Effects of methyl derivatives of the rexinoid UAB30 on methylnitrosourea (MNU) induced mammary cancers and on various indicators of toxicity

Abstract #1108

UAB30 is a RXR selective retinoid that has shown excellent activity in preventing chemically-induced mammary cancers with minimal toxicity (specifically not increasing serum triglycerides), and is currently undergoing a Phase I clinical trial. Several methyl analogs were synthesized in an attempt to obtain a more active preventive agent and to better understand the mechanisms of activity/toxicity of this class of agents. The retinoids included 4-methyl-UAB30, 5-methyl-UAB30, 6-methyl-UAB30, 7-methyl-UAB30, and 8-methyl-UAB30. The compounds were evaluated in the MNU-induced mammary cancer model using female Sprague-Dawley rats. MNU (75 mg/kg BW) was administered at 50 days of age and the animals observed for 120 days afterward for the development of ER+ mammary cancers. The retinoids were given at a dose of 200 mg/kg diet, except 7-methyl-UAB30 (given at 100 mg/kg diet). 4-Methyl-UAB30 and 7-methyl-UAB30 were highly active; reducing mammary cancer multiplicity by 74 and 61%, respectively. The retinoids 5-methyl-UAB30, 6-methyl-UAB30, and 8-methyl-UAB30 did not have chemopreventive activity in this model. The 8-methyl-UAB30 derivative actually caused a 108% increase in growth of the mammary cancers. As we have previously shown (Carcinogenesis 27, 1232-1239, 2006), serum triglycerides correlated with cancer preventive activity; i.e., high levels were observed with active retinoids. 4-methyl-UAB30 caused an initial large increase in body weight gain of the rats. Of interest, serum levels of 6-methyl-UAB30 and 7-methyl-UAB30 were high, while the other agents were low or could not be detected. All methyl derivatives caused varying decreases in liver retinyl palmitate levels. Structure-activity relationships are also being evaluated using crystallography of RXR/UAB-rexinoid complexes as a guide. Structure-based and dynamic-based approaches are used to facilitate the design of new rexinoids that fit into the LBD of RXRs. For example, 4-methyl-UAB30 had a 5-fold greater binding affinity to hRXR alpha LBD than 9-cis-retinoic-acid. These studies emphasize that minor modifications of a retinoid molecule can drastically change its absorption, metabolism, toxicity, binding affinities to receptors, and activity in preventing mammary cancers. (Supported by NCI Breast SPORE CA089019 and contract HHSN261200433001C).

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 1108.

Novel aromatic inhibitors of influenza virus neuraminidase make selective interactions with conserved residues and water molecules in the active site

The active site of type A or B influenza virus neuraminidase is composed of 11 conserved residues that directly interact with the substrate, sialic acid. An aromatic benzene ring has been used to replace the pyranose of sialic acid in our design of novel neuraminidase inhibitors. A bis(hydroxymethyl)pyrrolidinone ring was constructed in place of the N-acetyl group on the sialic acid. The hydroxymethyl groups replace two active site water molecules, which resulted in the high affinity of the nanomolar inhibitors. However, these inhibitors have greater potency for type A influenza virus than for type B influenza virus. To resolve the differences, we determined the X-ray crystal structure of three benzoic acid substituted inhibitors bound to the active site of B/Lee/40 neuraminidase. The investigation of a hydrophobic aliphatic group and a hydrophilic guanidino group on the aromatic inhibitors shows changes in the interaction with the active site residue Glu275. The results provide an explanation for the difference in efficacy of these inhibitors against types A and B viruses, even though the 11 active site residues of the neuraminidase are conserved.

Hydrophobic benzoic acids as inhibitors of influenza neuraminidase

Neuraminidase (NA) plays a critical role in the life cycle of influenza virus and is a target for new therapeutic agents. A new benzoic acid inhibitor (11) containing a lipophilic side chain at C-3 and a guanidine at C-5 was synthesized. The X-ray structure of 4-(N-acetylamino)-5-guanidino-3-(3-pentyloxy)benzoic acid in complex with NA revealed that the lipophilic side chain binds in a newly created hydrophobic pocket formed by the movement of Glu 278 to interact with Arg 226, whereas the guanidine of 11 interacts in a negatively charged pocket created by Asp 152, Glu 120 and Glu 229. Compound 11 was highly selective for type A (H2N2) influenza NA (IC50 1 microM) over type B (B/Lee/40) influenza NA (IC50 500 microM).

Bioactivity of peptide analogs of the neutrophil chemoattractant, N-acetyl-proline-glycine-proline

PURPOSE

The release of N-acetyl-proline-glycine-proline (PGP), a chemoattractant resulting from direct alkaline hydrolysis of corneal proteins, is believed to be the initial trigger for neutrophil invasion into the alkali-injured cornea. The purpose of this study is twofold: (1) to compare the activity of N-acetyl-PGP with the bioactivities of other similar synthetic peptides in an effort to uncover information about this chemoattractant molecule, and (2) to test these peptide analogs as potential antagonists of N-acetyl-PGP.

