Novel semisynthetic analogues of betulinic acid with diverse cytoprotective, antiproliferative, and proapoptotic activities

In light of the scientific promise of chemoprevention, there is an overwhelming need to develop new chemopreventive agents that are both effective and safe (1). One practical approach to this problem is to use natural products as a platform for drug development. Approximately half of the drugs currently used in the clinic are derived from natural products (2). The starting material for these drugs is already biologically active, with a defined three-dimensional structure and important functional groups. Ideally, chemical modification of these natural products can significantly enhance potency and maintain levels of toxicity that are often lower than many drugs developed by combinatorial chemistry. Indeed, by modifying oleanolic acid, we have developed a series of multifunctional triterpenoids that suppress inflammation, inhibit proliferation, and induce apoptosis of cancer cells and that are effective for both the prevention and treatment of cancer in experimental animals (3–9). Two of these compounds, 2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oic acid (CDDO) and its methyl ester (CDDO-Me), are currently in phase 1 clinical trials for the treatment of leukemia and solid tumors.

Because of our success with the oleanane triterpenoids, we have used a similar strategy, but with a different natural product as a starting platform. Birch bark has been used for medicinal purposes in many cultures, and two of the most active components in American paper birch (Betula papyrifera) bark, the triterpenoid alcohol betulin and the corresponding carboxylic acid betulinic acid (BA), together account for up to 20% of the dry weight of the bark (10). The discoveries that BA (a) slows the growth of the HIV (11) and (b) is a selective inhibitor of human melanoma cell growth, both in vitro and in vivo (12), have sparked interest in BA as a potential drug candidate (10, 13, 14). BA also suppresses growth and induces apoptosis of other human cancer cells, including brain cancer cells, neuroblastomas, ovarian carcinomas, and leukemias (reviewed in ref. 15), whereas normal cells such as dermal fibroblasts and peripheral blood lymphocytes are much less sensitive to growth inhibition by this agent (16). Despite a lack of toxicity in these studies, BA is a very weak antineoplastic agent; micromolar concentrations were needed to block cell proliferation in vitro, and doses as high as 250 mg/kg body weight were required to suppress the growth of melanomas in athymic mice (10).

To improve the potency of BA, dozens of structural modifications have been made to BA (reviewed in refs. 14, 17). In lupanes such as BA (Fig. 1), the cyclization of squalene proceeds with different cyclases than used in the oleananes (18), resulting in an E-ring with only five carbons and an isopropenyl group at C-19 in a unique steric conformation. Moreover, this isopropenyl group includes a double bond, which offers opportunities for chemical modifications not possible with oleanolic acid. We hypothesized that new BA derivatives would have some overlapping biological properties with the oleanane triterpenoids but would differ in their pharmacokinetics and structure-activity relationships. We have synthesized 15 novel BA derivatives with significant alterations to the A ring, to the isopropenyl group at C-19, and to the C-28 carboxylic acid position, and recently reported that these agents inhibit nitric oxide production in RAW cells stimulated with IFN-γ (19). Here, we will limit our description to six BA compounds (structures shown in Fig. 1; chemical names listed in Table 1), which revealed new and unexpected structure-activity relationships. Structures containing a cyano-enone functionality in ring A [cyano-betulonic acid (CBA, TP-291), the methyl ester of CBA (CBA-Me, TP-290), or the imidazolide of CBA (CBA-Im, TP-292)] are highly active at suppressing inflammation and inducing phase 2 cytoprotective enzymes, but substituting a methoxycarbonyl group for the cyano group yields compounds that are potent inducers of apoptosis in a variety of cancer cells.

