2016 Biomedical Postdoctoral Research Symposium


Tsc2-dependent mTORC1 activation deregulates lung homeostasis

Kseniya Obraztsova1, Ryan Rue1, Edward E. Morrisey1,2, and Vera P. Krymskaya1  
1Penn Center for Pulmonary Biology, Perelman School of Medicine, University of Pennsylvania; 2Institute for Regenerative Medicine, Perelman School of Medicine, University of Pennsylvania  

Tumor suppressor Tuberous sclerosis complex 2 (TSC2) is a key negative regulator of mammalian target of rapamycin (mTOR), a central controller of cell growth and metabolism in health and disease. Loss of TSC2 induces the constitutive activation of mTORC1 in rare lung disease pulmonary Lymphangioleiomyomatosis (LAM), which affects only women of childbearing age and characterized by lung destruction and progressive loss of pulmonary function. Little is known how TSC2 loss induces LAM and what is the LAM cell of origin. To determine cell-type specific effects of TSC2 loss and mTORC1 activation on lung homeostasis we generated new transgenic mice with targeted Tsc2 deletion in lung mesenchyme Tbx4-Cre+Tsc2flox/flox and lung epithelium Shh-Cre+Tsc2flox/flox. Both Tbx4-Cre+, Tsc2flox/flox and Shh-Cre+, Tsc2flox/flox mice were viable and fertile. Adult Tbx4-Cre+Tsc2flox/flox mice demonstrated mTORC1 activation in lung mesenchyme, alveolar enlargement, and lesion growth. Interestingly, adult Shh-Cre+, Tsc2flox/flox lungs showed modest hyperplastic phenotype, mTORC1 activation in lung epithelium, mis-localization of E-cadherin, and decreased immune cell counts. Using lung cell lineage separation from mice with targeted Tsc2 deletion in either lung mesenchyme (Tbx4-Cre+Tsc2flox/flox) or lung epithelium (ShhCreTsc2flox/flox), we found that Tsc2 deletion only in mesenchyme but not in epithelium upregulates Igf2 and Stat3, which are known to be required for LAM cell survival. Collectively, our study demonstrates that (1) Tsc2 loss induces lung cell lineage specific effects on signaling pathways (see in rare disease LAM), (2) Tsc2-dependent mTORC1 activation induces lung degeneration, and (3) Tsc2 is required for normal lung homeostasis.

back to the top

Neuroepigenetic remodeling in cocaine addiction associated transcription and behavior

Songjun Xu1,2, Marco Carpenter1,2, Sonia Lombroso1,2, Ajinkya Sase1,2, Qiwen Hu1,2, and Elizabeth Heller1,2

1Department of Systems Pharmacology and Translational Therapeutics, Perelman School of Medicine, University of Pennsylvania ; 2Penn Epigenetics Institute, Perelman School of Medicine, University of Pennsylvania

Numerous research has implicated epigenetic remodeling plays a key role in addiction pathologies, yet the precise molecular mechanisms underlying such behavior pattern still remain unclear. Here, we tested the notion that histone posttranslational modifications, specifically H3K36me3, directly mediates alternative splicing in brain, and the splicing isoforms generated in the process, in turn, facilitate drug seeking behaviors. Previous work from our lab highlighted strong correlation between H3K36me3 enrichment and the splicing complexity of expressed alternative isoforms. To further dissect this phenomenon, we injected HSV-SET2 along with a control virus with a catalytically-inactive SET2 (R195C) to induce global H3K36me3 enrichment. Then we confirmed H3K36me3 enrichment by western blot and quantitate mass-spectrometry. Next, we used ChIP- RNA-seq analysis of NAc injected with HSV-SET2 and control to identify H3K36me3 induced alternative isoform expressions. With this information, we attempt to identify key neuronal genes that show coincident cocaine-regulated H3K36me3 modification and alternative splicing. We also employed a dual-viral system of CRISPR/dCas9-SET2 to target one critical factor in drug addiction, gene FosB, to investigate the relevance between H3K36me3 enrichment and FosB spliced isoform delta FosB. Our preliminary result indicated HSV-SET2 injection mediated a noticeable increase in delta FosB expression, suggests both feasibility of the system as well as the strong correlation between SET2 mediated H3K36me3 enrichment and cocaine drug seeking behavior. Taken together, our work supported a mechanism for alternative splicing in brain in the context of cocaine addiction.

back to the top

Tree Structured Boosting: A Novel Patient Stratification Tool in the Era of Precision Medicine

Jose M. Luna1, Gilmer Valdes2, Abigail Berman1, Chimbu Chinniah1, Grace Ma1, Hann-Hsiang Chao1, Timothy D. Solberg2, Lyle H. Ungar3, Eric Diffenderfer1, and Charles B. Simone II4
1Department of Radiation Oncology, University of Pennsylvania; 2Department of Radiation Oncology, University of California, San Francisco; 3Department of Computer and Information Science, University of Pennsylvania; 4Department of Radiation Oncology, University of Maryland Medical Center

Decision trees (DT) are interpretable and, therefore, used extensively in medicine for stratifying patients. Current DT algorithms, however, are consistently outperformed in accuracy by other less-interpretable machine learning models, such as ensemble methods. We introduce Tree Structured Boosting (TSB), a novel framework that defines a continuum between additive models (AM) and full order interaction models (FOIM) for building DTs. We hypothesize that TSB produces interpretable DTs that retain interpretability while having accuracy similar to ensemble methods.We evaluated if TSB defines a continuum of DTs between FOIM and AM controlled via a single tunable parameter, so that we could improve the classification error of DTs. We tested the TSB algorithm on a clinical dataset formed by a cohort of 203 consecutive stage II-III locally advanced non-small cell lung cancer patients treated with radiotherapy to a median dose of 66.6/1.8 Gy to assess for radiation pneumonitis (RP) prediction. Overall, 17.7% patients developed grade≥2 RP. We evaluated 32 continuous and categorical features per patient grouped into risk factors, comorbidities, pretreatment imaging, stage, histology, radiation treatment, chemotherapy, and dosimetry. We compared TSB classification error with CART trees and the ensembles RUSBoost and Bagging.The performance of the TSB algorithm (AUC=0.61,p=0.03) is comparable with ensemble methods such as RUSBoost (AUC=0.62,p=0.08) and Bagging (AUC=0.63,p=0.01) and vastly superior than CART (AUC=0.54,p>0.1).We have demonstrated, both theoretically and empirically that TSB reveals intrinsic connections between AM and FOIM. The experiments reveal that TSB can outperform either AM and FOIM alone, therefore it exhibits better performance than traditional DTs.

back to the top

Defining the role of DLST in MYCN-driven neuroblastoma

Nicole Anderson1, Jacob Athoe2, Yun Zhou2, Ping Ren4, Ruijuan Xiu4, Xueli Pan2, Ting Tao4, Nina Wiechert4, Shizen Zhu5, A. Thomas Look4, Guoliang Qing3, M. Celeste Simon1, and Hui Feng2      
1AFCRI, University of Pennsylvania, Philadelphia, USA; 2Boston University School of Medicine, Boston, Massachusetts, USA; 3Dana-Farber Cancer Institute, Boston, Massachusetts, USA; 4Wuhan University School of Medicine, Wuhan, China; 5Mayo Clinic, Rochester, New York, USA

Neuroblastoma is the most common extracranial tumor in children and is responsible for ~15% of childhood cancer-related mortalities. Aberrant MYCN activity is found in ~30% of neuroblastoma cases and is associated with highly aggressive tumors and a poor prognosis. Metabolic reprogramming is a hallmark of cancer and represents a fundamental difference between cancer and normal cells in terms of how they utilize nutrients for energy production and macromolecule synthesis. In particular, MYCN-driven neuroblastoma cells are addicted to glutamine and use it to fuel the tricarboxylic acid (TCA) cycle. Upon entering the TCA cycle, the glutamine metabolite, -ketoglutarate is converted to succinyl-CoA by alpha-ketoglutarate dehydrogenase complex (KGDHC). We found that elevated expression of dihydrolipoamide S-succinyltransferase (DLST) predicts poor overall survival in human neuroblastoma patients. Additionally, immunohistochemical analysis of primary human neuroblastoma tumor samples illustrated that DLST protein levels are elevated in stage IV tumors, compared to the low-grade tumors. Utilizing a zebrafish model of MYCN-driven neuroblastoma, we have demonstrated that heterozygous loss of dlst significantly delays tumor onset. To determine the sensitivity of human neuroblastoma cells to DLST inhibition, we genetically inactivated DLST in a panel of MYCN-amplified neuroblastoma cell lines and found that DLST inactivation results in decreased cell viability and increased cell death. Taken together, our studies identified DLST as an important mediator of high-risk neuroblastoma with MYCN amplification and demonstrated that DLST inactivation can kill these cancer cells, which provides compelling evidence that the metabolic dependence of neuroblastoma cells on the TCA cycle.

back to the top

Cyclin E: Targeting cell cycle dependencies in CCNE1 amplified tumors     

Kai Doberstein1, Alison Karst2, Paul T Kroeger Jr.1, Paul Jones2, William Hahn2, and Ronny Drapkin1,2
1University of Pennsylvania, Perelman School of Medicine, Ovarian Cancer Research Center; 2Harvard Medical School, Dana-Farber Cancer Institute

Genomic instability is a hallmark of high grade serous ovarian carcinoma (HGSOC). It is estimated that approximately 50% of HGSOCs harbor a defect in the homologous recombination (HR) pathway of DNA repair. In contrast, the 20% that harbor CCNE1 amplifications appear to have an intact HR pathway. These tumors are associated with shorter overall survival and resistance to chemotherapy. Our previous data showed that CCNE1 overexpression occurs early in serous tumorigenesis. Importantly, in immortalized human fallopian tube secretory epithelial cells (FTSEC), constitutive Cyclin E overexpression imparts malignant characteristics to these cells. This leads to an accumulation of DNA damage and altered gene expression. However, Cyclin E overexpression alone was not capable of fully transforming the FTSECs. Therefore, in order to identify cooperating genetic alterations, we performed an in vitro gain-of-function (GOF) screen. One of the hits was the RAD51 paralog XRCC2, which is known to be involved in the HR DNA repair pathway and in fork protection. We could show that XRCC2 expression is induced in response to Cyclin E overexpression. Moreover, CCNE1-amplified tumors become dependent on XRCC2. The knock down of XRCC2 in Cyclin E overexpressing cells leads to a strong reduction in fork speed and fork recovery, and triggers cell death. Interestingly, we found that XRCC2 interacts with the minichromosome maintenance deficient 7 (MCM7) protein and appears to stabilize the replication fork through this interaction. Surprisingly, the downregulation XRCC2 led to a strong reduction in MCM7 protein expression. Further defining the factors that contribute to the XRCC2-MCM7 interaction at the replication fork may define novel vulnerabilities in CCNE1 amplified tumors.

However, we have only incomplete knowledge of the number and diversity of spidroins that exist in nature, and possess limited understanding of the molecular machinery involved in silk production. Therefore, to investigate spidroin genes systematically, we constructed the first annotated genome for an orb-weaving spider: the Golden Orb-Weaver (Nephila clavipes), which builds large intricate webs using an extensive repertoire of silks with diverse physical properties. Our catalog of 28 spidroins across all silk classes reveals an unprecedented 394 unique repeated coding motifs, including hundreds of novel sequences and higher-order repetitive cassettes. We associate spidroin sequences with silk gland-specific expression, and our data tests the long-held assumption that differentiated silk glands extrude spidroins exclusive to one silk class. We additionally find evidence of an alternatively spliced spidroin, a spidroin expressed only in venom glands, evolutionary mechanisms for spidroin diversification, and non-spidroins highly expressed in silk glands as candidates involved in silk production.

back to the top

The role of AHR in TCDD-mediated sex ratio distortion                  

Kristin M. Bircsak1,2, Andrew M Prantner1, Latresa Copes1, George Gerton1,2
1Center for Research on Reproduction and Women’s Health; 2Center of Excellence in Environmental Toxicology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, United States     

Paternal exposure to the persistent organic pollutant, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is associated with a female-predominant sex ratio. This research aims to determine the effect of TCDD on embryo sex ratio, localization of the TCDD-specific transcription factor aryl hydrocarbon receptor (AHR) in the testis, and the testicular expression of genes known to regulate sex ratio in mice. For the sex ratio experiments, WT C57BL/6 male mice were injected weekly with TCDD (wk 1: 2000 ng/kg, wks 2-21: 400 ng/kg) or corn oil vehicle and mated biweekly. Embryos were collected at gestation day 15.5 and genotyped for sex. For the AHR localization and gene expression experiments, mice were subjected to a similar treatment paradigm for a shorter duration (3 wks). The sex ratio (number of males/total number of embryos) was significantly lower in embryos from TCDD-treated males relative to vehicle-treated males by 20% (Vehicle: 0.517, TCDD: 0.0.448), indicating that paternal TCDD exposure resulted in a female biased sex ratio. Following 3 weeks of TCDD or vehicle exposure, AHR protein in the testis was similar between treatment groups and localized to the acrosomal region of round spermatids, the nucleus of elongating spermatids, and the cytoplasm of interstitial Leydig cells. Importantly, TCDD-treated males exhibited a 30% increase and a 20% decrease in the testis mRNA expression of genes involved in the determination of embryo genotypic sex, Sycp3-like Y-linked (Sly) and Slx, respectively. While Slx was not detected in the liver, Sly gene expression was enhanced 40% in the livers of TCDD-treated mice. Together, these data confirm that paternal exposure to TCDD distorts embryo sex ratio and begins to shed light on a possible mechanism by which this phenomenon may occur.

back to the top

Defining the mechanism of circular RNA nuclear export

Chuan Huang1, Jeremy E. Wilusz1
1Department of Biochemistry & Biophysics, University of Pennsylvania Perelman School of Medicine

Although it was long assumed that eukaryotic pre-mRNAs are almost always spliced to generate a linear mRNA, it is now clear that thousands of protein-coding genes can be non-canonically spliced (or “backspliced”) to produce circular RNAs (circRNAs). Most mature circRNAs accumulate in the cytoplasm; however, little is known about how circRNAs are exported from the nucleus as they lack many of the common signals used for mRNA export. Here, we show that the DExH/D box protein Hel25E (UAP56) is required for circRNA export. Depletion of Hel25E results in robust accumulation of circRNAs in Drosophila nuclei. Hel25E has previously been shown to be essential for bulk mRNA nuclear export by helping recruit the Nxt1-sbr receptor, which mediates transport through nuclear pores. Surprisingly, depletion of Nxt1 or sbr did not affect circRNA export, suggesting a separate pathway from canonical mRNA export. Together, these data suggest that circRNA export relies on a novel export pathway that utilizes Hel25E, but not the Nxt1-sbr receptor.

back to the top

A computational method to improve missing data imputation in Electronic Health Record        

Ruowang Li, Yong Chen, and Jason H. Moore
Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania

Clinical data of patients’ measurements and treatment history stored in electronic health record (EHR) systems are starting to be mined for better treatment options and unknown disease associations. A primary challenge associated with utilizing the EHR data is the considerable missing data of the patients. Failure to address this issue can introduce significant bias into the EHR research. Currently, imputation methods solely rely on correlation structures among the structured phenotype variables in the EHR. However, existing Genome-Wide Association Studies (GWAS) have demonstrated that many human phenotypes have varying degrees of heritability and some of these phenotypes are routinely being measured in the EHR, for example, low-density lipoprotein levels and blood pressures. Therefore, we developed a computational model that incorporates patients’ genetic information to perform EHR data imputation. We used individual Single Nucleotide Polymorphism’s association to phenotype variables in the EHR as input to construct a genetic score that quantifies the genetic contribution to the phenotype. The risk score along with phenotype variables correlation are then used as predictors to impute the missing values. To demonstrate the method performance, we applied our model to impute missing cardiovascular related measurements on 15,000 samples in the PennOmics data warehouse. We show that incorporating genetic information can significantly increase the accuracy of imputation. Compared to the state-of-the-art imputation methods, our method is the first EHR specific data imputation method that integrates patients’ genetic information.

back to the top

Metastasis in triple negative breast cancer is dependent on ΔNp63/CXCL2/CCL22-mediated recruitment of MDSCs

Sushil Kumar and Rumela Chakrabarti
Department of Biomedical Sciences, University of Pennsylvania, Philadelphia, PA

Triple-negative breast cancer (TNBC) represents 20% of all breast cancer cases and is the most aggressive subtype, with high rates of mortality due to very limited treatment options. Despite the use of cytotoxic chemotherapies, the overall patient survival is extremely poor due to the high rate of recurrence and metastatic spread. Our earlier studies demonstrate that ΔNp63, an oncogenic isoform of p63, is overexpressed in TNBC and is an important regulator of TNBC cancer stem cell (CSC) activity. In this study, we found that reduced levels of ΔNp63 in TNBC decrease tumor growth, progression, and metastasis, which is concomitant with reduced recruitment of myeloid-derived tumor suppressor cells (MDSCs). Moreover, RNA-seq and ChIP-seq analyses identified two ΔNp63-dependent chemokines, CXCL2 and CCL22, which could potentially recruit MDSCs to the primary tumor and metastatic sites. In support, inhibition of cognate CXCL2 and CCL22 receptors remarkably decreased metastatic events in mice with TNBC xenografts. Finally, we find that MDSCs secrete pro-metastatic factors such as MMP9 and Chitinase3-like1, highlighting a novel function for MDSCs in promoting CSC function in TNBCs. Together, our data demonstrate a unique crosstalk between ΔNp63+ TNBC cells enriched for CSC activity and MDSCs that promote metastasis and therefore may serve as a target for future combinatorial treatment of chemotherapy and immunotherapy.

back to the top

Poster Abstracts

1. Electrophysiological changes following brain injury in a large animal model of post traumatic epileptogenesis       

Alexandra V. Ulyanova1, Paul F. Koch 1, Christopher D. Adam 1, Robin J. Russo 1, D. Kacy Cullen 1,2, Douglas H. Smith 1, Victoria E. Johnson 1, H. Isaac Chen 1,2, and John A. Wolf 1,2
1Department of Neurosurgery, University of Pennsylvania; 2Corporal Michael J. Crescenz Veterans Affairs Medical Center

Traumatic brain injury (TBI) is a prevalent cause of acquired epilepsy, particularly post-traumatic epilepsy (PTE). A prominent clinical question has been the relative contribution of the different types of TBI to the development of epileptogenesis. While underlying mechanisms are still unknown, hippocampal excitability is often an outcome for both TBI and PTE. In order to examine changes in hippocampal circuitry, we performed in vivo electrophysiological recordings in porcine hippocampus following TBI. Oscillatory field potentials and single-unit neuronal activity were recorded in sham and injured animals, and location of 32-electrode silicon probes were confirmed histologically. In acute animals, current stimulation of entorhinal cortex induced sub-clinical epileptiform activity (sharp waves and paroxysmal depolarizing shifts) in a subset of injured animals. In all injured animals, interneurons located in pyramidal CA1 layer had their firing rate and action potential width reduced, while reduction in number of bursts and incorrect locking to theta oscillation closely followed development of hyperexcitability. To monitor these changes chronically in awake behaving animals, we developed and successfully implemented a wireless recording system. This system allows for 24/7 video and electrophysiological monitoring of animals, and is being implemented in sham, contusion, rotational, and contusion plus rotational injury models for comparison of epileptogenesis in these injury types. A post-mortem examination of neuropathology and axonal injury will be performed, with emphasis on the temporal lobe connections. Number and frequency of seizures and inter-ictal events will be correlated with the neuropathological outcomes to determine mechanistic underpinnings of PTE.

