Improving predictions of compound amenability for liquid chromatography-mass spectrometry to enhance non-targeted analysis.

Mass-spectrometry-based non-targeted analysis (NTA), in which mass spectrometric signals are assigned chemical identities based on a systematic collation of evidence, is a growing area of interest for toxicological risk assessment. Successful NTA results in better identification of potentially hazardous pollutants within the environment, facilitating the development of targeted analytical strategies to best characterize risks to human and ecological health. A supporting component of the NTA process involves assessing whether suspected chemicals are amenable to the mass spectrometric method, which is necessary in order to assign an observed signal to the chemical structure. Prior work from this group involved the development of a random forest model for predicting the amenability of 5517 unique chemical structures to liquid chromatography-mass spectrometry (LC-MS). This work improves the interpretability of the group's prior model of the same endpoint, as well as integrating 1348 more data points across negative and positive ionization modes. We enhance interpretability by feature engineering, a machine learning practice that reduces the input dimensionality while attempting to preserve performance statistics. We emphasize the importance of interpretable machine learning models within the context of building confidence in NTA identification. The novel data were curated by the labeling of compounds as amenable or unamenable by expert curators, resulting in an enhanced set of chemical compounds to expand the applicability domain of the prior model. The balanced accuracy benchmark of the newly developed model is comparable to performance previously reported (mean CV BA is 0.84 vs. 0.82 in positive mode, and 0.85 vs. 0.82 in negative mode), while on a novel external set, derived from this work's data, the Matthews correlation coefficients (MCC) for the novel models are 0.66 and 0.68 for positive and negative mode, respectively. Our group's prior published models scored MCC of 0.55 and 0.54 on the same external sets. This demonstrates appreciable improvement over the chemical space captured by the expanded dataset. This work forms part of our ongoing efforts to develop models with higher interpretability and higher performance to support NTA efforts.

Wikipedia on the CompTox Chemicals Dashboard: Connecting Resources to Enrich Public Chemical Data.

The online encyclopedia Wikipedia aggregates a large amount of data on chemistry, encompassing well over 20,000 individual Wikipedia pages and serves the general public as well as the chemistry community. Many other chemical databases and services utilize these data, and previous projects have focused on methods to index, search, and extract it for review and use. We present a comprehensive effort that combines bulk automated data extraction over tens of thousands of pages, semiautomated data extraction over hundreds of pages, and fine-grained manual extraction of individual lists and compounds of interest. We then correlate these data with the existing contents of the U.S. Environmental Protection Agency's (EPA) Distributed Structure-Searchable Toxicity (DSSTox) database. This was performed with a number of intentions including ensuring as complete a mapping as possible between the Dashboard and Wikipedia so that relevant snippets of the article are loaded for the user to review. Conflicts between Dashboard content and Wikipedia in terms of, for example, identifiers such as chemical registry numbers, names, and InChIs and structure-based collisions such as SMILES were identified and used as the basis of curation of both DSSTox and Wikipedia. This work also allowed us to evaluate available data for sets of chemicals of interest to the Agency, such as synthetic cannabinoids, and expand the content in DSSTox as appropriate. This work also led to improved bidirectional linkage of the detailed chemistry and usage information from Wikipedia with expert-curated structure and identifier data from DSSTox for a new list of nearly 20,000 chemicals. All of this work ultimately enhances the data mappings that allow for the display of the introduction of the Wikipedia article in the community-accessible web-based EPA Comptox Chemicals Dashboard, enhancing the user experience for the thousands of users per day accessing the resource.

Assembly and Curation of Lists of Per- and Polyfluoroalkyl Substances (PFAS) to Support Environmental Science Research.

