Fc-GDF15 glyco-engineering and receptor binding affinity optimization for body weight regulation.

GDF15 is a distant TGF-ß family member that induces anorexia and weight loss. Due to its function, GDF15 has attracted attention as a potential therapeutic for the treatment of obesity and its associated metabolic diseases. However, the pharmacokinetic and physicochemical properties of GDF15 present several challenges for its development as a therapeutic, including a short half-life, high aggregation propensity, and protease susceptibility in serum. Here, we report the design, characterization and optimization of GDF15 in an Fc-fusion protein format with improved therapeutic properties. Using a structure-based engineering approach, we combined knob-into-hole Fc technology and N-linked glycosylation site mutagenesis for half-life extension, improved solubility and protease resistance. In addition, we identified a set of mutations at the receptor binding site of GDF15 that show increased GFRAL binding affinity and led to significant half-life extension. We also identified a single point mutation that increases p-ERK signaling activity and results in improved weight loss efficacy in vivo. Taken together, our findings allowed us to develop GDF15 in a new therapeutic format that demonstrates better efficacy and potential for improved manufacturability.

A Mechanistic Site-Of-Action Model: A Tool for Informing Right Target, Right Compound, And Right Dose for Therapeutic Antagonistic Antibody Programs.

Quantitative modeling is increasingly utilized in the drug discovery and development process, from the initial stages of target selection, through clinical studies. The modeling can provide guidance on three major questions-is this the right target, what are the right compound properties, and what is the right dose for moving the best possible candidate forward. In this manuscript, we present a site-of-action modeling framework which we apply to monoclonal antibodies against soluble targets. We give a comprehensive overview of how we construct the model and how we parametrize it and include several examples of how to apply this framework for answering the questions postulated above. The utilities and limitations of this approach are discussed.

Small Molecule Proprotein Convertase Subtilisin/Kexin Type 9 (PCSK9) Inhibitors: Hit to Lead Optimization of Systemic Agents.

The optimization of a new class of small molecule PCSK9 mRNA translation inhibitors is described. The potency, physicochemical properties, and off-target pharmacology associated with the hit compound (1) were improved by changes to two regions of the molecule. The last step in the synthesis of the congested amide center was enabled by three different routes. Subtle structural changes yielded significant changes in pharmacology and off-target margins. These efforts led to the identification of 7l and 7n with overall profiles suitable for in vivo evaluation. In a 14-day toxicology study, 7l demonstrated an improved safety profile vs lead 7f. We hypothesize that the improved safety profile is related to diminished binding of 7l to nontranslating ribosomes and an apparent improvement in transcript selectivity due to the lower strength of 7l stalling of off-target proteins.

Assessing the Feasibility of Neutralizing Osteopontin with Various Therapeutic Antibody Modalities.

Osteopontin is a secreted glycophosphoprotein that is highly implicated in many physiological and pathological processes such as biomineralization, cell-mediated immunity, inflammation, fibrosis, cell survival, tumorigenesis and metastasis. Antibodies against osteopontin have been actively pursued as potential therapeutics for various diseases by pharmaceutical companies and academic laboratories. Many studies have demonstrated the efficacy of osteopontin inhibition in a variety of preclinical models of diseases such as rheumatoid arthritis, cancer, nonalcoholic steatohepatitis, but clinical utility has not yet been demonstrated. To evaluate the feasibility of osteopontin neutralization with antibodies in a clinical setting, we measured its physiological turnover rate in humans, a sensitive parameter required for mechanistic pharmacokinetic and pharmacodynamic (PK/PD) modeling of biotherapeutics. Results from a stable isotope-labelled amino acid pulse-chase study in healthy human subjects followed by mass spectrometry showed that osteopontin undergoes very rapid turnover. PK/PD modeling and simulation of different theoretical scenarios reveal that achieving sufficient target coverage using antibodies can be very challenging mostly due to osteopontin's fast turnover, as well as its relatively high plasma concentrations in human. Therapeutic antibodies against osteopontin would need to be engineered to have much extended PK than conventional antibodies, and be administered at high doses and with short dosing intervals.

Glyco-engineered Long Acting FGF21 Variant with Optimal Pharmaceutical and Pharmacokinetic Properties to Enable Weekly to Twice Monthly Subcutaneous Dosing.