2. Arhgef11 regulates RhoA-MLC mediated tight junction integrity through an alternative splice isoform controlled Esrp1       

Sungkyoung Lee, Benjamin Cipley, Tammy Yang, Natoya Peart ,and Russ P. Carstens
Department of Medicine, Perelman School of Medicine, University of Pennsylvania

Alternative splicing (AS) have important roles in post-transcriptional gene regulation and determining cell fate by providing genetic control of multiple critical steps in most vertebrate genes. The AS regulator Esrp regulates a large network of alternatively spliced genes involved in cell-cell adhesion, cytoskeletal dynamics, and EMT/MET. In mammalian development, the Esrp’s are required for patterning and organogenesis of multiple organ systems. Loss of the Esrps in the skin epidermis leads to defects in epidermal barrier function and hair loss. We profiled the global Esrp-mediated splicing regulatory program in epidermis, which revealed numerous Esrp splice targets are components of or are associated with tight junction (TJs). Accordingly, deletion of the Esrp1 in the epidermis and cultured keratinocyte cells revealed their functional requirement for TJs. Despite of significant phenotypes, knowledge about the signaling pathways involved in Esrp regulating TJs is still unclear. Here, we have explored that Esrp1 regulates the phosphoration of MLC by directly control alternative splicing of Arhgef11. Knockout Esrp1 in epithelial cells causes a switch from Arhgef11 epithelial isoform to mesenchymal isoform. We demonstrate that the mesenchymal, but not epithelial, isoform of Arhgef11 binds to inhibitory PAK4 thereby linking Esrp regulated splicing with enhanced RhoA activation, MLC phosphorylation, and maintenance of the epithelial TJ barrier. These results therefore demonstrate mechanisically and functionally how the expression of Esrp-regulated splicing targets contribute to epithelial barrier function and TJs maintenance.These results suggest that Esrp1 is the key regulator of TJs formation and have crucial roles in epidermal barrier function.

3. Ambidexterity and Alzheimer's Disease Risk     

Elisabeth E. Mlynarski1,2,3, Mitchell Tang1,2,3, Alexandre Amlie-Wolf1,2,3, Liming Qu1,2,3, Amanda Kuzma1,2,3, Laura Cantwell1,3, John M. Ringman4,5, Gerard D. Schellenberg1,3, and Li-San Wang1,2,3
1Penn Neurodegeneration Genomics Center, Perelman School of Medicine, University of Pennsylvania; 2Institute for Biomedical Informatics, Perelman School of Medicine, University of Pennsylvania; 3Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania; 4Memory and Aging Center, Keck School of Medicine of University of Southern California, Los Angeles, CA; 5Department of Neurology, University of Southern California, Los Angeles, CA

Previous studies have investigated the association between left handedness and Alzheimer’s disease, however these studies either excluded ambidextrous individuals, and/or did not include genome-wide genetic analyses. Therefore, we performed association analyses to examine the prevalence of ambidexterity and left handedness in late-onset Alzheimer's disease. Interestingly, our analysis revealed an association between Alzheimer’s disease and ambidexterity, but not left handedness. In fact, the results suggest that ambidexterity is protective for late-onset Alzheimer's disease (OR = 0.591, p = 9.26 x 10^-3). A multi stage GWAS was then performed for each of the four different handedness splits: ambidextrous vs left or right (A vs L|R), left vs right or ambidextrous (L vs R|A), right vs left or ambidextrous (R vs L|A), and left vs right (L vs R). In the discovery, only one locus in the A vs L|R analysis reached genome-wide significance (p < 10^-8). The one genome-wide significant signal (p = 4.07 x 10^-9), located at 10p12.33, replicated in the A vs L|R meta-analysis (p = 3.35 x 10^-9). The top SNP from the meta-analysis mapped to an intergenic region ~16 kb distal to the 3’ end of ST8SIA6. Notably, there were no significant signals for any of the other three handedness splits (L vs R|A, R vs L|A, and L vs R), and the few suggestive loci (p < 10^-5) that were identified in the discovery analysis did not replicate. Together, these results indicate that ambidexterity is associated with a lower risk of AD and the locus on 10p12.33 may play a role. Additional functional analyses are currently underway to identify the specific causal variant for the GWAS signal and elucidate the underlying genetic mechanisms.

4. An Optimized DNA encoded monoclonal antibody (DMAb) targeting OspA prevents transmission of Lyme disease     

Rianne Esquivel1*, Seleeke Flingai1*, Yang Wang2, Aurélie Kern3, Sangya Agarwal1, Zachary Schiller2, Kaori Graybeal2, Ami Patel1, Linden Hu3, Mark Klempner2#, David Weiner1#
*Co-first authors. #Co-corresponding authors
1The Vaccine Center, The Wistar Institute; 2MassBiologics, University of Massachusetts Medical School, Boston, MA; 3Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, MA

While Lyme disease is responsive with antibiotics, treatment of chronic Lyme disease presents obstacles. With no current Lyme vaccine for humans, healthcare workers are often left with no prevention options for Lyme endemic regions. It has been well established that antibodies are critical mediators in protection against Lyme and monoclonal antibody (mAb) treatment is a promising approach for preventing Lyme infection. However, mAb production has significant limitations including: dose, manufacturing, and cost. DNA encoded monoclonal antibody (DMAb) technology may provide an alternative to passive antibody therapy for generating rapid, antibody-based immunity against infectious diseases with the possibility of persistent protection for several months. DMAbs are highly optimized DNA plasmids engineered to encode mAbs for in vivo DMAb production. The antigen OspA, expressed by B. burgdorferi in the tick midgut presents an interesting protein target as the antibody must bind to OspA within the tick to prevent disease and is highly conserved. By studying anti-OspA human IgG antibody sequences with varying breadths of Borrelia neutralization capabilities, OspA affinities, and protective efficacies in mice, we engineered and optimized DMAb vectors. In vitro, we achieved high expression levels for optimized DMAb encoding MAB 319-44 exhibiting significant B. burgdorferi neutralization and in vivo expression of several ug/ml. Using DMAb 319-44 in a murine Lyme challenge model we were able to observe 80% protection versus the vector control group. Importantly, this is the first demonstration of a transmission blocking DMAb providing protection for the host and suggests additional study of this unique technology as a possible new tool for protection against tick borne infections.

5. Insulin-dependent maturation of newly generated olfactory sensory neurons after injury     

Akihito Kuboki1,2, Shu Kikuta3, Takashi Sakamoto3, Kaori Kanaya3, Kenji Kondo3, Nobuyoshi Otori2, Hiromi Kojima2, Johannes Reisert1, and Tatsuya Yamasoba3
1Monell Chemical Senses Center; 2Department of Otolaryngology, The Jikei University School of Medicine; 3Department of Otolaryngology, Graduate School of Medicine, University of Tokyo

Insulin is a peptide hormone that regulates glucose metabolism, but also has different regulatory roles as an extrinsic signals in the brain. However, it is poorly understood whether insulin signaling affects ongoing incorporation of newly generated olfactory sensory neurons (OSNs) to maintain homeostasis in the olfactory epithelium (OE). Here, we examined whether insulin signaling affects the dynamic incorporation of new OSNs in adult mice after injury. Mice were administered Streptozotocin (STZ) to ablate pancreatic β cells, resulting in hypoinsulinemia. Methimazole, an olfactotoxicity-inducing drug, was also intraperitoneally injected to ablate OSNs in the STZ-administered mice. Up to 7 days post-injury, there was no difference in the numbers of recovering mature and apoptotic OSNs between the STZ-administered and control (saline-administered) mice. However, between days 7 and 28, the STZ-administered mice showed remarkably fewer mature OSNs and more apoptotic OSNs than control mice. By day 28, control OE was restored to its pre-injury condition, while STZ-treated mice still had OEs that had not recovered. Consistent with fewer mature OSNs in the STZ-administered mice, electroolfactogram responses induced by odorants at day 28 following injury were significantly reduced compared with those in control mice. Furthermore, replenishment of insulin in STZ-administered mice during days 7 to 14 post-injury promoted the recovery of the OE. These results indicate that insulin signaling is involved in homeostatic regeneration of the OE following injury, and that newly generated OSNs have a high susceptibility to insulin signaling for their maturation following day 7 post-injury.

6. Impact of transcriptional errors on proteostasis and organismal health     

Suraiya Haroon1, Clark Fritsch2, Jean-Francois Gout4, Weiyi Li4, Michael Lynch4, and Marc Vermulst1,3
1Center for Mitochondrial and Epigenomic Medicine, Children's Hospital of Philadelphia; 2Cell and Molecular Biology, University of Pennsylvania; 3Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania; 4Department of Biology, Indiana University, Bloomington, IN.

Fidelity of DNA replication, transcription and translation is crucial for the propagation of life but these processes are rarely error-free. We know that DNA mutations can lead to cancer and aberrant translation can cause Amyotrophic Lateral Sclerosis (ALS). However, the effect of transcriptional fidelity on health is less understood. Currently, we are mostly aware of the effects of transcriptional error at specific disease causing loci. For example, mRNA mutations in β-amyloid precursor proteins generate toxic Aβ protein and cause Alzheimer’s disease. But, what is the cost of genome-wide transcriptional errors on health? Emerging technologies for assaying transcriptional errors and new animal models have given us the opportunity to answer this question. By studying the error-prone mutants of the RNA polymerase subunits rbp-1 and rpb-9, we have observed that they display disrupted proteostasis and reduced lifespan, which is further exacerbated by the removal of the chaperone ydj1p. These instrumental discoveries provide a clear cause-effect relationship between global transcriptional error, proteostasis and organismal heath. Expanding on these insights, we generated mutant alleles of ama-1, the homolog of rbp-1 in the multicellular organism, C. elegans. We observe that the milder allele does not affect the fitness of the worms. Strong mutant variants of TDP43 and SOD1 that cause ALS have no additional fitness cost in the presence of this mutant allele. However, a polyQ peptide, a model for Huntington’s disease, displayed increased fitness defect in the presence of the mild mutant ama-1 strain. These differential responses can be exploited to further dissect the relationship between transcriptional errors, proteostasis and fitness.

7. Is Adderall the Next Oxy? What role can social media mining play in the prescription medication abuse problem?     

Abeed Sarker1, Karen O'Connor1, Rachel Ginn2, Matthew Scotch2, Karen Smith3, Dan Malone4, Graciela Gonzalez1
1Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania; 2Department of Biomedical Informatics, Arizona State University, Scottsdale, AZ; 3Rueckert-Hartman College for Health Professions, Regis University, Denver, CO; 4Department of Pharmacy Practice and Science, University of Arizona, Tucson, AZ

Prescription medication (PM) abuse is a major public health crisis in the USA. The growing nature of this problem necessitates the implementation of novel monitoring strategies for investigating the prevalence and patterns of abuse of specific medications. Our aims were to assess the possibility of utilizing social media as a resource for automatic monitoring of PM abuse and to devise an automatic classification technique that can identify potentially abuse-indicating user posts. We collected Twitter user posts (tweets) associated with three abuse-prone medications (Adderall®, oxycodone, and quetiapine). We manually annotated 6400 tweets mentioning these three medications and a control medication (metformin) that is not the subject of abuse due to its mechanism of action. We performed quantitative and qualitative analyses of the annotated data to determine whether posts on Twitter contain signals of PM abuse. Finally, we designed an automatic supervised classification technique to distinguish posts containing signals of medication abuse from those that do not and assessed the utility of Twitter in investigating patterns of abuse over time. Our analyses show that clear signals of medication abuse can be drawn from Twitter posts and the percentage of tweets containing abuse signals are significantly higher for the three case medications (23%,12% and 5%) than the proportion for the control medication (metformin: 0.3 %). Our automatic classification approach achieves 82% accuracy overall (abuse class F-measure: 0.46). We found very rates of abuse for Adderall®. Our study indicates that social media can be a crucial resource for obtaining medication abuse-related information and automated techniques for analyzing the information must be developed.

8. Detecting Gestation Period Using Social Media Data Analysis     

Masoud Rouhizadeh1, Arjun Magge2, Ari Klein1, Abeed Sarker1, Graciela Gonzalez1
1Department of Epidemiology, Biostatistics, and Informatics, Perelman School of Medicine, University of Pennsylvania; 2Department of Biomedical Informatics, College of Health Solutions, Arizona State University, Scottsdale, AZ

Social media are unique resources to obtain insights into clinical cohorts. An important population for health monitoring are pregnant women who are protected from clinical trials. This work is a part of a larger study to identify a pregnancy cohort on Twitter and detect the medication effects on the mother and newborn. Here we focus on the temporal analysis as some outcomes depend on the period of intake. Our objectives include: (1) Developing text analysis tools for detecting gestation period using social media posts. (2) Identifying real mothers vs bots and identify discrepancies within pregnancy time frame. Data. We used the timelines from 73,800 Twitter users who have at least one pregnancy announcement tweet. We queried the tweets for pregnancy- and time-related keywords. Context. We hand-crafted linguistic patterns for 3 types of temporal anchors: pregnancy duration, due date, and birth announcement. We then extracted the dates according to the patterns. Tweet-Level Analysis. Our patterns capture 41,350 tweets from 16,810 timelines, each indicating an anchor to the gestation period. We could estimate the start or end date of pregnancy correctly, 99.2% of the time (based on our analysis of 1% of the matching tweets). User-Level Analysis. We cluster the estimated dates into bins of 30 days and found that users with 4+ bins are likely to be bots, 99.3% of the time. We applied heuristic analysis to resolve discrepancies, ending in ~15,000 users with validated estimate of the gestation period. The achieved precision is specifically important for health-related social media studies where selecting the accurate cohort is crucial. We plan to apply statistically-driven feature extraction to increase the coverage.

9. Development of a novel syngeneic model system for ovarian cancer research     

Sarah Stuckelberger1, Ronny Drapkin2
1,2University of Pennsylvania Perelman School of Medicine, Penn Ovarian Cancer Research Center

Ovarian cancer is the most lethal gynecologic malignancy and afflicts nearly a quarter of a million women each year worldwide. High-grade serous ovarian carcinoma (HGSOC) is the most common subtype and most are thought to arise from the fallopian tube (FT) secretory epithelial cell. Although a number of human models have been developed, no murine FT-derived cancer cell lines are available for ovarian cancer research. The most frequently used murine cancer cell line is the ID8 model, derived from the ovarian surface epithelium of a purebred C57Bl6 mouse. Recent genomic characterization of ID8 revealed that this line does not harbor the hallmarks of human HGSOC. Thus, we hypothesize that developing a murine FT-derived cell line that recapitulates the phenotypic and genomic hallmarks of human FT epithelial cells and HGSOC will enable studies in immunocompetent animal, addressing novel therapeutic combinations, especially immunotherapies. We will establish and immortalize a C57Bl/6 murine oviductal primary secretory epithelial cell line by knocking down the p19ARF gene using a stable shRNA. We will then transform immortalized C57Bl/6 murine oviductal cells by deleting genes (Tp53, Brca1 or Brca2, and Pten) that are typically altered in human HGSOC using the CRISPR-Cas9 technology. Successful transformation of oviductal cells will be monitored in vitro using proliferation and clonogenic assays as well as anchorage-independent growth capability. Tumorigenicity in vivo will be assessed by implanting cells into both immunocompromised and syngeneic mice. By developing transformed oviductal cells from purebred C57Bl6 mice, we will be able to assess the immune component of these tumors and determine how different tumor genetics influence immune infiltrates in a syngeneic system.