Details describing the synthesis of new BA analogues have been published (19, 20). Compounds were dissolved in DMSO, and controls containing equivalent concentrations of DMSO were included in all experiments. The Bax/Bak knockout fibroblasts (21) were provided by the late Stanley Korsmeyer (Harvard Medical School, Boston, MA); all other cell lines were obtained from American Type Culture Collection. To check for antiinflammatory activity or activation of phase 2 enzymes, cells were treated with triterpenoids and then analyzed by Western blotting with inducible nitric oxide synthase (iNOS) and heme oxygenase 1 (HO-1) antibodies, by flow cytometry for production of reactive oxygen species (ROS), or for specific enzyme activity of NAD(P)H quinone oxidoreductase (NQO1). For proliferation assays, cells were treated with compounds for 2 to 3 days, pulsed with ³H-thymidine for 2 h and counted. Apoptosis was analyzed by fluorescence-activated cell sorting using the TACS Annexin V-FITC Apoptosis Detection Kit (R&D Systems) or by immunoblotting with PARP antibodies. Additional details are included in the figure or table legends using previously reported methods (22–24).

Four CD-1 male mice were gavaged with 2 μmol of triterpenoid dissolved in DMSO or DMSO alone. Six hours later, solid tissues were harvested and blood was collected by cardiac puncture into heparinized tubes. Blood was centrifuged at 5,000 rpm for 5 min to isolate plasma. For extraction of plasma, three volumes of acetonitrile were added to the plasma, and the samples were vortexed. For extraction from solid tissues, 0.5 mL of acetonitrile was added to 200 to 400 mg tissue, and the tissue was homogenized on ice using a Tissuemiser (Fisher Scientific). Plasma and tissue samples were then centrifuged at 14,000 rpm for 10 min. The acetonitrile extracts were diluted 1:1 with 20 mmol/L of ammonium acetate (pH 7.4), centrifuged at 14,000 rpm for 5 min, and the supernatants (100 μL injection volume) loaded onto a Waters 2695 high-performance liquid chromatography. Triterpenoids were separated by reverse phase chromatography on a Waters XTerra MS C18 5 μm particle column using an 8 min gradient from 46% to 94% acetonitrile. Triterpenoids and their metabolites were detected using a single quadrupole mass spectrometer with electrospray ionization (Waters Micromass ZQ). Analysis was carried out using Waters MassLynx 4.1 software and standard curves for at least six concentrations per compound (serial dilutions starting at 2 μmol/L) were generated by adding the compound of interest to control plasma or to control tissue extracts. All calculated values were within the limits of the standards.

Because of the importance of inflammation in carcinogenesis (1, 25, 26), our primary screen for evaluating new potential chemopreventive agents is their ability to block the production of iNOS in response to inflammatory cytokines. As shown in Table 1, BA derivatives containing an added cyano-enone functionality (CBA, CBA-Me, and CBA-Im) are potent inhibitors of nitric oxide production in primary mouse macrophages stimulated with IFN-γ (with IC₅₀ values of 1 nmol/L). BA itself and the three triterpenoids lacking a cyano group in ring A are inactive in primary macrophages, although in RAW264.7 mouse macrophage-like cells, TP-295-297 can suppress the release of nitric oxide but are half a log to a log less active than TP-290-292 (19). CBA, CBA-Me, and CBA-Im also inhibit the induction of iNOS protein in a dose-dependent manner in RAW264.7 cells stimulated with IFN-γ (Fig. 2A) or lipopolysaccharide (data not shown), whereas again TP-295-297 are only active at higher concentrations. We recently reported that the ability of a set of oleanolic acid derivatives to suppress iNOS and to induce the NQO1 enzyme, part of the phase 2 response used by cells to deactivate electrophilic or oxidative stress, are tightly correlated (22). Here, for new BA derivatives, we show that the same structure-activity relationships observed in the iNOS assays are evident for the induction of the cytoprotective phase 2 enzymes NQO1 (Table 1) and HO-1 in vitro (Fig. 2B) and in vivo (19). One favorable outcome of activating the phase 2 response is a reduction of ROS. In cells pretreated with triterpenoids for 24 h, and then challenged with tert-butyl-hydroperoxide to induce ROS, CBA reduced ROS levels by 60% compared with control cells not pretreated with triterpenoids, whereas CBA-Im reduced ROS levels by 43% (Fig. 2C). BA alone reduced ROS levels by 50%, but at a 10-fold higher concentration than CBA and CBA-Im. CDDO was the most potent protector against ROS generation, in agreement with its high potency as an inducer of NQO1 and HO-1.

