Selective Activation of an Apoptotic Retinoid Precursor in Macrophage Cell Lines*

Advances in the understanding of the retinoid signaling mechanism has allowed the discovery of highly selective retinoids that activate only one specific receptor class, subtype, or signaling pathway. These novel compounds lack certain of the common retinoid toxicities and therefore suggest promising new approaches for therapeutic applications. We describe here a new compound, 6-[3-(1-adamantyl)-4-hydroxyphenyl]-2-naphthalene carboxylic acid methyl ester (MX84), that is selectively activated in macrophages, leading to killing of only macrophage monocyte type cells in vitro. We provide evidence that MX84 is an inactive precursor that is converted into an active apoptosis-inducing retinoid in macrophages. The macrophage activity is also secreted, and our data suggest that the secreted activity is a phospholipase D type activity. Our observation may lead to the development of molecules that are highly macrophage-selective apoptosis inducers in vivo and that could represent important novel therapeutics against diseases caused by excessive macrophage activity.

Advances in the understanding of the retinoid signaling mechanism has allowed the discovery of highly selective retinoids that activate only one specific receptor class, subtype, or signaling pathway. These novel compounds lack certain of the common retinoid toxicities and therefore suggest promising new approaches for therapeutic applications. We describe here a new compound, 6-[3-(1-adamantyl)-4-hydroxyphenyl]-2-naphthalene carboxylic acid methyl ester (MX84), that is selectively activated in macrophages, leading to killing of only macrophage monocyte type cells in vitro. We provide evidence that MX84 is an inactive precursor that is converted into an active apoptosis-inducing retinoid in macrophages. The macrophage activity is also secreted, and our data suggest that the secreted activity is a phospholipase D type activity. Our observation may lead to the development of molecules that are highly macrophage-selective apoptosis inducers in vivo and that could represent important novel therapeutics against diseases caused by excessive macrophage activity.
Retinoids exert their multiple biological effects through interaction with specific nuclear proteins, the retinoid receptors. These regulatory proteins belong to a large group of transcription factors, the steroid/thyroid hormone nuclear receptor superfamily. Two classes of retinoid receptors have been identified, the RARs 1 and RXRs, each of which has three subtypes, ␣, ␤, and ␥, and several isoforms. Cell type-and developmental stage-specific expression patterns of these receptor subtypes and isoforms are believed to contribute to distinct cell type and developmental stage gene expression patterns that are regulated by retinoids. Further amplifying the diversity of retinoidregulated gene expression patterns, these receptors bind as RXR/RAR heterodimers or RXR homodimers to specific DNA sequences in the promoter region of retinoid-responsive genes or interact with other transcription factors, such as activator protein-I (reviewed in Ref. 1).
Although natural retinoids are currently used in the treatment of certain skin diseases and in the treatment of acute promyelocytic leukemia (2)(3), their medical applications are restricted by multiple side effects, caused by the intimate and universal role of retinoids in regulation of gene expression, orchestrating cell proliferation, differentiation, and embryonic development (4 -6). This broad activity of the natural retinoids is due to their fairly indiscriminate binding to and activation of the various receptor subtypes. Thus, all-trans-retinoic acid (tRA) activates all RAR subtypes, whereas 9-cis-RA activates all RARs and, in addition, the RXRs. The more recent findings that some synthetic retinoids exert their biological effects by selective binding or activating only specific RAR subtypes (either ␣, ␤, or ␥) (7)(8)(9)(10) or only RXRs (11)(12)(13)(14) were encouraging. The first RAR subtype-selective compound, Adapalene, an RAR␤/RAR␥-selective retinoid, has now been approved in many countries for the topical treatment of acne, showing fewer side effects in this application than tRA. RXR-selective compounds were recently found to be effective in animal models for the treatment of type II diabetes (15).
In addition to receptor-selective compounds, synthetic retinoids could also be found to function in some of the retinoid signaling pathways but not in others. For instance, anti-AP-1selective compounds show transcriptional repression but not transcriptional activation activities (16). These types of compounds lack many of the typical retinoid activities, and thus toxicities, and may find applications as specific inhibitors of certain disease-associated genes, activated by the transcription factor activator protein-I. Another interesting class of novel retinoids has been recently identified that induce apoptosis (17)(18)(19)22). These molecules have been shown to specifically bind or transactivate RAR␥ (20 -22). The first apoptosis-inducing retinoids described, N-(4-hydroxyphenyl)-all-trans-retinamide and CD437 (Refs. 23-26 and 27-29, respectively), are, however, relatively indiscriminate apoptosis inducers and are thus very toxic in vivo at effective concentrations. 2 However, some of us showed recently that new apoptosis-inducing retinoids can be identified that show cell type selectivity (22). Importantly, one of the molecules identified by an in vitro screen was effective against a human non-small cell lung cancer xenograft in vivo, at concentrations at which it did not induce major side effects (22). * This work was supported by Grant CA55681 from the National Institutes of Health (to M. P.). 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.