Per- and polyfluoroalkyl substances (PFAS) are a class of man-made chemicals of global concern for many health and regulatory agencies due to their widespread use and persistence in the environment (in soil, air, and water), bioaccumulation, and toxicity. This concern has catalyzed a need to aggregate data to support research efforts that can, in turn, inform regulatory and statutory actions. An ongoing challenge regarding PFAS has been the shifting definition of what qualifies a substance to be a member of the PFAS class. There is no single definition for a PFAS, but various attempts have been made to utilize substructural definitions that either encompass broad working scopes or satisfy narrower regulatory guidelines. Depending on the size and specificity of PFAS substructural filters applied to the U.S. Environmental Protection Agency (EPA) DSSTox database, currently exceeding 900,000 unique substances, PFAS substructure-defined space can span hundreds to tens of thousands of compounds. This manuscript reports on the curation of PFAS chemicals and assembly of lists that have been made publicly available to the community via the EPA's CompTox Chemicals Dashboard. Creation of these PFAS lists required the harvesting of data from EPA and online databases, peer-reviewed publications, and regulatory documents. These data have been extracted and manually curated, annotated with structures, and made available to the community in the form of lists defined by structure filters, as well as lists comprising non-structurable PFAS, such as polymers and complex mixtures. These lists, along with their associated linkages to predicted and measured data, are fueling PFAS research efforts within the EPA and are serving as a valuable resource to the international scientific community.

DNA-Encoded Library Technology-Based Discovery, Lead Optimization, and Prodrug Strategy toward Structurally Unique Indoleamine 2,3-Dioxygenase-1 (IDO1) Inhibitors.

We report the discovery of a novel indoleamine 2,3-dioxygenase-1 (IDO1) inhibitor class through the affinity selection of a previously unreported indole-based DNA-encoded library (DEL). The DEL exemplar, spiro-chromane 1, had moderate IDO1 potency but high in vivo clearance. Series optimization quickly afforded a potent, low in vivo clearance lead 11. Although amorphous 11 was highly bio-available, crystalline 11 was poorly soluble and suffered disappointingly low bio-availability because of solubility-limited absorption. A prodrug approach was deployed and proved effective in discovering the highly bio-available phosphonooxymethyl 31, which rapidly converted to 11 in vivo. Obtaining crystalline 31 proved problematic, however; thus salt screening was performed in an attempt to circumvent this obstacle and successfully delivered greatly soluble and bio-available crystalline tris-salt 32. IDO1 inhibitor 32 is characterized by a low calculated human dose, best-in-class potential, and an unusual inhibition mode by binding the IDO1 heme-free (apo) form.

Discovery of next generation inhibitors of HIV protease.

Due to factors such as resistance and long-term side effects as well as dosing regimen-related adherence issues, HIV therapy is a constantly moving target. HIV-1 protease inhibitors had an immediate and dramatic impact on the outcome of HIV/AIDS when launched in late 1995, and the search for new and improved next generation molecules has been under way in many laboratories. At GlaxoSmithKline (GSK) and Vertex Pharmaceuticals, this effort focused on two key issues, patient compliance and viral resistance. Using a water-solubilizing prodrug approach, the pill-burden in delivering our protease inhibitor, amprenavir, was dramatically decreased. By eliminating the large amounts of excipients necessary for the original soft-gel formulation, fosamprenavir (Lexiva/Telzir) delivers the clinically efficacious dose of amprenavir with two compact tablets per dose, compared to eight gel capsules. Our efforts to overcome viral resistance to 1(st) generation protease inhibitors by further elaborating the SAR of the amprenavir and related scaffolds, led to successive and dramatic improvements in wild-type antiviral potencies, and ultimately to the discovery of "ultra-potent" molecules with very favorable overall resistance profiles. The selection of GW640385 (brecanvir--USAN approved only) as a clinical candidate and its progression into current phase 2 dose ranging studies represents the culmination of our effort toward next generation protease inhibitors.

Neuromuscular blocking activity and therapeutic potential of mixed-tetrahydroisoquinolinium halofumarates and halosuccinates in rhesus monkeys.

