Publications

T Cell Engager (TCE)s are an exciting therapeutic modality in immuno-oncology that acts to bypass antigen presentation and forms a direct link between cancer and immune cells in the Tumor Microenvironment (TME). TCEs are efficacious only when the drug is bound to both immune and cancer cell targets. Therefore, approaches that maximize the formation of the drug-target trimer in the TME are expected to increase the drug's efficacy. In this study, we quantitatively investigate how the concentration of ternary complex and its biodistribution depend on both the targets' specific properties and the design characteristics of the TCE, and specifically on the binding kinetics of the drug to its targets. A simplified mathematical model of drug-target interactions is considered here, with insights from the "three-body" problem applied to the model. Parameter identifiability analysis performed on the model demonstrates that steady state data, which is often available at the early pre-clinical stages, is sufficient to estimate the binding affinity of the TCE molecule to both targets. We used the model to analyze several existing antibodies, both clinically approved and under development, to explore their common kinetic features. The manuscript concludes with an assessment of a full quantitative pharmacology model that accounts for drug disposition into the peripheral compartment.

Iron dysregulation has been implicated in multiple neurodegenerative diseases, including Parkinson's disease (PD). Iron-loaded microglia are frequently found in affected brain regions, but how iron accumulation influences microglia physiology and contributes to neurodegeneration is poorly understood. Here we show that human induced pluripotent stem cell-derived microglia grown in a tri-culture system are highly responsive to iron and susceptible to ferroptosis, an iron-dependent form of cell death. Furthermore, iron overload causes a marked shift in the microglial transcriptional state that overlaps with a transcriptomic signature found in PD postmortem brain microglia. Our data also show that this microglial response contributes to neurodegeneration, as removal of microglia from the tri-culture system substantially delayed iron-induced neurotoxicity. To elucidate the mechanisms regulating iron response in microglia, we performed a genome-wide CRISPR screen and identified novel regulators of ferroptosis, including the vesicle trafficking gene SEC24B. These data suggest a critical role for microglia iron overload and ferroptosis in neurodegeneration.

In order to control highly-contagious and prolonged outbreaks, public health authorities intervene to institute social distancing, lock-down policies, and other Non-Pharmaceutical Interventions (NPIs). Given the high social, educational, psychological, and economic costs of NPIs, authorities tune them, alternatively tightening up or relaxing rules, with the result that, in effect, a relatively flat infection rate results. For example, during the summer of 2020 in parts of the United States, daily COVID-19 infection numbers dropped to a plateau. This paper approaches NPI tuning as a control-theoretic problem, starting from a simple dynamic model for social distancing based on the classical SIR epidemics model. Using a singular-perturbation approach, the plateau becomes a Quasi-Steady-State (QSS) of a reduced two-dimensional SIR model regulated by adaptive dynamic feedback. It is shown that the QSS can be assigned and it is globally asymptotically stable. Interestingly, the dynamic model for social distancing can be interpreted as a nonlinear integral controller. Problems of data fitting and parameter identifiability are also studied for this model. This letter also discusses how this simple model allows for a meaningful study of the effect of population size, vaccinations, and the emergence of second waves.

Social distancing as a form of nonpharmaceutical intervention has been enacted in many countries as a form of mitigating the spread of COVID-19. There has been a large interest in mathematical modeling to aid in the prediction of both the total infected population and virus-related deaths, as well as to aid government agencies in decision making. As the virus continues to spread, there are both economic and sociological incentives to minimize time spent with strict distancing mandates enforced, and/or to adopt periodically relaxed distancing protocols, which allow for scheduled economic activity. The main objective of this study is to reduce the disease burden in a population, here measured as the peak of the infected population, while simultaneously minimizing the length of time the population is socially distanced, utilizing both a single period of social distancing as well as periodic relaxation. We derive a linear relationship among the optimal start time and duration of a single interval of social distancing from an approximation of the classic epidemic SIR model. Furthermore, we see a sharp phase transition region in start times for a single pulse of distancing, where the peak of the infected population changes rapidly; notably, this transition occurs well before one would intuitively expect. By numerical investigation of more sophisticated epidemiological models designed specifically to describe the COVID-19 pandemic, we see that all share remarkably similar dynamic characteristics when contact rates are subject to periodic or one-shot changes, and hence lead us to conclude that these features are universal in epidemic models. On the other hand, the nonlinearity of epidemic models leads to non-monotone behavior of the peak of infected population under periodic relaxation of social distancing policies. This observation led us to hypothesize that an additional single interval social distancing at a proper time can significantly decrease the infected peak of periodic policies, and we verified this improvement numerically. While synchronous quarantine and social distancing mandates across populations effectively minimize the spread of an epidemic over the world, relaxation decisions should not be enacted at the same time for different populations.