§ Supported by a postdoctoral fellowship from Galderma Research, Inc.
Using the same cell-based antiproliferation/cell killing screen, we also discovered one molecule, MX84, that induced cell death in monocyte/macrophage-derived cell lines (mMn1, mMn2, and mMn3) but was inactive against more than 30 other cell lines, including solid tumor-derived cell lines. We show here that MX84 is inactive as a transcriptional activator in CV-1 cells but at the same concentration can activate gene transcription in macrophage type cells. Interestingly, in these latter cells, the compound is converted from an inactive retinoid precursor into an active retinoid that induces apoptosis. The monocyte/macrophage type cell lines investigated here were also found to secrete a protein into the medium that allows the activation of the retinoid. This cell-specific enzymatic conversion of an inactive retinoid precursor into an apoptosis-inducing compound may be exploited as a scheme for the developing of macrophage cell type-selective retinoids with apoptotic and/or other desirable activities.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfections, and MTT Test-CV-1, H661, J774A.1, LNCaP, and RAW264.1 cell lines used in this study were maintained in the recommended growth medium, either Dulbecco's modified Eagle's medium or RPMI 1640 medium (Irvine Scientific), supplemented with fetal calf serum (Irvine) and penicillin and streptomycin (Life Technologies, Inc.).
Transient transfection of CV-1 cells was performed as described elsewhere (30). CAT activity was determined using a CAT ELISA colorimetric enzyme immunoassay (Boehringer Mannheim) as recommended by the manufacturer. CAT activities were corrected for ␤-galactosidase activities.
RAW264.1 macrophage cells for transient transfection were maintained in RPMI 1640 medium (Irvine Scientific) supplemented with 5% fetal calf serum (Irvine) with penicillin and streptomycin (Life Technologies, Inc.). Cells were harvested and resuspended at approximately 2 ϫ 10 7 cells/ml in RPMI 1640 medium supplemented with 10% fetal calf serum. The cells were transfected with the pECE expression plasmids for RAR␥, RXR␣, and the TREpal-tk-CAT reporter. Using an electroporation method, aliquots of cells were transfected at room temperature with 30 g of plasmid DNAs per ml (5 ϫ 10 Ϫ6 g of the plasmid for RAR␥, 5 ϫ 10 Ϫ6 g of the plasmid for RXR␣, and 20 ϫ 10 Ϫ6 g of TREpal-tk-CAT) in a 0.4-cm electroporation cuvette at 300 V, 1800 microfarads using a BTX Pulser. After electroporation, cells were cultured for 24 h and grown for 12 h in the presence of either tRA, MX84, or CD437. CAT activity was determined using CAT ELISA, a colorimetric enzyme immunoassay as recommended by the manufacturer. CAT activities were normalized to total cell protein concentration to correct variations in cell survival caused by toxic effects of both DNAs (31) and retinoids.
For MTT tests, the J774A.1 and RAW 264.1 cells were seeded into 96-well plates at 6 ϫ 10 3 cells/well. Other cell lines were seeded at 1 ϫ 10 3 cells/well. Twenty-four hours later, either tRA or MX84 was applied at the desired concentrations (10 Ϫ11 to 10 Ϫ6 M). Following 6 days of incubation with the retinoids, MTT tests were performed using a MTT kit (Boehringer Mannheim).
Apoptotic Cell Death Assay-Apoptotic cell death detection was performed using the Cellular DNA Fragmentation ELISA kit (Boehringer Mannheim), essentially as recommended by the manufacturer. 3 ϫ 10 4 RAW 264.1 cells/well were used for the assay. RAW 264.1 cells were exposed for 15 h to either 10 Ϫ6 M MX84 or 10 Ϫ6 M CD437. 10 Ϫ6 M staurosporine, a non-retinoid inducer of apoptosis, was used as a positive control. Peptide fluoromethyl ketones ZVAD-fmk and ZDEVD-fmk were obtained from Enzyme Systems Products and were applied at 50 M concentrations simultaneously with the retinoids.
