (34). However, mice lacking PLD1 and PLD2 are viable, fertile, and have very benign phenotypes (5)(6)(7)(8). Thus, PLD may play distinct roles in different species. Accordingly, this information points to a need for further discussion about the true function of the PLD mammalian system.
It has been suggested that PLD1 or PLD2 ablation might be compensated for by the other isoform or other signaling enzymes that increase the formation or decrease the catabolism of PA (7). Thus, it can be assumed that PLDs have dispensable functions during development and in normal mouse physiology. However, PLD1 and PLD2 knock-out mice are protected under pathological conditions (5)(6)(7)(8). Although pharmacological inhibition of PLD1 and PLD2 would be well tolerated, it seems that these observations do not fit the plethora of functions ascribed to these genes well.
Because specific inhibitors for PLD were unavailable until recently, many PLD functional studies have employed primary alcohols to inhibit PLD-dependent generation of PA. However, more recent studies have raised concerns about off-target effects of primary alcohols, even when the tertiary alcohol is used as a control, and emphasized that the role of PLD in cell functions should be reevaluated (12,35,36).
It has been suggested that mice lacking PLD1, but not PLD2, incurred fewer lung metastases than wild-type mice, and thus PLD1 in the tumor microenvironment is critical for tumor growth and metastasis (8). These studies report complementary portions of the role of PLD1 and PLD2 in tumorigenesis and metastasis, indicating that a small molecule capable of inhibiting both PLD1 and PLD2 may be used in cancer therapeutics. Although small molecule PLD inhibitors appear to have some value in cell culture systems, their usefulness for PLD inhibition in vivo in animal models is less well established. Recent studies have shown that pharmacologically and genetically induced PLD inhibition had no obvious side effects (8,11); thus, such a safe therapy could be particularly advantageous in clinical practice. Accordingly, highly selective PLD inhibitors with greater potency need to be developed and analyzed to enable optimized drug delivery and bioavailability.
Triple-negative breast cancers (TNBC) are difficult to treat due to their negative hormone receptor and ErbB2/HER2 sta-tus. In addition, TNBC are aggressive because of their frequent recurrence and high metastatic potential (37). Ceramide transfer protein (CERT) was recently reported to determine the signaling output of the EGF receptor (EGFR/ErbB1), which is upregulated in TNBC (38). Reduced expression of CERT in TNBC is associated with alterations in plasma membrane organization and PLD2 activation (38). Heering et al. (38) suggested that the loss of CERT might trigger aberrant ligand-induced ErbB1 signaling through PLD2 activation, which may be relevant to the design of therapeutic interventions targeting TNBC. Choline kinase-␣ (ChK-␣) is up-regulated in several cancers and a major contributor to increased phosphocholine, which is known as a metabolic hallmark in various cancers (39). Choline generated by PLD activation is used as a substrate of ChK-␣. Recently, these two enzymes were found to be interactive, with depletion of ChK-␣ increasing PLD1 expression and vice versa in breast cancer cells and simultaneous depletion of both enzymes increasing apoptosis (40). Thus, ChK-␣ and PLD1 might be multiple target enzymes in choline phospholipid metabolism of breast cancer. Combined treatment with ChK-␣ inhibitor and PLD inhibitor will be more effective against breast cancer than individual treatments alone.

