Preclinical validation of a potent γ-secretase modulator for Alzheimer's disease prevention.

A potent γ-secretase modulator (GSM) has been developed to circumvent problems associated with γ-secretase inhibitors (GSIs) and to potentially enable use in primary prevention of early-onset familial Alzheimer's disease (EOFAD). Unlike GSIs, GSMs do not inhibit γ-secretase activity but rather allosterically modulate γ-secretase, reducing the net production of Aß42 and to a lesser extent Aß40, while concomitantly augmenting production of Aß38 and Aß37. This GSM demonstrated robust time- and dose-dependent efficacy in acute, subchronic, and chronic studies across multiple species, including primary and secondary prevention studies in a transgenic mouse model. The GSM displayed a >40-fold safety margin in rats based on a comparison of the systemic exposure (AUC) at the no observed adverse effect level (NOAEL) to the 50% effective AUC or AUCeffective, the systemic exposure required for reducing levels of Aß42 in rat brain by 50%.

Pharmacological and Toxicological Properties of the Potent Oral γ-Secretase Modulator BPN-15606.

Alzheimer's disease (AD) is characterized neuropathologically by an abundance of 1) neuritic plaques, which are primarily composed of a fibrillar 42-amino-acid amyloid-ß peptide (Aß), as well as 2) neurofibrillary tangles composed of aggregates of hyperphosporylated tau. Elevations in the concentrations of the Aß42 peptide in the brain, as a result of either increased production or decreased clearance, are postulated to initiate and drive the AD pathologic process. We initially introduced a novel class of bridged aromatics referred tγ-secretase modulatoro as γ-secretase modulators that inhibited the production of the Aß42 peptide and to a lesser degree the Aß40 peptide while concomitantly increasing the production of the carboxyl-truncated Aß38 and Aß37 peptides. These modulators potently lower Aß42 levels without inhibiting the γ-secretase-mediated proteolysis of Notch or causing accumulation of carboxyl-terminal fragments of APP. In this study, we report a large number of pharmacological studies and early assessment of toxicology characterizing a highly potent γ-secretase modulator (GSM), (S)-N-(1-(4-fluorophenyl)ethyl)-6-(6-methoxy-5-(4-methyl-1H-imidazol-1-yl)pyridin-2-yl)-4-methylpyridazin-3-amine (BPN-15606). BPN-15606 displayed the ability to significantly lower Aß42 levels in the central nervous system of rats and mice at doses as low as 5-10 mg/kg, significantly reduce Aß neuritic plaque load in an AD transgenic mouse model, and significantly reduce levels of insoluble Aß42 and pThr181 tau in a three-dimensional human neural cell culture model. Results from repeat-dose toxicity studies in rats and dose escalation/repeat-dose toxicity studies in nonhuman primates have designated this GSM for 28-day Investigational New Drug-enabling good laboratory practice studies and positioned it as a candidate for human clinical trials.

An in-vitro cocktail assay for assessing compound-mediated inhibition of six major cytochrome P450 enzymes.

An efficient screening assay was developed and validated for simultaneous assessment of compound-mediated inhibition of six major human cytochrome P450 (CYP) enzymes. This method employed a cocktail of six probe substrates (i.e., phenacetin, amodiaquine, diclofenac, S-mephenytoin, dextromethorphan and midazolam for CYP1A2, 2C8, 2C9, 2C19, 2D6 and 3A4, respectively) as well as individual prototypical inhibitors of the six CYP enzymes in human liver microsomes under optimized incubation conditions. The corresponding marker metabolites (i.e., acetaminophen, N-desethylamodiaquine, 4-OH-diclofenac, 4-OH-S-mephenytoin, dextrorphan and 1-OH-midazolam) in the incubates were quantified using LC-MS/MS methods either by an internal standard (IS) calibration curve or a simplified analyte-to-IS peak area ratio approach. The results showed that the IC50 values determined by the cocktail approach were in good agreement with those obtained by the individual substrate approach as well as those reported in the literature. Besides, no remarkable difference was observed between the two quantification approaches. In conclusion, this new cocktail assay can be used for reliable screening of compound-mediated CYP inhibition.

Pharmacokinetics of biotech drugs: peptides, proteins and monoclonal antibodies.