Thin-layer Chromatography-Confluent flasks of J774A.1 and RAW264.1 cells were incubated for 18 h in 5 ml of growth medium (Dulbecco's modified Eagle's medium and RPMI 1640, respectively) in the presence of 10 Ϫ6 M MX84. The growth medium was then collected, floating cells were removed by centrifuging, and 2 ml of methanol/ methylene chloride (10:90) were added. After vigorous vortexing, the organic phase was separated by centrifugation, and 0.8 ml of the organic phase was concentrated to 10 l using a Speed Vac SC100 (Savant Instruments, Inc.) and analyzed by TLC. For the assay, 1 l of the concentrate was loaded onto 2 ϫ 7-cm TLC plate silica gel 60 (EM Separations Technology, Gibbstown, NJ) with a preconcentration zone. As a running solvent, methanol/methylene chloride (10:90) mix was used. After a run of 3 min, the plates were analyzed with a Shimadzu CS-9301 PC dual wavelength flying spot scanning densitometer.
Enzymatic Assays and Inhibition Studies-To obtain a concentrated solution of the macrophage-secreted high molecular weight activation enzyme(s) (converting MX84 into an active compound), RAW264.1 cells were washed three times with PBS and then incubated in PBS for 6 h at 37°C, 6% CO 2 . The PBS was collected, floating cells were removed by spinning, and the supernatant was concentrated using an Ultrafree-15 Centrifugal Filter Device, Biomax-100K NMWL membrane (Millipore Corp., Bedford, MA).
All commercially available enzymes used in this study were purchased from Sigma.
Phospholipases A2 from bovine pancreas (50 units) and porcine pancreas (50 units) were incubated with MX84 as above in 20 mM Tris-HCl pH 8.0. Phospholipase A2 isoenzymes A2-I and A2-III from Laticauda semifasciata (50 and 40 units, respectively) were incubated for 20 h at 25°C in 0.2 ml of 20 mM Tris-HCl, pH 8.9, containing 5 ϫ 10 Ϫ6 M MX84. Phospholipase B from Vibrio species (10 units), phospholipase D type I from cabbage (100 units), phospholipase D type III from peanut (100 units), phospholipase D type IV from cabbage (100 units), and phospholipase D type VI from Streptomyces chromofuscus (100 units) were tested for the esterase activity toward MX84 as above, except that buffers were used that were either 20 mM Tris-HCl, pH 8.0, or 20 mM MES, pH 5.6, depending on the pH value recommended for the enzymes. Phospholipases C from B. cereus (20 units) and from C. perfingens (20 units) were incubated with MX84 as above in 20 mM Tris-HCl pH 7.3.
Cholesterol esterases from bovine pancreas and from Pseudomonas fluorescens (5 units of each) were incubated with MX84 for 20 h at 37°C in PBS in the presence of 0.5% taurocholate.
For inhibition studies, 0.5 ml of PBS containing 10 l of the concentrated high molecular weight macrophage-secreted proteins was preincubated for 3 h at 37°C in the presence of different inhibitors. Then, MX84 was added into each tube to a concentration of 2 ϫ 10 Ϫ6 M, and the mixes were incubated for 20 h at 37°C. Following the incubation, the mixtures were vigorously vortexed with 0.3 ml of methanol/methylene chloride (10:90) mix, and the organic phase was collected, concentrated, and analyzed by TLC as described above.
The following phospholipase inhibitors were purchased from Alexis Biochemicals (San Diego, CA). The antitumor drug suramin, which was reported to inhibit PLD activity (32), was tested at 100 M, 1 mM, and 5 mM concentrations. The antiviral and antitumoral xanthogenate compound D609, known as an inhibitor of phospholipase C (PLC) and PLD (32,33), was tested at 100 M, 300 M, and 1 mM. The aminosteroid U73122, a PLA2 and PLC inhibitor, which was described also as a PLD inhibitor (32,34), was tested at 1, 5, and 100 M concentrations. A potent inhibitor of PLC, the antitumor ether lipid sn-ET-18-OCH3, which also was reported to interfere with PLD activity (35,36), was tested at 10, 50, and 100 M.