10. The Weakest of Weak Protien Interactions: Reverse Micelles as a platform for Hydration Dynamics and Drug Discovery       

Brian Fuglestad1, Nicole E. Kerstetter (1), Christine Jorge (1), Patricia A. Jennings (2), Nataniel V. Nucci (3), and A. Joshua Wand (1)
1Johnson Research Foundation and Department of Biochemistry and Biophysics, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA; 2Department of Chemistry and Biochemistry, University of California San Diego, La Jolla, CA; 3Department of Physics & Astronomy and Department of Biomedical and Translational Sciences, Rowan University, Glassboro, New Jersey, USA

Despite being among the most prevalent interactions that proteins encounter, the nature of those between water and protein is elusive and has long been an open problem in biophysics. A recently developed technology in the Wand laboratory, Reverse Micelle (RM) encapsulation, has allowed protein hydration to be measured at atomic resolution for the first time. The principle behind the technology is simple; by eliminating nearly all water within a sample and replacing it with alkane, the hydrating water that remains at the surface of the protein is resolved and measurable via NMR spectroscopy. This was applied for the first time to ubiquitin and revealed clustering of slow water in the interface that engages in protein-protein interactions. The next application was to Interleukin-1β (IL-1β), a cytokine that engages its receptor to initiate inflammatory response. Hydration dynamics measurements revealed a stark clustering of slow water in the receptor-binding interface of IL-1β, which is expected to enhance to the affinity of IL-1β to the receptor complex through released solvent entropy upon binding. The reverse micelle technology is now being applied to measurement of another weak protein interaction; that between a protein drug target and small-molecule fragments in a drug screen. Theory predicts that binding fragments to protein within RMs is enhanced ~100 fold as compared to standard methods, which was confirmed experimentally. This greatly enhances detection of fragment hits, which are used as initial building blocks for drug design. IL-1β is important in inflammatory disease, but lack of a binding pocket renders drug screening with current methods nearly impossible. IL-1β has been screened against a fragment library and several fragments were found to bind.

back to the top

11. The role of innate immune responses in Zika virus cellular tropism

Jillian N. Whelan, Ruth Elliott, Susan R. Weiss   
Department of Microbiology, Perelman School of Medicine, University of Pennsylvania

Since the 2015 onset of the Zika virus (ZIKV) outbreak, novel modes of transmission, neurological symptoms, and viral persistence within human tissues have been reported, highlighting our poor understanding of ZIKV pathogenesis and cellular tropism. We evaluated ZIKV replication and innate immune responses in various human and mouse primary cells, and assessed whether degree of innate immune function dictated cell-type specific susceptibility to ZIKV. We found that in primary mouse cells, ZIKV replicated in neurons and astrocytes, but not in bone marrow-derived macrophages or microglia. ZIKV did, however, replicate in human monocyte-derived macrophages. Since ZIKV replication in mice was limited to cell types lacking a robust antiviral response, we used CRISPR-Cas9 to genetically disrupt targeted host genes in the human A549 cell line to elucidate the innate immune pathways antagonized by ZIKV in a species-dependent manner, thereby permitting ZIKV infection of human but not mouse macrophages. Knockout of type I IFN proteins, namely MAVS, STAT1, or STAT2, had no effect on ZIKV replication compared to that of the parental A549 WT cells, suggesting ZIKV-inhibition of type I IFN responses during infection. Host type I and type III IFN and ISG expression induced by ZIKV in WT cells was also unaffected by KO of STAT1 or STAT2. Additionally, ZIKV activated the antiviral OAS/RNase L pathway, however OAS3 or RNase L knockouts did not enhance ZIKV replication compared to that in WT cells, which is not consistent with increased replication of other RNase L-activating viruses including another flavivirus WNV, in the KO cells, indicating ZIKV-mediated inhibition of RNase L function downstream of its activation.

back to the top

12. Noninvasive Monitoring of Critical Closing Pressure with Near-infrared Light

Wesley B. Baker1, Ashwin B. Parthasarathy2, Kimberly P. Gannon3, Venkaiah C. Kavuri2, Michael T. Mullen3, David R. Busch4, John A. Detre3, Arjun G. Yodh2, Daniel J. Licht4, Ramani Balu3, W. Andrew Kofke1
1Department of Anesthesiology and Critical Care, 2Department of Physics and Astronomy, and 3Department of Neurology, Perelman School of Medicine, University of Pennsylvania; 4Division of Neurology, Children's Hospital of Philadelphia

The critical closing pressure (CrCP) of the cerebral circulation depends on both tissue intracranial pressure and vasomotor tone. CrCP defines the arterial blood pressure (ABP) at which cerebral blood flow approaches zero, and their difference (ABP - CrCP) is an accurate estimate of cerebral perfusion pressure. Here we demonstrate a novel noninvasive technique for continuous monitoring of CrCP at the bedside. The methodology combines optical diffuse correlation spectroscopy (DCS) measurements of pulsatile cerebral blood flow in arterioles with concurrent ABP data during the cardiac cycle. Together, the two waveforms permit calculation of CrCP via the two-compartment Windkessel model for flow in the cerebral arterioles. Measurements of CrCP by optics (DCS) and transcranial Doppler ultrasound (TCD) were carried out in 18 healthy adults; they demonstrated good agreement (R = 0.66, slope = 1.14). Additionally, a potentially useful and rarely measured arteriole compliance parameter was derived from the phase difference between ABP and DCS arteriole blood flow waveforms. The measurements are well suited for long-term continuous monitoring of CrCP and assessment of arteriole compliance in the clinic.

back to the top

13. Detecting Mentions of Personal Medication Intake in Twitter

Ari Z. Klein, Abeed Sarker, Masoud Rouhizadeh, Karen O'Connor, and Graciela Gonzalez
Department of Biostatistics, Epidemiology, and Informatics, Perelman School of Medicine, University of Pennsylvania

In biomedical informatics, social media has been used for drug safety surveillance. However, studies that mine information (e.g., adverse drug reactions) from user posts that mention medications typically do not distinguish posts that actually indicate the user’s intake, perhaps because deriving this information poses challenges in natural language processing (NLP). Without this distinction, social media will remain a noisy data source and untapped for medication-related cohort studies. Towards improving and expanding the use of social media, we will present a publicly available, annotated corpus that can be used to train machine learning systems to automatically detect mentions of personal medication intake in Twitter. To build the corpus, we pre-processed a collection of tweets and queried them for 55 medications, including generated spelling variants. Two annotators annotated 10,260 tweets, with overlapping annotations for 10% of the tweets. Their inter-annotator agreement was κ = 0.88 (Cohen’s Kappa). To account for the linguistic idiosyncrasies of how Twitter users might express their medication intake, each tweet is from a unique user. To demonstrate the utility of the annotated corpus as a training set for supervised classification, we performed experiments using several machine learning algorithms. We used a stratified 80-20 (training/test) split of the annotated data and only word n-grams (n = 1, 2, 3) as features, following standard pre-processing. The baseline results suggest that this annotated corpus can be used for training automated classification systems, and our error and feature analyses of the best performing classifier—Support Vector Machines (F = 0.67; Accuracy = 73.4%)—provide insights for improving the performance of the system in future work.

back to the top

14. EvoBic: an AI-based parallel biclustering algorithm for reproducible pattern discovery in big data

Patryk Orzechowski1,2, Moshe Sipper1,3, Xiuzhen Huang4, and Jason H. Moore1       
1Institute for Biomedical Informatics, University of Pennsylvania, Philadelphia, PA; 2Department of Automatics and Biomedical Engineering, AGH University of Science and Technology, Krakow, Poland; 3Department of Computer Science, Ben-Gurion University, Beer-Sheva, Israel; 4Department of Computer Science, Arkansas State University, Jonesboro, AR

EvoBic is a next-generation biclustering algorithm based on artificial intelligence (AI). It is probably the first biclustering algorithm to capture with high accuracy six most challenging patterns in noisy and complex data and one of the very few parallel methods on the market. The algorithm outperforms multiple established biclustering methods in terms of recovery and relevance on both synthetic and genomic datasets. The algorithm is also up to 12 times faster than the most accurate method and maintains much higher reproducibility of the Gene Ontology (GO) terms findings across different datasets, independently of the platform and annotation. The proposed method may soon become a broadly accepted standard for complex data analysis in general and for gene expression in particular.

back to the top

15. Quantitative Doppler Vascularity Improves Computer-Based Sonographic Diagnosis of Breast Cancer      

Laith R Sultan1, Zhen Chen1, Santosh S Venkatesh2, Emily F Conant1, Chandra M Sehgal1
1Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States; 2Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA, United States

Assess the role of quantitative ultrasound Doppler features in enhancing the performance of machine learning methods for breast cancer diagnosis. 8 grayscale computer-based features describing lesion morphology and margin sharpness were extracted from ultrasound images of solid breast masses. Quantitative Doppler features including vascular fraction area, flow velocity and flow volume indices were measured in 3 regions corresponding to lesion center, lesion rim, and surrounding parenchyma. Features were used independently to train logistic regression classifiers. Cross-validation was performed using the leave-one-out method. Dispersion of regression coefficients from the mean were measured to assess the learning ability of each cross-validation. Cases with highest dispersion represented weak learning (low confidence diagnosis) and were pruned. The diagnostic performance was measured by the area under ROC (Az). Sensitivity (Se) and specificity (Sp) were measured at the Youden index threshold for each ROC curve Age of patients with benign lesions (45.9 ± 13.4) was significantly different from malignant (57. 8 ± 12.1) (P < 10e-7). Morphologic features with age showed an Az of 0.85 ± 0.04. When Doppler measurements were included, the performance of ML improved markedly (Az = 0.89 ± 0.03. The improvement in diagnosis was significant, P = .04. Diagnostic performance improved to (AUC = 0.96), P = .01 after pruning of 18 weak learning (low confidence) cases. Adding Doppler vascularity features to sonographic morphologic features markedly improves the diagnostic performance of machine learning algorithms for breast cancer diagnosis.

back to the top

16. Investigating the role of the posterior paraventricular thalamic nucleus in habituation to stress

Corbett BF, Luz S, Siu V and Bhatnagar S
Department of Anesthesiology, Children's Hospital of Philadelphia

In the field of stress neurobiology, habituation is defined as a decreasing and learned response to a familiar stressor over time. Disrupted habituation is a signature of post-traumatic stress disorder (PTSD), causing devastating effects for those afflicted. In rats, we model habituation using the repeated restraint paradigm. Repeated exposure to this moderate stressor increases the expression of immediate early genes in certain brain regions, induces the production of stress-related hormones, and elicits struggle behavior. All of these responses are highest on day 1 of restraint and attenuate over time. We have previously identified the posterior paraventricular thalamic nucleus (pPVT) as a crucial brain region that regulates habituation. However, the underlying molecular mechanisms and specific neural connections between the pPVT and other brain regions that mediate habituation are unknown. Using Designer Receptors Exclusively Activated by Designer Drugs (DREADDs), we investigating the role of the pPVT in regulating the stress response. We demonstrated that chemogenetic inhibition of the pPVT attenuated habituation. Here show that in the pPVT, expression of activity-regulated cytoskeleton-associated protein (Arc), which regulates neuronal plasticity, is increased following 1 day of restraint and its knockdown attenuates habituation. We hypothesize that Arc-mediated alterations in neuronal plasticity in the pPVT are a critical mechanism underlying habituation of the stress response. Our findings offer new insight into the role of the pPVT in mediating the stress response and are among the first to provide a possible molecular mechanism of stress habituation.

back to the top

17. The serotonin transporter polymorphism 5-HTTLPR-LL genotype is associated with aortic valve insufficiency

Elisa Salvati1, Juan B Grau2,3, Nancy Rioux3, Robert J Levy1, and Giovanni Ferrari4,5
1Department of Pediatrics, The Children's Hospital of Philadelphia; 2University of Ottawa Heart Institute, ON, Canada; 3Valley Hospital Heart Center, Ridgewood, NJ; 4Perelman School of Medicine, University of Pennsylvania School of Medicine; 5Columbia University, New York City, NY.

Serotonin (5HT) is a neurotransmitter involved in a variety of physiological functions. Elevated levels of 5HT have been associated with valvulopathies. 5-HT functions are mediated by the serotonin transporter (SERT), that has two allelic forms, short (S) or long (L), due to a polymorphism in the promoter region (5-HTTLPR). The LL genotype has been associated with either a gain or loss of SERT function. Diminished SERT expression results in enhanced 5HTR signaling and downstream events involving up-regulation of TGF-beta related pathways. We hypothesized that patients with severe aortic insufficiency (AI) requiring surgery have a higher than expected frequency of LL, compared to aortic stenosis (AS) patients. The echocardiogram and surgical reports of 72 patients undergoing aortic valve replacement for isolated AI or AS were reviewed. Patients with i) isolated AS (N=18), or ii) AI in the setting of bicuspid aortic valve (N=18), or iii) AI and tricuspid aortic valve (N=36) were considered. 5-HTTLPR genotype was determined using genomic DNA extracted from buffy coats followed by high throughput fragment analysis. The allele frequency in the AI population (N= 54) was SS= 5.6%, LS= 53.7 %, LL 40.7 %. Within this group, the distribution of alleles in the TAV patients was SS= 5.5%, LS= 55.5%, LL= 38.9%, and in the BAV patients it was SS= 5.6%, LS= 50.0 %, LL= 44.4%. 5-HTTLPR in patients affected by AS follows a different distribution with SS genotype being the most frequent (SS= 39.9%, LS= 44.4%, LL= 16.7%). 5-HTTLPR-LL was significantly increased in AI versus AS (p<0.001). The 5-HTTLPR-LL genotype is more prevalent in AI compared to AS. This may indicate an imbalance in 5HTR signaling that contributes to the pathophysiology of aortic regurgitation.

back to the top

18. Vesicular Monoamine Transporter 2 (VMAT2) Binding in Human Dorsal Raphe

Megan L. Fitzgerald1,2, Suham Kassir2, Mihran Bakalian2, Mark D. Underwood2, J. John Mann2, Victoria Arango2
1Department of Radiology, Perelman School of Medicine, University of Pennsylvania, 2Columbia University and New York State Psychiatric Institute, New York, NY

The vesicular monoamine transporter 2 (VMAT2) is responsible for the transport of monoamines serotonin, dopamine, norepinephrine, and histamine into synaptic vesicles. While the role of this 12-transmembrane domain transporter protein has been well-studied and mapped in striatal dopamine-rich brain regions, less is known about its expression patterns in the human dorsal raphe nucleus (DRN )—the region that contains the majority of serotonin-synthesizing neurons in the brain. We therefore sought to map the distribution of VMAT2 in the human DRN. Autoradiography was performed in postmortem human brain sections using tritiated dihydrotetrabenazine ([3H]DHTBZ), which binds specifically to VMAT2. VMAT2 expression throughout the brainstem was mapped in a single drug-free subject who did not have a psychiatric disorder and died suddenly from natural causes. 24 sections were collected throughout the brainstem. In the brainstem, VMAT2 binding overlapped with the immunoautoradiographic localization of tryptophan hydroxylase, the rate-limiting enzyme in the biosynthesis of serotonin. VMAT2 binding was also evident within the substantia nigra, which is critically involved in dopamine function, and the locus coeruleus, principal site of norepinephrine synthesis. Furthermore, two sections representative of rostral and caudal levels of the DRN, were analyzed for VMAT2 binding from 38 additional subjects (27 male, 11 female). VMAT2 binding was measured in the dorsal, ventral, ventrolateral, interfascicular , and caudal subnuclei of the DRN as well as the median raphe nucleus. VMAT2 binding did not differ by sex or race. These findings are consistent with the critical role of VMAT2 in serotonin transport and highlight VMAT2 in the brainstem as potentially important to mood regulation.

back to the top

19. Utrophin Genome Editing for Duchenne Muscular Dystrophy (DMD)

Kasturi Sengupta1, Manoj K. Mishra2, David W. Fredrick3, Tejvir S. Khurana1
1Department of Muscle Physiology, 2Department of Anesthesiology, Medical College of Wisconsin, Milwaukee, 3GlaxoSmithKline

Duchenne Muscular dystrophy (DMD) is a severe muscle-degenerative genetic disease caused by dystrophin gene mutations, leading to a nonfunctional truncated dystrophin protein. Currently, a number of dystrophin based approaches such as gene therapy using Adeno-associated virus (AAV) based-dystrophin gene delivery, stem cells and dystrophin exon skipping using splice-skipping 2OMePS and morpholino site blocking oligonucleotides (SBOs) are in various stages of preclinical and clinical development. An exciting new development has been to use genome editing rather than SBOs to achieve exon skipping to circumvent the potential toxicity of SBOs. It has been previously demonstrated that muscle dystrophin protein, which becomes truncated in DMD, can be functionally compensated by utrophin, a homologous member of the spectrin superfamily. An alternative approach to dystrophin-based DMD therapy, is by upregulating the expression of Utrophin, the autosomal homolog of dystrophin.
Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and CRISPR-associated protein 9 (Cas9) genome editing strategies are proving extremely valuable in terms of developing therapeutic strategies for treating genetic diseases. Earlier work from our lab identified the role for microRNAs, including let-7c, in suppressing the utrophin transcript by binding to the 3' UTR, suggesting that means of inhibiting this mechanism might result in therapeutic upregulation of utrophin expression. Here we designed the CRISPR-Cas9 to delete the binding site of those inhibitory microRNAs, with the aim of upregulating utrophin protein expression and rescue dystrophic phenotype.
We screened numbers of short guide RNAs to target the SaCas9 to edit out the let-7c and other microRNA target site from the 3’ UTR of Utrn gene. The CRISPR Cas9 was cloned in a single plasmid and transfected in human (HEK293T) cell line and mouse (C2C12) cell line. Effective genome editing was screened by a simple PCR amplification from the transfected cells genomic DNA. The selected short guide RNA pairs were cloned along with SaCas9 in a plasmid for generation of AAV vector. The SaCas9 and guide RNA pair containing AAV was used to transduce the HEK293T and hiPSC (human induced pluripotent stem cell) cell line.
The CRISPR-Cas9 based strategy was used successfully to delete ~500 bp region from the 3’UTR of the Utrn gene in order to remove the binding sites for microRNAs let-7c, miR-150, miR-133b, and miR-196b. Following in vitro validation of the approach in mouse (C2C12) and human (HEK293T) cell lines, the CRISPR construct was packaged into AAV viral particle and used to transduce the HEK293T and hiPSC cell line. We have plan to estimate the endogenous utrophin protein upregulation after genome editing. A major advantage of utrophin based gene editing approaches is that rather than a small subset of DMD patients that could benefit using exon-skipping by genome editing or SBOs, utrophin upregulation promises to benefit all DMD patients. In future we have plan to package the CRISPR-Cas9 targeting mouse genome to AAV and deliver in mdx mouse model of DMD. The ability of AAV CRISPR-Cas9 based Utrophin genome editing strategy to upreguate utrophin and rescue dystrophic pathophysiology in the mdx mouse model of DMD, in vivo will be examined.