Genomic Alternation of PLD1 Gene in Cancer
Despite the dramatic advances in the field being driven by genome sequencing, somatic mutations or genomic alternations (rearrangements, copy number variations) at the PLD gene loci have not been widely observed in cancer. The Cancer Genome Atlas (TCGA) indicates that the PLD1 gene is mutated, amplified, and up-regulated in several types of cancers, including lung squamous cell carcinoma, ovarian serous cystadenocarcinoma, breast invasive carcinoma, uterine corpus endometrioid carcinoma, bladder urothelial carcinoma, and hepatocellular carcinoma. Accordingly, although changes in PLD expression may occur in some cancers, the mechanisms involved might be driven by environmental or genetic factors that constitute the primary cause of the disease.
Although PLD inhibition may have a place in cancer therapy, this approach might not be targeting the primary cause of the disease. A possible exception may be the 3q26 amplicon. Indeed, the 3q26 locus is frequently amplified in various types of cancer (ovarian, breast, and non-small cell lung cancers) and is correlated with poor prognosis and an invasive phenotype (41,42). PLD1 was identified as a resident gene in the minimal common amplified region of 3q26 (42). However, the information relating copy number variation at this locus with PLD1 mRNA levels is not yet available in the TCGA. Although PLD1 has not been suggested as a transforming gene because overexpression of PLD1 did not show highly invasive capability in immortalized mammary epithelial cells (41), it is speculated that other genes at 3q26, including PLD1, may contribute to other cancer-related phenotypes. Overexpression of PLD1 may result in cancerous phenotypes via activation of various mitogenic or oncogenic signaling pathways. However, it is important to know how the very high expression levels needed to see these effects relate to changes in PLD1 expression as a result of 3q26 locus amplification.
PKC and SOX2 were recently reported to be coamplified on chromosome 3q26, as well as to cooperate to activate hedgehog (Hh) signaling and drive a stem-like phenotype in lung squamous cell carcinoma (LSCC) (44). Aberrant Hh signaling has been implicated in the initiation and progression of various cancer subtypes. PKC activates the Hh signaling pathway to stimulate oncosphere growth. Interestingly, RNA sequencing (RNA-Seq) data have revealed that knockdown of PKC decreased expression of PLD1 and PLD2, which was increased in the oncosphere of lung squamous cell carcinoma (43). Furthermore, PLD inhibitors suppressed anchorage-independent growth of glioma stem cells (44). Thus, it is possible that PLD plays a role in stem-like cells and regulates the Hh signaling pathway.

MicroRNA Targeting PLD in Cancer
MicroRNA (miR), endogenous non-coding small strands of RNA 19 -25 nucleotides in length, regulate gene expression through binding to the 3Ј-untranslated region (UTR) of target mRNAs, which eventually results in either mRNA degradation or translational repression. MiR is important in the development and progression of various types of cancer. Treatment of hypopharynx cancer cells with paclitaxel increased miR-29b-1*, which putatively mediated the decreased expression of PLD1 (45). miR-203 is reduced in various types of cancer, and acts as a tumor-suppressive microRNA by directly targeting certain oncogenes (46,47). Expression of miR-203 is significantly lower in high World Health Organization (WHO) grade glioma tissues than low WHO grade glioma tissues and normal brain tissues (48). miR-203 inhibited the proliferation and invasion of glioma cells, at least in part, by targeting 3Ј-UTR of PLD2 and suppressing its expression (48). Moreover, PLD2 has been identified as a regulator of cell survival in glioblastoma through its regulation of the prosurvival kinase Akt (44). Thus, PLD2 as a new target of miR-203 may be a potential therapeutic candidate for the treatment of glioblastoma.

Autoregulation of PLD Activity Is Coupled to Selective Induction of PLD1 Expression via NFB
Despite the importance of the duration and amplitude of PLD signaling in cancer, mechanisms that regulate PLD expression remain poorly understood. Conversion of transient activation of signaling pathways to long term changes in cellular behavior is accomplished through changes in gene expression patterns. Despite the extensive information regarding the regulation of PLD activity, the long term effects of PLD activation on PLD expression have not yet been elucidated. It is well known that protein kinase C (PKC) stimulates PLD activity (49). Physiological activators of PLD (PDGF, EGF, or IL-1␤) and phorbol-12-myristate-13-acetate, a PKC activator, enhanced the expression of PLD1, but not PLD2 (50). Phorbol-12-myristate-13-acetate and PDGF enhance binding of two functional NFB sites to PLD1 promoter, suggesting that PLD1 is a downstream target of NFB (50,51). Depletion of PLD1 or PLD2 inhibited PDGF-induced PLD1 expression (51). Additionally, ectopic expression of wild-type PLD1 and PLD2, but not cat-alytically inactive PLD isozymes, increased the mRNA level and promoter activity of PLD1, but not PLD2. PA also increased PLD1 expression via the NFB elements (52). Moreover, inhibitors selective for PLD1 and/or PLD2 (VU0155056, VU0155069, and VU0285655-1) suppressed the expression of PLD1 stimulated by the growth factor, but not that of PLD2 (52). Kang et al. (52) suggested that enzymatic activities originating from both PLD1 and PLD2 are required for selective induction of PLD1 via the positive feedback loop.
Helicobacter pylori induces chronic gastric inflammation and has been defined as a carcinogen leading to gastric cancer (53). H. pylori cagPAI (cag pathogenicity island) is a strain-specific locus that encodes a secretion system responsible for mediation of the translocation of bacterial virulence factor cagA protein into host epithelial cells (54). Infection of gastric cancer cells with cagA-positive H. pylori led to selective induction of PLD1 expression via cagA-dependent activation of NFB (55), and H. pylori and expression of cagA increased the binding of NFB to the PLD1 promoter. Additionally, expression of PLD1 was elevated in H. pylori-infected human gastric cancer tissues. Considering the direct link between cagA and the PLD1 signaling pathway, cagA-positive H. pylori-induced NFB activation might lead to a risk of gastric cancer via upregulation of PLD1 expression. Mutated H-Ras selectively increased PLD1, but not PLD2, via an interaction between Sp1 and its binding site in the 5Ј promoter of PLD1 (56). Because mutated H-Ras is known to increase PLD activity, H-Ras can enhance PLD1 expression via NFB. Autoregulation of PLD activity is coupled to selective induction of PLD1 expression via NFB and contributes to cancer progression through increased MMP up-regulation and invasion (Fig. 1). Thus, PLD inhibitor may be useful for cancer therapy via inhibition of both PLD1 expression and PLD activity. However, it cannot exclude the possibility that such double inhibition would affect other key and necessary functions, such as membrane maintenance.