METHODS

The polarization assay was used to measure the potential chemotactic response of human neutrophils to peptides. Bioactivity was expressed as the peptide concentration required to produce 50% neutrophil polarization (EC50). Antagonist activity was expressed as the peptide concentration required to produce 50% inhibition (ID50) of polarization activated by N-acetyl-PGP.

RESULTS

Peptide bioactivities (EC50) were ranked as follows: APGPR (0.34 mM) > N-acetyl-PGP (0.5 mM) > N-(PGP)4-PGLG (3 mM) = t-Boc-PGP (3 mM) > N-acetyl-PG (3.4 mM) > N-methyl-PGP (15 mM) = PGP (15 mM) > peptides without detectable activity (t-Boc-PGP-OMe, N-acetyl-P, PG, PGG, GP, GG and gly-pro-hyp). Peptides with no detectable bioactivity were tested as potential antagonists of neutrophil polarization induced by N-acetyl-PGP. Gly-Pro-Hyp inhibited N-acetyl-PGP activation of polarization at 20 mM (ID50). No other synthetic peptide demonstrated a capacity for inhibition.

CONCLUSIONS

The minimum requirement to elicit bioactivity was the presence of PGP alone or derivatives of PG in which the N-terminal proline is blocked. Using this approach, active and inactive mimetic peptides of N-acetyl-PGP were produced. The most active peptide, APGPR, was equal to or slightly greater than N-acetyl-PGP, suggesting that more potent analogs might be designed. Gly-pro-hyp was the only inactive peptide analog to inhibit the chemoattractant.

Stereoselective synthesis of a conformationally defined cyclohexyl carnitine analogue that binds CPT-1 with high affinity

Carnitine (1, 3-hydroxy-4-trimethylammoniobutyrate) is important in mammalian tissue as a carrier of acyl groups. In order to explore the binding requirements of the carnitine acyltransferases for carnitine, we designed conformationally defined cyclohexyl carnitine analogues. These diastereomers contain the required gauche conformation between the trimethylammonium and hydroxy groups but vary the conformation between the hydroxy and carboxylic acid groups. Here we describe the synthesis and biological activity of the all-trans diastereomer (2), which was prepared by the ring opening of trans-methyl 2,3-epoxycylohexanecarboxylate with NaN3. Racemic 2 was a competitive inhibitor of neonatal rat cardiac myocyte CPT-1 (K(i) 0.5 mM for racemic 2; K(m) 0.2 mM for L-carnitine) and a noncompetitive inhibitor of neonatal rat cardiac myocyte CPT-2 (K(i) 0.67 mM). These results suggest that 2 represents the bound conformation of carnitine for CPT-1.

Design of benzoic acid inhibitors of influenza neuraminidase containing a cyclic substitution for the N-acetyl grouping

A 2-pyrrolidinone ring containing a single hydroxymethyl side chain effectively replaces the N-acetylamino group of 4-(N-acetylamino)-3-guanidinobenzoic acid, a low micromolar inhibitor of influenza neuraminidase. This novel structural template affords new opportunities to evolve more potent benzoic acid inhibitors.

Potent inhibition of influenza sialidase by a benzoic acid containing a 2-pyrrolidinone substituent

On the basis of the lead compound 4-(N-acetylamino)-3-guanidinobenzoic acid (BANA 113), which inhibits influenza A sialidase with a Ki of 2.5 microM, several novel aromatic inhibitors of influenza sialidases were designed. In this study the N-acetyl group of BANA 113 was replaced with a 2-pyrrolidinone ring, which was designed in part to offer opportunities for introduction of spatially directed side chains that could potentially interact with the 4-, 5-, and/or 6-subsites of sialidase. While the parent structure 1-(4-carboxy-2-guanidinophenyl)pyrrolidin-2-one (8) was only a modest inhibitor of sialidase, the introduction of a hydroxymethyl or bis(hydroxymethyl) substituent at the C5' position of the 2-pyrrolidinone ring resulted in inhibitors (9 and 12, respectively) with low micromolar activity. Crystal structures of these inhibitors in complex with sialidase demonstrated that the substituents at the 5'-position of the 2-pyrrolidinone ring interact in the 4- and/or 5-subsites of the enzyme. Replacement of the guanidine in 12 with a hydrophobic 3-pentylamino group resulted in a large enhancement in binding to produce an inhibitor (14) with an IC50 of about 50 nM against influenza A sialidase, although the inhibition of influenza B sialidase was 2000-fold less. This represents the first reported example of a simple, achiral benzoic acid with potent (low nanomolar) activity as an inhibitor of influenza sialidase.