back to the top

20. Revealing the Notch-driven transcriptional landscape in triple negative breast cancer

Jelena Petrovic1, Yeqiao Zhou1, John Johnson2, George Georgakilas1, Golnaz Vahedi2, Warren Pear1 and Robert B. Faryabi1
1Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania; 2Department of Genetics, Perelman School of Medicine, University of Pennsylvania

Triple-negative breast cancer (TNBC) is an aggressive disease that is incurable with conventional chemotherapy. 10% of TNBC patients harbor mutations in genes coding for Notch receptors. Importantly, elevated Notch signaling is specifically enriched in and associated with worse survival in the TNBC subtype. Despite this, the Notch targets in TNBC are poorly understood. To create a high confidence map of the oncogenic Notch-driven regulome in TNBC, we performed an integrative and functional genome-wide analysis of Notch-driven targets in Notch-mutated TNBC cell lines by combining high-resolution genome-wide chromatin topology data (HiChIP), dynamic mapping of Notch transcriptional complex members, active and repressive chromatin marks (ChIP-seq), as well as evaluation of dynamic transcription (RNAseq) in presence and absence of Notch signals. One important oncogene and critical Notch target in TNBC is MYC. Our integrative analysis produced a detailed map of Notch-dependent MYC regulatory network in TNBC. The TNBC-specific MYC regulatory region is significantly more complex compared to other known Notch-driven malignancies. It constitutes of a broad multi-domain active enhancer region, directly bound by the members of Notch transcriptional complex (NTC) at multiple locations. These novel enhancers are sensitive to acute changes in Notch activation and contact the MYC promoter via tumor-specific chromatin interactions. We showed that genetic perturbation of these NTC by CRISPR/Cas9 approach was sufficient to repress MYC expression. To this end, elucidating the Notch regulatory landscape and particularly the mechanism by which Notch regulates MYC expression in TNBC could guide personalized treatment options for breast cancer patients with Notch-driven tumors.

back to the top

21. Finding biomarkers in a pan-cancer multi-omics data set with inteGREAT

Gregory W. Schwartz1and Robert B. Faryabi1,2,3
1Department of Pathology and Laboratory Medicine; 2Abramson Family Cancer Research Institute; and 3Center for Personalized Diagnostics, Perelman School of Medicine at the University of Pennsylvania

With data collection encompassing multiple aspects of biological systems from the transcriptome to the proteome, there is an increasing need for algorithms to integrate different levels of information. While most analyses focus only on a particular aspect of a system, for instance the transcriptome, these analyses may miss critical details and limits the spectrum of answerable questions. We present here "inteGREAT", a novel method of integration of multi-omics data. inteGREAT considers all information from each level to measure conservation between different sources using random walkers and / or cosine similarity. This technique allows for reporting of each individual molecule based on all relationships with other molecules. Importantly, inteGREAT detects differential biomarkers between conditions. For instance, inteGREAT can rank the molecules are the most confidently conserved between the transcriptome and proteome in a disease and different from the normal condition, effectively pinpointing the genetic biomarkers of a disease. We applied inteGREAT to the transcriptome and proteome data sets from the TCGA pan-cancer cohort. To evaluate inteGREAT beyond simulations, we first focused on differential integrative analysis between the luminal and basal breast cancer subtypes. We ranked the canonical biomarker estrogen receptor 1 (ESR1) as the second most differential molecule between these subtypes, along with other high ranking targets such as GATA3 and CA12. Expanding our analysis to a larger pan-cancer set, we are applying inteGREAT to breast, ovarian, colon, and rectal cancers. Comparative methods using multiple sources will be instrumental in capturing the underlying complexity of biological systems and improve identification of drivers of disease states.

back to the top

22. Noncoding Recurrent Mutations in Pediatric Acute Myeloid Leukemia

Bing He1,2, and Kai Tan1,2
1Division of Oncology and Center for Childhood Cancer Research, Children’s Hospital of Philadelphia; 2Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia

Acute myeloid leukemia (AML) is a heterogeneous disease that accounts for approximately 20% of acute leukemias in children and adolescents. To date, the molecular pathogenesis of AML for different subtypes of AML has not been completely defined, limiting our ability to develop targeted therapies. Despite great advances of mapping somatic mutations in AML patients, much work remains to be focused on the alterations affect protein-coding genes. However, over 98% of the genome is made up of noncoding regions. The role of noncoding variants in tumorigenesis is currently likely underestimated. Several recent studies on the structural variants near cis regulatory regions have shed more light on the importance of noncoding variants in cancer. Novel mechanisms of enhancer hijacking for oncogene activation have been elucidated. The inactivated cancer genes were found to link to active enhancers via translocation, disruption of topologically associating domain (TAD) boundary, or tandem duplication of TAD boundary. Such enhancer hijacking events have been found in the patients of multiple tumor types. Here we developed a computational framework to identify causal noncoding mutations that disrupt enhancer regulation of oncogenes and tumor suppressor genes. By systematically analyzing whole genome sequencing (WGS) and RNA-Seq data for 152 pediatric AML patients, we identified a set of causal mutations that disrupt enhancer-promoter interactions and cause deregulation of the target genes in pediatric AML patients. The predicted causal mutations will be experimentally validated and using both cell line models and patient-derived xenograft models of pediatric AML.

back to the top

23. Non-AUG translation is resistant to multiple common protein synthesis inhibitors

Michael G. Kearse and Jeremy E. Wilusz
Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania

Aberrant translation from non-AUG start codons (e.g., CUG, GUG) has been recently implicated as a driving force in cancer malignancy and multiple neurodegenerative diseases. To study how non-AUG translation may be regulated differently than canonical AUG-initiated translation, we designed luciferase and eGFP reporters with various start codons for expression in cultured mammalian cells. Surprisingly, we found that non-AUG translation is selectively resistant to multiple protein synthesis inhibitors. While treatment with puromycin and lactimidomycin inhibited non-AUG translation, the well-characterized elongation inhibitors cycloheximide, anisomycin, and emetine were each ineffective. In fact, the amount of protein expressed from non-AUG start codons increased over time. This is in stark contrast to canonical AUG-initiated translation, which declined over time due to protein turnover, as expected. By altering the 5’ and 3’ ends of the reporters, we find that this inhibitor-resistant form of non-AUG translation does not require the 3’ poly(A) tail, but does require a scanning 40S for initiation. Specifically, insertion of a strong hairpin within the 5’ leader is sufficient to cause non-AUG translation to be sensitive to the inhibitors. Further control experiments confirmed that this phenotype is cell type-, reporter-, and promoter-independent. Together, these data clearly demonstrate that non-AUG translation is regulated uniquely from canonical AUG translation. Given that variations in rRNA and r-protein composition/modification have been previously reported to cause inhibitor resistance, we propose that ribosomes translating non-AUG encoded ORFs are fundamentally distinct from those translating AUG encoded ORFs, explaining the dichotomous sensitivity to the inhibitors.

back to the top

24. Targeted neuroepigenetic editing of cdk5 regulates behavior

Ajinkya Sase1, Snehajyoti Chatterjee1,2, Brandon Santhumayor1, Rozalyn Wood1, Sonia Lombroso1, Ted Abel1,2, and Heller EA1
1Department of Systems Pharmacology and Translational Therapeutics, Perelman School of Medicine, University of Pennsylvania; 2Departments of Molecular Physiology and Biophysics, University of Iowa Carver College of Medicine, Iowa City, IA

Chronic stress and memory deficits mount an enormous toll on the global population. While epigenetic modifications driving gene expression related to stress and memory are widely documented, identification of the precise molecular mechanisms by which they regulate specific gene expression remains elusive. Cyclin dependent kinase 5 (Cdk5) is known to regulate both these behaviors and magnitude of these behaviors. Although there is much evidence on the function of Cdk5 protein, very little is known about the regulation of Cdk5 gene expression following stress and learning. We hypothesize that histone modification(s) of Cdk5 is sufficient to regulate its expression, and influence behavioral responses to chronic stress in both male and female animals. We thus sought to determine the spatiotemporal regulation of Cdk5 expression by both chronic unpredictable stress (CUS) and fear conditioning (FC) in mice. We have found that indeed Cdk5 levels vary by phase of FC. Specifically, we measure a significant increase in Cdk5 mRNA levels after memory retrieval, and a decrease after acquisition and consolidation phases. We then utilized targeted epigenetic editing to examine the behavioral and biochemical consequences of Cdk5 histone modifications in CUS & FC. This work will provide a model of stress and memory-evoked chromatin remodeling at Cdk5, and the causal relevance of Cdk5 transcriptional regulation to stress- and memory- induced behavior. Finally, the identification of such precise mechanisms in stress-mediated gene expression allows for the development of targeted therapeutic treatments.

back to the top

25. Egg-adapted influenza vaccines elicit human antibodies focused on the egg-adapted receptor binding site of hemagglutinin

Tyler A. Garretson1, Benjamin S. Chambers1, Arnold S. Monto2, Scott E. Hensley1
1Department of Microbiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; 2Department of Epidemiology, University of Michigan School of Public Health, Ann Arbor, MI

Influenza virus hemagglutinin (HA) binds to terminally-linked sialic acids on target cells. Some broadly neutralizing antibodies target the HA receptor binding site (RBS). Human influenza viruses possess a relatively conserved HA RBS that binds efficiently to α-2,6-linked sialic acids, whereas the HA RBS of avian influenza viruses binds efficiently to α-2,3-linked sialic acids. Human influenza viruses passaged in eggs often acquire mutations in the HA RBS that promote binding to α-2,3-linked sialic acids. Raymond et al. recently isolated monoclonal antibodies specific for the egg-adapted HA RBS from humans following vaccination with an egg-adapted H1N1 vaccine strain. To determine if these types of antibodies are elicited by vaccination, we completed antigenic analyses using sera collected from 189 individuals pre- and post-vaccination with the 2015-16 seasonal influenza vaccine, along with individuals immunized with the same vaccine component in 2009-10. We completed hemagglutination-inhibition (HAI) serological assays to assess antibody binding to the HA of A/California/7/2009 wild-type strain, X-179A vaccine strain, and a strain possessing only the RBS Q226R mutation. We found that 5% of vaccinated individuals in 2015-16 had antibody titers that were >4-fold higher to the vaccine strain X-179A compared to WT. The same percentage of individuals studied from 2009-10 also had this X-179A vaccine-focused response. We confirmed that the mutation at Q226R, which is located within the RBS of the X-179A HA, was responsible for the antigenic differences between X-179A and WT. These studies highlight the need to develop better vaccine antigens that do not rely on production in eggs.

back to the top

26. “Silent code”: diverse functions of actin isoforms are defined by their nucleotide, rather the amino acid sequence.

Pavan Vedula, Satoshi Kurosaka, Nicolae Adrian Leu, Junling Wang, and Anna Kashina
Department of Biomedical Sciences, Veterinary School of Medicine, University of Pennsylvania

β‐ and γ‐cytoplasmic‐actin are nearly indistinguishable in their amino acid sequence, but are encoded by different genes that play non‐redundant biological roles. The key determinants that drive their functional distinction are unknown. Here we tested the hypothesis that β- and γ-actin functions in vivo are defined by their nucleotide, rather than their amino acid sequence, using targeted editing of the mouse genome. Although previous studies have shown β-actin is essential for cell migration and embryogenesis, we demonstrate here that γ-actin protein expressed off the β-actin gene is functionally capable of substituting for β-actin without any significant impact on organism survival. Our data suggest that the differences in in vivo functions of β- and γ-actin actin are determined solely by their nucleotide sequence.

back to the top

27. Myeloperoxidase as an Immunomodulator in Intraoperative Photodynamic Therapy

Richard W. Davis IV1, Emma Snyder1, Joann Miller1, Astero Klampatsa2, Steve Maceyko2, Steve Albelda2, Keith Cengel1, Theresa M. Busch1
1Department of Radiation Oncology and 2Department of Medicine, Perelman School of Meidicine, University of Pennsylvania

Treatment of malignant pleural mesothelioma remains palliative in nature, and consists of surgical resection in order to achieve local control. More recently, surgical procedures which spare the lung (radical pleurectomy) have been coupled with photodynamic therapy (PDT) of residual disease to achieve better local control. Due to increasing evidence of the effects of surgery on both host immunity and residual tumor cells, we investigated the contribution of injuries sustained during surgery to efficacy of photodynamic therapy in a mouse model of malignant mesothelioma. We previously observed that surgical injury prior to PDT eradicates long-term response in vivo. As it relates to PDT outcome, we are now investigating the predisposition of neutrophil profiles in the presence and absence of surgical injury. Our results demonstrated neutrophils influx the tumor and lymph node sooner when PDT was preceded by surgical injury, as demonstrated by higher neutrophil counts in the respective tissue. Through in vivo imaging of luminol chemiluminescence as a marker of neutrophil activation, we show a role of neutrophil-secreted myeloperoxidase activity in producing long-term response after PDT. However, myeloperoxidase deposition in the lymph node is known to suppress dendritic cell migration, activation, and antigen uptake. Thus, we are currently investigating if early influx and activation of neutrophils in tumor draining lymph node results in a loss of establishment of PDT-mediated immunity. Taken together, these studies will describe potential immunomodulatory roles for myeloperoxidase in responses to intraoperative PDT.

back to the top

28. The Integrator complex cleaves nascent RNAs to regulate transcription of inducible protein-coding genes 

Deirdre C. Tatomer1, David Bailout2, Nathan Elrod2, Telmo Henriques3, Dongming Liang1, Karen Adelman3, Eric J. Wagner2, Sara Cherry4, Jeremy E. Wilusz1  
1Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania; 2Department of Biochemistry and Molecular Biology, The University of Texas Medical Branch, Galveston, TX; 3Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA; 4Department of Microbiology, Perelman School of Medicine, University of Pennsylvania

Cells maintain homeostasis by responding to their environment through regulated transcriptional programs. In Drosophila, the presence of excess metals triggers import of the MTF-1 transcription factor to the nucleus, where it activates expression of the Metallothionein genes that then act to sequester the metals. To reveal further insights into how Metallothionein A (MtnA) activation is regulated, we performed a high-throughput RNAi screen and surprisingly found that depletion of the Integrator complex (INT) results in “superinduction” of the MtnA promoter. INT has a well-established role in 3' end processing of snRNAs, but we find that it is also directly recruited to the MtnA promoter upon metal stress. Using reporter genes driven by the MtnA promoter, we demonstrate that transcriptional up-regulation upon INT depletion is not affected by the coding ORF, splicing signals, nor the mRNA 3’ end processing signals. A number of small RNAs are generated from the 5’ ends of the endogenous MtnA gene and the reporter in an INT-dependent manner. Expression of these small RNAs increases upon depletion of the RNA exosome, and the MtnA promoter fails to induce when these RNA degradation/processing components are depleted from cells. These results suggest that INT cleaves nascent transcripts derived from the MtnA promoter, and degradation of these small RNAs is required for MtnA induction. When Integrator is depleted from cells, these small RNAs are not produced and we hypothesize that more polymerases continue into the productive elongation phase, leading to the “superinduction” phenotype. ChIP-seq identified a number of other gene promoters bound by INT, suggesting that this complex may play a broad role in regulating early steps in transcription of protein-coding genes.

back to the top

29. The Smell of Sickness: How Inflammatory Signals and Metabolic Pathways Influence Body Odor

Patrick Millet1, Bruce Kimball1,2, Gary Beauchamp1
1Monell Chemical Senses Center; 2USDA National Wildlife Research Center

Body odors of often contain biological information about the producer’s identity and physiological state. In mice and many other species, this includes information regarding health status. However it is currently unknown how much detailed health information is present in body odor. Previous research demonstrates that trained biosensor mice can discriminate between body odors induced by different immune stimuli. We investigated select cytokine signals and metabolic changes commonly associated with immune activation to determine the role each played in disease-related body odors. Mice trained to recognize lipopolysaccharide (LPS)-induced body odor generalized to (identified as similar) body odor from conspecifics treated with the inflammatory cytokine TNF; however, mice failed to generalize to a downstream inflammatory cytokine, IL-1β, or the compensatory anti-inflammatory cytokine IL-10. We further found that some of the LPS-induced changes to metabolism could be induced by 48 hours of fasting, even in the absence of an inflammatory stimulus. This period of fasting makes the resulting body odor indistinguishable from LPS-induced odor, as measured by conspecific biosensors. Analysis of body odors present in urine by GC/MS headspace techniques detected changes in endogenous volatile organic compounds resulting from each treatment. LDA modeling of these volatile metabolites suggest similarities between the volatile profiles of LPS, TNF, and IL-1β treated mice, although the three treatments were readily differentiated from each other. Together, our findings indicate that chemometric analysis of body odors can reveal complex information regarding the health status of an individual. These data advance our goal of using volatile metabolite analysis as a medical diagnostic tool.

back to the top

30. Dynamic microtubules specify local delivery and capture of presynaptic cargo at en passant synapses 

Pedro Guedes-Dias, Jeffrey J. Nirschl, Nohely Abreu, Mariko Tokito, and Erika L.F. Holzbaur         
Department of Physiology, Perelman School of Medicine, University of Pennsylvania