Blocking cell proliferation and inducing apoptosis are important properties of chemopreventive and chemotherapeutic agents (1, 27), and BA alone is selectively cytostatic and cytotoxic in a wide variety of human tumor lines, independent of p53 status, with IC₅₀ values for the induction of apoptosis ranging from 5 to 20 μg/mL (14, 16). The new BA derivatives inhibit the growth of Jurkat and U937 leukemia cells, RPMI 8226 myeloma cells (Fig. 3), and MCF-7 breast cancer cells (data not shown), and are markedly more potent than BA alone. In all four cell lines, TP-295 is the most effective at inhibiting cell proliferation, but TP-296, CBA-Im, and CBA-Me (data not shown) are also extremely effective at concentrations of 1 μmol/L. At slightly higher concentrations (3–5 μmol/L), compounds with a methoxycarbonyl enone functionality in ring A (TP-295, TP-296) are extremely potent inducers of apoptosis in U937 human leukemia cells, RPMI 8226 human myeloma cells, and H358 human lung cancer cells (Fig. 4A and B). TP-295 and TP-296 also induce apoptosis in PNET-2 and SH-SY-5Y human neuroblastoma cells (data not shown), but these cells are more resistant, as concentrations between 3 and 5 μmol/L are needed to induce apoptosis. Apoptosis was confirmed in all of these cell lines by both Annexin V staining and PARP cleavage. Although all six BA compounds induced PARP cleavage in RPMI 8226 myeloma cells (Fig. 4B), TP-295, TP-296, and to a lesser extent, TP-297, were the most potent compounds among all cell types. At concentrations of 5 to 10 μmol/L, BA alone was inactive, whereas CBA-Im was the most active compound among the three derivatives of BA (CBA-Me, CBA, and CBA-Im) containing a cyano group in ring A. Remarkably, TP-295 and TP-296 also induce apoptosis in Bax/Bak^−/− fibroblasts (Fig. 4C), which lack two important components of the intrinsic (mitochondrial-dependent) apoptotic pathway (21). Defects in this pathway are often found in chemoresistant leukemias (28), and Bax/Bak^−/− fibroblasts are highly resistant to apoptosis induced by anisomycin and chemotherapeutic agents. Anisomycin does not include apoptosis in Bax/Bak^−/− fibroblasts (Fig. 4C) but does induce PARP cleavage in other cell lines (Fig. 4B). BA, CDDO, and TP-290-292 (data not shown) are also inactive in Bax/Bak^−/− cells at equivalent concentrations.

To compare the tissue levels of the four most promising BA derivatives (CBA-Im, TP-290; CBA-Me, TP-290, TP-295, and TP-296), male CD-1 mice were gavaged with either DMSO or 2 μmol compound; this dose of DMSO or BA derivative did not cause any signs of distress or toxicity. After 6 h, blood and tissues were harvested, extracted, and analyzed by liquid chromatography mass spectroscopy. As shown in Table 2, the levels of the new BA analogues were highest in the liver (0.3–10.3 μmol/kg) and lung (0.2–1.6 μmol/kg), and detectable levels were also found in the plasma (0.2–1.2 μmol/L) and brain (0.014–0.2 μmol/kg). The tissue levels of these new BA compounds were markedly higher than the levels of BA itself, whereas the levels of all of the lupane triterpenoids were significantly higher than the levels found with the similar oleanane triterpenoid CDDO. Notably, the levels of CBA-Im were 2- to 10-fold higher in all of the tissues than the other three BA compounds; tissue levels of CBA-Im were also dose-responsive and still detectable 24 h after gavage of 1 μmol CBA-Im (data not shown).