Background Whole-genome sequencing data are available from several large studies across a variety of diseases and traits. However, massive storage and computation resources are required to use these data, and to achieve sufficient power for discoveries, harmonization of multiple cohorts is critical. Objectives The Accelerating Medicines Partnership Parkinson's Disease program has developed a research platform for Parkinson's disease (PD) that integrates the storage and analysis of whole-genome sequencing data, RNA expression data, and clinical data, harmonized across multiple cohort studies. Methods The version 1 release contains whole-genome sequencing data derived from 3941 participants from 4 cohorts. Samples underwent joint genotyping by the TOPMed Freeze 9 Variant Calling Pipeline. We performed descriptive analyses of these whole-genome sequencing data using the Accelerating Medicines Partnership Parkinson's Disease platform. Results The clinical diagnosis of participants in version 1 release includes 2005 idiopathic PD patients, 963 healthy controls, 64 prodromal subjects, 62 clinically diagnosed PD subjects without evidence of dopamine deficit, and 705 participants of genetically enriched cohorts carrying PD risk-associated GBA variants or LRRK2 variants, of whom 304 were affected. We did not observe significant enrichment of pathogenic variants in the idiopathic PD group, but the polygenic risk score was higher in PD both in nongenetically enriched cohorts and genetically enriched cohorts. The population analysis showed a correlation between genetically enriched cohorts and Ashkenazi Jewish ancestry. Conclusions We describe the genetic component of the Accelerating Medicines Partnership Parkinson's Disease platform, a solution to democratize data access and analysis for the PD research community. © 2021 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society. This article is a U.S. Government work and is in the public domain in the USA.

The efficiency and effectiveness of airport security can be improved through fast and high-fidelity detection of concealed threat objects. If detected objects can be accurately characterized, the number of false alarms will be reduced, whereas terrorist threats can be more quickly noticed and apprehended. This article aims to accurately characterize the physical properties of weak dielectric objects (such as explosive threats) covering a conductive plane (such as the human body) using a single-frequency millimeter-wave (mm-wave) radar system with a Fresnel aperture. While the conventional synthetic aperture radar (SAR) method requires data over a broad bandwidth, our proposed method works even at a single frequency and is particularly suitable for focused-beam CW mm-wave sensing systems. This method is validated with experimental data.

We consider a nonlinear SISO system that is a cascade of a scalar "bottleneck entrance" with a stable positive linear system. In response to any periodic inflow, all solutions converge to a unique periodic solution with the same period. We study the problem of maximizing the averaged throughput via controlled switching. We compare two strategies: 1) switching between a high and low value, and 2 using a constant inflow equal to the prescribed mean value. We show that no possible switching policy can outperform a constant inflow rate, though it can approach it asymptotically. We describe several potential applications of this problem in traffic systems, ribosome flow models, and scheduling at security checks.

We propose a fast and accurate method of finding the thickness and dielectric permittivity of weak dielectric objects using a focused continuous-wave radar security scanning system, to characterize objects that may be explosive threats. Building on the previously developed ray-based Virtual Source model for sensing scattering from slabs, this work presents an inversion algorithm using two observable CW millimeter wave probe parameters by considering expected small variations of dielectric slab thickness over its surface. The determination works reasonably well with a simulated dielectric covered conductive ground plane as a representation of body worn explosives.

We develop a Virtual Source Model which is an intuitive, non-iterative, forward model to analyze the differential scattering of a dielectric covered conductive ground plane. This simple model is well suited for focused beam, nearfield mm-wave sensing of body-worn explosives for person scanning security systems. Ray analysis using the method of images for multiple interfaces yields good, experimentally-validated results.