In contrast to the classical terminal synapse at the distal end of the axon, the majority of synapses in the central nervous system are established en passant along axonal processes, juxtaposed to uniformly polarized microtubule arrays. Presynaptic cargo are transported along microtubules by KIF1A and dynein motors, however, the mechanisms specifying their targeting to en passant synapses are not well understood. Here, we investigate the role of local cytoskeletal determinants and vesicular motors in this process using live-neuron fluorescence microscopy and in vitro single-molecule approaches. We found that presynaptic regions are hotspots of high microtubule dynamicity, and anterogradely moving presynaptic vesicles pause preferentially at these sites. We found that dampening microtubule dynamics with low-dose nocodazole was sufficient to abolish the preferential retention of anterograde-moving vesicles at presynapses. In contrast, sites of lysosome pausing were stochastically distributed along the axon, with low rates of capture at synaptic regions. This difference suggests that the high dynamicity of presynaptic microtubules particularly affects presynaptic cargo transport, potentially acting as a cue to halt KIF1A processivity. Using single-molecule assays, we found that the lysosome motor kinesin-1 binds equally well to either dynamic or stable microtubules. In contrast, the presynaptic vesicle motor KIF1A has a significantly lower affinity to dynamic microtubules, thus revealing a mechanistic basis for the high vesicle pausing and retention rates observed at presynaptic sites in neurons. Together, our results unveil a model whereby the interplay between local microtubule cues and cargo-specific motors allows precise delivery of appropriate cargo to the presynapse.

back to the top

31. Designing transmembrane redox active proteins for action potential sensing  

Sohini Mukherjee, Martin Iwanicki, and Bohdana Discher      
Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania

Our complex behavior is encoded by action potentials, which are changes in electrical potential along the membrane of the nerve cells. To measure these potentials, many voltage-sensitive indicators have been already developed. However, their in vivo applications have not been fully realized due to various combinations of insensitivity, slow kinetics, heavy capacitative loading, phototoxicity, or lack of genetic targetability. The main goal of this work is to design highly improved optical voltage sensors based on genetically encoded de novo designed synthetic redox proteins, called maquettes (Maq), are being developed to bind a chain of redox cofactors to form an intra-protein electron transfer pathway. Electric field changes across the membrane will lead to a change in the oxidation state of the Maq cofactor positioned adjacent to a fluorescent protein (FP). We are currently testing the sensitivity of fluorescent proteins to report the oxidation state of a very commonly used cofactor, heme. Due to the ideal spectral overlap of the fluorescence of the FP (mOrange2) with the absorption spectra of the reduced heme, we expect a good FRET only for reduced heme. Our second design involves fusing Redox-Sensitive Red Fluorescent Protein, rxRFP, to the voltage sensitive Maq to initiate fast electron transfer between the Maq and rxRFP that will cause a change in the rxRFP fluorescence signal. To optimize the kinetics, sensitivity, and brightness of our genetically targetable sensors, we are working first with water-soluble Maq-FP constructs. The successful design principles will be integrated into a transmembrane construct AM-1, which contains chain of 3 hemes and are expected to achieve a speed comparable to the best organic voltage sensitive dyes when fused to FPs.

back to the top

32. The role of VEGFB in fetal and placental vascularization

Sneha R. Mani, Chantae Sullivan-Pyke, Teri Ord, Christos Coutifaris, and Monica Mainigi
Center for Research on Reproduction and Women's Health, University of Pennsylvania

Assisted reproductive technologies (ART) increase the incidence of adverse pregnancy outcomes. We previously showed that superovulation (SO) - a common procedure in ART - impacts fetal and placental weight and causes vascular changes. The vascular endothelial growth factor (VEGF) family of proteins, specifically VEGFA and PLGF, are essential for pregnancy. However, manipulating their expression leads to systemic effects limiting therapeutic utility. VEGFB’s role is ambiguous, though studies suggest functions specifically in pathologic conditions. This led us to hypothesize that VEGFB could impact vascularization during pregnancy. We first examined VEGFB localization in mice through gestation. VEGFB is expressed cytoplasmically in the preimplantation uterus, at E6.5 when fetal vasculogenesis begins and in E18.5 placenta. In humans, VEGFB is expressed in endothelial and trophoblast cells at term. We next examined the effect of SO on vascular gene expression. In SO placenta at E18.5, VegfB was specifically upregulated. Since placental vasculogenesis is believed to be under maternal control, we examined uteri and also found increased VEGFB. To mechanistically explore VEGFB’s role in placental development, we added recombinant VEGFB to a trophoblast cell line, and found a suppression of VEGFA mRNA and protein levels. This study demonstrates that SO leads to upregulation of maternal and fetal VEGFB which may be responsible for vascular changes following ART. In vitro studies show a regulatory effect of VEGFB on VEGFA suggesting that VEGFB could alter the bioavailability of VEGFA and affect early vascularization. VEGFB is non-essential in normal development, making it an attractive target for therapeutic intervention to prevent disorders of placentation.

back to the top

33. Characterization of the mechanically-induced shape change of erythrocytes into polyhedrocytes

Valerie Tutwiler1, Alexander R. Mukhitov1, Alina D. Peshkova2, Giang Le Minh2, Jacqueline Vicksman1, Chandrasekaran Nagaswami1, Rusten I. Litvinov1, John W. Weisel1
1Department of Cell and Developmental Biology, Perelman School of Medicine, University of Pennsylvania; 2Institute of Fundamental Medicine and Biology, Kazan Federal Research University, Russia

Polyhedral structures have been observed in nature since the time of Aristotle; however, most research has focused on Platonic and Archimedean solids rather than irregular polyhedra. Recently, these tightly packed cells were observed in contracted blood clots, where the highly deformable erythrocytes were compressed into the core of the clot and formed a tessellated network. This resulted in a shape change from the normal biconcave cell to a polyhedral shape; thus, the term polyhedrocytes was coined. We use histology, transmission electron microscopy, scanning electron microscopy and confocal microscopy to visualize and quantify the mechanical deformation of erythrocytes. We found that shape change is linked to the presence of contractile forces generated by the platelets and directly correlated with platelet count and activity. Inhibition of contractile proteins or ability to propagate force resulted in softer clots with fewer polyhedrocytes. Three-dimensional reconstruction of erythrocytes from contracted clots allowed us to compare bioconcave cells, polyhedrocytes and intermediate forms. There was no change in the surface area, volume, or sphericity of the cells following shape change. The polyhedrocytes became less oblate and more prolate than the biconcave cells. The polyhedrocytes had a total number of faces from 10 to 16 with a median of 13 faces. The faces were made up of polygons with 3 to 6 sides, with the majority having 4 sides. Taken together, these results point to the importance of studying erythrocyte shape change as a result of clot contraction into a tightly packed network, since these structures could provide a key function of replicating the barrier function of the endothelial layer post injury and/or influence the outcome of thrombotic conditions.

back to the top

34. CA novel functional role for local striatal inhibitory circuits during goal-directed behavior

Elizabeth N. Holly, Felicia Davatolhagh, Kyuhyun Choi, and Marc V. Fuccillo
Department of Neuroscience, Perelman School of Medicine, University of Pennsylvania

Deficits in goal-directed behavior are a hallmark of most neuropsychiatric diseases. The dorsomedial striatum (DMS) is a key mediator of goal-directed actions. The DMS striatal low-threshold-spiking (LTSI) interneuron population is sparse, yet exhibits strong synaptic control over spiny projection neurons. Thus, LTSIs are poised to play a key role in regulating goal-directed behavior, although no behavioral function of LTSIs has been reported. We first virally overexpressed Kir2.1, an inwardly rectifying potassium channel, to decrease excitability of LTSIs in the DMS. Mice were then trained in a self-initiated operant task consisting of temporally discrete initiation, choice and reward periods. Mice were first trained in a two-alternative forced choice task to respond on one lever. After 3 consecutive days >50 rewards, the contingency reversed, rewarding the previously inactive lever. Mice were then run in a dynamic serial reversal task, in which the active lever reversed after 8/10 correct responses. Reduced activity of LTSIs enhanced acquisition of the initial contingency, decreased the number of trials to reverse in the initial single reversal, and improved performance on the first day of the serial reversal task. To further explore the temporal specificity of LTSI involvement in this task, we used halorhodopsin-mediated optogenetic inhibition specifically during the reward phase. Similar to Kir2.1 overexpression, optogenetic inhibition of LTSIs selectively during reward delivery accelerated task acquisition. Together, these data suggest striatal LTSI activity during reward processing may inhibit acquisition of new behavioral contingencies. Current work focuses on the striatal circuit mechanisms by which this interneuron population modulates goal-directed choice.

back to the top

35. Mapping projection populations to striatal neurons: Integrating anatomical and physiological perspectives.

Kyuhyun Choi1, Elizabeth Holly1, Felicia Davatolhagh1,2, Kevin Beier3, and Marc V. Fuccillo1
1Department of Neuroscience, and 2Neuroscience Graduate Group, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; 3Department of Psychiatry and Behavioral Sciences, Stanford University Medical School, Stanford, CA

The dorsomedial striatum (DMS), a key site of reward-sensitive motor output, receives extensive afferent input from neurons in cortex, thalamus and midbrain. It is presently unclear whether these projections exhibit specific connectivity to the diverse cell types within DMS microcircuits. We compared the anatomical and physiological connectivity of projection neurons that synapse on three major DMS populations - D1 spiny projection neurons (SPN), parvalbumin-positive fast-spiking interneurons (PV+ FSIs) and somatostatin-positive low-threshold-spiking interneurons (SST+ LTSI) - using cell-type specific rabies virus (RV) tracing and optogenetic-mediated projection neuron recruitment. Retrograde tracing revealed target cell type-specific patterns for the major DMS inputs, including orbitofrontal cortex, anterior cingulate/secondary motor cortex (ACC/M2), and parafascicular nucleus of the thalamus. While rabies virus methods have been regularly used to assess anatomical connectivity, it is unclear how this information corresponds to measures of synaptic connectivity. To explore this, we compared our anatomical data with novel afferent-specific measures of synaptic strength for ACC/M2 and parafascicular thalamic projections. We found a significant divergence between cell type-specific anatomical connectivity and measures of excitatory synaptic strength. Furthermore, we observed that these distinctions were further modified according to the region of projection neuron origin. Taken together, this suggests a model for striatal wiring in which postsynaptic cell type imposes diversity of physiological connectivity on a “skeleton" of non-specific anatomical contacts – an arrangement that is largely obscured by rabies virus tracing approaches.

back to the top

36. Contributions of PRP28 and SRPK2 to regulation of alternative splicing          

Sean P. McClory1, Michael J. Mallory1, Changwei Shao2, Jinsong Qiu2, Xiang-dong Fu2, Kristen W. Lynch1    
1Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; 2Department of Cell and Molecular Medicine, University of California at San Diego, San Diego, California

Alternative splicing produces distinct mRNA isoforms from a single gene, consequently greatly increasing the diversity of the proteome. Regulation of alternative splicing is critical for cellular growth and development. Alternative splicing is regulated by factors that bind to pre-mRNA to promote or repress the assembly of the spliceosome at the 3’ and 5’ splice sites (3’/5’ss) of alternative exons. Spliceosome assembly is a multistep process, involving numerous rearrangements of the interacting RNA and protein components of the complex, and each step presents an opportunity to regulate alternative splicing. PRP28 is an RNA helicase that participates in spliceosome assembly by promoting the exchange of U1 small nuclear ribonuclear particle (snRNP) for U6 snRNP at the 5’ss. In human cells, PRP28 is phosphorylated by the kinase SRPK2. We knocked down PRP28 or SRPK2 in Jurkat T-cells using short-hairpin (sh) RNAs and used RASL-seq to look for changes in a set of ~5000 alternative splicing events. Upon KD of PRP28, a subset of 89 alternative exons displayed significant changes in inclusion. Those exons that were more included upon KD of PRP28 contained weaker 5’ss sequences that generally diverged from the consensus motif suggesting that PRP28 may repress these weaker splice sites. Exons that were less included upon KD of PRP28 contained 5’ss that were generally like those of unaffected alternative exons. Effects of SRPK2 on alternative splicing appear limited to a subset of PRP28-repressed exons, where SRPK2 also represses inclusion. The correlation between SRPK2 and PRP28 suggests that SRPK2’s effects on alternative splicing are mediated through its phosphorylation of PRP28 and that phosphorylation of PRP28 plays a role in its proofreading of weak 5’ss sequences.

back to the top

37. The oncogenic function of MN1 in acute myeloid leukemia

Simone S. Riedel, Molly C. Kingsley, Clara Libbrecht, Taylor Pastuer, and Kathrin M. Bernt
Division of Oncology, Children's Hospital of Philadelphia

Meningioma-1 (MN1) is a transcriptional coactivator that is overexpressed at high levels in a subset of Acute Myeloid Leukemia (AML) patients who have a particularly poor outcome. 70-80% of patients with MN1-high AML die within 2 years of diagnosis. MN1 has strong oncogenic potential in mouse models and induces an aggressive leukemia as a single hit. In comparison to other highly aggressive murine leukemia models, such as MLL-fusion driven AML, MN1 leukemia stands out with significantly shorter latency. Yet, very little is known about the mechanism how MN1 induces leukemia. MN1 is a large protein with no homology to any other known protein. We showed that its overexpression induces a leukemogenic gene program that is in part dependent on the methyltransferase DOT1L. Early results from DOT1L directed clinical trials demonstrate that DOT1L inhibition alone will not be curative. Therefore, we explored the MN1 interaction network using proximity dependent biotinylation followed by mass spectrometry. We identified the MLL2 (KMT2B) and MLL3 (KMT2C) COMPASS-like complexes as well as MLL1 (KMT2A) as potential MN1 interaction partners. We show that KO of mll1 in MN1-driven leukemia significantly increases latency. With this screen we also verified a previously described interaction of MN1 with CBP/P300 and show that it is functionally important. Inhibition of the cbp/p300 bromodomain and HAT domain induces cell cycle arrest and apoptosis in murine MN1-driven leukemia. We are currently working on validating further interaction partners. The goal is to identify interaction partners that are functionally important and to ultimately understand the molecular mechanisms of MN1 oncogenic function.

back to the top

39. A proteomic approach to identify complement components in the African malaria mosquito Anopheles gambiae

Sutopa Dwivedi1, Gregory L Sousa1, Sarah Sneed1, Elizabeth Edgerton1, Akhilesh Pandey2, and Michael Povelones1      
1Department of Pathobiology, School of Veterinary Medicine, University of Pennsylvania; 2Johns Hopkins University, School of Medicine, Baltimore, MD

Anopheles gambiae is one of the most efficient vectors of malaria in the world. Despite its efficiency, A. gambiae mounts a potent immune response against malaria parasites (Plasmodium) that drastically limits infection. The mosquito’s complement system is the main arm of the immune defense machinery. This pathway, present in the mosquito’s blood (hemolymph) forms a protective barrier against bacterial and fungal infections. Our goal is to identify new mosquito complement proteins and characterize their function to develop novel ways to disrupt transmission. We have developed a new method to study complement components by proteomic analysis of mosquito proteins isolated from E. coli surface after exposure to the hemolymph. Using state of the art high resolution LTQ-Orbitrap Fusion mass spectrometer, the hemolymph surface extracted material has been analyzed from knockdown of two complement proteins (TEP1 and LRIM1) and controls. Label-free quantitative proteomics was performed on the extracted samples to reveal hemolymph proteins which show differential protein expression on the bacterial surfaces isolated from control mosquitoes compared to the TEP1 or LRIM1 knockdowns. Data analysis provided abundance profiles of 674 proteins using SIEVE search algorithm. We have selected a subset of these proteins for screening using RNAi-mediated gene silencing for changes in E. coli. Preliminary silencing data on a serine protease and a leucine-rich repeat protein suggest roles in bacterial clearance and melanization, two effector functions linked to mosquito complement activation. Given the correspondence between anti-bacterial and anti-plasmodial defense, we will screen the candidates for intensity and prevalence upon infection by rodent and human malaria parasites.

back to the top

40. Adenovirus co-opts host factors to mediate m6A modification of viral RNAs important for late gene production      

Alexander M. Price1, Katharina E. Hayer2, Nandan S. Gokhale3, Alexa B. McIntyre4, Chris E. Mason4, Stacy M. Horner3, Matthew D. Weitzman1,5
1Pathology and Laboratory Medicine, The Children's Hospital of Philadelphia; 2Biomedical and Health Informatics, The Children's Hospital of Philadelphia; 3Department of Molecular Genetics and Microbiology, Duke University Medical Center, Durham, NC; 4Department of Physiology and Biophysics, Weill Cornell Medicine, New York, NY; 5Department of Pathology and Laboratory Medicine, University of Pennsylvania Perelman School of Medicine

The discovery of reversible N6-methyladenosine (m6A) modification of RNA has fundamentally altered our view of the central dogma of molecular biology. M6A is a chemical modification added post-transcriptionally to RNA, where it has been implicated in such diverse processes as RNA splicing, nuclear export, stability, and translation. Since small DNA viruses that replicate in the nucleus have to employ cellular machinery to transcribe and translate their gene products, these viruses have developed ways to harness cellular RNA processing pathways. Shortly after the discovery of m6A on human RNA, it was demonstrated that Adenovirus RNAs are also marked by m6A. However, the effect of m6A modifications on Adenovirus has never been deciphered, and no progress has been made in understanding how these so-called “epitranscriptomic” modifications impact the life cycle of DNA viruses. Using methylated RNA Immunoprecipitation and Sequencing (meRIP-Seq), I have identified the site-specific locations of m6A modifications within the Adenovirus transcriptome. I have also shown that during infection Adenovirus recruits host factors involved in methylating RNA and binding methylated RNA to sites of active viral transcription. Furthermore, knock down of these m6A-specific methyltransferases, METTL3 and METTL14, reduces viral late protein expression and the production of infectious virus particles. These data demonstrate that m6A modification of viral RNA is important during the late stage of Adenoviral infection. These experiments are the first to show how a DNA virus co-opts cellular epitranscriptomic machinery to mediate viral RNA biogenesis.