Structure-activity relationships in rhesus monkeys for a novel mixed-onium class of ultra-short-acting nondepolarizing tetrahydroisoquinolinium neuromuscular blockers (NMBs) are described. Bis-onium chlorofumarate 20a with (1R,2S)-benzyltetrahydroisoquinolinium groups was a potent lead compound (ED(95) = 0.079 mg/kg) with an ultra-short duration of NMB effect (7.1 min) and a selectivity index (SI: defined as a ratio of the cardiovascular threshold dose to the ED(95)) similar to that of mivacurium (3). The mean threshold dose for cardiovascular effects with 20a was ca. 20 times its ED(95) value (SI = 20). A novel mixed-onium analogue of 20a was prepared by replacing the benzyltetrahydroisoquinolinium group distal to the fumarate chlorine atom with a (1S,2R)-phenyltetrahydroisoquinolinium moiety. The resulting mixed-onium chlorofumarate 24a displayed good NMB potency (ED(95) = 0.063 mg/kg), ultra-short duration of action (5.6 min) and an improved selectivity index (SI = 57). Several other mixed-onium derivatives containing octanedioate (25a; ED(95) = 0.103 mg/kg), difluorosuccinate (27c; ED(95) = 0.056 mg/kg), and fluorofumarate (28a; ED(95) = 0.137 mg/kg) linkers were also potent, ultra-short-acting NMBs with good to excellent selectivity index values (SI = 37-96). Octanedioate 25a was longer acting at higher doses compared to difluorosuccinate 27c and chlorofumarate 24a. Durations of NMB effect following a 0.4 mg/kg bolus dose (100% block) of 25a, 27c, and 24a were 16.9, 13.0, and 10.0 min, respectively. Recovery time for mixed-onium chlorofumarate 24a following a 1 h continuous infusion at 10-20 microg/kg/min (95-100% block) was ca. 5 min which is similar to that observed following a 0.2 mg/kg bolus dose of this compound and indicates a lack of cummulative effects. Preliminary studies with chlorofumarate 24a in whole human blood revealed that mixed-onium thiazolidine 29 was the major metabolite and that plasma cholinesterases do not play the primary role in duration of NMB effect. The NMB properties of 24a in rhesus monkeys led to its clinical evaluation as a possible alternative to succinylcholine.

Discovery of a potent boronic acid derived inhibitor of the HCV RNA-dependent RNA polymerase.

A boronic acid moiety was found to be a critical pharmacophore for enhanced in vitro potency against wild-type hepatitis C replicons and known clinical polymorphic and resistant HCV mutant replicons. The synthesis, optimization, and structure-activity relationships associated with inhibition of HCV replication in a subgenomic replication system for a series of non-nucleoside boron-containing HCV RNA-dependent RNA polymerase (NS5B) inhibitors are described. A summary of the discovery of 3 (GSK5852), a molecule which entered clinical trials in subjects infected with HCV in 2011, is included.

Discovery of 6,7-dihydro-5H-pyrrolo[2,3-a]pyrimidines as orally available G protein-coupled receptor 119 agonists.

GPR119 is a 7-transmembrane receptor that is expressed in the enteroendocrine cells in the intestine and in the islets of Langerhans in the pancreas. Indolines and 6,7-dihydro-5H-pyrrolo[2,3-a]pyrimidines were discovered as G protein-coupled receptor 119 (GPR119) agonists, and lead optimization efforts led to the identification of 1-methylethyl 4-({7-[2-fluoro-4-(methylsulfonyl)phenyl]-6,7-dihydro-5H-pyrrolo[2,3-d]pyrimidin-4-yl}oxy)-1-piperidinecarboxylate (GSK1104252A) (3), a potent and selective GPR119 agonist. Compound 3 showed excellent pharmacokinetic properties and sufficient selectivity with in vivo studies supporting a role for GPR119 in glucose homeostasis in the rodent. Thus, 3 appeared to modulate the enteroinsular axis, improve glycemic control, and strengthen previous suggestions that GPR119 agonists may have utility in the treatment of type 2 diabetes.