Comparative molecular field analysis of hydantoin binding to the neuronal voltage-dependent sodium channel

Comparative molecular field analysis (CoMFA), a 3-D QSAR technique, is widely used to correlate biological activity with observed differences in steric and electrostatic fields. In this study, CoMFA was employed to generate a model, based upon 14 structurally diverse 5-phenylhydantoin analogues, to delineate structural and electrostatic features important for enhanced sodium channel binding. Correlation by partial least squares (PLS) analysis of in vitro sodium channel binding activity (expressed as log IC50) and the CoMFA descriptor column generated a final non-cross-validated model with R2 = 0.988 for the training set. The final CoMFA model explained the data better than a simpler correlation with log P (R2 = 0.801) for the same training set. The CoMFA steric and electrostatic maps described two general features that result in enhanced binding to the sodium channel. These include a preferred 5-phenyl ring orientation and a favorable steric effect resulting from the C5-alkyl chain. This model was then utilized to accurately predict literature sodium channel activities for hydantoins 14-20, which were not included in the training set. Finally the hydantoin CoMFA model was used to design the structurally novel alpha-hydroxy-alpha-phenylamide 21. Synthesis and subsequent sodium channel evaluation of compound 21 (predicted IC50 = 9 microM, actual IC50 = 9 microM), a good binder to the sodium channel, established that the intact hydantion ring is not necessary for efficient binding to this site. Thus alpha-hydroxy-alpha-phenylamides may represent a new class of ligands that bind with increased potency to the sodium channel.

Synthesis of 2-[6-(2,4-dinitrophenoxy)hexyl]oxiranecarboxylic acid: a selective carnitine palmitoyltransferase-1 inhibitor

Carnitine palmitoyltransferases 1 and 2 (CPT-1 and CPT-2) catalyze the transfer of long chain fatty acids between carnitine and coenzyme A. Unlike CPT-2, CPT-1 exists in at least two isoforms with different physical and kinetic properties. Liver and skeletal muscle each contain a different isoform of CPT-1. Cardiac muscle contains both isoforms, and the minor component is identical to the isoform found in the liver. 2-[6-(2,4-Dinitrophenoxy)hexyl]oxiranecarboxylic acid (2) was reported to be a selective inhibitor for the liver isoform of CPT-1. A synthesis of 2 is described here which involves the reaction of diethyl malonate with 1-bromo-6-phenoxyhexane.

Conformationally defined retinoic acid analogues. 4. Potential new agents for acute promyelocytic and juvenile myelomonocytic leukemias

We recently synthesized several conformationally constrained retinoic acid (RA) analogues [8-(2'-cyclohexen-1'-ylidene)-3, 7-dimethyl-2,4,6-octatrienoic acids with different alkyl substituents at 2' (R1) and 3' (R2) positions on the cyclohexene ring] (Muccio et al. J. Med. Chem. 1996, 39, 3625) as cancer chemopreventive agents. UAB8 (R1 = Et; R2 = iPr), which contains sufficient steric bulk at the terminal end of the polyene chain to mimic the trimethylcyclohexenyl ring of RA, displayed biological properties similar to those of RA. To explore the efficacy of this retinoid in acute promyelocytic leukemia (APL) and juvenile myelomonocytic leukemia (JMML), we evaluated UAB8 isomers in in vitro assays which measure the capacity of retinoids to inhibit aberrant myeloid colony growth from blood or bone marrow cells obtained from human JMML patients and in assays measuring the potential of retinoids to differentiate NB4 cells (an APL cell line). Both (all-E)- and (13Z)-UAB8 were 2-fold more active than RA in the NB4 cell differentiation assay; however, only (all-E)-UAB8 had comparable activity to the natural retinoids in the JMML cell assays. These results were compared to the biological effectiveness of a new retinoid, UAB30 [8-(3', 4'-dihydro-1'(2'H)-naphthalen-1'-ylidene)-3,7-dimethyl-2,4, 6-octatrienoic acid], which had different nuclear receptor binding and transactivational properties than UAB8. Relative to (all-E)-RA and (all-E)-UAB8, (all-E)-UAB30 bound well to RARalpha but did not activate transcription-mediated RARalpha homodimers, even though it was effective in RARbeta- and RARgamma-mediated transactivational assays. In APL assays, this retinoid had much reduced activity and was only moderately effective in JMML assays and in cancer chemoprevention assays.

Design of aromatic inhibitors of influenza virus neuraminidase

Structure-based drug design, a terminology used to describe rational drug design by complementing the structure, spatially and chemically, of the target macromolecule, is rapidly developing as one of the innovative approaches to drug discovery. A growing volume of protein structure data and new techniques of protein structure determination make this all possible. The method of structure-based drug design and a specific example of the design of influenza virus neuraminidase is briefly presented. A whole new class of influenza virus neuraminidase inhibitors has been designed that can potentially be developed as antiinfluenza drugs.

Regulation of hepatic lecithin:retinol acyltransferase activity by retinoic acid receptor-selective retinoids