back to the top

41. Rapid down-regulation of FV-short by activated protein C is evaded by its association with TFPIα

Ayombil F.1, Petrillo L.1, and Camire M. Rodney2
1The Children's Hospital of Philadelphia; 2University of Pennsylvania Perelman School of Medicine

Factor V (FV) East Texas bleeding disorder is characterized by the abundance in plasma of an active, FV-short (missing residues 756-1458) form, an alternatively spliced, truncated, B-domainless variant of the FV gene. This bleeding phenotype is partly explained by compensatory high levels of the plasma inhibitor, tissue factor pathway inhibitor-alpha (TFPIα), and its tight interaction via its C-terminal basic region (BR) to the exposed acidic region 2 (AR2) on FV-short. The result is impaired thrombin generation and prolonged clot time in the affected patients. Physiologically, activated protein C (APC) down-regulates active forms of FV such as FVa and FV-short. The present study investigates the mechanism by which TFPIα-FV-short complex regulates the susceptibility of this active molecule to proteolytic inactivation by APC. Using full length TFPIα or a peptide derived from the BR of TFPIα, we monitored the discontinuous proteolysis and inactivation of FV-short in the presence of APC in vitro. Proteolytic products were visualized by Western blotting using FV specific antibodies. APC cleaves both FVa and FV-short with similar profiles. However, densitometric analyses showed that both TFPIα and the BR peptide significantly inhibited proteolysis of FV-short with >80% of the starting material remaining after 15 min of incubation with APC compared to the controls. Using FV-short and other B-domainless forms (FV810-R306Q, and R506Q) we found that TFPIα and the BR peptide blocked cleavage at R506 suggesting the BR somehow obscures the ability of APC to engage FV-short at this site. TFPIα and the BR peptide had no effect at R306 or R679. Consistent with data showing a high affinity interaction between TFPIα-BR and AR2 of FV-short, TFPIα or the BR has no impact on FVa (lacks AR2) inactivation by APC. Together, these data show the FV-short-TFPIα complex is resistant to APC proteolysis. The findings are consistent with data from our laboratory showing that the internal BR of FV has a major impact on APC-mediated inactivation of this molecule. Thus, not only does TFPIα block the hemostatic function of FV-short it also protects it from down-regulation by the protein C pathway.

back to the top

42. Genetic ablation of NAD biosynthesis in skeletal muscle results in improved glucose tolerance and paradoxically increase

Timothy S. Luongo, David W. Frederick, William J. Quinn, and Joseph A. Baur
Department of Physiology and Institute for Diabetes, Obesity, and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA

Nicotinamide adenine dinucleotide (NAD) is required for generation of energy via glycolysis or oxidative phosphorylation and also acts as a cofactor for a multitude of enzymes. NAD levels have been shown vary dynamically in response to physiological or pathological states, and both genetic and pharmacological interventions to enhance NAD concentrations appear to have beneficial effects. Our lab has previously altered NAD biosynthesis in skeletal muscle by increasing or decreasing nicotinamide phosphoribosyltransferase (Nampt) expression. We observed that increasing NAD content did not result in an overt phenotype nor did it enhance skeletal muscle performance. In contrast, depletion of NAD levels leads to progressive skeletal muscle wasting triggered by mitochondrial dysfunction, which was rescued by nicotinamide riboside. Further characterization of skeletal muscle Nampt knockout (mNKO) mice has revealed some unexpected observations. First, mNKO mice had increased poly-ADP-ribosylation. This was striking as we expected NAD depletion would result in decreased PARP activity since it is an NAD-dependent process. Our results raise the possibility that there is an unknown feed-forward process arising from the reduction of NAD that enhances PARP activity. Equally surprising, mNKO mice had increased glucose tolerance. It was previously reported that NAD supplementation had anti-diabetic effects. However, we observed that depletion of NAD augmented glucose uptake. We hypothesize that AMPK is being activated independent of insulin signaling leading to increased Glut4 translocation. These findings raise new questions into how NAD signaling is contributing to muscle physiology.

back to the top

43. Characterization of the interaction between FVshort and TFPIα responsible for the East Texas bleeding disorder          

Teodolinda Petrillo and Rodney M. Camire          
Division of Hematology, The Children's Hospital of Philadelphia, Philadelphia, PA

The esophageal lumen is lined by a stratified squamous epithelium comprised of proliferative basal cells that differentiate while migrating towards the lumThe East Texas bleeding disorder (ET-BD) is a rare inherited autosomal disorder linked to a single point mutation in F5 gene that leads to the expression of high levels of Factor V (FV)short. It is a FV splice variant present at low levels in normal plasma that lacks most of the B-domain, including the basic region (BR). In FV, BR in concert with the acidic region (AR) keeps FV inactive. Deletion of BR yields an active molecule in absence of proteolysis and the addition of the BR in trans inhibits FV procoagulant function. The C-terminal of tissue factor pathway inhibitor (TFPIα) is homologous to BR (TFPI-BR) and forms a complex with FVshort. In ET-BD, the FVshort-TFPIα complex increases TFPIα level in blood and contributes to the bleeding diathesis. In this study we examine how TFPIα via its BR regulates FV and FVshort. In a purified system, FVshort and activated FV (FVa) were functionally equivalent when assembled in prothrombinase. FVshort activity was blocked in the presence of TFPI-BR while there was no effect on FVa. Fluorescence anisotropy experiments revealed that TFPI-BR bound with high affinity to FVshort and with no detectable binding to FVa. In a thrombin generation assay (TGA) using FVshort, both TFPI-BR and TFPIα reduced TG. In contrast to FVshort, there is no detectable binding of TFPI-BR to plasma-derived FV. However the BR fragment reduced TG in normal plasma. Simultaneous fluorescence and western blotting experiments revealed that TFPIα-BR only binds partially cleaved forms of FV and binding is eliminated once AR is removed. These data show that TFPIα through its BR downregulates TG by engaging forms of FV with an available AR, such as FVshort and partially proteolyzed FV and AR removal eliminates its ability to impact FV procoagulant function.

back to the top

44. Correlative Live-cell and Super Resolution-STORM Imaging assess the Tubulin-Code

Nitin Mohan1, Ione Verdeny Villanova2, Joseph Borbely2, Angel Sandoval2 and Melike Lakadamyali1,2
1Department of Physiology, Perelman School of Medicine, University of Pennsylvania; 2ICFO-The Institute of Photonic Sciences, Barcelona, Spain

Intracellular transport steered by molcular motors and cytoskeletal network is tightly regulated in space and time for vital cell functioning and survival. Tubulin-code hypothesis proposes that post-translational modifications (PTMs) of tubulin could be one such regulatory mechanisms. Here, we investigate the regulation of motor driven cargo transport by Acetylation and Detyrosination in living cells using correlative single particle tracking (SPT) and super resolution microscopy (STORM). Lysosomes and Autophaghagic vesicles are tracked with high temporal resolution (100-ms), the cells are then fixed and immunostained in situ to visualize the modified and unmodified tubulin using multi-colour STORM imaging with a spatial resolution of 20-nm. Thus by mapping the vesicular trajectories on to the distinct microtubule populations, we elucidate the differential cargo dynamics on post-translationally modified microtubules with high spatiotemporal resolution. Our results show that about 30% of microtubules in epithelial cells are both acetylated and detyrosinated. Stationary lysosomes are preferentially anchored on this microtubule sub-population. Further, motile lysosomes on the detyrosinated-acetylated MTs have shorter run-lengths and higher pause frequencies with respect to lysosome on unmodified MTs. The differences in lysosome motility was due to the detyrosination and not the acetylation state of this MT sub-population. We observed similar results for autophagosomes, suggesting that the impact on motility is likely a general phenomenon affecting diverse sets of organelles and vesicles. Our results suggest that the PTMs are recognized by microtubule motors differentially and modified motility could be cue for organelle interactions such as lysosome-autophagosome fusion.

back to the top

45. A Window of Opportunity for Prevention: Increase in Heart Rate Prior to Challenging Behavior in Preschoolers with Autism

Heather J. Nuske1, Emma Finkel1, Darren Hedley2, Melanie Pellecchia1, Valentina Parma3, John D. Herrington4, David Mandell1, and C. Dissanayake4
1Center for Mental Health Policy and Services Research, Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania, USA; 2Olga Tennison Autism Research Centre, School of Psychology & Public Health, La Trobe University, AUSTRALIA; 3Scuola Internazionale Superioredi Studi Avanzati, ITALY; 4Center for Autism Research, The Children's Hospital of Philadelphia, USA

Many children with autism present with challenging behaviors, which are often associated with stress. The most effective approach to reducing challenging behavior requires appropriate intervention before onset of such behaviors. However, understanding the triggers is difficult due to limited communication abilities in this population. As increase in heart rate is a well-established physiological measure of emotional stress, this may provide important insights on the triggers to challenging behavior. Our aim was to identify physiological precursors to challenging behavior episodes in preschoolers with autism. Whilst wearing a lightweight wireless electrocardiogram (Biopac BioNomadix®), 41 children with autism aged 2-4 years were administered tasks from the Laboratory Temperament Assessment Battery. The tasks were designed to mimic everyday life experiences in which children would need to regulate low-level stress (e.g. waiting for a snack, interacting with a stranger). Coders blind to hypotheses coded challenging behaviors including aggression, self-injury, property destruction, loud noises and noncompliance. Analysis of heart rate data (BPM) indicated a Mean of 21% (SD: 10%, Range: 5 – 34%) increase in BPM from baseline, at a Mean of 58 secs (SD: 22 secs, Range: 21 – 110 secs) prior to the onset of the challenging behavior episode. Data indicate that children with autism show clear physiological reactivity prior to the onset of challenging behaviors, thus identifying an important window of opportunity during which caregivers and teachers may intervene. Wearable biosensors may provide a useful means by which stress of children can be communicated, leading to efforts to teach emotional regulation in situations that matter most, such as the classroom.

back to the top

46. Mechanism of 3’UTR Dependent CELF2 mRNA Expression

Rakesh Chatrikhi, Michael J. Mallory, and Kristen W. Lynch
1Laboratory for Structural, Physiologic and Functional Imaging, Department of Radiology, University of Pennsylvania; 2Maternal and Department of Biochemistry and Biophysics, University of Pennsylvania School of Medicine

More than half of human genes use alternative cleavage and polyadenylation (APA) to generate alternative 3’ untranslated regions (3’UTRs). Altered 3’UTR has various functional consequences in cells. However, why the alternative 3’UTRs have evolved and their biological roles in mRNA stability and translational efficiency, are not entirely understood. CELF2 (CUGBP, Elav-like family member 2) is an RNA binding protein that is implicated in playing a major role in shaping the transcriptome of T-cells. Previous results have shown that 3’UTR of CELF2 undergoes both regulated splicing and APA upon T-cell stimulation. However, the regulation of CELF2 APA, and the relationship between 3’UTR identify and protein expression, has not been determined. Here I propose a multipronged approach to uncover the mechanisms controlling CELF2 3’UTR identity and the consequences of 3’UTR regulation in controlling CELF2 expression. We previously identified five different 3’UTR sequences of CELF2 mRNA in T-cells that result from APA and intron retention. We first show that CELF2 protein binds to the 3’ splice site of the regulated 3’UTR intron. Consistently, we used a quantitative reporter gene assay to examine whether individual 3’UTR sequences can functionally regulate CELF2 expression. We find that 3’UTRs that exclude the intron promote reporter gene expression in both T-cells and HeLa cells, suggesting that regulation of the individual 3’UTR is conserved across tissues. Finally, we are utilizing CRISPR-Cas9 to modify 3’UTR identity in T-cells, to interrogate the role of 3’UTR identity in the endogenous gene. Understanding regulation of CELF2 expression has a broad impact given its role in controlling gene expression in T-cells as well as in the development and function of muscle and neurons.

back to the top

47. Crk Adapter Proteins Coordinate LFA-1 Dependent Cytoskeletal Responses and Mechanosensing in T cells

Nathan H. Roy1,2, Joanna L. MacKay3, Tanner F. Robertson1,2, Daniel A. Hammer3,4, Janis K. Burkhardt1,2
1Department of Pathology and Laboratory Medicine, Children's Hospital of Philadelphia Research Institute and 2Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; 3Department of Bioengineering, University of Pennsylvania and 4Department of Chemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, PA

T cell entry into inflamed tissue involves firm adhesion, cell spreading, and migration, all of which depend on integrin activity and cytoskeletal responses. While it is clear that these events involve “outside-in” signals from engaged integrins, very little is known about the signaling pathways leading to cytoskeletal reorganization. Here, we show that T cells from mice lacking both Crk and CrkL exhibit defects in LFA-1 induced actin polymerization and cell spreading. Crk/CrkL deficient T cells fail to form an actin-rich leading edge, and migrate more slowly and less directionally on ICAM-1 than WT controls. In addition, we find that while WT T cells respond to changes in substrate stiffness with altered spreading and actin dependent phosphorylation of the force-sensitive protein CasL, this response is defective in Crk/CrkL deficient T cells. Analysis of LFA-1 dependent signaling pathways reveals that Crk proteins interact with and promote the tyrosine phosphorylation of Cbl family ubiquitin ligases in T cells spreading on immobilized ICAM-1. This phosphorylation event allows Cbl to serve as a scaffold for the PI3K p85 regulatory subunit, thereby promoting PI3K dependent cytoskeletal remodeling. These findings are consistent with a model in which LFA-1 engagement triggers a cytoskeletal regulatory signaling cascade involving Crk proteins, Cbl, and PI3K, thereby setting into motion actin-driven force production needed for CasL-dependent mechanotransduction.

back to the top

48. Detection of MET Exon 14 skipping in Non-Small Cell Lung Cancer (NSCLC) via RNA-based anchored multiplexed PCR    

Carmela Paolillo1,2, Amanda Oran1,2, Jason N. Rosenbaum1,2, Jennifer JD Morrissette1,2, Robyn Sussman1,2, Kojo Elenitoba-Johnson1,2        
1Center for Personalized Diagnostics, and 2Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania

Recurrent actionable Mesenchymal-Epithelial Transition (MET) tyrosine kinase receptor exon 14 skipping alterations (METex14) occur in up to 7% of non-small cell lung cancers (NSCLCs). METex14 alterations have been described as promising targets for small-molecule kinase inhibitors and monoclonal antibody therapies. Data published in the last decade, highlight the need for comprehensive next-generation sequencing (NGS)-based genomic profiling detection of such variants. However, most clinical NGS targeted oncology panels are DNA based, and limited to coverage of hotspots or exons, thus rendering them inadequate at detecting the diverse composition of variants that lead to METex14. Here we describe the detection of METex14 by Anchored Multiplex PCR. 38 individual NSCLC formalin-fixed paraffin-embedded NSCLC tumors were analyzed on a custom NGS translocation panel, which uses anchored primers specific to known fusion partners and universal primers, allowing for the detection of both known and novel fusion partners and breakpoints. Total nucleic acid was run on the translocation panel and DNA from the same samples was also analyzed on a 153-gene targeted exome solid tumor panel. Across the 38 samples analyzed, we detected three METex14 (~7%). These data were also compared to the matched DNA sequencing data. A splice site mutation in exon 14 of MET was detected in only 1/3 samples (p.D1028Y). Sequencing reads from the DNA oncology panel do cross into the intron boundaries surrounding exons 13, 14 and 15 of MET, but no variants were detected that would predict METex14 in the other two METex14 samples. Together, these data highlight the need for RNA-based assays for the detection of an actionable oncogenic isoform of MET which is often missed by traditional DNA-based assays.

back to the top

49. Comparison of ALK, RET, and ROS1 Gene Fusion Detection Methods in Non-Small Cell Lung Cancer

Amanda R. Oran1,2, Joseph Grubb1,2, David B. Lieberman1,2, Robyn T. Sussman1,2, Jennifer J.D. Morrissette1,2, Jason N. Rosenbaum1,2, Kojo S.J. Elenitoba-Johnson1,2
1Center for Personalized Diagnostics, and 2Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania

Gene fusions of tyrosine kinases (TKs) ALK, RET, and ROS1 are recurrent drivers of non-small cell lung cancer (NSCLC). Patients with these fusions respond to tyrosine kinase inhibitor (TKI) therapies. Genomic rearrangements are traditionally detected by fluorescence in situ hybridization (FISH). Expression of the protein may be detected by immunohistochemistry (IHC). Both FISH and IHC may have significant false positive rates. Next-generation sequencing (NGS) techniques are increasingly being adopted as a high-throughput method for detection of genetic abnormalities. During clinical validation of an NGS anchored multiplex PCR (NGS-AMP) assay, we examined cases which showed discordance between NGS-AMP and FISH/IHC. Fifty-four NSCLC samples analyzed for ALK, RET, or ROS1 fusion by FISH/IHC were sequenced on an NGS-AMP fusion panel and on a targeted DNA-NGS panel. Of the 54 samples analyzed, NGS-AMP results from 46 (85%) cases were concordant with FISH or IHC. Seven (13%) cases were positive for ALK, RET, or ROS1 detection by FISH or IHC, but displayed no evidence of translocation by NGS-AMP. Among the discordant cases, three (6%) samples displayed mutations which are thought to be mutually exclusive with TK fusions. In two (4%) discordant cases, the FFPE tissue samples prepared for NGS and FISH came from different blocks of the same surgical specimen. Our data suggest high concordance (85%) between NGS-AMP and FISH or IHC. Further analysis of discordant cases suggests a potential false positive rate of FISH/IHC of 6% in this cohort, and up to 95% accuracy of the NGS-AMP findings. Overall, these data indicate that NGS-AMP is a highly sensitive and specific method for detection of TK fusions in NSCLC, and suggest potential advantages to the data provided by NGS methods.