Synthesis of novel anilinoquinolines as c-fms inhibitors.

A novel series of potent substituted anilinoquinolines were discovered as c-fms inhibitors. The potency could be manipulated upon modification of the C4 aniline and C7 aryl functionality. Pharmacokinetic analysis identified a metabolically stable analog suitable for further investigative work.

Acyclic, orally bioavailable ketone-based cathepsin K inhibitors.

Starting from a potent ketone-based inhibitor with poor drug properties, incorporation of P(2)-P(3) elements from a ketoamide-based inhibitor led to the identification of a hybrid series of ketone-based cathepsin K inhibitors with better oral bioavailability than the starting ketone.

Aza-stilbenes as potent and selective c-RAF inhibitors.

The synthesis of several novel aza-stilbene derivatives was carried out. The compounds were tested for their c-RAF enzyme inhibition. Compound 27 possesses significant potency against c-RAF and demonstrates selectivity over other protein kinases. A hypothesis for the binding mode, activity, and selectivity is proposed.

Novel, potent P2-P3 pyrrolidine derivatives of ketoamide-based cathepsin K inhibitors.

Starting from a potent pantolactone ketoamide cathepsin K inhibitor discovered from structural screening, conversion of the lactone scaffold to a pyrrolidine scaffold allowed exploration of the S(3) subsite of cathepsin K. Manipulation of P3 and P1' groups afforded potent inhibitors with drug-like properties.

Pyrazolopyridine antiherpetics: SAR of C2' and C7 amine substituents.

A novel series of potent pyrazolo[1,5-a]pyridine inhibitors of herpes simplex virus 1 replication have been identified. Several complimentary synthetic methods were developed to allow facile access to a diverse set of analogs from common late stage intermediates. Detailed examination of the amine substituents at the C2' position of the pyrimidine and C7 position of the core pyrazolopyridine is described. The antiviral data suggests that non-polar amines are preferred for optimal activity. Additionally, the 2' position has been shown to require an NH group to retain activity levels similar to that of the gold standard acyclovir.

P2-P3 conformationally constrained ketoamide-based inhibitors of cathepsin K.

An orally bioavailable series of ketoamide-based cathepsin K inhibitors with good pharmacokinetic properties has been identified. Starting from a potent inhibitor endowed with poor drug properties, conformational constraint of the P(2)-P(3) linker and modifications to P(1') elements led to an enhancement in potency, solubility, clearance, and bioavailability. These optimized inhibitors attenuated bone resorption in a rat TPTX hypocalcemic bone resorption model.

Novel arylsulfonamides possessing sub-picomolar HIV protease activities and potent anti-HIV activity against wild-type and drug-resistant viral strains.

A novel series of P1' chain-extended arylsufonamides was synthesized and evaluated for wild-type HIV protease inhibitory activity and in vitro antiviral activity against wild type virus and two protease inhibitor-resistant mutant viruses. All of the compounds showed dramatic increases in enzyme activity as compared to the currently marketed HIV protease inhibitors amprenavir, indinavir, and nelfinavir. In addition, significant improvements in antiviral potencies against wild type and the two mutant viruses were also realized.

Inhibition of wild-type and mutant human immunodeficiency virus type 1 proteases by GW0385 and other arylsulfonamides.