With the advances in recombinant DNA biotechnology, molecular biology and immunology, the number of biotech drugs, including peptides, proteins and monoclonal antibodies, available for clinical use has dramatically increased in recent years. Although pharmacokinetic principles are equally applicable to the large molecule drugs and conventional small molecule drugs, the underlying mechanisms for the processes of absorption, distribution, metabolism and excretion (ADME) of large molecule drugs are often very different from that of small molecule drugs. Therefore, a good understanding of the ADME processes of large molecule drugs is essential in support of the development of therapeutic biologics. The purpose of this article is to review the current knowledge of the ADME processes that govern the pharmacokinetics of biotech drugs. The challenges encountered by orally administered peptide and protein drugs, and the nature of lymphatic absorption after subcutaneous administration will be discussed. In addition, molecular mechanisms of biodistribution, metabolism and renal excretion of biotech drugs will also be discussed. Finally, approaches used for prediction of human pharmacokinetics of protein drugs will be briefly discussed.

Applications and limitations of genetically modified mouse models in drug discovery and development.

Genetically modified mouse models in which a specific gene is removed or replaced have proven to be powerful tools for identification/validation of target gene and scientific understanding of molecular mechanisms underlying drug-induced toxicity through mechanistic studies. In spite of the advantage, there are significant limitations of genetically modified mouse models. Modification of a given gene does not always result in the anticipated phenotype. In some instances, phenotypes of targeted mouse mutants were not those predicted from the presumed function of the given genes, while other null mutants revealed no apparent defects. Furthermore, the phenotypic outcome can be influenced by many environmental and genetic factors. Therefore, interpretation of the significance of the findings from studies using genetically modified mouse models is not always as straightforward as one would expect, especially when desire is to extrapolate the findings to humans. Interestingly, many humanized mouse models have been generated for evaluating the function and regulation of cytochrome P450 (CYP) enzymes. Our fascination with humanized animals dates back to ancients. For example, the Great Sphinx of Giza, a large half-human and half-lion statue, is believed to have been built by Egyptians about 4500 years ago. Although the creation of humanized animals that carry a particular human CYP gene provides useful tools for scientific understanding of the function and regulation of the CYP enzyme, these humanized mouse models are not so useful in prediction of human pharmacokinetics in a quantitative sense. Accordingly, it is important to keep in mind that an animal engineered to express a human gene and its protein is still an animal.

CSF as a surrogate for assessing CNS exposure: an industrial perspective.

For drugs that directly act on targets in the central nervous system (CNS), sufficient drug delivery into the brain is a prerequisite for drug action. Systemically administered drugs can reach CNS by passage across the endothelium of capillary vasculatures, the so-called blood-brain barrier (BBB). Literature data suggest that most marketed CNS drugs have good membrane permeability and relatively high plasma unbound fraction, but are not good P-glycoprotein (P-gp) substrates. Therefore, it is important to use the in vitro parameters of P-gp function activity, membrane permeability and plasma unbound fraction as key criteria for lead optimization during the early stage of drug discovery. Evidence from preclinical and clinical studies suggests that drug concentration in cerebrospinal fluid (CSF) appears to be reasonably accurate in predicting unbound drug concentration in the brain. Therefore, CSF can be used as a useful surrogate for in vivo assessment of CNS exposure and provides an important basis for the selection of drug candidates for entry into development. However, it is important to point out that CSF drug concentration is not always an accurate surrogate for predicting unbound drug concentration in the brain. Depending on the physicochemical properties of drugs and the site/timing of CSF sampling, the unbound drug concentration at the biophase within the brain could differ significantly from the corresponding CSF drug concentration.

Pharmacokinetic and pharmacodynamic variability: a daunting challenge in drug therapy.

Patients vary considerably in their response to drug therapy. A drug that proves to be pharmacologically effective in some patients at a given dose may be ineffective or even toxic in others. The interindividual variability in drug response represents a major challenge in drug therapy, particularly for drugs with narrow therapeutic index. The intensity and duration of a drug action are determined not only by pharmacokinetic processes, but also by pharmacodynamic processes. Therefore, the variability in drug response is a result of the variability in either pharmacokinetic or pharmacodynamic processes, or a combination of both. The purpose of this paper is to review the sources that contribute to pharmacokinetic and pharmacodynamic variability. Although the main focus will be on the genetic variability, the impact of environmental factors on drug response will also be discussed. Finally, the application and limitation of the concept of personalized medicine will be briefly discussed.

Transporter-mediated drug interactions: clinical implications and in vitro assessment.