The microsomal enzyme LRAT esterifies retinol and has been implicated in the hepatic storage of vitamin A. Previously, we showed that hepatic LRAT activity is negligible during vitamin A deficiency and that all-trans-retinoic acid (all-trans-RA) rapidly induces the activity of liver LRAT in retinoid-deficient rats. In the present studies, we have examined the ability of natural and synthetic retinoids to induce liver LRAT activity in retinoid-deficient rats. The natural retinoids retinol, all-trans-RA (100 microg), 9-cis-RA, or equal molar amounts of other retinoids were injected ip and LRAT specific activity was measured in liver homogenates 17-18 h later. In retinoid-deficient rats, liver LRAT activity was extremely low [0.13 +/- 0.03 pmol retinyl ester (RE)/min/mg liver protein, mean +/- SE]. The natural retinoids retinol and all-trans-RA strongly induced LRAT activity (12.71 +/- 1.09 and 13.10 +/- 1.55 pmol RE/min/mg, respectively), whereas 9-cis-RA induced a lower level of LRAT activity (3.96 +/- 1.88 pmol RE/min/mg, P < 0.001 vs all-trans-RA). The retinoic acid receptor (RAR)-selective analog (RAR pan-agonist) all-trans-UAB8 and the RAR-alpha-selective retinoid Am580 also strongly induced LRAT activity. In contrast, neither RXR-selective agonists nor retinoids having a retro structure were active. For retinoids with significant RAR-alpha binding activity there was a strong direct correlation between receptor binding in vitro and the ability to induce hepatic LRAT activity in vivo (r2 = 0.920). These data implicate the RARs in the induction of hepatic LRAT and suggest a predominant role for RAR-alpha-active ligands.

Guanidinobenzoic acid inhibitors of influenza virus neuraminidase

The active site of influenza virus neuraminidase (NA) is formed by 11 universally conserved residues. A guanidino group incorporated into two unrelated NA inhibitors was previously reported to occupy different negatively charged sites in the NA active site, A new inhibitor containing two guanidino groups was synthesized in order to utilize both sites in an attempt to acquire a combined increase in affinity. The X-ray crystal structures of the complexes show that the expected increase in affinity could not be achieved even though the added guanidino group binds to the negatively charged site as designed. This suggests that the ligand affinity to the target protein is contributed both from ligand-protein interactions and solvation/conformation energy of the ligand.

Effects of log P and phenyl ring conformation on the binding of 5-phenylhydantoins to the voltage-dependent sodium channel

Binding to the neuronal voltage-dependent sodium channel (NVSC) was evaluated for 12 5-phenylhydantoins which systematically varied either log P and/or 5-phenyl ring orientation. The linear correlation of log P with in vitro sodium channel binding activity (log IC50) for hydantoins 1-12 and diphenylhydantoin (DPH) (r2 = 0.638) suggested that simple partitioning into the lipid phase is important but not sufficient to account for the effects of hydantoins on the NVSC. Comparisons among different hydantoins with the same log P but different low-energy phenyl ring orientations revealed that, in addition to log P, the correct 5-phenyl orientation is important for efficient binding.

Conformationally defined 6-s-trans-retinoic acid analogs. 3. Structure-activity relationships for nuclear receptor binding, transcriptional activity, and cancer chemopreventive activity

We recently demonstrated that conformationally defined 6-s-trans-retinoic acid (RA) analogs were effective in the prevention of skin papillomas (Vaezi et al. J. Med. Chem. 1994, 37, 4499-4507) and selective agonists for nuclear receptor binding and activation (Alam et al. J. Med. Chem. 1995, 38, 2302-2310). In order to probe important structure-activity relationships, we evaluated a homologous series of four 6-s-trans-retinoids that are 8-(2'-cyclohexen-1'-ylidene)-3,7-dimethyl-2,4,6-octatrienoic acids with different substituents at 2' (R2) and 3' (R1) positions on the cyclohexene ring. UAB1 (R1 = R2 = H), UAB4 (R1 = R2 = Me), UAB7 (R1 = Me, R2 = iPr), and UAB8 (R1 = Et, R2 = iPr) contain alkyl R groups that mimic, to different extents, portions of the trimethylcyclohexenyl ring of RA. Both 9Z- and all-E-isomers of these retinoids were evaluated in binding assays for cellular retinoic acid-binding proteins (CRABP-I and CRABP-II), a nuclear retinoic acid receptor (RAR alpha), and a nuclear retinoid X receptor (RXR alpha). The all-E-isomers of UAB retinoids bound tightly to CRABPs and RAR alpha, the binding affinity of the all-E-isomer increased systematically from UAB1 to UAB8, and binding for the latter was comparable to that of all-E-RA. In contrast to RA, the (9Z)-UAB retinoids were at least 200-fold less active than the all-E-isomers in binding to RAR alpha. The (9Z)-UAB isomers exhibited increasingly stronger binding to RXR alpha, and (9Z)-UAB8 was nearly as effective as (9Z)-RA in binding affinity. The retinoids were also evaluated in gene expression assays mediated by RAR alpha and RXR alpha homodimers or RAR alpha/RXR alpha heterodimers. Consistent with the binding affinities, the (all-E)-UAB retinoids activated gene transciption mediated by RAR alpha homodimers or RAR alpha/RXR alpha heterodimers, while the (9Z)-UAB isomers activated only the RXR alpha homodimer-mediated transcription. The all-E- and 9Z-isomers of the UAB retinoids were further evaluated for their capacity to prevent the induction of mouse skin papillomas. When compared to RA, only the (all-E)-UAB retinoids containing bulky R1 and R2 groups were effective in this chemoprevention assay. (9Z)-RA displayed equal capacity as RA to prevent papillomas, while the 9Z-isomers of the UAB retinoids were much less effective. Taken together, these studies demonstrate that the cyclohexenyl ring substituents of 6-s-trans-UAB retinoids are important for their biological activities and that the chemopreventive effect of the all-E-isomers of these retinoids correlates well with their capacity to bind to RARs and activate RAR/RXR-mediated transcription.