Divalent ion (Ca 2ϩ and Mg 2ϩ ) requirements for the enzymatic activity were studied in 20 mM Tris-HCl buffer, pH 7.5, using 1, 5, and 10 mM concentrations of the cations.

Identification of a Macrophage-selective Retinoid
In the last few years, novel selective retinoids and retinoidrelated molecules have been identified that show potential for medical applications. A particularly interesting class of retinoids is molecules that induce apoptosis. However, the first such molecules described appear to induce apoptosis fairly indiscriminately. 2 Recently, we were able to demonstrate that selective apoptosis inducers with good effectivity/toxicity profiles in vivo could be identified (22). Using a high throughput biological activity assay, we analyzed more than a thousand retinoids for their ability to inhibit cell growth or induce cell killing on a large panel of cell types (22). In brief, cell cultures were incubated with 2 ϫ 10 Ϫ6 M concentrations of the various retinoids for periods of 5 days, after which the percentages of live cells were determined using a colorimetric (MTT) assay. Among more than 1000 retinoids analyzed, we found one compound, MX84, that under these conditions effectively killed macrophage/monocyte type cells but did not kill more than 40 other cell lines investigated. An excerpt of the screening results is shown in Fig. 1, in which MX84 is compared with the strong apoptosis inducer CD437 and the natural retinoid tRA. As can be seen, CD437 indiscriminately induces cell growth arrest/cell killing in all cell types shown, whereas MX84 is active only in monocyte/macrophage type cell lines J774A.1 and RAW264.1 under the conditions used. tRA showed a significant growth inhibitory effect only on few of the cell lines, including F-9 cells, as well as HL-60 cells. In these cell types, tRA is known to induce differentiation, which includes cell growth arrest. Interestingly, MX84 appears not to be a differentiation-inducing agent because it shows no activity in F-9 cells and PC19 cells. To verify the screening results, titration experiments were carried out comparing the growth inhibitory/cell killing effects of MX84 and tRA at various concentrations on several cell lines. As can be seen (Fig. 2, a and b) at 10 Ϫ6 M, MX84 is a very effective inhibitor of two monocyte/macrophage cell lines J774A.1 and RAW264.1 but not of the non-small cell human lung cancer cell line H661 or the prostate cancer cell line LNCaP (Fig. 2, c and d, respectively). On this latter cell line, tRA showed a weak inhibitory effect at 10 Ϫ6 M, consistent with published results, whereas MX84 did not. Thus, under the in vitro conditions used, MX84 is a selective inhibitor of monocyte/ macrophage type cell lines.

MX84 Induces Apoptosis in Macrophage Type Cells
To characterize the mode of action of this new retinoid in macrophages, we applied 10 Ϫ6 M concentrations of either MX84 or the apoptosis-inducing retinoid CD437 to RAW 264.1 cells. After incubation with MX84, the macrophage cultures demonstrated morphological changes characteristic of cells undergoing apoptosis (data not shown), such as formation of membrane-bound vesicles and cell shrinkage due to condensation of cytoplasm; these were similar to the changes observed in cells treated with CD437 or with staurosporine, which are known inducers of apoptosis. Furthermore, using a cellular DNA fragmentation ELISA kit (Boehringer Mannheim), we found that MX84 induced apoptosis in these cells almost as efficiently as CD437 (Fig. 3), whereas tRA showed minimum basal activity (not shown).
Peptide fluoromethyl ketones ZVAD-fmk, an inhibitor of caspases with selectivity toward caspase 1 (32), and ZDEVDfmk, a specific inhibitor of caspase 3, completely protected the macrophages from MX84-induced or CD437-induced apoptotic cell death (data not shown). We have previously shown that retinoid-induced apoptosis/cell killing can be inhibited by specific caspase inhibitors (38). Thus, our observation here that the antimacrophage activity of MX84 is inhibited by caspase inhibitors provides further proof that MX84 is an apoptosisinducing retinoid in macrophages.

Cell Type-specific Conversion of MX84 into a Transcriptional Activator
It is generally believed that retinoids require a carboxyl terminus to efficiently bind and transactivate their nuclear receptors. Consistent with this, we observed that MX84 was very inefficient in activating a retinoid-responsive reporter gene in CV-1 cells, when compared to tRA, a cell line commonly used to measure transactivation capacities of nuclear receptor ligands. In contrast, CD437, known to be a potent transcriptional activator (39), induced the retinoid-responsive reporter gene expression even more strongly than tRA when applied at same concentration in CV-1 cells (See Fig. 4a). When we carried out transactivation studies with the macrophage type cell line RAW264.1, we observed that MX84 behaved similarly to CD437, leading to induction of a retinoid-responsive reporter gene in the presence of RAR␥. Thus, although MX84 is inactive in CV-1 cells in terms of induction of apoptosis and in functioning as a transcriptional activator, it gains both functions in macrophage type cells (Fig. 4b).