Although they are less frequently compared with the reported cases of CYP-mediated drug interactions, clinically significant transporter-mediated drug interactions, which are mainly based on efflux transporter or P-glycoprotein data, have been reported. Unlike the CYP-mediated drug interactions that can be readily defined by inhibition or induction of CYP enzymes, the evidence for the so-called transporter-mediated drug interactions is often less conclusive. The difficulty in defining transporter-mediated drug interactions is due mainly to the interplay between transporters and drug-metabolizing enzymes in drug disposition, and the lack of specific and potent inhibitors for each transporter and enzyme. An important lesson learned from animal studies is that transporter inhibition has a much greater impact on the tissue distribution of drugs than on the systemic exposure of drugs measured in plasma. The potential risk of transporter-mediated drug interactions might be underestimated if only plasma concentrations are monitored.

Complicating factors in safety testing of drug metabolites: kinetic differences between generated and preformed metabolites.

This paper aims to provide a scientifically based perspective on issues surrounding the proposed toxicology testing of synthetic drug metabolites as a means of ensuring adequate nonclinical safety evaluation of drug candidates that generate metabolites considered either to be unique to humans or are present at much higher levels in humans than in preclinical species. We put forward a number of theoretical considerations and present several specific examples where the kinetic behavior of a preformed metabolite given to animals or humans differs from that of the corresponding metabolite generated endogenously from its parent. The potential ramifications of this phenomenon are that the results of toxicity testing of the preformed metabolite may be misleading and fail to characterize the true toxicological contribution of the metabolite when formed from the parent. It is anticipated that such complications would be evident in situations where (a) differences exist in the accumulation of the preformed versus generated metabolites in specific tissues, and (b) the metabolite undergoes sequential metabolism to a downstream product that is toxic, leading to differences in tissue-specific toxicity. Owing to the complex nature of this subject, there is a need to treat drug metabolite issues in safety assessment on a case-by-case basis, in which a knowledge of metabolite kinetics is employed to validate experimental paradigms that entail administration of preformed metabolites to animal models.

In vitro and in vivo CYP3A64 induction and inhibition studies in rhesus monkeys: a preclinical approach for CYP3A-mediated drug interaction studies.

In this study, induction and inhibition of rhesus monkey CYP3A64 versus human CYP3A4 were characterized in vitro, and the corresponding pharmacokinetic consequences were evaluated in rhesus monkeys. In monkey hepatocytes, rifampin markedly induced CYP3A64 mRNA (EC50 = 0.5 microM; Emax = 6-fold) and midazolam (MDZ) 1'-hydroxylase activity (EC50 = 0.2 microM; Emax = 2-fold). Compound A (N-[2(R)-hydroxy-1(S)-indanyl-5-[2(S)-(1,1-dimethylethylaminocarbonyl)-4-[(furo[2,3-b]pyridin-5-yl)-methyl]piperazin-1-yl]-4(S)-hydroxy-2(R)-phenylmethylpentanamide), a known potent and mechanism-based inhibitor of CYP3A4, strongly inhibited the formation of 1'-hydroxy MDZ by recombinant CYP3A64 in a concentration- and time-dependent manner (KI = 0.25 microM; k(inact) = 0.4 min(-1)). Similar corresponding results also were obtained with human CYP3A4 in the presence of rifampin or compound A. In rhesus monkeys, MDZ exhibited a relatively high metabolic clearance (primarily via 1'-hydroxylation followed by glucuronidation) and a low hepatic availability (Fh = 16%). Consistent with the induction of hepatic metabolism of a high-clearance compound, pretreatment with rifampin (18 mg/kg p.o. for 5 days) did not significantly affect the i.v. kinetics of MDZ, but caused a pronounced reduction (approximately 10-fold) in the systemic exposure to MDZ and, consequently, its Fh following intrahepatic portal vein administration (i.pv.) of MDZ. A comparable extent of the pharmacokinetic interaction also was obtained after a 1.8 mg/kg rifampin dose. Also consistent with the in vitro CYP3A64 inhibition finding, compound A (6 mg/kg i.v.) markedly increased (10-fold) the i.pv. administered MDZ exposure. At the doses studied, plasma concentrations of rifampin or compound A reached or exceeded their respective in vitro EC50 or KI values. These findings suggest the potential applicability of the in vitro-in vivo relationship approach in rhesus monkeys for studying CYP3A-mediated interactions in humans.

CYP induction-mediated drug interactions: in vitro assessment and clinical implications.