back to the top

50. Antibody responses in children with symptomatic and asymptomatic influenza H3N2 infection during the 2014-2015 season  

Sigrid Gouma and Scott E. Hensley
Department of Microbiology, Perelman School of Medicine, University of Pennsylvania

The majority of H3N2 influenza viruses that circulated during the 2014-2015 season were antigenically distinct from the A/Texas/50/2012 H3N2 vaccine strain. Here, we characterized antibody responses against A/Texas/50/2012 H3N2 and a representative 2014 H3N2 strain (A/Colorado/15/2014) in 53 children before and after PCR-confirmed H3N2 infection during the 2014-2015 influenza season. We also examined sera from 97 age-matched ‘controls’ who did not report influenza-like illness during the 2014-2015 season. Antibody responses were measured using the focus reduction neutralization test and expressed as geometric mean ND90 values. Pre-season antibody titers against A/Texas/50/2012 were similar among children in the ‘infected’ and ‘control’ groups (ND90 93 versus ND90 147, p=0.098). Pre-season antibody titers against A/Colorado/15/2014 were slightly lower in ‘infected’ children compared to ‘control’ children (ND90 10 versus ND90 13, p=0.018). As expected, antibody titers against A/Colorado/15/2014 rose in the sera of ‘infected’ individuals collected after the 2014-2015 season (ND90 10 versus ND90 82, p<0.005). Surprisingly, anti-A/Colorado/15/2014 antibody titers in sera collected after the 2014-2015 season also significantly increased in 25.8% of children in our ‘control’ group, all of whom did not report influenza-like illness during the season. This suggests that a substantial number of children in our ‘control’ group had an asymptomatic or very mild influenza infection. We are actively characterizing antibody responses in sera isolated from these children to identify potential antibodies that might confer protection against symptomatic influenza infection.

back to the top

51. Characterization of the patterns of healthy cortical myelination

Nicolas Honnorat and Christos Davatzikos
Deartment of Radiology

A combination of T1 and T2 weighted magnetic resonance imaging (MRI) can be used to estimate the amount of myelin in the brain. This approach allowed, for instance, to associate schizophrenia with pronounced myelin abnormalities. However, a better modeling of myelin variations in a reference healthy population would help quantifying and understanding these abnormalities. In this work, we exploit the data available for the hundred unrelated subjects of the Human Connectome Project to investigate healthy cortical myelination. For each subject, we partition the cortex into 360 regions, where we compute average myelination, cortical thickness (CT), and local cortical area (CA). For each measurement, the patterns of variations across the population were revealed by computing the correlations, across the population, between the regional measurements. We compared the matrices obtained by computing the Spearman correlation between their extra diagonal entries. Our results indicate that strong and significant patterns of myelin variations exist in the healthy HCP population: 19.83% of absolute correlations are larger than 0.4 (p-value<3.73 10^-5). The variations are symmetric between the two hemispheres. Moreover, the patterns observed for myelin, CT, and CA are strongly correlated (Spearman correlations: 0.46 and 0.41, p-values<10^-15). Spearman correlation between the myelin patterns and the average of the variations observed for CT and CA is even stronger: 0.56 (p-value<10^-20). These results suggest that a pattern overlapping CT and CA variability underlies healthy cortical myelin variations. We hope that removing this healthy variability will help refine myelin abnormalities reported in the literature, such as the abnormalities detected for schizophrenic patients.

back to the top

52. Potential of TEM-1 Targeted Probe in Intraoperative Image-Guided Surgery and Metastases Detection   

Yujue Wang, Jie Min, Qingxin Liu, Xiaohui Peng, and Chunsheng Li     
Department of Ovarian Cancer Research Center, University of Pennsylvania

Background: Incomplete tumor excision may result in a higher risk of cancer recurrence and a lower subsequent patient survival. The Image-guided surgery (IGS) has been clinically available in detecting tumor margins during surgery. For real-time intraoperative imaging, this technique relies on the specific tumor-targeted probe. We previously synthesized TEM-1 specific human antibody fragment scFv78 and its derivatives by genetic engineering. Methods: The specific human TEM-1 antibody fragment 78F2 and 78Fc as control were conjugated to the Near-infrared Fluorescence fluorophore Vivotag750. Expression of TEM-1 in human sarcoma tissue was confirmed by immunohistochemisty and human sarcoma cell lines by flow cytometry. The cell-surface binding activity of the probe in A673-Luc was measured by a live cell-based ELISA assay. The fluorescence NIR signal guided resection during cancer surgery with the probe was explored by using bioluminescence imaging system. An invasive intramuscular mouse model was evaluated established to the performance of the probe in detecting mini-metastasis sites and metastasis sentinel lymph node. Results: This study showed over 95% positive in all patients and at high level among ~76% of patients. Its expression is mostly by malignant cells, and perivascular and stromal cells. In vitro and in vivo experiments were proved that the 750-78F2 had specific binding of TEM-1 in sarcoma tissue and cell line. The 750-78F2 probe was stable in serum and showed preserved binding capacity. Conclusion: TEM-1 is overexpressed in the sarcoma tissue array and a few cell lines. Therefore, the study showed a potential of TEM-1 specific targeted NIR-fluorescent probe 750-78F2 in real-time in vivo imaging.

back to the top

53. Entropy in protein-ligand interactions

José A. Caro, Kyle W. Harpole, Vignesh Kasinath, Jackwee Lim, Jeffrey Granja, Kathleen G. Valentine, Kim A. Sharp, and A. Joshua Wand
Department of Biochemistry & Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA

Molecular recognition by proteins is fundamental to biology. Specific interactions at the interface (ΔHbinding) and the release of solvating water (TΔSsolvation) are often assumed to govern binding energetics. The role of conformational entropy (TΔSconf), on the other hand, has remained poorly understood and difficult to measure experimentally. Recent developments in nuclear magnetic resonance spectroscopy (NMR) have suggested an underlying relationship between TΔSconf and dynamics measurements of fast (ps-ns) motions in proteins. To test this idea in a quantitative and general way, we curated a dataset of 28 different protein-ligand complexes, spanning mM-nM affinities and a variety of ligands (NA, peptides, sugars, etc.). For these 28 complexes, the NMR relaxation data were either in the literature or measured by us. The catalog of entropic terms was completed with calorimetric data to obtain TΔStotal, and with structural data to calculate TΔSsolv from changes in accessible surface area. We use the largest dataset available to date to empirically calibrate a correlation between side chain motions and the underlying entropy. The resulting “entropy meter” indicates that TΔSconf can contribute favorably or unfavorably to binding and can often drive binding reactions. The correlation also recapitulates the rotational and translational entropy component, TΔSr-t, expected from theory, and allows for refinement of the contributions from TΔSsolv. Application of the “entropy meter” to barnase-DNA and its hydrated interface reveals a surprising role for specific hydration that escapes the generic surface area calculation used for TΔSsolv. Supported by grants from the NIH, NIH F32, The Mathers Foundation and NSF.

back to the top

54. GFP-labeled tau transgenic mice develop Alzheimer's disease-like pathology following CNS injections of tau fibrils

Garrett S. Gibbons, Rachel A. Banks, Bumjin Kim, Hong Xu, Lakshmi Changolkar, Susan N. Leight, Dawn M. Riddle, Chi Li, Ronald J. Gathagan, Hannah J. Brown, Bin Zhang, John Q. Trojanowski, and Virginia M.-Y. Lee
Center for Neurodegenerative Disease Research, Perelman School of Medicine, University of Pennsylvania       

Alzheimer's disease (AD) is characterized by neuropathological hallmarks comprised of intraneuronal tangles of tau protein and extracellular plaques consisting of amyloid-beta peptide that spread throughout the brain in a predictable pattern during disease progression. Mounting evidence supports a model of cell-to-cell transmission of pathological "seeds" of tau protein that recruit normally soluble tau into self-templated aggregates. Human AD brain-derived tau paired helical filaments (PHFs) injected into the CNS of wiltype and mutant tau transgenic mice seed the formation of AD-like tau inclusions that spread from the site of injection to neuroanatomically connected brain regions. We hypothesize that a GFP-labeled tau transgenic mouse model will recapitulate the seeded aggregation of tau following CNS injection of tau fibrils. Here we report the characterization of a novel mouse model expressing mutant human tau tagged with GFP (T40PL-GFP). We injected either tau PHFs or recombinant tau preformed fibrils (PFFs) into the hippocampus of T40PL-GFP mice and evaluated the seeded recruitment of tau into pathological inclusions by immunohistochemistry. T40PL-GFP protein and endogenous mouse tau were both recruited into neuronal inclusions by either tau fibril and spread throughout the brain in manner consistent with neuroanatomical connections opposed to transgene expression levels. This study validates the use of GFP-labeled tau as a model for investigating mechanisms underlying the seeded transmission of tau pathology as well as tau focused drug discovery to identify disease modifying therapies for AD.

back to the top

55. The role of the TIPE proteins in intestinal injury and inflammation

Jason R Goldsmith, Terry Cathopoulis, Amanda Boggs, Nina Spikofsky, Honghong Sun, Youhai Chen
Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania  

TNFAIP8 (tumor necrosis factor-α-induced protein 8-like-0, also known as TIPE0) is one of 4 members of the highly homologous TIPE family of proteins, which regulate cell survival, inflammation, and immune migration, and are involved in PI3K/Akt signaling. TIPE0 loss also appears to affect cell survival through altered TNF signaling. In contrast to our previous findings with acute colitis, TIPE0 knockout results in markedly decreased intestinal ischemia/reperfusion (I/R) injury. This was associated with increased intestinal epithelial cell (IEC) pAKT levels and increased IEC and intestinal leukocyte pSTAT3 levels. IEC pAKT is known to be protective against IEC injury, as is pSTAT3; the latter of which is driven by innate lymphoid cell IL-22 production. Ex vivo, TIPE0-/- enteroids were found to be resistant to TNF-mediated cell death, but displayed altered morphology while maintaining normal differentiation. In contrast, knockout of immune-specific TIPE2 (TNFAIP8-like-2), which is protective against DSS colitis, showed only modest effects on intestinal I/R injury. Protection from injury was only modest, with significantly less enhancement of IEC pAKT signaling but equivalent enhancement of pSTAT3 signaling. TIPE2-/- organoids exhibited aberrant growth and cohesion, preventing analysis of resistance to TNF-mediated cell death. Together, this data suggests that TIPE proteins regulate both immune and IEC responses to intestinal injury and inflammation, given the differences between global TIPE0 and immune-specific TIPE2. Further work must be done to isolate the important cell compartments and exact mechanisms of action involved in TIPE-mediated protection from I/R injury.

back to the top

56. Pediatricians’ Vaccine Attitudes and Practices before and after the Disneyland Measles Outbreak

Salini Mohanty1, Alison Buttenheim1, Kristen A. Feemster2,3,4, Charlotte A. Moser2, Robert I. Field5,6, Michael Yudell5, Renee M. Turchi5,7, and Amy Carroll-Scott5          
1Department of Family and Community Health, School of Nursing, University of Pennsylvania; 2Vaccine Education Center, Children's Hospital of Philadelphia; 3Center for Pediatric Clinical Effectiveness and PolicyLab, Children's Hospital of Philadelphia; 4Division of Infectious Diseases, The Children's Hospital of Philadelphia; 5Dornsife School of Public Health, Drexel University; 6Thomas R. Kline School of Law, Drexel University; 7Department of Pediatrics, St. Christopher’s Hospital for Children      

Background: From December 2014 through April 2015, the United States experienced a multi-state measles outbreak originating at a Disneyland theme park in California that received national media coverage. The objective of this study was to describe how pediatricians were impacted by and responded to the outbreak. Methods: We conducted three repeated cross-sectional, online surveys in 2014 (before the outbreak), 2015, and 2016 (after the outbreak) among members of three state chapters of the American Academy of Pediatrics: Delaware, Hawaii, and Pennsylvania. We assessed pediatricians’ level of willingness and length of time comfortable delaying the MMR (measles-mumps-rubella) vaccine before and after the outbreak. We also measured reported changes in the frequency of alternative immunization schedule requests and creation of office immunization policies due to the measles outbreak. Results: The sample included 304 pediatricians in 2014, 270 in 2015, and 221 in 2016. Multivariate logistic regression models showed no significant changes in willingness or comfort delaying the MMR vaccine before and after the outbreak. Shortly after the measles outbreak, 38% of pediatricians reported fewer requests for alternative immunization schedules and 20% created stricter office immunization policies. A subsample of pediatricians reported changes to clinical practices as a result of the outbreak including administering the MMR vaccine earlier in the recommended timeframe and taking extra precautions in waiting rooms. Conclusion(s): Our results suggest that the Disneyland measles outbreak did not lead to significant changes in attitudes or practices among pediatricians, but did modestly affect alternative immunization requests and office immunization policies.

back to the top

57. Neuroimaging of diffuse axonal and vascular injury in chronic traumatic brain injury         

Margalit Haber1, Franck Amyot2,3, Kelly Flesher2, Kimbra Kenney2,3,4, Carol Moore3, Erika Silverman1, Yi-Yu Chou3, Dzung L. Pham3, Eric Wasserman4, Ramon Diaz-Arrastia1 
1University of Pennsylvania, Department of Neurology, Philadelphia, PA; 2Uniformed Services University of the Health Sciences, Department of Neurology, Rockville, MD; 3Center for Neuroscience and Regenerative Medicine; 4National Institute of Neurological Disorders and Stroke, NIH, Bethesda, MD

Traumatic Brain Injury (TBI) results in diffuse axonal injury (DAI) and diffuse vascular injury (DVI). Both DAI and DVI result from inertial shearing forces, and the two terms are often used interchangeably, the spatial relationships between DAI and DVI have not been carefully studied. Multimodal magnetic resonance imaging (MRI) can help to distinguish these injury mencahisms: diffusion tensor imaging (DTI) provides information about axonal integrity, while arterial spin labeling (ASL) and functional Blood Oxygen Level Dependent imaging (BOLD) with hypercapnia challenge, reflect cerebral blood flow (CBF) and cerebrovascular reactivity (CVR) respectively. Chronic TBI participants (n=27) and age- and education-matched healthy controls (n=15) underwent multimodality MRI. The Freesurfer image analysis suite (MGH, Harvard, MA) was used to segment each MP-RAGE image into regions of interest (ROIs). Mean values of mean diffusivity (MD), fractional anisotropy (FA), CBF, and CVR were extracted for each ROI. Additionally, maps were normalized into a common space (MNI Atlas) and z-score maps were generated based on a pool of healthy controls. Normality of an ROI/voxel was determined based on z-score (abnormal MD: z-score>2.5; abnormal FA, CBF, and CVR: z-score<-2.5). Abnormal ROIs in one modality were not predictive of abnormalities in another modality. Approximately 9-10% of abnormal voxels for CVR and CBF also show an abnormal voxel value for MD, while only 1% of abnormal CVR and CBF voxels show a concomitant abnormal FA value. These data indicate that chronic TBI patients display two distinct endophenotypes: microstructural tissue/axonal injury and vascular injury that are spatially independent.

back to the top

58. Longitudinal assessment of cerebrovascular dysfunction after Traumatic Brain Injury using fNIRS  

Michael Sangobowale1, Franck Amyot2,3, Hasan Ayaz4, Yerram Pratusha Reddy4, Erika Silverman1, Tawny Meredith-Duliba1, Megan Moyer1, Danielle Sandsmark1, Ramon Diaz-Arrastia1
1University of Pennsylvania, Department of Neurology; 2Uniformed Services University of the Health Sciences, Department of Neurology, Rockville, MD; 3Center for Neuroscience and Regenerative Medicine; 4Drexel University, School of Biomedical Engineering

TTraumatic cerebrovascular injury (TCVI) is common after TBI is responsible for a significant portion of TBI-related disability. Development of therapies targeting TCVI will require reliable and inexpensive biomarkers to measure vascular dysfunction and document target engagement. Cerebrovascular reactivity (CVR) is impaired following TBI and methods which reliably and non-invasively measure CVR are available, making CVR an attractive candidate predictive and pharmacodynamic biomarker for TCVI-directed therapies. This study was designed to assess CVR longitudinally after TBI in humans using functional Near Infrared Spectroscopy (fNIRS) from the acute to the subacute and chronic stages. We also studied the effect of treatment with a phosphodiesterase 5 inhibitor, sildenafil citrate, in order to assess the utility of fNIRS as a pharmacodynamic biomarker in future clinical trials. 13 participants with complicated mild TBI were studied in the acute (within 72 after injury), subacute (14 days), and chronic (6 months) stages, in addition to 8 age- matched healthy controls (HC), who were studied once. CVR was assessed by measuring the changes in oxygenated hemoglobin (ΔHbO) and deoxygenated hemoglobin (ΔHbR) concentration produced by mild hypercapnia (5% CO2). The change in CVR before and one hour after the administration of single dose of sildenafil citrate (60 mg orally) was also assessed. Sildenafil administration did not result in an increase in CVR in HC (p=1.66) whereas TBI patients show a clear potentiation of the CVR response at 72hr (p=.058), 2 weeks (p=0.04) and 3 months (p=0.01) post injury. These findings support the hypothesis that vascular injury represents a distinct endophenotype following TBI and PDE5 inhibition as a potential therapy for TCVI.

back to the top

59. Contrasting properties of active and inactive hippocampal dentate granule cells

S.A. Park1, F.-C. Hsu1, H. Takano1, I. Petrof1, D.A. Coulter1,2,3
1Neurology, Children's Hospital of Philadelphia; 2Pediatrics, Children's Hospital of Philadelphia; 3Neuroscience, University of Pennsylvania