We present a novel non-iterative model based on ray analysis to characterize non-metallic, weak dielectric objects (like threat objects) on the surface of a highly conducting background (like the human body) using a focused continuous millimeter-wave sensor. For a simple constant thickness dielectric slab on a ground plane, there are five primary scattering phenomena that must be considered.

Objectives: Tividenofusp alfa is an intravenous (IV) enzyme replacement therapy comprising iduronate 2-sulfatase enzyme fused to an engineered transferrin receptor (TfR)-binding Fc domain [1]. Tividenofusp alfa is the first FDA-approved medicine designed to leverage the TfR to enable delivery across the blood–brain barrier and address both central nervous system (CNS) and somatic manifestations of mucopolysaccharidosis type II (MPS II), a rare X-linked lysosomal disease. This analysis characterized the population pharmacokinetics (PK) and pharmacodynamics (PD) of tividenofusp alfa in participants with MPS II, linking exposure to heparan sulfate (HS) levels in cerebrospinal fluid (CSF) and urine, and evaluating covariate effects. Methods: Population PK and sequential PK/PD modeling were performed using data from a Phase 1/2 study in 47 participants with MPS II (NCT04251026) [2], including serum PK and longitudinal CSF and urine HS concentrations following weekly IV infusions of tividenofusp alfa. The M3 method [3] was used to handle observations below the quantification limit. Semi-quantitative anti-drug antibody (ADA) titer levels were used to characterize ADA impact on PK. Covariate analysis was conducted based on FDA guidance [4]. Effect compartment models were used for CSF and urine HS biomarkers to characterize CNS and peripheral effects of tividenofusp alfa, and simulated PD data were normalized to age-dependent normal ranges [5]. The final PK/PD models were used to investigate exposure/response relationships for probability of normalization of PD biomarkers via simulation. Results: The model that best captured PK was a two-compartment model incorporating linear clearance, and two nonlinear clearance mechanisms [6] representing target-mediated drug disposition (TMDD) and ADAs. Key covariates included age, body weight, and genotype. Post hoc PK simulations showed no significant differences across covariate subgroups for maximum concentration, and differences in area under the curve were within interparticipant variability. For the PK/PD model of CSF HS, increased ADA titers were associated with a prolonged effect compartment half-life (t1/2), and lower baseline HS levels were associated with decreased maximal effect (Imax). For the urine HS model, Imax decreased with both decreasing age and baseline HS levels and half-maximal inhibitory concentration increased with higher ADA titers. Lower baseline urine HS levels were also associated with shorter effect compartment t1/2. For both CSF and urine, the effect of baseline HS levels on Imax suggests a lower limit defined by normal ranges. Based on the population PK/PD simulations, higher serum tividenofusp alfa concentrations were associated with greater reductions of CSF and urine HS from baseline, with maximum effect achieved at 15 mg/kg weekly. Conclusions: A population PK model incorporating TMDD- and ADA-related clearances, together with effect compartment PK/PD models, characterized tividenofusp alfa exposure and PD biomarkers. Simulated exposure differences across covariate subgroups were not statistically significant. Population PK/PD simulations supported the maintenance dose of 15 mg/kg for the accelerated approval of tividenofusp alfa. Citations : [1] Arguello A et al. J Exp Med 2022;219:e20211057. [2] Muenzer J et al. N Engl J Med 2026;394:39–50. [3] Beal SL. J Pharmacokinet Pharmacodyn 2001;28:481–504. [4] FDA. 2022. Available from: https://www.fda.gov/media/128793/download (Accessed December 14, 2023). [5] Herber CB, et al. Mol Genet Metab 2026:109862. [6] Gibiansky L, Gibiansky E. J Pharmacokinet Pharmacodyn 2010;37:323–46. Keywords : Anti-drug antibody (ADA), target-mediated drug disposition (TMDD), transferrin receptor (TfR)

The effective delivery of therapeutics to the brain is a significant challenge when treating central nervous system (CNS) disorders. One possible approach to overcome this challenge is to utilize human transferrin receptor (TfR) to deliver biotherapeutic agents such as enzymes and proteins across the blood-brain barrier (BBB) to the CNS. Transport vehicle (TV) platform consists of an anti-human TfR binding domain engineered in the Fc region that is designed to enable BBB crossing after intravenous (IV) administration.