Cytochrome P450 (CYP) induction-mediated interaction is one of the major concerns in clinical practice and for the pharmaceutical industry. There are two major issues associated with CYP induction: a reduction in therapeutic efficacy of comedications and an induction in reactive metabolite-induced toxicity. Because CYP induction is a metabolic liability in drug therapy, it is highly desirable to develop new drug candidates that are not potent CYP inducer to avoid the potential of CYP induction-mediated drug interactions. For this reason, today, many drug companies routinely include the assessment of CYP induction at the stage of drug discovery as part of the selection processes of new drug candidates for further clinical development. The purpose of this article is to review the molecular mechanisms of CYP induction and the clinical implications, including pharmacokinetic and pharmacodynamic consequences. In addition, factors that affect the degree of CYP induction and extrapolation of in vitro CYP induction data to in vivo situations will also be discussed. Finally, assessment of the potential of CYP induction at the drug discovery and development stage will be discussed.

Tissue distribution and pharmacodynamics: a complicated relationship.

With few exceptions, drugs exert their effects not within the plasma compartment, but in the defined target tissues. The process of drug distribution to the active site constitutes the "link-bridge" of the pharmacokinetic/pharmacodynamic (PK/PD) relationship. In spite of the importance of drug distribution as a key factor in determining pharmacologic response, research on drug distribution has historically received much less attention than that of absorption, metabolism, and excretion. The negligence of research on drug distribution is due mainly to the inaccessibility of the target tissues for obvious ethical reasons. In addition, lack of reliable experimental tools to assess the distribution process is also a major contributing factor. Because of this negligence, drug distribution has been referred to as "the forgotten relative in clinical pharmacokinetics." Although recent advances in molecular biology have led to the identification of many drug transporters, many of the processes of drug distribution are still not fully understood. The primary aim of this article is to provide new insight into the mechanisms of drug distribution, with an attempt to describe the relationship between the drug distribution and pharmacologic response. In addition, the factors that affect the processes of drug distribution will also be reviewed. Also, validity of some key assumptions that are used to relate the processes of tissue distribution with pharmacologic activity will be discussed.

Orally bioavailable highly potent HIV protease inhibitors against PI-resistant virus.

Efforts directed to identifying potent HIV protease inhibitors (PI) have yielded a class of compounds that are not only very active against wild-type (NL4-3) HIV virus but also very potent against a panel of PI-resistant viral isolates. Chemistry and biology are described.

A series of 5-(5,6)-dihydrouracil substituted 8-hydroxy-[1,6]naphthyridine-7-carboxylic acid 4-fluorobenzylamide inhibitors of HIV-1 integrase and viral replication in cells.

Introduction of a 5,6-dihydrouracil functionality in the 5-position of N-(4-fluorobenzyl)-8-hydroxy-[1,6]naphthyridine-7-carboxamide 1 led to a series of highly active HIV-1 integrase inhibitors. These compounds displayed low nanomolar activity in inhibiting both the strand transfer process of HIV-1 integrase and viral replication in cells. Compound 11 is a 150-fold more potent antiviral agent than 1, with a CIC(95) of 40 nM in the presence of human serum. It displays good pharmacokinetics when dosed in rats and dogs.

Interconversion pharmacokinetics of simvastatin and its hydroxy acid in dogs: effects of gemfibrozil.