Pharmacological administration of FGF21 analogues has shown robust body weight reduction and lipid profile improvement in both dysmetabolic animal models and metabolic disease patients. Here we report the design, optimization, and characterization of a long acting glyco-variant of FGF21. Using a combination of N-glycan engineering for enhanced protease resistance and improved solubility, Fc fusion for further half-life extension, and a single point mutation for improving manufacturability in Chinese Hamster Ovary cells, we created a novel FGF21 analogue, Fc-FGF21[R19V][N171] or PF-06645849, with substantially improved solubility and stability profile that is compatible with subcutaneous (SC) administration. In particular, it showed a low systemic clearance (0.243 mL/hr/kg) and long terminal half-life (~200 hours for intact protein) in cynomolgus monkeys that approaches those of monoclonal antibodies. Furthermore, the superior PK properties translated into robust improvement in glucose tolerance and the effects lasted 14 days post single SC dose in ob/ob mice. PF-06645849 also caused greater body weight loss in DIO mice at lower and less frequent SC doses, compared to previous FGF21 analogue PF-05231023. In summary, the overall PK/PD and pharmaceutical profile of PF-06645849 offers great potential for development as weekly to twice-monthly SC administered therapeutic for chronic treatment of metabolic diseases.

Liver-Targeted Small-Molecule Inhibitors of Proprotein Convertase Subtilisin/Kexin Type 9 Synthesis.

Targeting of the human ribosome is an unprecedented therapeutic modality with a genome-wide selectivity challenge. A liver-targeted drug candidate is described that inhibits ribosomal synthesis of PCSK9, a lipid regulator considered undruggable by small molecules. Key to the concept was the identification of pharmacologically active zwitterions designed to be retained in the liver. Oral delivery of the poorly permeable zwitterions was achieved by prodrugs susceptible to cleavage by carboxylesterase 1. The synthesis of select tetrazole prodrugs was crucial. A cell-free in vitro translation assay containing human cell lysate and purified target mRNA fused to a reporter was used to identify active zwitterions. In vivo PCSK9 lowering by oral dosing of the candidate prodrug and quantification of the drug fraction delivered to the liver utilizing an oral positron emission tomography 18 F-isotopologue validated our liver-targeting approach.

Once-weekly administration of a long-acting fibroblast growth factor 21 analogue modulates lipids, bone turnover markers, blood pressure and body weight differently in obese people with hypertriglyceridaemia and in non-human primates.

AIMS: To assess the safety, tolerability, pharmacokinetics and pharmacodynamics of PF-05231023, a long-acting fibroblast growth factor 21 (FGF21) analogue, in obese people with hypertriglyceridaemia on atorvastatin, with or without type 2 diabetes. METHODS: Participants received PF-05231023 or placebo intravenously once weekly for 4 weeks. Safety (12-lead ECGs, vital signs, adverse events [AEs], laboratory tests) and longitudinal weight assessments were performed. Blood samples were collected for pharmacokinetic and pharmacodynamic analyses. Cardiovascular safety studies were also conducted in telemetered rats and monkeys. Blood pressure (BP; mean, systolic and diastolic) and ECGs were monitored. RESULTS: A total of 107 people were randomized. PF-05231023 significantly decreased mean placebo-adjusted fasting triglycerides (day 25, 33%-43%) and increased HDL cholesterol (day 25, 15.7%-28.6%) and adiponectin (day 25, 1574 to 3272 ng/mL) across all doses, without significant changes in body weight (day 25, -0.45% to -1.21%). Modest decreases from baseline were observed for N-terminal propeptides of type 1 collagen (P1NP) on day 25, although C-telopeptide cross-linking of type 1 collagen (CTX-1) increased minimally. Systolic, diastolic BP, and pulse rate increased in a dose- and time-related manner. There were 5 serious AEs (one treatment-related) and no deaths. Three participants discontinued because of AEs. The majority of AEs were gastrointestinal. PF-05231023 increased BP and heart rate in rats, but not in monkeys. CONCLUSIONS: Once-weekly PF-05231023 lowered triglycerides markedly in the absence of weight loss, with modest changes in markers of bone homeostasis. This is the first report showing increases in BP and pulse rate in humans and rats after pharmacological administration of a long-acting FGF21 molecule.

Examination of the Human Cytochrome P4503A4 Induction Potential of PF-06282999, an Irreversible Myeloperoxidase Inactivator: Integration of Preclinical, In Silico, and Biomarker Methodologies in the Prediction of the Clinical Outcome.

