BACKGROUND:The protease BACE1 is a major drug target for Alzheimer's disease, but chronic BACE1 inhibition is associated with non-progressive cognitive worsening that may be caused by modulation of unknown physiological BACE1 substrates. METHODS:To identify in vivo-relevant BACE1 substrates, we applied pharmacoproteomics to non-human-primate cerebrospinal fluid (CSF) after acute treatment with BACE inhibitors. RESULTS:Besides SEZ6, the strongest, dose-dependent reduction was observed for the pro-inflammatory cytokine receptor gp130/IL6ST, which we establish as an in vivo BACE1 substrate. Gp130 was also reduced in human CSF from a clinical trial with a BACE inhibitor and in plasma of BACE1-deficient mice. Mechanistically, we demonstrate that BACE1 directly cleaves gp130, thereby attenuating membrane-bound gp130 and increasing soluble gp130 abundance and controlling gp130 function in neuronal IL-6 signaling and neuronal survival upon growth-factor withdrawal. CONCLUSION:BACE1 is a new modulator of gp130 function. The BACE1-cleaved, soluble gp130 may serve as a pharmacodynamic BACE1 activity marker to reduce the occurrence of side effects of chronic BACE1 inhibition in humans.

This paper describes the structure-activity-relationships of novel fluoroalkyl substituents at the C2 position of iminothiazine dioxide beta secretase inhibitors. Key discoveries include reduced amidine basicity and its effect on Pgp, cell potency, and efficacy in various preclinical in vivo efficacy animal models. Findings from these structure-activity-relationships are discussed.

More than 20 y ago, we developed an animal model for chronic and continuous collection of cerebrospinal fluid (CSF) from conscious rhesus macaques. Since our previous publication in 2003, we have successfully implanted 168 rhesus macaques using this approach. Our experience enables us to provide up-to-date information regarding the model, including refine- ments to our implant design, reductions in maintenance, and new procedures for dealing with contamination. The results of our experiences have reduced the number of surgeries required and helped to increase the longevity of the implant, with some functioning for more than 18 y. Building on our success in rhesus macaques, we attempted to develop similar animal models in the African green monkeys and dogs but have been unable to develop reliable chronic models for CSF collection in these species.

Disrupted tau proteostasis and transneuronal spread is a pathological hallmark of Alzheimer’s disease. Neurodegenerative diseases remain an unmet medical need and novel disease modifying therapeutics are paramount. Our objective was to develop a mechanistic mathematical model to enhance our understanding of tau antibody pharmacokinetics and pharmacodynamics in animals and humans. A physiologically-based pharmacokinetic-pharmacodynamic (PBPK-PD) modeling approach was employed to support the preclinical development and clinical translation of therapeutic antibodies targeting tau for the treatment of Alzheimer’s disease. The pharmacokinetics of a tau antibody was evaluated in rat and non-human primate microdialysis studies. Model validation for humans was performed using publicly available clinical data for gosuranemab. In-silico analyses were performed to predict tau engagement in human brain for a range of tau antibody affinities and various dosing regimens. PBPK-PD modeling enabled a quantitative understanding for the relationship between dose, affinity, and target engagement, which supported lead candidate optimization and predictions of clinically efficacious dosing regimens.

We describe successful efforts to optimize the in vivo profile and address off-target liabilities of a series of BACE1 inhibitors represented by 6 that embodies the recently validated fused pyrrolidine iminopyrimidinone scaffold. Employing structure-based design, truncation of the cyanophenyl group of 6 that binds in the S3 pocket of BACE1 followed by modification of the thienyl group in S1 was pursued. Optimization of the pyrimidine substituent that binds in the S2'-S2″ pocket of BACE1 remediated time-dependent CYP3A4 inhibition of earlier analogues in this series and imparted high BACE1 affinity. These efforts resulted in the discovery of difluorophenyl analogue 9 (MBi-4), which robustly lowered CSF and cortex Aβ40 in both rats and cynomolgus monkeys following a single oral dose. Compound 9 represents a unique molecular shape among BACE inhibitors reported to potently lower central Aβ in nonrodent preclinical species.