The arylsulfonamide derivatives described herein were such potent inhibitors of human immunodeficiency virus type 1 (HIV-1) protease (enzyme, E) that values for the inhibition constants (K(i)) could not be determined by conventional steady-state kinetic techniques (i.e., the minimal enzyme concentration usable for the activity assay was much greater than the value of the dissociation constant). Consequently, two alternative methods were developed for estimation of K(i) values. The first method employed kinetic determinations of values for k(1) and k(-1), from which K(i) was determined (k(-1)/k(1)). The second method was a competitive displacement assay used to determine binding affinities of other inhibitors relative to that of GW0385. In these assays, the inhibitor of unknown affinity was used to displace [(3)H]GW0385 from E.[(3)H]GW0385. From the concentration of E.[(3)H]GW0385 at equilibrium, the concentrations of enzyme-bound and free inhibitors were calculated, and the ratio of the K(i) value of the unknown to that of GW0385 was determined (K(i,unknown)/K(i,GW0385)). The values of k(1) were calculated from data in which changes in the intrinsic protein fluorescence of the enzyme associated with inhibitor binding were directly or indirectly monitored. In the case of saquinavir, the fluorescence changes associated with complex formation were large enough to monitor directly. The value of k(1) for saquinavir was 62 +/- 2 microM(-1) s(-1). In the case of GW0385, the fluorescence changes associated with complex formation were too small to monitor directly. Consequently, the value of k(1) was estimated from a competition experiment in which the effect of GW0385 on the binding of E to saquinavir was determined. The value of k(1) for GW0385 was estimated from these experiments to be 137 +/- 4 microM(-1) s(-1). Because E.[(3)H]GW0385 was stable in the standard buffer at room temperature for greater than 33 days, the value of the first-order rate constant for dissociation of E.[(3)H]GW0385 (k(-1)) could be estimated from the time-course for exchange of E.[(3)H]GW0385 with excess unlabeled GW0385. The value of k(-1) calculated from these data was (2.1 +/- 0.1) x10(-6) s(-1) (t(1/2) = 91 h). The K(i) value of wild-type HIV-1 protease for GW0385, calculated from these values for k(1) and k(-1), was 15 +/- 1 fM. Three multidrug resistant enzymes had K(i) values for GW0385 that were less than 5 pM.

Preclinical pharmacology of GW280430A (AV430A) in the rhesus monkey and in the cat: a comparison with mivacurium.

BACKGROUND: No replacement for succinylcholine is yet available. GW280430A (AV430A) is a representative of a new class of nondepolarizing neuromuscular blocking drugs called asymmetric mixed-onium chlorofumarates. It undergoes rapid degradation in plasma by chemical hydrolysis and inactivation by cysteine adduction, resulting in a very short duration of effect. The neuromuscular, cardiovascular, and autonomic pharmacology of GW280430A is compared herein with that of mivacurium. METHODS: Adult male rhesus monkeys and adult male cats were anesthetized with nitrous oxide-oxygen-halothane and chloralose-pentobarbital, respectively. The neuromuscular blocking properties of GW280430A and mivacurium were compared at a stimulation rate of 0.15 Hz in the extensor digitorum of the foot (monkey) and the tibialis anterior (cat). Sympathetic responses were assayed in the cat in the nictitating membrane preparation, and vagal effects were evaluated in the cat via observation of bradycardic responses after stimulation of the cervical right vagus nerve. RESULTS: GW280430A and mivacurium were equipotent in the monkey (ED95 was 0.06 mg/kg in each case). GW280430A was half as potent as mivacurium in the cat. The total duration of action of GW280430A was less than half that of mivacurium in the monkey; recovery slopes were more than twice as rapid. The 25-75% recovery index of GW280430A did not vary significantly after various bolus doses or infusions, averaging 1.4-1.8 min in the monkey, significantly shorter than the same time interval (4.8-5.7 min) for mivacurium. Dose ratios for autonomic versus neuromuscular blocking properties in the cat were greater than 25 for both GW280430A and mivacurium. The ratio ED Hist:ED95 Neuromuscular Block in the monkey was significantly greater (approximately 53 vs. 13) for GW280430A, indicating approximately four times less relative prominence of the side effects of skin flushing and decrease of blood pressure, which are associated with release of histamine. CONCLUSIONS: These experiments show a much shorter neuromuscular blocking effect and much-reduced side effects in the case of GW280430A vis-à-vis mivacurium. These results, together with the novel chemical degradation of GW280430A, suggest further evaluation in human subjects.