Murine toxicology and pharmacology of UAB-8, a conformationally constrained analog of retinoic acid

(2E, 4E, 6E)-8-[3'-Ethyl-2'-(1-methylethyl)-2'-cyclohexen-1'-ylidene] -3, 7-dimethyl-2,4,6-octatrienoic acid (UAB-8) has potent activity in preventing papillomas on the skin of mice similar to that determined in a previous study for the homolog containing one less carbon atom. To evaluate the toxicological profile for UAB-8, relative to all-trans-retinoic acid (RA), female mice were dosed by oral gavage for 29 days with amounts of 0.05, 0.1, or 0.2 mmol/kg/day. For the two compounds, the effects on body weights were similar. Mice dosed with UAB-8, however, had a lower incidence of clinical signs of toxicity (alopecia, scaly skin, and limping). At necropsy, bone fractures, skin abnormalities, and splenomegaly were observed in some mice dosed with RA but not in any dosed with UAB-8. Lymph node hyperplasia was noted in some mice dosed with either dose of RA but only in those dosed with the highest dose of UAB-8. All dose levels of RA produced microscopic lesions in the bones of mice; only the highest dose of UAB-8 had this effect. RA and UAB-8 had similar effects on chondrogenesis in cultures of cells from mouse limb buds, an indication of comparable teratogenic effects. For mice dosed i.v. (10 mg/kg), there was a saturated phase of elimination of RA from plasma (Km = 0.61 microgram/ml and Vmax = 2572 micrograms/hr); no such phase was noted when UAB-8 was administered. UAB-8 had values for t1/2 alpha and t1/2 beta of 0.47 and 17.1 hr, respectively. Relative to RA, UAB-8 has a favorable toxicological profile and different pharmacokinetics.

A strategy for theoretical binding constant, Ki, calculations for neuraminidase aromatic inhibitors designed on the basis of the active site structure of influenza virus neuraminidase

Neuraminidase (NA) is one of the two major surface antigens of influenza virus. It plays an indispensable role in the release and spread of progeny virus particles during infection. NA inhibitors reduce virus infection in animals. To improve the clinical efficacy of NA inhibitors, we have begun the design of non-carbohydrate inhibitors based on the active site structure of NA. The approach is an iterative process of ligand modeling and electrostatic calculations followed by chemical synthesis of compounds, biological testing, and NA-inhibitor complex structure determination by X-ray crystallography. A strategy has been developed to calculate Ki for newly designed inhibitors. The calculations using the DelPhi program were performed for carbohydrate inhibitors and three preliminary benzoic acid inhibitors of neuraminidase (BANA) that have been synthesized and shown to bind to the active site of NA in the crystal structure. The calculated Kis of these inhibitors have an enlightening agreement with their in vitro biological activities. This demonstrates that the calculations produce informative results on the affinity of modeled inhibitors. GRID maps were also calculated and several pockets were identified for accepting possible new ligands. The calculated Kis for newly designed ligands suggest that these potential compounds will have high inhibitory activities.

4-(Acetylamino)-3-hydroxy-5-nitrobenzoic acid

The 4-(acetylamino)-3-hydroxy-5-nitrobenzoic acid molecule, C9H8N2O6, a designed inhibitor for the influenza virus neuraminidase protein, crystallizes as hydrogen-bonded dimers. The dihedral angles of the substituent groups with respect to the planar phenyl moiety are 5.0 (3) degrees for the carboxyl group, 45.0 (2) degrees for the nitro group and 37.3 (1) degrees for the acetylamino substituent. The crystal structure is stabilized by intermolecular hydrogen bonding.

Structure-based inhibitors of influenza virus sialidase. A benzoic acid lead with novel interaction

Influenza virus sialidase is a surface enzyme that is essential for infection of the virus. The catalytic site is highly conserved among all known influenza variants, suggesting that this protein is a suitable target for drug intervention. The most potent known inhibitors are analogs of 2-deoxy-2,3-didehydro-N-acetylneuraminic acid (Neu5Ac2en), particularly the 4-guanidino derivative (4-guanidino-Neu5Ac2en). We utilized the benzene ring of 4-(N-acetylamino)benzoic acids as a cyclic template to substitute for the dihydropyran ring of Neu5Ac2en. In this study several 3-(N-acylamino) derivatives were prepared as potential replacements for the glycerol side chain of Neu5Ac2en, and some were found to interact with the same binding subsite of sialidase. Of greater significance was the observation that the 3-guanidinobenzoic acid derivative (equivalent to the 4-guanidino grouping of 4-guanidino-Neu5Ac2en), the most potent benzoic acid inhibitor of influenza sialidase thus far identified (IC50 = 10 microM), occupied the glycerol-binding subsite on sialidase as opposed to the guanidino-binding subsite. This benzoic acid derivative thus provides a new compound that interacts in a novel manner with the catalytic site of influenza sialidase.