PURPOSE: To characterize the pharmacokinetics of simvastatin (SV) and simvastatin acid (SVA), a lactone-acid pair known to undergo reversible metabolism, and to better understand mechanisms underlying pharmacokinetic interactions observed between SV and gemfibrozil. METHODS: Pharmacokinetic studies were conducted after intravenous administration of SV and SVA to dogs pretreated with a vehicle or gemfibrozil. In vitro metabolism of SVA in dog hepatocytes as well as in vitro hepatic and plasma conversion of SV/SVA were investigated in the absence and presence of gemfibrozil. RESULTS: In control animals, the irreversible elimination clearances of SV (CL10) and SVA (CL20) were 10.5 and 18.6 ml min(-1) kg(-1), respectively. The formation clearance of SVA from SV (CL12 = 4.8 ml min(-1) kg(-1)) was 8-fold greater than that of SV from SVA (CL21 = 0.6 ml min(-1) kg(-1)), and the recycled fraction was relatively minor (0.009). In gemfibrozil-treated animals, CL10 was essentially unchanged, whereas CL12, CL20, CL21, and recycled fraction were significantly decreased to 2.9, 9, 0.14 ml min(-1) kg(-1), and 0.003, respectively. In control dogs, values for real volume of distribution at steady state (Vss,real) of SV (2.3 L kg(-1)) were much larger than the corresponding values of SVA (0.3 L kg(-1)). Gemfibrozil treatment did not affect Vss,real of either SV or SVA. In dog hepatocytes, gemfibrozil modestly affected the formation of CYP3A-mediated oxidative metabolites (IC50 > 200 microM) and beta-oxidative products (IC5) approximately 100 microM), but markedly inhibited the glucuronidation-mediated lactonization of SVA and the glucuronidation of an SVA beta-oxidation product (IC50 = 18 microM). In in vitro dog and human liver S9 and plasma, hydrolysis of SV to SVA was much faster than that of SVA to SV. Gemfibrozil (250 microM) had a minimal inhibitory effect on the hydrolysis of either SV to SVA or SVA to SV in dog and human liver S9, but had a significant ( approximately 60%) inhibitory effect on the SV to SVA hydrolysis in both dog and human plasma. CONCLUSIONS: In dogs, the interconversion process favored the formation of SVA and was less efficient than the irreversible elimination processes of SV and SVA. Treatment with gemfibrozil did not affect the distribution of SV/SVA, but rather affected the elimination of SVA and the SV/SVA interconversion processes. Gemfibrozil decreased CL20 and CL21 likely via its inhibitory effect on the glucuronidation of SVA, and not on the CYP3A-mediated oxidative metabolism of SV or SVA, the beta-oxidation of SVA, nor the SVA to SV hydrolysis. The decrease in CL12 might be due in part to the inhibitory effect of gemfibrozil on SV to SVA hydrolysis in plasma. Similar rationales may also be applicable to studies in humans and/or other statin lactone-acid pairs.

Hepatic uptake of the novel antifungal agent caspofungin.

Caspofungin (CANCIDAS, a registered trademark of Merck & Co., Inc.) is a novel echinocandin antifungal agent used in the treatment of esophageal and invasive candidiases, invasive aspergillosis, and neutropenia. Available data suggest that the liver is a key organ responsible for caspofungin elimination in rodents and humans. Caspofungin is primarily eliminated by metabolic transformation; however, the rate of metabolism is slow. Accordingly, it was hypothesized that drug uptake transporters expressed on the basolateral domain of hepatocytes could significantly influence the extent of caspofungin uptake and subsequent elimination. In this study, experiments ranging from perfused rat livers to heterologous expression of individual hepatic uptake transporters were utilized to identify the transporter(s) responsible for the observed liver-specific uptake of this compound. Data from perfused rat liver studies were consistent with the presence of carrier-mediated caspofungin hepatic uptake, although this process appeared to be slow. To identify a relevant hepatic uptake transporter, we developed novel Tet-on HeLa cells expressing OATP1B1 (OATP-C, SLC21A6) and OATP1B3 (OATP8, SLC21A8), whose target gene can be overexpressed by the addition of doxycycline. A modest but statistically significant uptake of caspofungin was observed in cells overexpressing OATP1B1, but not OATP1B3. Taken together, these findings suggest that OATP1B1-mediated hepatic uptake may contribute to the overall elimination of this drug from the body.

Metabolism-based drug-drug interactions: what determines individual variability in cytochrome P450 induction?

Individual variability in cytochrome P450 (P450) induction comprises an important component contributing to the difficulties in assessing and predicting metabolism-based drug-drug interactions in humans. In this article, we outline the major factors responsible for the individual variability in P450 induction, including variable transporter activity and metabolism of inducers in vivo, genetic variations of P450 genes and their regulatory regions, genetic variations of receptors and regulatory proteins required for induction, and different physiological and environmental elements. With a better understanding of the major determinants in P450 induction and a profile of the phenotypes of these determinants in each individual, it is believed that the individual variability in induction-mediated drug-drug interactions can be adequately evaluated.

Interactions of human P-glycoprotein with simvastatin, simvastatin acid, and atorvastatin.