The propensity for CYP3A4 induction by 2-(6-(5-chloro-2-methoxyphenyl)-4-oxo-2-thioxo-3,4-dihydropyrimidin-1(2H)-yl)acetamide (PF-06282999), an irreversible inactivator of myeloperoxidase, was examined in the present study. Studies using human hepatocytes revealed moderate increases in CYP3A4 mRNA and midazolam-1'-hydroxylase activity in a PF-06282999 dose-dependent fashion. At the highest tested concentration of 300 µM, PF-06282999 caused maximal induction in CYP3A4 mRNA and enzyme activity ranging from 56% to 86% and 47% t0 72%, respectively, of rifampicin response across the three hepatocyte donor pools. In a clinical drug-drug interaction (DDI) study, the mean midazolam Cmax and area under the curve (AUC) values following 14-day treatment with PF-06282999 decreased in a dose-dependent fashion with a maximum decrease in midazolam AUC0-inf and Cmax of ∼57.2% and 41.1% observed at the 500 mg twice daily dose. The moderate impact on midazolam pharmacokinetics at the 500 mg twice daily dose of PF-06282999 was also reflected in statistically significant changes in plasma 4ß-hydroxycholesterol/cholesterol and urinary 6ß-hydroxycortisol/cortisol ratios. Changes in plasma and urinary CYP3A4 biomarkers did not reach statistical significance at the 125 mg three times daily dose of PF-06282999, despite a modest decrease in midazolam systemic exposure. Predicted DDI magnitude based on the in vitro induction parameters and simulated pharmacokinetics of perpetrator (PF-06282999) and victim (midazolam) using the Simcyp (Simcyp Ltd., Sheffield, United Kingdom) population-based simulator were in reasonable agreement with the observed clinical data. Since the magnitude of the 4ß-hydroxycholesterol or 6ß-hydroxycortisol ratio change was generally smaller than the magnitude of the midazolam AUC change with PF-06282999, a pharmacokinetic interaction study with midazolam ultimately proved important for assessment of DDI via CYP3A4 induction.

Physiologically relevant binding affinity quantification of monoclonal antibody PF-00547659 to mucosal addressin cell adhesion molecule for in vitro in vivo correlation.

BACKGROUND AND PURPOSE: A monoclonal antibody (PF-00547659) against mucosal addressin cell adhesion molecule (MAdCAM), expressed as both soluble (sMAdCAM) and trans-membrane (mMAdCAM) target forms, showed over 30-fold difference in antibody-target KD between in vitro (Biacore) and clinically derived (KD,in-vivo ) values. Back-scattering interferometry (BSI) was applied to acquire physiologically relevant KD values which were used to establish in vitro and in vivo correlation (IVIVC). EXPERIMENTAL APPROACH: BSI was applied to obtain KD values between PF-00547659 and recombinant human MAdCAM in buffer or CHO cells and endogenous MAdCAM in human serum or colon tissue. CHO cells and tissue were minimally processed to yield homogenate containing membrane vesicles and soluble proteins. A series of binding affinities in serum with various dilution factors was used to estimate both KD,in-vivo and target concentrations; MAdCAM concentrations were also measured using LC-MS/MS. KEY RESULTS: BSI measurements revealed low KD values (higher affinity) for sMAdCAM in buffer and serum, yet a 20-fold higher KD value (lower affinity) for mMAdCAM in CHO, mMAdCAM and sMAdCAM in tissue. BSI predicted KD,in-vivo in serum was similar to clinically derived KD,in-vivo , and the BSI-estimated serum sMAdCAM concentration also matched the measured concentration by LC-MS/MS. CONCLUSIONS AND IMPLICATIONS: Our results successfully demonstrated that BSI measurements of physiologically relevant KD values can be used to establish IVIVC, for PF-00547659 to MAdCAM despite the lack of correlation when using Biacore measured KD and accurately estimates endogenous target concentrations. The application of BSI would greatly enhance successful basic pharmacological research and drug development.

A Long-Acting FGF21 Molecule, PF-05231023, Decreases Body Weight and Improves Lipid Profile in Non-human Primates and Type 2 Diabetic Subjects.

FGF21 plays a central role in energy, lipid, and glucose homeostasis. To characterize the pharmacologic effects of FGF21, we administered a long-acting FGF21 analog, PF-05231023, to obese cynomolgus monkeys. PF-05231023 caused a marked decrease in food intake that led to reduced body weight. To assess the effects of PF-05231023 in humans, we conducted a placebo-controlled, multiple ascending-dose study in overweight/obese subjects with type 2 diabetes. PF-05231023 treatment resulted in a significant decrease in body weight, improved plasma lipoprotein profile, and increased adiponectin levels. Importantly, there were no significant effects of PF-05231023 on glycemic control. PF-05231023 treatment led to dose-dependent changes in multiple markers of bone formation and resorption and elevated insulin-like growth factor 1. The favorable effects of PF-05231023 on body weight support further evaluation of this molecule for the treatment of obesity. Longer studies are needed to assess potential direct effects of FGF21 on bone in humans.