In this study, we developed translational pharmacokinetic (PK) models for TV molecules that are designed to deliver enzymes and proteins across the BBB, e.g., iduronate-2-sulfatase (IDS). The PK model employs a mathematical formalism to gain insights into the mechanisms impacting systemic PK and uptake of these molecules into the brain and identify appropriate approaches for translation of PK from nonclinical species to human.

The model is calibrated to serum and brain PK data from nonclinical studies in mice and monkeys, and serum PK from participants in our ongoing clinical studies where available. The serum PK component of this mechanistic model includes a standard two-compartment representation with a non-specific clearance, non-linear TfR-mediated clearance, and a non-linear mechanism to account for other receptor-mediated clearance in serum. The brain PK component of this model includes terms representing non-specific uptake as well as TfR-mediated uptake across the BBB, and a linear term to represent the loss of biotherapeutic agent in the brain via cellular uptake and other clearance mechanisms. This model allows for the translation of PK data of a given molecule from mice and cynomolgus monkeys to humans. Using the model, we identified the appropriate translational approach for each parameter based on three categories of parameters: 1) molecule-independent and species specific, 2) molecule specific and allometrically scaled across species, and 3) molecule-independent and allometrically scaled across species. The data not used for model calibration is used to test and confirm the translation of the clearance mechanisms across the three species. In addition, we conducted local and global sensitivity analysis of the calibrated model parameters to ensure a robust identifiable model structure for the data.

Overall, the presented PK model plays a critical role in understanding the processes impacting PK and distribution of therapeutics to the brain and optimizing the design of preclinical studies for development of future therapeutic candidates for central nervous system disorders. Furthermore, the model can inform the clinical development and first-in-human (FIH) dose selection for current and future TV-enabled molecules for delivery of enzymes and proteins.

Objective: Here, for the first time, we apply a genome-wide polygenic risk score approach using 6.2 million variants to compute a Parkinson's Disease (PD) genome-wide polygenic risk score (PD-GPRS).

Background: PD is the second most common and fastest-growing neurological disorder. Polygenic Risk Scores (PRS) using hundreds to thousands of PD-associated variants support polygenic heritability.

Design/Methods: PD-GPRS was computed via the LDPred algorithm using a European Genome Reference Panel and PD-GWAS summary statistics published by Nalls et al. 2019. PD-GPRS validation and testing used Accelerating Medicines Partnership – Parkinson’s Disease (AMP-PD) and FinnGen Consortia genomic data from 1,654 PD Cases and 79,123 Controls.

Results: PD odds for the top 8%, 2.5%, and 1% of PD-GPRS were three-, four-, and seven times greater compared with lower percentiles, respectively (p<1e-10). PD age of onset and MDS-UPDRS motor scores also differed by PD-GPRS decile. Enrichment for phagosome related, dopamine signaling, immune related, and neuronal signaling pathways was found for genes nearest high PD-GPRS variants identified by MAF analysis.

Conclusions: PD-GPRS offers a promising screening tool to identify high-risk individuals for preventive lifestyle or new drug therapy trials.

ABSTRACT Alpha-synuclein (SNCA) aggregates are pathological hallmarks of synucleinopathies, neurodegenerative disorders including Parkinson’s Disease (PD) and Lewy Body Dementia (LBD). Functional networks are not yet well-characterized for SNCA by CNS cell type. We investigated cell-specific differences in SNCA expression using Allen Brain Database single-nucleus RNA-seq data from human Middle Temporal Gyrus (MTG, 15,928 nuclei) and Anterior Cingulate Cortex (ACC, 7,258 nuclei). Weighted gene co-expression analysis (WGCNA) and hierarchical clustering identified a conserved SNCA co-expression module. Module genes were highly conserved (p < 10 −10 ) and most highly expressed in excitatory neurons versus inhibitory neurons and other glial cells. SNCA co-expression module genes from ACC and MTG regions were then used to construct a protein-protein interaction (PPI) network, with SNCA empirically top hub. Genes in the SNCA PPI network were compared with genes nearest single nucleotide polymorphisms linked with PD risk in genome-wide association studies. 16 genes in our PPI network are nearest genes to PD risk loci (p < 0.0006) and 55 genes map within 100kb. Selected SNCA PPI network genes nearest PD risk loci were disrupted by CRISPR knock out gene editing for validation of network functional significance; disruption of STK39, GBA, and MBNL2 resulted in significantly elevated intracellular SNCA expression.