Alzheimer's disease (AD) is the most common neurodegenerative dementia, with the Aβ peptides continuing as a top target for therapeutic intervention, either by reducing Aβ accumulation (secretase inhibition) or facilitating Aβ clearance/neutralization (immunotherapy). Oligomers have emerged as the most toxic Aβ species. Using our Aβ oligomer-specific CSF assay (J. Neurosci. 2014, 34(8):2884-97), we recently demonstrated a potential diagnostic ability of Aβ oligomers in early AD (MCI), with a robust 3-5-fold increase in oligomers in AD (N=63) vs. healthy controls (N=54). Herein, we further decipher the role of Aβ oligomers in AD, and suggest their novel use as a marker in evaluating drug/oligomer pharmacodynamics in preclinical and clinical models of APP processing. To expand our study of Aβ oligomers as a pharmacodynamic biomarker in response to secretase inhibition, we treated cisterna magna-ported Rhesus monkeys (N=6, 7-12 years old) with a Merck BACE inhibitor (BACEi: MBI-5; 30 or 125 mpk), gamma secretase inhibitor (GSi: MK-0752; 240 mpk) or vehicle in a four-way crossover design. CSF draws were taken up to 96 hrs post-treatment. We observed Aβ oligomers in vehicle-treated Rhesus’ CSF at concentrations similar to the healthy human CSF (∼1 pg/mL). Treatment with secretase inhibitors resulted in a significant (∼5-fold) reduction in oligomers in Rhesus CSF in a time- and dose-dependent fashion, with greatest inhibition between 9 and 31 hrs post-treatment. The kinetics and magnitude of response of oligomers differed compared with Aβ monomers and sAPPα/sAPPβ, enabling a revised model of Rhesus central nervous system APP processing that includes a role for Aβ oligomers. Together, these findings demonstrate potential utility of Aβ oligomers as not only a diagnostic and prognostic tool for human AD but, importantly, as a novel marker of pharmacodynamic response to secretase inhibitor therapy. Further, the novel oligomer-specific assay is sensitive enough to quantify/monitor Aβ oligomer levels in Rhesus CSF before and after secretase inhibitor therapy, suggesting it can also be used in human clinical trials of BACE inhibitors for the treatment of AD. Future studies will explore the correlation of CSF oligomers with cognitive decline in novel cross-sectional/longitudinal AD and healthy control cohorts.

BACE, a β-secretase, is an attractive potential disease-modifying therapeutic strategy for Alzheimer's disease (AD) as it results directly in the decrease of amyloid precursor protein (APP) processing through the β-secretase pathway and a lowering of CNS amyloid-β (Aβ) levels. The interaction of the β-secretase and α-secretase pathway-mediated processing of APP in the rhesus monkey (nonhuman primate; NHP) CNS is not understood. We hypothesized that CNS inhibition of BACE would result in decreased newly generated Aβ and soluble APPβ (sAPPβ), with increased newly generated sAPPα. A stable isotope labeling kinetics experiment in NHPs was performed with a (13)C6-leucine infusion protocol to evaluate effects of BACE inhibition on CNS APP processing by measuring the kinetics of sAPPα, sAPPβ, and Aβ in CSF. Each NHP received a low, medium, or high dose of MBI-5 (BACE inhibitor) or vehicle in a four-way crossover design. CSF sAPPα, sAPPβ, and Aβ were measured by ELISA and newly incorporated label following immunoprecipitation and liquid chromatography-mass spectrometry. Concentrations, kinetics, and amount of newly generated APP fragments were calculated. sAPPβ and sAPPα kinetics were similar, but both significantly slower than Aβ. BACE inhibition resulted in decreased labeled sAPPβ and Aβ in CSF, without observable changes in labeled CSF sAPPα. ELISA concentrations of sAPPβ and Aβ both decreased and sAPPα increased. sAPPα increased by ELISA, with no difference by labeled sAPPα kinetics indicating increases in product may be due to APP shunting from the β-secretase to the α-secretase pathway. These results provide a quantitative understanding of pharmacodynamic effects of BACE inhibition on NHP CNS, which can inform about target development.

Inhibition of Beta-site APP-cleaving enzyme 1 (BACE1), the β-secretase in the central nervous system (CNS), is an attractive potential disease modifying therapeutic strategy for Alzheimer's disease (AD) as it decreases amyloid precursor protein (APP) processing through the amyloidogenic pathway and CNS β-amyloid peptide levels. The interaction of the β-secretase (amyloidogenic) and α-secretase (non-amyloidogenic) pathway-mediated processing of APP in the rhesus monkey, non-human primate (NHP) CNS is not understood. We hypothesized that CNS inhibition of BACE1 would result in decreased newly-generated amyloid β peptide (Aβ) and sAPPβ, with increased sAPPα. A Stable Isotope Labeling Kinetic (SILK) experiment in NHPs was performed with a 13 C 6 -Leucine infusion protocol to evaluate the effects of a BACE1 inhibitor on CNS processing of APP by measuring the production rates and clearance rates of sAPPα, sAPPβ, and Aβ in cerebrospinal fluid (CSF). Each NHP received vehicle, low dose (10 mg/kg), medium dose (30 mg/kg), and a high dose (125 mg/kg) of the BACE1 inhibitor MBI-5 in a 4-way crossover design. CSF sAPPα, sAPPβ, and Aβ were measured by both ELISA and newly-incorporated label following immunopurification and liquid-chromatography-mass spectrometry. Concentrations, production rates, clearance rates, and total amount of newly-generated APP fragments were calculated. Models were developed to further explore the data and results. Vehicle-treated NHP Aβ production and clearance rates were similar to human measures. sAPPβ and sAPPα production and clearance rates were similar to each other, but much slower than Aβ. The newly-synthesized sAPPα to sAPPβ ratio in NHP CSF is approximately 3:1. Inhibition of BACE1 by MBI-5 resulted in a dose-dependent decrease in newly-labeled CSF sAPPβ and Aβ, without an observable change in newly-labeled CSF sAPPα, while the concentrations of sAPPβ and Aβ both decreased and sAPPα increased as measured by ELISA. While effects on the estimated, newly-synthesized Aβ and sAPPβ levels were directionally consistent to those measured by ELISA, the magnitude of AUC reductions in the newly-synthesized APP fragments, captured during steady state plasma 13 C leucine and peak MBI-5 levels, was greater. These results of BACE1 inhibition in the primate CNS suggest a common pool of APP for both the α-secretase and β-secretase pathways.