Pharmacokinetics (PK), pharmacodynamics (PD) and integrated PK/PD modeling of a novel long acting FGF21 clinical candidate PF-05231023 in diet-induced obese and leptin-deficient obese mice.

Pharmacological administration of fibroblast growth factor 21 (FGF21) improves metabolic profile in preclinical species and humans. FGF21 exerts its metabolic effects through formation of beta-klotho (KLB)/FGF receptor 1c FGFR1c complex and subsequent signaling. Data from various in vitro systems demonstrate the intact C- and N-terminus of FGF21 is required for binding with KLB, and interaction with FGFR1c, respectively. However the relative roles of the termini for in vivo pharmacological effects are unclear. Here we report PF-05231023, a long-acting FGF21 analogue which is unique in that the half-life and subcutaneous (s.c.) bioavailability of the intact C-terminus are significantly different from those of the intact N-terminus (2 vs. 22 hr for half-life and 4~7 vs. ~50% SC bioavailability). Therefore, this molecule serves as a valuable tool to evaluate the relative roles of intact C-terminus vs. N-terminus in in vivo pharmacology studies in preclinical species. We determined the effects of PF-05231023 administration on body weight (BW) loss and glucose reduction during an oral glucose tolerance test (OGTT) following SC and intravenous (i.v.) administration in diet-induced obese (DIO) and leptin-deficient obese (ob/ob) mice, respectively. Our data show that the intact N-terminus of FGF21 in PF-05231023 appears to be sufficient to drive glucose lowering during OGTT and sustain BW loss in DIOs. Further, PK/PD modeling suggests that while the intact FGF21 C-terminus is not strictly required for glucose lowering during OGTT in ob/ob mice or for BW reduction in DIO mice, the higher potency conferred by intact C-terminus contributes to a rapid initiation of pharmacodynamic effects immediately following dosing. These results provide additional insight into the strategy of developing stabilized versions of FGF21 analogs to harness the full spectrum of its metabolic benefits.

Comprehensive mechanism-based antibody pharmacokinetic modeling.

Pharmacokinetic models of antibody distribution and dynamics are useful for predicting and optimizing therapeutic behavior. Targeted antigens are produced and distributed in various tissues in specific patterns in disease phenotypes. Existing models leave out significant mechanistic detail which would enable an understanding of how to modify therapeutics in an optimal manner to allow appropriate tissue penetration in either a healthy or diseased state. The model presented here incorporates additional complexity such as diffusion through endothelial barriers, differential transcytosis properties, FcRn-mediated recycling, and incorporates these properties in an organ-specific manner. This creates a platform which can be expanded upon to include understanding of the effect of target on therapeutic distribution and clearance, differences in dynamics during a diseased versus healthy state, differential dose strategies, and mechanistic translation between animal models and human disease state. This model represents a superior alternative to typical and potentially over-simplified scaling strategies utilized in most existing physiologically-based pharmacokinetic models. Ultimately, this will enable better therapeutic design and greater pharmacological effects.

Two additive mechanisms impair the differentiation of 'substrate-selective' p38 inhibitors from classical p38 inhibitors in vitro.

BACKGROUND: The success of anti-TNF biologics for the treatment of rheumatoid arthritis has highlighted the importance of understanding the intracellular pathways that regulate TNF production in the quest for an orally-available small molecule inhibitor. p38 is known to strongly regulate TNF production via MK2. The failure of several p38 inhibitors in the clinic suggests the importance of other downstream pathways in normal cell function. Recent work has described a 'substrate-selective' p38 inhibitor that is able to preferentially block the activity of p38 against one substrate (MK2) versus another (ATF2). Using a combined experimental and computational approach, we have examined this mechanism in greater detail for two p38 substrates, MK2 and ATF2. RESULTS: We found that in a dual (MK2 and ATF2) substrate assay, MK2-p38 interaction reduced the activity of p38 against ATF2. We further constructed a detailed kinetic mechanistic model of p38 phosphorylation in the presence of multiple substrates and successfully predicted the performance of classical and so-called 'substrate-selective' p38 inhibitors in the dual substrate assay. Importantly, it was found that excess MK2 results in a stoichiometric effect in which the formation of p38-MK2-inhibitor complex prevents the phosphorylation of ATF2, despite the preference of the compound for the p38-MK2 complex over the p38-ATF2 complex. MK2 and p38 protein expression levels were quantified in U937, Thp-1 and PBMCs and found that [MK2] > [p38]. CONCLUSION: Our integrated mechanistic modeling and experimental validation provides an example of how systems biology approaches can be applied to drug discovery and provide a basis for decision-making with limited chemical matter. We find that, given our current understanding, it is unlikely that 'substrate-selective' inhibitors of p38 will work as originally intended when placed in the context of more complex cellular environments, largely due to a stoichiometric excess of MK2 relative to p38.

