The goal of the Kaeberlein lab is to understand the underlying molecular mechanisms of aging and how they can be perturbed in order to beneficially alter the aging process. To do so, we use the unicellular model S. cerevisiae or brewers yeast. In order to expedite the process of studying aging in large numbers of cells, the Kaeberlein lab has been using microfluidic devices that immobilize yeast cells throughout their lives so that visual data can be collected without manual interference of the cells. The question being investigated here is whether a neural network can be trained to automate the scoring tasks required for processing the image data generated by these microfluidic devices. Using python and the Yolov3 neural network architecture, I have created an AI-based screening tool to quickly analyze data generated with these microfluidic devices, such that the pipeline as a whole produces lifespan data for about 120 cells at a time, using stacks of image files as raw data. This work is significant as it provides insights into the issues of training a neural network on similar objects, and more importantly, insights into how these issues can be mitigated. In addition to increasing the speed with which data can be scored, this system will also increase the capacity for scoring experiments with large numbers of experimental groups. The system itself can also be used in order to expedite biology research.

Approximately 10-20% of adolescents globally struggle with mental health conditions. Due to multiple physical, social and emotional changes, adolescents are particularly vulnerable to mental health illnesses. Untreated or inadequately treated mental health conditions can be exacerbated by stigma, discrimination, and isolation. The inability to effectively address mental health problems can manifest both short-term and long-term consequences into adulthood, impairing both quality of life and having negative impacts on health outcomes and mortality. Mental Health for Every Adolescent (MHEA) is a programme we, as students at the University of Washington, created as an early intervention to destigmatize and educate students about mental health. Through this project, we sought to understand the impact and need for community based mental health intervention for young adults. We partnered with faculty to create curricula that addressed stress, anxiety, depression and other mental health related topics. We then recruited volunteers through social media advertising and virtually trained them to facilitate workshops. Using feedback we obtained from local stakeholders at our workshop sites, we modified our curricula to ensure they were culturally appropriate and sensitive. Over the course of 3 years, we completed 74 workshops across 3 countries - India, the United Arab Emirates and the US - and we reached more than 3100 middle school and high school students. We found that the needs of students varied across communities and therefore, the workshops had to be tailored appropriately. In April 2020, the MHEA team leveraged local relationships and published a COVID-specific guide aimed at adolescents which was launched at 6 school districts in the United States. The MHEA project describes a successful implementation of a community-based mental health intervention for children and young adults. Programmatic materials developed can be disseminated and incorporated into middle and high-school curriculum. Future work will evaluate the impact on participants in the program.

Currently, in the United States and worldwide, the population is aging, living longer, and the increase in people living with dementia (PLWD) will continue to rise. Many PLWD receive care from their primary care clinicians, who face unique challenges caring for this population given decline in cognitive function, overall health and unpredictable behavior requiring more attention. Gaps in care could be addressed through the development of a cadre of dementia nurse specialists, as much of dementia care is within the scope of nursing practice. However, interest in the aging population is one of the least desirable disciplines to work for new nurses as well as currently practicing nurses. The purpose of this study was to evaluate nursing students’ and practicing nurses’ perceived knowledge and attitudes about providing care for people who are living with dementia and their family caregivers. Based on existing literature, we identified the factors contributing to the attitudes and conducted semi-structured focus group interviews with nursing students and current practicing nurses. Quantitative and qualitative data was analyzed to identify themes and factors that would encourage nursing students and practicing nurses to receive specialty training in dementia. This research will aid in development of a multimodal dementia education training program for nurses in ambulatory care.