Benzoic acid inhibitors of influenza virus neuraminidase

A strategy was developed to design non-carbohydrate inhibitors of influenza virus neuraminidase. Using an iterative cycle of modeling, synthesis, biological testing and X-ray crystallography structure determination, a series of inhibitors based on benzoic acid were produced. The refined structures of three compounds complexed with neuraminidase are reported. The results demonstrate the success of this structure-based drug-design strategy.

Conformationally defined cyclohexyl carnitine analogs

Three diastereoisomers of racemic (3-carboxy-2-hydroxy-1-cyclohexyl)trimethylammonium chloride [C10H20NO3+.Cl-; (1S,2S,3S) (2), (1R,2S,3S) (3) and (1S,2R,3S) (4)] were designed as rigid analogs for different low-energy conformational states of carnitine [(1), (3-carboxy-2-hydroxy-1-propyl)trimethylammonium chloride]. Structures (2)-(4) all assume a chair conformation in the solid state, in which the bulky trimethylammonio group occupies the equatorial position. As such, the orientations about C2-C3 in (2), (3) and (4) are all essentially the same as that found for (1) in the solid state (torsion angles for C1-C2-C3-N1 near 180 degrees), while the orientations about C1-C2 in (2)-(4) are such that each diastereoisomer contains a different one of the three possible low-energy staggered conformations predicted for (1) in solution. Comparisons between (1) and (2)-(4) in the solid state revealed that diastereoisomers (2), (3) and (4) provide rigid models for the major low-energy conformations of carnitine.

Structures of aromatic inhibitors of influenza virus neuraminidase

Neuraminidase (NA), a surface glycoprotein of influenza virus, is a potential target for design of antiinfluenza agents. The crystal structure of influenza virus neuraminidase showed that in the active site 11 residues are universally conserved among all strains known so far. Several potent inhibitors based on the carbohydrate compound 2-deoxy-2,3-didehydro-D-N-acetylneuraminic acid (DANA) have been shown to bind to the conserved active site and to reduce virus infection in animals when administered by nasal spray. Inhibitors of this type are, however, rapidly excreted from physiological systems and may not be effective in order to provide long-time protection. A new class of specific NA inhibitors, which are benzoic acid derivatives, has been designed on the basis of the three-dimensional structure of the NA-DANA complex and modeling of derivatives of 4-(acetylamino)benzoic acid in the NA active site. Intermediates were synthesized and were shown to moderately inhibit the NA activity and to bind to the NA active site as predicted. These rudimentary inhibitors, 4-(acetylamino)-3-hydroxy-5-nitrobenzoic acid, 4-(acetylamino)-3-hydroxy-5-aminobenzoic acid, and 4-(acetylamino)-3-aminobenzoic acid, and their X-ray structures in complexes with N2 (A/Tokyo/3/67) and B/Lee/40 neuraminidases have been analyzed. The coordinates of such inhibitors complexed with NA were used as the starting model for further design of more potent benzoic acid inhibitors. Because the active site residues of NA are invariant, the designed aromatic inhibitors have the potential to become an antiviral drug against all strains of influenza virus.

A conformationally defined 6-s-trans-retinoic acid isomer: synthesis, chemopreventive activity, and toxicity

A conformationally defined retinoic acid analog (1) which contains a dimethylene bridge to maintain the 6-s-trans orientation for two terminal double bonds in the polyene chain was synthesized. A Reformatsky reaction was utilized to extend the polyene chain of the starting enone, which provided exclusively the 9Z-configuration for the intermediate aldehyde. A Horners-Emmons condensation with this aldehyde then produced retinoic acid analogs with both 9Z- and 9Z,13Z-configurations. An I2-catalyzed isomerization of the intermediate 9Z-aldehyde yielded the all-E-aldehyde, which was olefinated as above to yield the (all-E)- and (13Z)-retinoic acid analogs of 1. Each configurational isomer of 1 was evaluated for its ability to inhibit the binding of retinoic acid to CRABP (chick skin) and to inhibit the chemical induction of ornithine decarboxylase in mouse skin. In each assay (all-E)-1 was the most active isomer, and this activity was comparable to or better than that for (all-E)-retinoic acid. (all-E)-1 and (13Z)-1 were both shown to be equally effective as (13Z)-retinoic acid in suppressing the proliferation of human sebaceous cells in vitro. (all-E)-1 was further evaluated for its ability to prevent the induction of mouse skin papillomas and to induce signs of vitamin A toxicity in mice. The cancer chemopreventive activity of (all-E)-1 was comparable to that of (all-E)-retinoic acid, and the toxicity was comparable to or slightly better than that of the natural vitamin.