Conversion of MX84 into an Active Retinoid
At least two different hypotheses appear reasonable to explain the cell type-specific activity of MX84: (i) the retinoid signaling machinery differs in these cells from most other cell types, containing cofactors/coactivators that allow MX84 to be active, or (ii) MX84 is enzymatically converted in these cells into an active compound. In Fig. 5, the formulas of tRA, MX84, and CD437 are compared. It is apparent that the methyl ester MX84 could be converted into a free acid end compound. This free acid would in fact be identical to the apoptosis-inducing retinoid CD437, which functions in macrophage type cells as an inducer of apoptosis and as a transcriptional activator.
To examine this hypothesis, we incubated 10 Ϫ6 M MX84 overnight with macrophage cell lines. When lysates of these cells were subsequently analyzed by thin-layer chromatography (TLC), we observed that a large portion of MX84 was converted to a compound migrating identical to CD437 on TLC plates (Fig. 6). In contrast, incubation of MX84 with the nonmacrophage cell line CV-1 resulted in very little conversion, and no conversion was seen with H661 cells (not shown).
To investigate whether the MX84 conversion could be associated with secretory functions of macrophage type cells, we added 10 Ϫ6 M MX84 to "macrophage-conditioned" medium, i.e. medium in which the macrophage cell lines had been grown overnight. Both macrophage cell lines, J774A.1 and RAW 264.1, were found to secrete the MX84 conversion activity into the medium (Fig. 6). Thus, MX84 is activated in macrophage type cells, and in addition, this activating activity is secreted by these cells.

Specificity of the Macrophage Activation Enzyme(s)
We observed that only macrophage/monocyte type cells excreted an MX84 activation activity into the growth medium. We took advantage of this observation to analyze the specificity of the excreted "esterase" activity in some detail. This was of interest because defining the selectivity of the MX84 converting/activating activity should allow the future design of compounds that are most selectively activated by macrophages in vivo. In this respect, it is important to note that a considerable variety of enzymes contain or can be expected to contain "esterase" activities, which would allow for the conversion of the MX84 (a methyl ester) into a carboxylic acid. Potential converting enzymes include certain proteases and phospholipases, as well as cholesterol esterases and lipases. To test for the presence of these enzymes, we analyzed MX84 conversion in RAW264.1 supernatants. For this, the cells were cultured for 6 h in PBS, after which the secreted proteins were concentrated using an Ultrafree-15 centrifuge device.
Protease Inhibitors-Use of known effective concentrations of various protease inhibitors (Table I) revealed that the activity that converts MX84 in RAW 264.1 cells could be completely inhibited by the thiol-blocking agents pCMPS and DTNB. The activity was also inhibited by the amino peptidase inhibitor bestatin, by the chymotrypsin inhibitor chymostatin, and by the serine protease inhibitor Pefabloc SC, suggesting a process involving the conversion of protein precursor to the active form of protein. No inhibition of the esterase activity was observed either with effective concentrations of several other inhibitors of serine, cysteine, and aspartate proteases, metalloendopeptidases, and metalloproteases, nor with turkey egg white trypsin inhibitor. Sodium fluoride, even at the highest concentration tested (10 mM), was unable to inhibit the esterase activity, as well as the C-1 esterase inhibitor and PMSF, (unless very high concentrations of the latter were used (33 and 57% inhibition by 3 and 10 mM PMSF, respectively). The esterase activity was also not inhibited by the caspase inhibitors ZVAD-fmk and ZDEVD-fmk. Overall, these data suggest that a cysteine residue is more likely to be part of the active site than a serine residue. In addition, because certain proteases, such as cys- teine and serine protein inhibitors, did not inhibit the conversion of MX84, proteases appear to be unlikely candidates.