Ewing Sarcoma Fusion Protein, EWS-Fli and Fli, Selectively Increase PLD2 Expression via the ETS Binding Motif
The reduced EWS-Fli-1 expression down-regulated PLD2 protein expression, resulting in repression of PDGF-mediated cell growth signaling in Ewing sarcoma (EWS) cells (57). EWS-Fli1, which is a fusion gene resulting from a chromosomal translocation t (11;22, q24;q12) found in EWS and primitive neuroectodermal tumors, encodes a transcriptional activator and promotes cellular transformation (58). Fli-1 genes are members of the erythroblast transformation-specific (ETS) family of transcription factors. Overexpression of EWS-Fli or Fli increases expression of PLD2 (60). PLD2, which is mainly involved in the increased PLD activity in EWS tumors (59), is a novel target of EWS-Fli-1 as well as Fli-1 transcription factor on the ETS binding site that functions as the principal promoter of the PLD2 gene. Other ETS transcription factors such as PU.1, ETS1 and ETS 2 could induce PLD2 gene expression, suggesting PLD2 as a general target of the ETS protein family. Thus, EWS-Fli1 might play a role in regulation of tumor proliferationsignaling enzymes via PLD2 expression in Ewing sarcoma cells. TC135 cells, an EWS cell line, showed predominant expression of PLD2, whereas PLD1 protein and gene expression were very weak. Although PLD1 promoter also contains six candidate ETS binding site motifs within the 5Ј promoter region, PLD1 was not responsive to EWS-Fli-1 or Fli-1 transcription factors. The condition in histone acetylation around the PLD1 genome could be one possible mechanism of the insensitivity of PLD1 transcription to members of the ETS family of proteins. Kikuchi et al. (59) suggested that trichostatin A, an inhibitor of histone deacetylase, could increase PLD1 expression in EWS cells. However, expression of PLD1 gene by histone deacetylase inhibitor needs further analysis. EWS-Fli1 plays a role in PDGFinduced signaling by regulating PLD2 expression in EWS cells (57,59). Based on our finding that PDGF-induced PLD activation is associated with PLD1 induction in breast cancer cells, regulation of the expression of each PLD isoform gene might be cell type-specific. Thus, differential expression of PLD isozymes might be mediated via a distinct signaling pathway or through different transcription factors dependent on the physiological status of cells or cell type (Fig. 1).

Wnt/␤-Catenin Signaling Up-regulates Expression of Both PLD1 and PLD2 Isozymes via TCF
Aberrant activation of Wnt/␤-catenin signaling followed by hyperactivation of T-cell transcription factor (TCF)-dependent transcription of target genes is linked to various cancers. Recently, both PLD1 and PLD2 were identified as target genes of Wnt/␤-catenin signaling via binding of TCF4/␤-catenin in their promoters (21,22,60). ␤-Catenin and TCF4 were shown to elevate expression and activity of PLD isozymes, whereas PLD isozymes were found to act as a positive feedback regulator of Wnt signaling, which subsequently promotes Wnt-driven anchorage-independent growth of colorectal cancer cells (Fig.  1). PLD enhances expression of the MMP-2 gene by increasing the DNA binding activity of NFB and Sp1, and then promotes glioma cell invasion (27). Thus, it is believed that PLDs can promote tumor phenotype independence of TCF-induced genes. PLD activity is required for ␤-catenin/TCF-4 transcriptional activation, which occurs via increased formation of the ␤-catenin/TCF-4 complex (21). A PLD inhibitor that suppresses binding of TCF-4 to ␤-catenin may confer a clinical benefit during treatment of Wnt/␤-catenin-driven malignancies. Further studies will enhance our understanding of the specific mechanisms by which increased expression of PLD contributes to the development and progression of cancer.