Cellular imaging predictions of clinical drug-induced liver injury.

Drug-induced liver injury (DILI) is the most common adverse event causing drug nonapprovals and drug withdrawals. Using drugs as test agents and measuring a panel of cellular phenotypes that are directly linked to key mechanisms of hepatotoxicity, we have developed an in vitro testing strategy that is predictive of many clinical outcomes of DILI. Mitochondrial damage, oxidative stress, and intracellular glutathione, all measured by high content cellular imaging in primary human hepatocyte cultures, are the three most important features contributing to the hepatotoxicity prediction. When applied to over 300 drugs and chemicals including many that caused rare and idiosyncratic liver toxicity in humans, our testing strategy has a true-positive rate of 50-60% and an exceptionally low false-positive rate of 0-5%. These in vitro predictions can augment the performance of the combined traditional preclinical animal tests by identifying idiosyncratic human hepatotoxicants such as nimesulide, telithromycin, nefazodone, troglitazone, tetracycline, sulindac, zileuton, labetalol, diclofenac, chlorzoxazone, dantrolene, and many others. Our findings provide insight to key DILI mechanisms, and suggest a new approach in hepatotoxicity testing of pharmaceuticals.

Analysis of mechanistic pathway models in drug discovery: p38 pathway.

Mechanistic models of signal transduction have emerged as valuable tools for untangling complex signaling networks and gaining detailed insight into pathway dynamics. The natural extension of these tools is for the design of therapeutic strategies. We have generated a novel computational model of lipopolysaccharide-induced p38 signaling in the context of TNF-alpha production in inflammatory disease. Using experimental measurement of protein levels and phospho-protein time courses, populations of model parameters were estimated. With a collection of parameter sets, reflecting virtual diversity, we step through analysis of the p38 signaling pathway model to answer specific drug discovery questions regarding target prioritization, inhibitor simulation, model robustness and co-drugging. We demonstrate that target selection cannot be assessed independently from inhibitor mechanism of action and is also linked with robustness to cellular variability. Finally, we assert that in the face of parameter uncertainty one can still uncover consistent findings that can guide drug discovery efforts.

Stochastic gene expression out-of-steady-state in the cyanobacterial circadian clock.

Recent advances in measuring gene expression at the single-cell level have highlighted the stochastic nature of messenger RNA and protein synthesis. Stochastic gene expression creates a source of variability in the abundance of cellular components, even among isogenic cells exposed to an identical environment. Recent integrated experimental and modelling studies have shed light on the molecular sources of this variability. However, many of these studies focus on systems that have reached a steady state and therefore do not address a large class of dynamic phenomena including oscillatory gene expression. Here we develop a general protocol for analysing and predicting stochastic gene expression in systems that never reach steady states. We use this framework to analyse experimentally stochastic expression of genes driven by the Synechococcus elongatus circadian clock. We find that, although the average expression at two points in the circadian cycle separated by 12 hours is identical, the variability at these two time points can be different. We show that this is a general feature of out-of-steady-state systems. We demonstrate how intrinsic noise sources, owing to random births and deaths of mRNAs and proteins, or extrinsic noise sources, which introduce fluctuations in rate constants, affect the cell-to-cell variability. To distinguish experimentally between these sources, we measured how the correlation between expression fluctuations of two identical genes is modulated during the circadian cycle. This quantitative framework is generally applicable to any out-of-steady-state system and will be necessary for understanding the fidelity of dynamic cellular systems.

Probing polymerization forces by using actin-propelled lipid vesicles.

Actin polymerization provides a powerful propulsion force for numerous types of cell motility. Although tremendous progress has been made in identifying the biochemical components necessary for actin-based motility, the precise biophysical mechanisms of force generation remain unclear. To probe the polymerization forces quantitatively, we introduce an experimental system in which lipid vesicles coated with the Listeria monocytogenes virulence factor ActA are propelled by actin polymerization. The polymerization forces cause significant deformations of the vesicle. We have used these deformations to obtain a spatially resolved measure of the forces exerted on the membrane using a model based on the competition between osmotic pressure and membrane stretching. Our results indicate that actin exerts retractile or propulsive forces depending on the local membrane curvature and that the membrane is strongly bound to the actin gel. These results are consistent with the observed dynamics. After a slow elongation of the vesicle from a spherical shape, the strong bonds between the actin gel and the membrane rupture if the retractile forces exceed a critical value, leading to a rapid release of the vesicle's trailing edge.