3-Amino-5,5-dimethylhexanoic acid. Synthesis, resolution, and effects on carnitine acyltransferases

The selective inhibition of individual carnitine acyltransferases may be useful in the therapy of diabetes and heart disease. Aminocarnitine (3) is a weak competitive inhibitor (K(i) = 4.0 mM) for carnitine acetyltransferase (CAT), although the N-acetyl derivative 4 is about 165 times more potent (K(i) = 0.024 mM) than 3. Compound 3 is also a potent competitive inhibitor for carnitine palmitoyltransferases 1 and 2 (CPT-1 and CPT-2) (IC50 for CPT-2 = 805 nM). We synthesized 3-amino-5,5-dimethylhexanoic acid (7) and its N-acetyl derivative (8) as isosteric analogs of 3 and 4 that lack the quaternary ammonium positive charge. Like 3 and 4, compounds 7 and 8 were competitive inhibitors of CAT with significantly different potencies, but in this case, 8 (K(i) = 25 mM) was 10 times less potent than 7 (K(i) = 2.5 mM). R-(-)-7 and S-(+)-7 were stereoselective inhibitors of CAT (K(i) = 1.9 and 9.2 mM, respectively). Racemic 7 was a weak competitive inhibitor of CPT-2 (K(i) = 20 mM) and had no effect on CPT-1. These results are consistent with differences among the carnitine-binding sites on carnitine acyl-transferases that may be useful in selective inhibitor design. Furthermore, the data suggest that the quaternary ammonium positive charge of carnitine may be important for the proper orientation of carnitine and its analogs in the binding site.

Bicyclic hydantoins with a bridgehead nitrogen. Comparison of anticonvulsant activities with binding to the neuronal voltage-dependent sodium channel

The anticonvulsant activity of diphenylhydantoin (DPH or phenytoin) is consistent with its actions on the neuronal voltage-dependent sodium channel. To further elucidate the binding requirements for this site, we synthesized several hydantoin analogs and evaluated these in in vitro sodium channel-binding and/or in vivo whole animal anticonvulsant assays. 5-Pentyl-5-phenylhydantoin (8), the most potent binder to the sodium channel in this study, had the same affinity as DPH (IC50 = 40 microM), revealing that one phenyl ring is sufficient for good interactions. Since our previous studies with monophenyl-substituted bicyclic 2,4-oxazolidinediones suggested that N3-alkylation and the conformational constraint of a 5-alkyl substituent over one face of the oxazolidinedione ring improved activity, we synthesized two examples of analogous bicyclic hydantoins. However, the bicyclic hydantoins were much less potent binders to the neuronal voltage-dependent sodium channel than their monocyclic counterparts. The binding activity for the more potent bicyclic hydantoin, 1,8-diaza-9,10-dioxo-7-phenylbicyclo[5.2.1]decane (4) (IC50 = 427 microM), was comparable to that of the ring-opened, N3-methylated monocyclic hydantoin model, 5-butyl-3-methyl-5-phenylhydantoin (9) (IC50 = 285 microM), and these were 8-11 times less potent than the monocyclic model 8, which contains a free imide NH. Furthermore, 5-butyl-5-phenylhydantoin (7; IC50 = 103 microM) was less potent than 8, suggesting that increased log P may enhance binding. Thus, unlike 2,4-oxazolidinediones, N3-alkylation of hydantoins dramatically decreases sodium channel-binding activity. Bicyclic hydantoin 4 was nevertheless a good anti-MES anticonvulsant in mice (ED50 = 86 mg/kg), although this activity likely results from mechanisms other than interactions at the neuronal voltage-dependent sodium channel. Compound 4 was also relatively neurotoxic (TD50 = 124 mg/kg). These results suggest that the binding of hydantoins to the sodium channel may be enhanced by (a) a free imide NH group and (b) an increased log P. Furthermore, 2,4-oxazolidinediones and hydantoins must either orient differently in the same binding site or interact with different sites on the neuronal voltage-dependent sodium channel.

Effects of C-methylated carnitine analogs on rates of mitochondrial fatty acid oxidation

Carnitine analogs containing one and/or two methyl substituents on the alpha-, beta-, or gamma-carbon were evaluated in isolated rat liver mitochondria for their effects on fatty acid oxidation. Their abilities to either support, in the absence of carnitine, or inhibit, in the presence of carnitine, carnitine-dependent fatty acid oxidation were determined by the conversion of radiolabeled [1-14C]palmitic acid to acid-soluble radiolabeled products. None of the methylcarnitine analogs were observed to be significant inhibitors of palmitate oxidation at concentrations (1.0 mM) up to ten times that for L-carnitine. The two diastereomers of D,L-4-methylcarnitine, however, were able to support palmitate oxidation in the absence of carnitine, and rates were roughly 40% of that obtained with equimolar (0.1 mM) L-carnitine.