The hippocampus is a critical mediator of spatial learning and navigation. The dentate gyrus contributes through its sparse activation properties to aid in pattern separation. Using 2-photon imaging in awake mice navigating a virtual environment, we found less dentate granule cells (DGCs) activate at lower frequencies than neighboring CA1 pyramidal neurons. We explored the intrinsic and local circuit mechanisms mediating this sparse DGC activation. Using fosTRAP (targeted recombination in active populations) transgenic mice crossed to a tdTomato reporter mouse to label active DGCs in vivo following exposure of mice to an enriched environment and 4-hydroxytamoxifen injection, we recorded from hippocampal slices prepared from these same mice. As expected, fosTRAP labeled DGCs were sparse; on average, 7.2 cells were active per dentate blade in a given slice. To understand the intrinsic and synaptic properties that differentiate active (tdTomato+) from inactive (unlabeled) DGCs, we used whole-cell patch-clamp to record intrinsic properties as well as synaptic potentials elicited by perforant path stimulation. Counterintuitively, active DGCs had a larger IPSC amplitude, shorter IPSC time-to-peak, and a longer EPSC time-to-peak compared to neighboring inactive DGCs. Further tests revealed active DGCs had a lower input resistance and a higher action potential threshold, both also inconsistent with elevated excitability. These results suggest that DGCs activate in vivo primarily due to a difference in the afferent input, and not to local circuit processing. Continued investigation of the synaptic connections and microcircuit properties may further clarify how active and inactive DGCs process their synaptic inputs to generate their characteristic behavior.

back to the top

60. Lef1-dependent hypothalamic neurogenesis inhibits anxiety

Yuanyuan Xie1,8, Dan Kaufmann2, Matthew J. Moulton3, Samin Panahi1, John A. Gaynes1, Harrison N. Watters1, Dingxi Zhou1,7, Hai-Hui Xue4, Camille M. Fung5, Edward M. Levine6,8, Anthea Letsou3, KC Brennan2, and Richard I. Dorsky1
1Department of Neurobiology and Anatomy, University of Utah, Salt Lake City, UT;2Department of Neurology, University of Utah, Salt Lake City, UT; 3Department of Human Genetics, University of Utah, Salt Lake City, UT; 4Department of Microbiology, Carver College of Medicine, University of Iowa, Iowa City, IA; 5Division of Neonatology, Department of Pediatrics, University of Utah School of Medicine, Salt Lake City, UT; 6Department of Ophthalmology and Visual Sciences, John A. Moran Eye Center, University of Utah, Salt Lake City, Utah; 7School of Life Sciences, Peking University, Beijing, China; 8Current address: Department of Neuroscience, Mahoney Institute for Neurosciences, Perelman School for Medicine, University of Pennsylvania (Y.X.); Department of Ophthalmology and Visual Sciences, Vanderbilt University School of Medicine, Nashville, TN (E.M.L).

While innate behaviors are conserved throughout the animal kingdom, it is unknown whether common signaling pathways regulate the development of neuronal populations mediating these behaviors in diverse organisms. Here we demonstrate that the Wnt/ß-catenin effector Lef1 is required for the differentiation of anxiolytic hypothalamic neurons in zebrafish and mice, although the identity of Lef1-dependent genes and neurons differ between these two species. We further show that zebrafish and Drosophila have common Lef1-dependent gene expression in their respective neuroendocrine organs, consistent with a conserved pathway that has diverged in the mouse. Finally, orthologs of Lef1-dependent genes from both zebrafish and mouse show highly correlated hypothalamic expression in marmosets and humans, suggesting co-regulation of two parallel anxiolytic pathways in primates. These findings demonstrate that during evolution, a transcription factor can act through multiple mechanisms to generate a common behavioral output, and that Lef1 regulates circuit development that is fundamentally important for mediating anxiety in a wide variety of animal species.

back to the top

61. Mutant Huntingtin Disrupts CELF1-MBNL1 Regulation in Patient Brain and Model Systems   

Shyam Ramachandran1, Gregory G Cajka1, Stephanie L Coffin1, Matthew J Sowada1, Sergey Akimov2, Megan S Keiser1, Christopher A Ross2, Beverly L Davidson1,3
1The Raymond G. Perelman Center for Cellular and Molecular Therapeutics, The Children’s Hospital of Philadelphia; 2Department of Psychiatry and Neuroscience, Johns Hopkins University School of Medicine; 3The Department of Pathology & Laboratory Medicine, University of Pennsylvania.

Huntington’s disease (HD) is caused by a polyglutamine-repeat expansion in exon 1 of huntingtin (HTT). The mutation confers gain of function phenotypes on the HTT protein, as well as on transcripts arising from the HTT locus. Additionally, recent analyses show differential gene expression and alternative splicing in HD brain, with RNA-binding proteins as candidate mediators of that misregulation. We show the presence of mutant huntingtin (mHTT) alters the transcript and protein levels of the RNA binding proteins Muscleblind-like Splicing regulator 1 (MBNL1) and CUGBP/Elav-Like Family Member 1 (CELF1), with CELF1 more stable and abundant in HD tissues and cells, and MBNL1 levels reduced. Moreover, the MBNL1 that is present has partitioned to a less soluble species that co-localizes with mHTT transcripts. CELF1s stability in HD cells results from a combination of a MBNL1-dependent switch in poly(A) site preference on the transcript, and a PKR- and PKC-dependent hyperphosphorylation of the protein. Knockdown of mHTT rescues the misregulation of the CELF1-MBNL regulatory loop, providing further support for a gain-of-function role for mHTT transcripts. With respect to HD pathogenesis, we rescue medium spiny neuron survival with CELF1 inhibition or MBNL1 overexpression in HD-derived neurons, while restoration of normal MBNL1 levels in striatum rescues the motor deficit in mouse models of HD.

back to the top

62. Hepatic PPP1R3B deficiency induces reduced hepatic insulin signaling resulting in insulin resistance and fatty liver   

Swapnil V. Shewale1,2, Minal B. Mehta1,2, John S. Millar1, Nicholas J. Hand2, and Daniel J. Rader1,2,3      
1Department of Cancer Biology, and 2Department of Genetics, Perelman School of Medicine, University of Pennsylvania; 3NIH National Institute on Aging, Bethesda, MD; 4Department of Neurobiology and Anatomy, Drexel University College of Medicine

Polymorphisms near PPP1R3B locus are associated with multiple cardio metabolic traits including plasma total, HDL and LDL-cholesterol, glucose, insulin, alkaline phosphatase and NAFLD. We developed Liver-specific Ppp1r3b knockout (Ppp1r3bΔhep) mice to investigate whether hepatic PPP1R3B may be the causative gene underlying the GWAS associations. Ppp1r3bΔhep mice display near absence of liver glycogen due to increased accumulation of its inactive substrate: p-GS in the liver and result in increased plasma non-esterified fatty acids (NEFAs) and ketones upon acute fasting mimicking human glycogen storage disease type 0 (GSD0). Upon high sucrose diet feeding, Ppp1r3bΔhep mice developed excess adiposity, systemic glucose intolerance, hyper-insulinemia and insulin resistance and were characterized by hepatic TG and TC accumulation, increased glycolytic and lipogenic gene expression and increased VLDL-TG secretion despite compensatory activation of liver PPARα signaling pathway after 4hr. fasting. Reciprocal phenotypes were observed in the AAV-Ppp1r3b overexpressing (OE) mice. Additionally, Ppp1r3bΔhep livers were characterized by reduced p-GSK3b as well as p-AKT/total irrespective of ad-libitum fed state, acute 4hr. fasting or 12 hr. fasting followed by 3 hr. re-feeding indicative of reduced insulin signaling and inefficient adoptive responsiveness to fasting and re-feeding. Additionally, Ppp1r3bΔhep mice displayed concurrent increase in plasma insulin and hepatic de-novo lipogenesis during the 6 hr. re-feeding phase. These phenotypes are likely due to reduced p-AMPKa/b/total and resultant excess mTORC1 activation in Ppp1r3bΔhep livers. In summary, we conclude that hepatic PPP1R3B acts as regulator of whole body carbohydrate and lipid compartmentalization through fine tuning.

back to the top

63. Foxl1-Expressing Telocytes Constitute the Intestinal Stem Cell Niche      

Michal Shoshkes-Carmel1, Kirk J. Wangensteen1, Yue J. Wang1, Beata Toth2, Shalev Itzkovitz2, and Klaus H. Kaestner1     
1Department of Genetics and Center for Molecular Studies in Digestive and Liver Diseases, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; 2Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel 

Stem cell niches provide essential signals and growth factors to sustain proliferation and self-renewal of stem cells in continuously self-renewing organs such as the skin and intestine. We identify large but rare mesenchymal cells in the gut expressing the winged-helix transcription factor forkhead box l1 (Foxl1) using antibody staining and genetic lineage labeling. These cells are telocytes which have a unique cell structure with long processes that extends hundreds of μm. Foxl1+ telocytes cover the entire gut epithelium from crypt base into the villus tips and are expressing key signaling pathway molecules such as members of the Wnt, BMP, Shh, FGF and TGF gene families in a localized fashion. Ablation of Foxl1+ telocytes or key signaling molecules within these cells causes loss of proliferating cells in the crypt compartment, rapid crypt collapse, and death of the mutant mice within a few days. Thus, Foxl1+ telocytes constitute the intestinal stem cell niche which is absolutely required for stem cell function.

back to the top

64. Evaluating LRRK2 as a therapeutic target in Parkinson’s disease

Michael X. Henderson, Raizel M. Sandler, Bin Zhang, Hannah Brown, and Virginia M.-Y. Lee
Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine

Parkinson’s disease (PD) is the most common neurodegenerative movement disorder. Patients with this disease experience rigidity, resting tremors, and slowness of movement; and 80% of patients will develop dementia. The clinical diagnosis of PD is confirmed post-mortem by the presence of intracytoplasmic inclusions termed Lewy bodies, which consist primarily of the synaptic protein α-synuclein. While α-synuclein is thought to be pathogenic in this largely sporadic disease, mutations in several genes can increase the lifetime risk of developing disease. The most commonly mutated gene in PD is leucine-rich repeat kinase (LRRK2). LRRK2 is a broadly-expressed protein with unclear function. It has kinase, GTPase and scaffolding domains and has been implicated in membrane trafficking and cytoskeletal modeling. The most common PD-linked mutation, G2019S, leads to elevated kinase activity, so LRRK2 inhibitors have been the subjects of intense drug development for PD. However, the lack of a reliable preclinical disease model for LRRK2 dysfunction has been a major challenge for the field. We developed cell and animal models to evaluate whether mutant LRRK2*G2019S expression exacerbates the extent or spread of α-synuclein pathology. We then utilized these models to test the efficacy of several top LRRK2 inhibitors to affect α-synuclein pathology. Our data represent important preclinical observations that will inform the development of LRRK2 inhibitors for use in PD patients with and without mutations in LRRK2.

back to the top

65. Basis for FOXA1 dynamic chromatin scanning        

Jonathan Lerner and Ken Zaret          
University of Pennsylvania, Perelman School of medicine, Department of Cell and Developmental Biology.

During cell fate changes, Pioneer transcription factors bind to genomic sites in silent chromatin and render them accessible for other transcription factors. FOXA are pioneer factors inducing liver fate in endodermal cells during development. Ectopic expression of FOXA, with HNF1A and HNF4A, results in the direct conversion of fibroblasts to a hepatic cell lineage. The pioneer function of FOXA1 is related both to a high affinity for nucleosomes and to a slow chromatin scanning, as measured in FRAP assays, in comparison to non-pioneer factors. Even though the genome-wide interaction of FOXA1 with chromatin has been assayed previously, the precise contribution of the biochemical properties of FoxA1 in its scanning and binding to chromatin sites in live cells has yet to be precisely understood, as well as the influence of cell lineage context (chromatin landscape, co-binding factors). Here, using well-characterized mutants of FOXA1, we show that non-specific DNA binding provides the highest contribution to the pioneer factor’s slow scanning, and that specific-DNA binding and histone interaction provides lesser but significant contributions. FRAP studies also revealed that FOXA1 exhibits unusual chromatin scanning dynamics in an ectopic cell lineage context, that is slowly recovered after 48h. In conclusion, specific and non-specific DNA binding and histone interaction capacities of FOXA1 are essential for its characteristic pioneer chromatin scanning. In an ectopic cell lineage context, FOXA1 exhibits impaired dynamics, possibly contributing to the lack of efficiency of the cell conversion process.

back to the top

66. HFD-Induced enhancer remodeling modulates the reprogramming of hepatic gene transcription        

Dongyin Guan1,2, Patricia C. Borck1,2, Romeo Papazyan1,2, Bin Fangsup1,2, Yuxiang Zhang1,2, Erika Briggs1,2, David Steger1,2, Sean Armour1,2, Jarrett Remsberg1,2, Mitchell A. Lazar1,2,3          
1Institute for Diabetes, Obesity, and Metabolism, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; 2Division of Endocrinology, Diabetes, and Metabolism, Department of Medicine; 3Department of Genetics, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA.

Circadian rhythms are self-sustained oscillations that are internally generated and function to anticipate and adapt to the 24-h period of the solar day. These rhythms comprise an auto-regulated negative transcription feedback loop which integrates internal genetic information and environmental signals to regulate DNA binding activities of transcription factors and control gene oscillating expression. Evidences of night shift workers, individuals with sleep disorders and genetically engineered mouse model of circadian rhythm genes support virtual roles of circadian rhythm in metabolism. However, the complex transcriptional and global epigenomic regulation of environment signals on circadian rhythms remain elusive. Global Run-On sequencing (GRO-seq) measures nascent transcription of both pre-mRNAs and enhancer RNAs (eRNAs) at a genome-wide level, making it a unique tool for unraveling complex gene regulation mechanisms in vivo. Here, we use 12 weeks high fat diet (HFD)-fed mice as model to determine the direct effect of HFD on nascent RNA transcription via GRO-seq to every 3-hour sampled liver. We have observed a genome-wide scale of enhancer remodeling and circadian transcription reprogramming in gene body regions. We are currently dissecting these gain and loss circadian transcription of enhancer RNAs and genes to understand the underlying mechanisms responsible for the re-constitute circadian transcriptional rhythm in response to HFD-induced metabolic disruption.

back to the top

67. Yap1 regulate receptor of hyaluronan-mediated motility (RHAMM) to modulate sarcoma cell proliferation        

Shuai Ye and T.S.K Eisinger-Mathason
Department of Pathology and Lab Medicine, University of Pennsylvania

Soft tissue sarcomas of the muscle (STS) are rare heterogeneous tumors originating from mesenchymal progenitor cells. Treatment options for these patients are limited to surgery, radiotherapy and chemotherapy. Therefore, the discovery of novel targets and mechanisms is critical. Sarcomas are highly desmoplastic and contain large amounts of deposited extracellular matrix (ECM). Hyaluronan (HA) is a major component of ECM and is known to promote tumorigenesis in other contexts. Expression of the receptor for hyaluronan-mediated motility receptor (HMMR/RHAMM) is tightly controlled in normal tissues. However, it is elevated in many tumors included soft-tissue sarcoma. Recently RHAMM expression has been linked to the Hippo pathway. Yap1, a transcription co-activator is a key downstream effector of Hippo pathway. Inactivation of this pathway promotes nuclear localization of Yap1 that drives proliferation. Whereas, the identity of Yap1’s critical downstream transcriptional targets in mesenchymal tumors remains elusive, in the present study, downregulation Yap1 decreases RHAMM protein level in murine and human sarcoma cell lines. ShRNA silencing of RHAMM reduced cell proliferation and ectopic expression of RHAMM enhanced proliferation. Importantly, we have discovered that a combination of JQ1, a BET family inhibitor, and Vorinostat (SAHA), a histone deacetylase (HDAC) inhibitor decreases both Yap1 and RHAMM. Consistent with my in vitro findings, silencing of RHAMM significantly reduced tumor growth in our allograft mouse model. Hence, Yap1 may regulate sarcoma cell growth via control of RHAMM. The present study suggests that RHAMM is a critical downstream effector of Yap1 and that JQ1/SAHA treatment exerts anticancer effects at least partly through the Yap1/RHAMM axis.

back to the top

68. Design and Synthesis of PC-PLC Selective Self Quenching Near-Infrared Fluorescing Probes

Benjamin K. Liebov, E. James Delikatny, and Anatoliy V. Popov          
Department of Radiology

The design and synthesis of PC-PLC targeted fluorescent probes could lead to a new and improved technique for the diagnosis and treatment of breast cancer. PC-PLC is strongly up-regulated in epithelial ovarian and breast carcinoma cell lines compared with non-tumor counterparts. In contrast to normal cells, PC-PLC accumulates primarily on the outer leaflet of the cancer cell plasma membrane where the enzyme reveals abnormally high activity. Amongst a series of breast cancer cell lines the highest level of PC-PLC activity was found in the triple negative highly-metastatic breast cancer cell line MDA-MB-231. These data suggest that PC-PLC may be an effective target for breast cancer therapy, particularly for tumors with high metastatic potential. In order to design a PC-PLC selective probe, a phospholipid is synthesized that contains a bulky substituent hindering the sn2-position to prevent attack by PLA2, a ubiquitous phospholipase. For this purpose, the phospholipid phosphatidylethanolamine (PtdEtn) is chosen and the bulky moiety at its sn-2 position can be a porphyrin e.g., pyropheophorbide a (Pyro) or bacteriochlorin e6 (Bchl), which can play three roles: a bulky substituent, NIR dye, and a photosensitizing agent. For PC-PLC-activatable probes, a quencher BHQ-3 or QSY21 is attached to the nitrogen containing head group of PtdEtn. Non-quenched probes based on phosphatidylcholine (PtdCho) with a porphyrin moiety (Pyro or Bchl) attached at the sn-2 position are also being studied. The probes have been tested in vitro for chemical stability serum stability, enzymatic cleavage and optical imaging of breast cancer cells.