Phospholipase Inhibitors-To evaluate possible phospholipase involvement in MX84 activation in macrophages, specific phospholipase inhibitors were tested. Data obtained with several concentrations of specific phospholipase inhibitors are summarized in Table II. The conversion activity was not inhibited by 100 nM MAFP, an inhibitor that is specific for the calcium-dependent phospholipase A2 (PLA2), or by 100 nM HELSS, which is specific for the calcium-independent PLA2. Only when these compounds were applied at 1 M concentra-tions could some inhibition be observed. The activity was affected by the PLD inhibitor 1,10-phenanthroline; however, it was not inhibited by 2,3-diphosphoglycerate, another known PLD inhibitor. In contrast, the antitumor drug suramin, which  has been reported to inhibit PLD activity (32), was found to inhibit the macrophage activity in a dose-dependent manner. The antitumor ether lipid sn-ET-18-OCH3, which has also been reported to interfere with PLD activity (35,36), was also a potent inhibitor of PLC. The aminosteroid U73122, an inhibitor of PLA2, PLC, and PLD (32,34), also suppressed the conversion of MX84 into CD437. The antiviral and antitumoral xanthogenate compound D609, known as an inhibitor of PLC and PLD (32,33), was effective only at highest concentrations tested, also suggesting the possible involvement of a PLD-like activity. Further important information was obtained when we carried out experiments with L-␣-phosphatidylcholine and L-␣lysophosphatidylcholine, which are both natural substrates for phospholipases, and as such, when supplied in excess over MX84, they could serve as competitive inhibitors. When a ϳ10fold molar excess of L-␣-phosphatidylcholine was present in the reaction mix, the conversion of MX84 into CD437 was inhibited up to 81%. L-␣-lysophosphatidylcholine was a less effective inhibitor for the reaction, demonstrating 36% inhibition under the same conditions, confirming that a PLD indeed might exercise esterase activity on MX84 in RAW 264.1 cells.
To obtain further information, additional studies with commercially available hydrolases were carried out. For instance, we found that under conditions described above, the phospholipase D type III from peanut and phospholipase D type IV from cabbage both were active toward MX84 (41 and 20% conversion, respectively, as was determined by Shimadzu CS-9301 PC dual wavelength flying spot scanning densitometer), as well as phospholipase B from Vibrio species (14%). However, phospholipase D type I from cabbage and phospholipase D type VI from S. chromofuscus were not able to cleave the ester bond in MX84. Phospholipases A2 from bovine and porcine pancreas, phospholipase A2 isoenzymes from L. semifasciata, and phospholipases C from B. cereus and from C. perfingens also were not active toward MX84, nor toward cholesterol esterases from bovine pancreas and P. fluorescens. An esterase from porcine liver known to have chymotrypsin-like activity completely cleaved the ester bond of MX84, whereas trypsin was inactive. A lipase from Candida rugosa did convert MX84 effectively into CD437 (32% conversion). Whereas a lipase from human pancreas failed to convert the compound. Taken together, a phospholipase D is the most likely candidate to represent the MX84 converting activity excreted by macrophages. DISCUSSION With few exceptions, the cell lines studied (80 -90% of cell lines) were found to be sensitive to 10 Ϫ6 M concentrations of the RAR␥-selective, apoptosis-inducing retinoid CD437. In contrast, only monocyte/macrophage cell lines were sensitive to 10 Ϫ6 M of MX84. In addition to the important role of these cell types in the immunological defense, excessive macrophage activity has been associated with a number of widespread diseases such as arthritis, multiple sclerosis, septic shock, and so forth. Control of macrophage populations could therefore provide an important approach for the treatment and management of those diseases. We describe here a compound, MX84, that from among more than 1000 compounds tested showed highly selective antimacrophage activity in vitro. We analyzed the mechanisms by which MX84 can exert this highly selective killing activity. We observed that in macrophage type cells, MX84 behaves as a transcriptionally active retinoid, but it is inactive in other cell types, such as CV-1 cells. In addition, MX84 induces apoptosis in macrophages. Chemical analysis of macrophage extracts revealed that MX84 is converted into a compound that shows migrations on TLC plates identical to those of the apoptosis-inducing compound CD437. This ob-served conversion is not by itself that surprising, because chemically, MX84 is a methyl ester of CD437 (see Fig. 5). What is surprising is the high degree of cell type selectivity for this conversion observed in vitro. The high degree of selectivity, however, is not expected to hold in vivo, because esterases are quite ubiquitous (consistent with our preliminary in vivo results). To eventually be able to obtain a compound that is selectively active against macrophage type cells in vivo, it is not only important to analyze the mechanism by which MX84 exerts its activity in these cells, but also to determine the nature of the MX84 activating enzymes. Our observation that all macrophage type cell lines secreted the specific activity may be of importance and is consistent with previous reports and the known characteristics of these cell types.