In summary, we have found that new BA derivatives containing a cyano-enone functionality in ring A potently inhibit the induction of iNOS and induce the phase 2 enzymes NQO1 and HO-1. Substituting a methoxycarbonyl group for the cyano group yields BA compounds that induce apoptosis in cancer cells and in Bax/Bak^−/− cells. Moreover, concentrations of the BA compounds that are active in vitro can be achieved in vivo, especially with CBA-Im. This newly discovered structure-activity relationship suggests that various BA derivatives could have unique applications. For example, inhibition of chronic inflammation and induction of the Nrf2/antioxidant response element–dependent phase 2 system are important pathways in chemoprevention (1, 25, 29), and the fact that tissue levels of CBA-Im are high enough to induce HO-1 in the liver and lung suggest that this compound should be tested for the prevention of liver and lung cancer (6, 7). In contrast, TP-295 should be more effective for the treatment of cancer, especially for leukemia (8).

Despite the promising but diverse biological activities of BA and its derivatives, its molecular targets and mechanisms of action are not known. To our knowledge, this is the first report that analogues of BA both suppress the induction of iNOS in response to inflammatory cytokines (Table 1; Fig. 2A) and induce phase 2 cytoprotective enzymes such as HO-1 and NQO1 (Table 1; Fig. 2B). These two cytoprotective responses to synthetic oleanane triterpenoids are tightly correlated (22) and are regulated by the transcription factor Nrf2. Nrf2 is normally sequestered by its repressor Keap1 and targeted for ubiquitinylation and proteasomal degradation. In response to oxidative stress or a direct chemical interaction between an inducer and its highly reactive sulfhydryl groups, Keap1 loses its ability to repress Nrf2, which then undergoes nuclear translocation and binds to an antioxidant response element on the promoter of phase 2–responsive genes (30, 31). Future experiments will explore whether CBA-Im or CBA-Me directly interacts with Keap1 to induce phase 2 enzymes and block the induction of iNOS. At high concentrations, BA alone can activate macrophages to induce proinflammatory cytokines (32). We have observed a similar bifunctional response with the oleanane triterpenoids, with low concentrations reducing oxidative stress and high concentrations increasing the production of ROS.

Indeed, at higher concentrations than are needed for blocking the induction of iNOS and activating the phase 2 response, BA derivatives induce apoptosis in a wide variety of cancer cells (Fig. 4) and at much lower concentrations than are required for BA (14). Although CBA (TP-290) and CBA-Me with an isopropyl group at C-19 are cytotoxic in A549 lung cancer cells and B16 melanoma cells (20), the five new BA derivatives shown in Fig. 1 are more potent than CBA in our proliferation and apoptosis assays. As for mechanism, BA acts directly on mitochondria to induce loss of membrane potential (33). In contrast, TP-295 and TP-296, which contain a methoxycarbonyl group in ring A, can induce apoptosis in Bax/Bak^−/− cells (Fig. 4C), which lack two key components of the mitochondrial-dependent apoptotic pathway, suggesting that these new BA compounds can also activate apoptosis by the extrinsic death receptor–mediated pathway. Future experiments will attempt to identify the proximate molecular target(s) for TP-295, TP-296, and CBA-Im that starts the apoptotic cascade.

The important biological activities and enhanced potency of these new BA derivatives suggest that further studies are needed to elucidate their molecular mechanisms of action and to determine their potential use for prevention or treatment of cancer and other diseases. Indeed, in addition to its potential anticancer activities (10, 17), a number of BA compounds are active against HIV and other viruses, bacteria, malaria, and helminthes (13, 15, 17), and the new BA derivatives described here should also be tested for these applications.

We thank John Pezzuto for supplying BA, the late Stanley Korsmeyer for supplying Bax/Bak^−/− fibroblasts, and Megan Padgett for assistance with the manuscript.