PURPOSE: In this study, P-glycoprotein (P-gp) mediated efflux of simvastatin (SV), simvastatin acid (SVA), and atorvastatin (AVA) and inhibition of P-gp by SV, SVA, and AVA were evaluated to assess the role of P-gp in drug interactions. METHODS: P-gp mediated efflux of SV, SVA, and AVA was determined by directional transport across monolayers of LLC-PK1 cells and LLC-PK1 cells transfected with human MDR1. Inhibition of P-gp was evaluated by studying the vinblastine efflux in Caco-2 cells and in P-gp overexpressing KBV1 cells at concentrations of SV, SVA, and AVA up to 50 microM. RESULTS: Directional transport studies showed insignificant P-gp mediated efflux of SV, and moderate P-gp transport [2.4-3.8 and 3.0-6.4 higher Basolateral (B) to Apical (A) than A to B transport] for SVA and AVA, respectively. Inhibition studies did not show the same trend as the transport studies with SV and AVA inhibiting P-gp (IC50 -25-50 microM) but SVA not showing any inhibition of P-gp. CONCLUSIONS: The moderate level of P-gp mediated transport and low affinity of SV, SVA, and AVA for P-gp inhibition compared to systemic drug levels suggest that drug interactions due to competition for P-gp transport is unlikely to be a significant factor in adverse drug interactions. Moreover, the inconsistencies between P-gp inhibition studies and P-gp transport of SV, SVA, and AVA indicate that the inhibition studies are not a valid means to identify statins as Pgp substrates.

Using real-time quantitative TaqMan RT-PCR to evaluate the role of dexamethasone in gene regulation of rat P-glycoproteins mdr1a/1b and cytochrome P450 3A1/2.

Cytochromes P450 (CYPs) and p-glycoproteins (Pgps) are believed to play important roles in drug absorption, metabolism, and elimination. Numerous drugs and environmental chemicals can modulate expression of these two classes of genes in different species. The present study investigated the effect of dexamethasone (Dex) on gene expression on both message and protein levels of mdr1a, mdr1b, CYP3A1, and CYP3A2 in small intestine, colon, liver, kidney, and brain microvessels of the rats treated orally with Dex at 1 or 20 mg/kg/day for 3 days. The basal expression of mdr1a mRNA was highest in the brain microvessels followed by colon, small intestine, liver, and kidney, and mdr1b mRNA was highest in the brain microvessels followed by kidney, liver, colon, and small intestine. After Dex treatment, mdr1a mRNA was increased by 5.5- and 10.7-fold in the small intestine, decreased extensively by 85-90% in the liver, and showed little or no change in the colon, kidney, and brain microvessels compared to the control rats. A similar pattern was observed for mdr1b mRNA. CYP3A1 mRNA was increased in all tissues examined. CYP3A2 mRNA was not significantly changed with the exception that at 20 mg/kg CYP3A2 mRNA was increased 5- and 30-fold in the colon and kidney. In general, Western blot analyses were consistent with mRNA changes. CYP3A protein expression was increased in all tissues examined. The disparity of the impact of Dex on the CYP 3A and Pgp expression in these studies suggest that the regulation of Pgp expression is very complex and is difficult to predict solely based on the PXR response to xenobiotics.

A naphthyridine carboxamide provides evidence for discordant resistance between mechanistically identical inhibitors of HIV-1 integrase.

The increasing incidence of resistance to current HIV-1 therapy underscores the need to develop antiretroviral agents with new mechanisms of action. Integrase, one of three viral enzymes essential for HIV-1 replication, presents an important yet unexploited opportunity for drug development. We describe here the identification and characterization of L-870,810, a small-molecule inhibitor of HIV-1 integrase with potent antiviral activity in cell culture and good pharmacokinetic properties. L-870,810 is an inhibitor with an 8-hydroxy-(1,6)-naphthyridine-7-carboxamide pharmacophore. The compound inhibits HIV-1 integrase-mediated strand transfer, and its antiviral activity in vitro is a direct consequence of this ascribed effect on integration. L-870,810 is mechanistically identical to previously described inhibitors from the diketo acid series; however, viruses selected for resistance to L-870,810 contain mutations (integrase residues 72, 121, and 125) that uniquely confer resistance to the naphthyridine. Conversely, mutations associated with resistance to the diketo acid do not engender naphthyridine resistance. Importantly, the mutations associated with resistance to each of these inhibitors map to distinct regions within the integrase active site. Therefore, we propose a model of the two inhibitors that is consistent with this observation and suggests specific interactions with discrete binding sites for each ligand. These studies provide a structural basis and rationale for developing integrase inhibitors with the potential for unique and nonoverlapping resistance profiles.