Our inhibition studies revealed that the macrophage-secreted esterase activity was completely inhibited by thiol-blocking agents, such as pCMPS and DTNB. In addition, the enzymatic activity was inhibited in part by such agents as the amino peptidase inhibitor bestatin, the chymotrypsin inhibitor chymostatin, and the serine protease inhibitor Pefabloc SC (Table I). PMSF, however, did not inhibit the activity (unless used at very high concentrations). Taken together, these data suggest the possible involvement of a cysteine residue(s) in a single active site, and, less likely, the involvement of a certain serine residue(s). Noteworthy, however, is that no inhibition of the esterase activity was observed with several other inhibitors for cysteine and serine proteases, such as antipain-dihydrochloride, aprotinin, E-64, leupeptin, and turkey egg white trypsin inhibitor, as well as with peptide inhibitors for caspases.
Studies with commercially available enzymes indicate that hydrolases of several different classes could be involved in the MX84 ester bond cleavage. These enzymes include phospholipases, chymotrypsin-like esterases, or lipases. These results are consistent with the observation that PLD activity was seen in J774 cells, RAW264 cells, and in all other myeloid cell lines. In addition, constitutive secretion of the PLD activity into the medium by J774 cells has been reported (40). Not unexpectedly, our preliminary studies show that three additional monocyte/macrophage cell lines tested (WEHI-3, WEHI-265, and P388) secreted MX84 conversion activity.
The activity in RAW264.1 macrophages was inhibited by the PLD inhibitor 1,10-phenanthroline in a dose-dependent manner. The antitumor drug suramin, previously reported to inhibit PLD activity (32), also demonstrated dose-dependent inhibition of the macrophage-associated esterase activity, as did an antiviral and antitumoral xanthogenate compound D609, known as an inhibitor of PLC and PLD (32,33), and an aminosteroid U73122, an inhibitor of PLA2, PLC, and PLD (32,34). The antitumor ether lipid sn-ET-18-OCH3, a potent inhibitor of PLC that is also reported to interfere with PLD activity (35,36), was also found to be an inhibitor of the RAW264.1 macrophage-associated esterase activity. However, the MX84 ester bond cleavage was not inhibited in our experiments by PLD inhibitor 2,3-diphosphoglycerate. This may suggest a highly specific phospholipase as the MX84 converter.
To further test this hypothesis, we tested whether the naturally occurring substrates for phospholipases, L-␣-phosphatidylcholine, and L-␣-lysophosphatidylcholine, could function, when in excess, as inhibitors of the MX84 ester bond cleavage. Indeed, we found that both L-␣-phosphatidylcholine and L-␣lysophosphatidylcholine, when present at a ϳ10-fold molar excess over MX84, were efficient inhibitors of the conversion, with L-␣-phosphatidylcholine being the more potent inhibitor. In contrast, these two substrates were unable to inhibit the esterase from porcine liver under similar conditions. Thus, these data are consistent with and support the hypothesis that the methyl ester MX84 is converted into CD437 by an esterase activity of a macrophage-secreted phospholipase.
Expression of a 100-kDa phospholipase A2 (PLA2) has been reported for several macrophage cell lines, including J774 (41) and RAW264 (42). Both calcium-independent and calcium-dependent activities of high molecular weight PLA2 in the macrophages were described (43)(44)(45)(46), as well as Ca 2ϩ -independent lysophospholipase activity of the enzyme purified from the RAW264.7 cells (42). However, in our hands, phospholipases A2 from different sources did not possess MX84 conversion activity. Consistent with this is our observation that the conversion activity was not affected either by HELSS or by MAFP, inhibitors for calcium-independent and calcium-dependent phospholipases A2, respectively (unless very high concentrations of these PLA2 inhibitors were used).
Thus, our analyses provide certain clues to the nature of the MX84 conversion enzyme that can now be tested in vitro and in vivo by designing other MX84 derivatives. Possibly our most important observation, however, is that along with an extensive search for receptor subtype-specific, cell-specific, and apoptosis-inducing retinoids, the design of inactive retinoid precursors that can be activated by cell type-specific enzymes can be used as an alternative strategy to develop effective and cell type-specific retinoids for the treatment of diseases such as inflammation and cancer.