PHILADELPHIA AND DELAWARE Using qPCR to Quantify the Presence of CART T Cells in vivo

Overview

Multiple sclerosis is a disorder in which the immune system attacks the protective covering of the nerve cells in the central nervous system (CNS). Bcells may contribute to disease progression. Therapies that deplete CNS- resident Bcells is a potential treatment for MS. Recent clinical studies have shown that decreasing peripheral B lymphocytes in MS patients reduces disease progression. However, these methods may not efficiently eradicate Bcells in the CNS. CAR Tcells are a new, potent cancer therapy that also target Bcells. T o efficiently deplete CNS Bcells, we tested CAR Tcells in a mouse model of MS. However, we saw no difference in the disease in mice treated with or without CAR Tcells. T o determine why we didn’t see a benefit, we developed an assay to test whether the CAR Tcells stuck around (engrafted) or not. We validated a qPCR assay first using CAR plasmids and then with blood spiked with CAR Tcells. Using this validated qPCR assay on blood samples from the MS study, we found that there was no difference in signal between the non-CAR Tcell and CAR Tcell treated groups. Based on the blood results so far, we were unable to detect CAR Tcells in the MS samples. Therefore, this may indicate that the CAR Tcells did not engraft, explaining the lack of effect on MS. Modeling Freshwater Microplastics and the Effects of Anthropogenic & Watershed Factors Using Machine Learning SydneyBlu Garcia-Yao Harriton High School, Bryn Mawr, PA Mentor Dr. Timothy Maguire, Academy of Natural Sciences Microplastics are a contaminant with significant potential for harm as they are associated with adsorbed heavy metal toxicity, persistent organic pollutants, and harmful pathogens. With plastic production continuing to rise and the need for policy and mitigation efforts, modeling has become an important tool. Past research on modeling microplastics explores transport dynamics, whereas this study uses anthropogenic and watershed factors to predict freshwater microplastics via machine learning. The study was executed in the Delaware River Basin with 115 samples from 36 sites. 500mL of each sample was filtered and microplastics visually identified. The average microplastic concentration per 500mL was 5.0 ± 3.5. Microplastic data was also normalized by discharge and upstream watershed area to calculate the transport of microplastics load per day per km2, with an average value of 17.7 ± 64.7. Then, data on watershed and anthropogenic factors was compared spatially to each sampling site. This found statistically significant (p < .05) relationships between microplastics per day and Strahler stream order (R = -.34), slope (R = .52), and number of upstream wastewater treatment plants (R = .43). Microplastics per 500mL had statistically significant correlations with days since January 1 (R = .46) and discharge (R = -.22). Three models were then created (linear, gradient boosting, random forest), and both the machine learning models outperformed the linear model, with the random forest predicting microplastic transport per day having an R = .73. This study shows that machine learning is a viable method for using anthropogenic and watershed characteristics to model microplastics.