SUMMARY Parkinson’s Disease (PD) is the second most common and fastest-growing neurological disorder. Polygenic Risk Scores (PRS) using hundreds to thousands of PD-associated variants support polygenic heritability. Here, for the first time, we apply a genome-wide polygenic risk score approach using 6.2 million variants to compute a PD genome-wide polygenic risk score (PD-GPRS) via the LDPred algorithm. PD-GPRS validation and testing used Accelerating Medicines Partnership – Parkinson’s Disease (AMP-PD) and FinnGen Consortia genomic data from 1,654 PD Cases and 79,123 Controls. PD odds for the top 8%, 2.5%, and 1% of PD-GPRS were three-, four-, and seven times greater compared with lower percentiles, respectively (p In Brief In Han and Teeple et al., Parkinson’s Disease inherited risk is quantified by a genome-wide polygenic risk score (PD-GPRS) approach using 6.2 million variants and data from 80,777 individuals. For the top 2.5% and 1% of PD-GPRS, individuals had five- and seven-fold greater odds of PD, respectively. PD-GPRS was found to be associated with overall PD risk, earlier age of onset, and MDS-UPDRS motor scores. Genes nearest to variants observed at higher frequencies among high-GPRS individuals are enriched for PD-implicated pathways. HIGHLIGHTS - Parkinson’s Disease genome-wide polygenic risk score (PD-GPRS) calculated from 6.2 million variants identifies individuals with inherited clinically significant increased neurodegeneration risk. - Top percentile PD-GPRS individuals were found to have up to seven-fold greater odds of PD and earlier age at PD diagnosis. - PD-GPRS scores correlated with all-subjects cohort mean MDS-UPDRS motor scores. - Pathway analysis of genes adjacent to frequently occurring variants in the high PD-GPRS population identified polygenic risk contributions for variations in PD-implicated pathways including dopamine signaling, immune responses, and autophagy pathways.

We consider a compartmental model for ribosome flow during RNA translation called the RFM. This model includes a set of positive transition rates that control the flow from every site to the consecutive site. It has been shown that when these rates are time-varying and jointly T-periodic every solution of the RFM converges to a unique periodic solution with period T. In other words, the RFM entrains to the periodic excitation. In particular, the protein production rate converges to a unique T-periodic pattern. From a biological point of view, one may argue that the average of the periodic production rate, and not the instantaneous rate, is the relevant quantity. Here, we study a problem that can be roughly stated as: can periodic rates yield a higher average production rate than constant rates? We rigorously formulate this question and show via simulations, and rigorous analysis in one simple case, that the answer is no.

Theory and practice are the two fundamental tools in engineering and scientific research. With a great increase of quantitative experiments in biological systems over the past decades, mathematical modeling is able to enhance predictions and generate new hypotheses. A “good” model of a system, that is expected to reproduce the experimental observations, is capable of making predictions outside the previous experimental settings. However, the accuracy of predictions based on mathematical models highly depends on the assumptions used to model the system. The objective of this study is to explore possible approaches to deploy such models in order to find new hypotheses to be tested in future experimental settings. From the lens of control and decision-making, a few biological systems relevant to chemotherapy, immunotherapy, and epidemics are considered in this work. Models are analyzed numerically and analytically in order to enhance the outcome of the system with a new control/decision. A new dosing plan for chemotherapy is identified and evaluated via in-silico experiments to optimally reduce the tumor volume at the end of the plan. The new dosing plan consists of two doses starting with a small dose at the beginning of the plan and an increased dose after a few weeks. Unlike traditional chemotherapy plans currently used, the proposed plan is neither a maximum tolerated dose, nor a metronomic/intermittent plan. Moreover, epidemic models under social distancing guidelines are studied. Considering a single interval social distancing based on the start time and the duration of the social distancing shows a linear relationship between optimal timing of the social distancing. Models analyzed in this work are generic and applicable to wide a range of applications.