Selective Induction of PLD1 by Inflammatory Cytokines
Expression of PLD1 has been shown to be up-regulated in a variety of inflammatory and autoimmune diseases, including acute pancreatitis (61), peritonitis (62), brain ischemia (63), rheumatoid arthritis (64), and Alzheimer disease (65). Thus, PLD may be involved in various inflammatory diseases. It has been reported that PLD2 expression is not affected in activated monocytes, and it appears to be a constitutive enzyme in circulating monocytes (66). Conversely, PLD1 is an inducible protein selectively induced during cell activation (52), suggesting different roles of PLD isoforms in mononuclear phagocytes. Therefore, selective induction of PLD1 by pathogen activation might play a relevant role in the late phase of immune response and in chronic inflammation. Microarray analysis during the transcriptional profiling of peripheral blood mononuclear cells associated with acute pancreatitis demonstrated that PLD1 was the top up-regulated gene (61), suggesting that it is a suitable molecular marker of acute pancreatitis. Genetic silencing of PLD1 effectively blocked the cytokine/chemokine production, vascular permeability, and leukocyte recruitment triggered by tumor necrosis factor ␣ (TNF␣) in an in vivo peritonitis model (62). Based on these findings, blockade of PLD1 has potential clinical implications for improving acute inflammatory conditions. PLD1 also plays a critical role in IL-1␤-induced synoviocyte activation and the progression of chronic inflammatory arthritis (64). PLD1 expression and activity in the synovium or synoviocytes were found to be higher in rheumatoid arthritis (RA) patients than in osteoarthritis patients. RA is a chronic inflammatory disease characterized by the progressive destruction of articular cartilage and bone in the chronic phase. PLD1 expression correlates well with the severity of RA; thus, abnormal up-regulation of PLD1 may contribute to the pathogenesis of chronic arthritis. Increased expression of PLD1, but not PLD2, was triggered by proinflammatory cytokines such as IL-1␤, TNF␣, and IL-6 in RA synoviocytes, but not in osteoarthritis synoviocytes. IL-1␤ induced PLD1 expression via the NFB and ATF-2 pathways. IL-1␤ enhances proinflammatory mediators, angiogenic factors, and proliferation and migration via PLD1-mediated NFB or HIF-1␣ activation and the FoxO3a inactivation signaling pathway in RA synoviocytes ( Fig. 2A). Osteoclastogenesis plays an important role in joint destruction in RA rheumatoid arthritis. It was recently suggested that osteoclastogenesis may occur through the expression of PLD1-induced receptor activator of nuclear factor-B ligand (RANKL) in rheumatoid synoviocytes stimulated by IL-15 (67). Moreover, the PLD signaling pathway is involved in the lung cancerderived IL-8-induced increase in osteoclastogenesis (68). Thus, regulation of PLD expression in RA patients may provide a new treatment strategy for inhibition of the bone destruction caused by this disease, as well as RA. IL-1␤ or TNF␣ has been reported to induce PLD1 expression by increasing the expression of cal-cium/calcineurin-regulated NFATc2/NFAT1 (69). It has also been suggested that PLD1 promoter contains putative NFAT binding sites. Taken together, these results indicate that proinflammatory cytokines selectively induce PLD1 expression via NFB, ATF2, or NFATc2 transcription factors ( Fig. 2A).