Carnitine acyltransferase enzymic catalysis requires a positive charge on the carnitine cofactor

3-Hydroxy-5,5-dimethylhexanoic acid (HDH) is an analogue of carnitine which differs only in the substitution of a quaternary carbon atom for the quaternary ammonium nitrogen. Thus HDH is isosteric with carnitine but lacks the quaternary ammonium positive charge. Racemic HDH, each of its enantiomers, and the O-acetyl derivative (Ac-HDH) were evaluated as alternate substrates and inhibitors for several carnitine acyltransferases. HDH and Ac-HDH are not substrates for carnitine acetyltransferase (CAT) at concentrations up to 10 mM, suggesting that the positive quaternary ammonium charge on carnitine is essential for CAT catalysis. However, HDH competitively inhibits CAT (Ki = 8.3 mM), carnitine palmitoyltransferase-I (CPT-I) (Ki = 3.6 mM), and CPT-II (Ki = 2.8 mM). Ac-HDH is also a competitive inhibitor of CAT when assayed in the reverse direction (Ki = 4.1 mM). Similarly, R-(+)-HDH and S-(-)-HDH are not substrates for CAT, but they are stereoselective competitive inhibitors (Ki = 20.3 and 7.5 mM for the R and S enantiomers, respectively). Stereoselective inhibition by HDH is even more dramatic with CPT-I, since S-(-)-HDH inhibits CPT-I (Ki = 1.4 mM) but R-(+)-HDH has no effect in concentrations up to 5 mM. As with CAT, HDH is a stereoselective inhibitor of CPT-II, and the Ki values for S-(-)- and R-(+)-HDH are 2.2 and 6.7 mM, respectively. Since the observed Ki values are significantly larger than the Km for carnitine, the positive charge on carnitine must also be important, but not essential, for binding to the carnitine site on carnitine acyltransferases.

Sodium channel binding and anticonvulsant activities of hydantoins containing conformationally constrained 5-phenyl substituents

As a preliminary investigation of the importance of the aromatic ring orientation in interactions of 5-phenylhydantoins with the anticonvulsant site on the neuronal voltage-sensitive sodium channel, two isomeric hydantoins containing conformationally constrained phenyl rings and their monocyclic analogues were synthesized. One, a spirohydantoin (2) derived from alpha-tetralone, contains the plane of the phenyl ring in an orientation approximately perpendicular to that for the hydantoin ring. The other, a tricyclic hydantoin (4) derived from tetrahydroisoquinoline, contains the plane of the phenyl ring in an orientation roughly coplanar with that for the hydantoin ring. These compounds were evaluated in sodium channel binding and whole animal (mice) anticonvulsant assays. In both assays, 4 was significantly more potent than 2, suggesting that the anticonvulsant receptor site on the voltage-sensitive sodium channel may require a specific aromatic ring orientation.

Sodium channel binding and anticonvulsant activities for the enantiomers of a bicyclic 2,4-oxazolidinedione and monocyclic models

Racemic 7-phenyl-9,10-dioxo-1-aza-8-oxabicyclo[5.2.1]decane (1), a bicyclic 2,4-oxazolidinedione that we previously reported was a possible sodium channel anticonvulsant, was resolved into its enantiomeric forms, the absolute configurations were determined, and the stereoisomers were evaluated for relative sodium channel binding and whole animal anticonvulsant activities. Similar studies were carried out with two monocyclic models, 5-ethyl-5-phenyl-2,4-oxazolidinedione (2) and 5-ethyl-3-methyl-5-phenyl-2,4-oxazolidinedione (3). None of these isomers exhibited stereoselective effects in the sodium channel assay, and only modest enantioselectivities were observed for 2 and 3 in the anticonvulsant assays. (R)-(-)-1 was, however, 4 times more toxic than (S)-(+)-1 in the rotorod test, and due to its larger protective index, (S)-(+)-1 exhibited greater therapeutic potential than either (R)-(-)-1 or racemic 1.

Anticonvulsant activities of phenyl-substituted bicyclic 2,4-oxazolidinediones and monocyclic models. Comparison with binding to the neuronal voltage-dependent sodium channel

8,9-Dioxo-6-phenyl-1-aza-7-oxabicyclo[4.2.1]nonane (1) and 9,10-dioxo-7-phenyl-1-aza-8-oxabicyclo[5.2.1]decane (2), examples of anti-Bredt bicyclic 2,4-oxazolidinediones, were investigated as anticonvulsants in mice. Compound 2 was the more potent (anti-MES ED50 = 66 mg/kg), and its in vivo anti-MES effect was consistent with its in vitro potency of binding to the voltage-sensitive sodium channel (IC50 = 160 microM for the inhibition of binding of [3H]BTX-B), suggesting that 2 may be a new class I anticonvulsant. Several partial structures of 2, either monocyclic lactams or monocyclic 2,4-oxazolidinediones, were also evaluated in these assays, but no correlation was observed between sodium channel binding and anti-MES effects. A significant finding was that monocyclic 5-alkyl-5-phenyl-2,4-oxazolidinediones provided relatively potent, nontoxic, broad-spectrum anticonvulsants.

Synthesis and anticonvulsant activity of some substituted lactams and amides

Thirteen derivatives of 3-phenyl-2-piperidinone were synthesized and evaluated for anticonvulsant activity. The most active compounds from this group included two simple lactams, 3-hydroxy-1-methyl-3-phenyl-2-piperidinone and 3-methoxy-3-phenyl-2-piperidinone, and two N-ethoxycarbonyl lactams, 1-(ethoxycarbonyl)-3-hydroxy-3-phenyl-2-piperidinone and 1-(ethoxycarbonyl)-3-methoxy-3-phenyl-2-piperidinone, whose anticonvulsant activity was comparable to or better than that for valproic acid. Four related acyclic amides were also prepared, but these were essentially inactive as anticonvulsants.