In Alzheimer's disease, one of the biochemical changes reported in the CNS is the elevation of the endogenous cholinergic and glutamatergic antagonist kynurenic acid (KYNA). KYNA is a metabolite within the kynurenine pathway that if chronically elevated could impair cognition. This pathway is one of the metabolic pathways for tryptophan degradation and also contains kynurenine (KYN), an important metabolite. Steady state levels of KYNA in the CNSare determined by several factors, including peripheral levels of KYN and TRP. In particular, KYN has been shown to readily cross the blood-brain barrier. Here we sought to understand both inter- and intra-individual variation of TRP, KYN, and KYNA in both plasma and CSF in rhesus macaques. Further, since TRP levels can be readily modulated by dietary manipulation, we monitored their responses to normal and high tryptophan containing food. Twelve cisterna magna ported rhesus macaques were used in these studies, eight males and four females with ages ranging from four to nineteen years old. Monkeys were divided into two groups and a crossover design was used, with those given TRP-enriched food considered the ‘treatment’ group and normal food the ‘control’. Each monkey was given TRP-enriched food twice and normal food twice with three to five days passing between studies. On a given day, plasma was sampled three times prior to ingestion, while CSF was sampled twice. CSF was collected through the indwelling port in the cisternamagna. Pre-ingestion values were used to determine baseline values for inter-and intra-individual comparisons, with four measurements per individual. Metabolites were quantified by LC/MS-MS. For plasma, both intra- and inter-individual variation for the three metabolites were determined. For CSF, values were only determined for KYNA. In general intra-individual variation ranged between 10-20% (CV) and was comparable to inter-individual variability. Two exceptions were plasma KYN and CSF KYNA, both of which had a clear age-dependent increase. In response to ingestion, there was a transient increase in all three metabolites in the plasma, with plasma KYNA showing the largest relative change. This peripheral change corresponded to a 20% increase in KYNA in the CSF. Lastly, plasma KYN showed the strongest correlation with CSF KYNA (r2=0.463, p < 0.0001). Plasma KYN and CSF KYNA show age-dependent changes in the plasma and CSF. Further, peripheral levels of KYN appear to impact CSF KYNA levels, consistent with evidence that KYN crosses the blood brain barrier. Understanding the normal variation of these metabolites will help clarify their function in neurodegenerative diseases.

The primary pathophysiology of peripheral artery occlusive disease is associated with impaired perfusion to the lower extremities. The lack of effective pharmacologic agents to treat this disease emphasizes the need for well-characterized animal models that can be used to evaluate the efficacy of emerging therapies. A major limitation with the current animal models of peripheral artery occlusive disease is that the variety of surgical methods employed to reduce peripheral blood flow produce differences in the severity and time course of the resting and reserve blood flow deficits. Furthermore, the methods used to evaluate the restoration of peripheral flow are often not suitable for serial measurements. This study used laser Doppler imaging to serially evaluate resting blood flow and the development of a functional collateral circulation after the induction of hind limb ischemia in the rat. Reserve blood flow was assessed by measuring hyperemic blood flow in the hind paw after temporary arterial occlusion. The magnitude of the hyperemic response was found to be dependent upon both the duration of arterial occlusion and the measurement time after release of the occluder. After ligation of the common iliac artery, but at a time when resting blood flow was reestablished, hyperemic tests unmasked a sustained deficit in reserve blood flow capacity that persisted for at least 14 days. Therefore, the use of a noninvasive vascular occluder and laser Doppler imaging represents a sensitive and consistent technique to measure peripheral blood flow status to assess the development of functional collateral vessels. These findings will enhance the ability to effectively study pharmacologic therapies aimed at promoting the growth and development of collateral vessels.