Role of PLD2 Expression in Inflammatory Cytokine-induced Migration of Immune Cells
IL-8, a potent chemoattractant, increases activity and expression of PLD2 and enables PLD2 to function as a facilitator for migration of immune cells (70). Leukocytes infiltrate into tissues at the site of inflammation and stay for prolonged periods of time until the insult is cleared, or remain longer in the case of chronic inflammation. Speranza et al. (71) recently demonstrated that, in the early stages of inflammation, activation of leukocytes by MCP-1/CCL2 induces the binding of Rac2 to PLD2 in the cytosol, which leads to stimulation of chemotaxis. Later in the process, Rac2 translocates into the nucleus and activates Sp1 via an unknown mechanism and then represses PLD2 expression. The subsequently reduced PLD2 activity inhibits the ability of cells to undergo chemotaxis. Although MCP-1 can activate ␤-catenin and increase PLD2 expression at earlier times in the process, Rac2 dominates at later times, exerting a negative effect on PLD2 expression. These findings suggest that if this negative effect is sustained, it could induce prolonged immobilization of leukocytes from normal functionality to the pathological stages of chronic inflammation (Fig. 2B).

Compounds Down-regulating PLD Expression
PLD isozymes are downstream transcriptional target molecules of NFB, Sp1, TCF4, EWS-Fli or Fli, ATF-2, and NFATc2, which contribute to carcinogenesis or inflammation. As a result, compounds suppressing PLD expression might be useful in sensitizing resistant cancers and reducing inflammation. PLD inhibition abolishes growth factors and inflammatory cytokine-induced PLD1 expression, which induces proliferation, migration, invasion, angiogenesis, and chronic autoimmune inflammatory arthritis; therefore, selective PLD inhibitors may have therapeutic potential as anti-cancer and anti-inflammatory agents. Several compounds are known to suppress inflammation and proliferation of cancer cells via down-regulation of PLD. Triptolide, a natural product used in traditional Chinese medicine, has a myriad of therapeutic uses against inflammation and autoimmune disease (72). Rebamipide, a mucosal protective agent, has been used clinically for treatment of gastritis and peptic ulcers (73). Triptolide and rebamipide suppressed the expression and activity of both PLD1 and PLD2 isozymes in various cancer cells, which was followed by inhibition of the proliferation of cancer cells (74,75). Triptolide-or rebamipide-mediated PLD1 repression was due to inhibition of NFB activity. However, it is likely that the compounds inhibit PLD2 expression via a pathway distinct from that of PLD1 because NFB is not responsible for PLD2 expression. PLD isozymes might be a potential therapeutic target of triptolide and rebamipide. Triptolide has been shown to interfere with a number of transcription factors at the molecular level, including NFB, p53, NFAT, HSF-1, and xeroderma pigmentosum group B protein (XPB) (76 -79), whereas rebamipide might contribute to the anti-tumorigenic effect of gastric cancer cells via inhibition of the H. pylori cagA-NFB-PLD1 signaling pathway (55). Quercetin and caffeic acid phenethyl ester are natural compounds that have also been shown to exert a broad range of pharmacological effects, including anti-oxidant and anti-inflammatory activities. Quercetin and caffeic acid phenethyl ester abolished PLD1 expression via inhibition of NFB transactivation, leading to inhibition of the invasion and proliferation of glioma cells (80,81). These compounds mentioned above are known to indirectly inhibit PLD activity, but are inadequate as chemical probes because they have many molecular targets in various signaling pathways. Novartis Pharmaceuticals identified a psychotropic agent, halopemide, by high throughput screening, which was used as a starting point for PLD inhibitor development (10). Brown and colleagues (9) developed highly selective PLD1 and PLD2 inhibitors based on this halopemide scaffold. PLD-selective inhibitors will undoubtedly lead to new opportunities in investigations of PLDs and enable development of potential and therapeutic drugs for treatment of diseases related to PLD dysregulation such as cancer and inflammation that function by suppressing both the activity and the expression of PLD.

Conclusion and Future Perspectives
In the last few years, the function of mammalian PLD has been investigated in greater detail because of the development of selective PLD inhibitors and PLD knock-out mice. As a result, we now have a much better molecular understanding of the multiple roles of PLD. Increasingly, PLD has emerged as a critical modulator of cancer and inflammatory responses, and its expression and activities have been linked to multiple diseases, including cancer, inflammation, neurodegeneration, hypertension, thrombosis, and ischemic stroke. However, key questions still remain. For example, we still have a poor understanding of the molecular mechanisms differentially regulating PLD isozyme expression and of the precise role of each isoform. Additionally, it is not clear whether PLD isozyme is regulated in a coordinated fashion or separately in cancer and inflammation. However, future design of the next generation of isoformspecific therapeutics for the treatment of cancer and other diseases looks promising.