Alzheimer’s Disease (AD) is a neurodegenerative disease that is the most common cause of dementia. One hallmark of AD pathology is hyperphosphorylation of Tau protein. Tau is a neuronal-specific protein that stabilizes microtubules. Hyperphosphorylation of Tau leads to loss of its normal function and promotes aggregation into neurotoxic fibrillary tangles. While Tau aggregation is well-documented, the exact role of Tau loss-of-function in AD pathogenesis remains uncharacterized. The goal of our project is to determine the mechanism by which Tau loss-of-function contributes to AD pathogenesis. We hypothesize that Tau loss-of-function contributes to AD pathogenesis by activating the cellular stress response in neurons, characterized by DNA damage, stress granule formation, and the upregulation of heat shock proteins, chaperones that stabilize and refold proteins damaged by cellular stress. To test this hypothesis, we cultured neural progenitor cells (NPCs) and neurons derived from human induced pluripotent stem cells (hiPSCs). We generated three cell lines: one in which the gene encoding Tau was knocked out (Tau KO), another in which Tau expression was knocked down (shTau), and a control line. To determine whether genes implicated in the cellular stress response are upregulated in Tau KO neurons, we used RNAseq and RT-PCR. Then, we used immunocytochemistry to detect protein markers of cellular stress in NPCs and neurons from all three lines. Preliminary results indicate that several genes encoding proteins involved in the cellular stress response are upregulated in Tau KO neurons, including the small heat shock protein HSPB8 and BAG3, a gene that is upregulated in the aging rodent brain and regulates the autophagic response. By immunostaining, we show that dsRNA aggregation, which may indicate stress granule formation, is more prevalent in Tau KO cell lines than control. By elucidating the role of Tau loss-of-function in AD pathogenesis, this research could inform therapeutic targets for AD.

The Wnt pathway is a critical regulator of osteoblast function. Activating and inactivating mutations in the Wnt pathway lead to high and low bone mass phenotypes in humans and animals, respectively. To evaluate the role of Wnt signaling in facial skeletal development, a transgenic strain of mice (daβcatOT) was generated in which the Wnt intracellular mediator, β-catenin, was constituitively expressed in osteocytes and mature osteoblasts. The craniomaxillofacial (CMF) tissues of daβcatOT mice and their littermate controls were collected at two time points: one where mice had reached sexual maturity e.g., 2 months old and the other where mice had reached adulthood e.g., 4 months old. Thirty-eight anatomically distinct landmarks were identified on three-dimensional volumetric renderings of control and daβcatOT (each N=8) head skeletons. Landmarks typically corresponded to intersections of skeletal elements and were predominantly located at CMF growth sites. Through the open-source visualization software Drishti, we performed 28 separate measurements. Morphometric analyses of the dataset were conducted using Microsoft Excel. Inter-rater reliability was qualitatively assessed to ensure reproducibility. Two tailed T-tests were performed, which revealed statistically significant (p<0.01) differences in the lengths of the maxillary, premaxillary, and mandibular bones (p<0.01) which comprise the jaw skeleton. Equivalent analyses revealed limited asymmetries. The resulting facial phenotype of daβcatOT mice was a largely symmetrical but severely flattened midface, accompanied by a widened, retruded mandible. Comparative analyses revealed that the daβcatOT facial anomalies phenocopy an exceedingly rare human malformation known as craniodiaphyseal dysplasia. Ongoing analyses of daβcatOT tissues are focused on the molecular/cellular basis for the facial malformations caused by constitutive activation of Wnt/β-catenin signaling and in doing so, gain more insights into how the CMF skeleton normally develops.
 

Microglia, the resident immune cells of the brain, become activated and mediate neuroinflammatory responses in response to traumatic brain injury (TBI). This neuroinflammation can be detrimental to the health of the brain; thus, inhibition of this natural response can benefit TBI patients. Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) are a technological advancement that allow manipulation of specific cellular signaling pathways. The hM4D DREADD gene was inserted in a viral construct downstream of the CD68 promoter, which is markedly upregulated in activated microglia following TBI. This construct is activated with clozapine n-oxide (CNO) to downregulate secondary messengers in activated microglia and thus attenuates the inflammatory response by reducing microglial activation and proliferation after injury. After viral transfection to deliver the construct to cells, rats underwent a controlled cortical impact (CCI) - an experimental model of TBI - and were treated with a CNO injection at varied intervals after injury. After the rat brains were dissected and sectioned, we used immunohistochemistry techniques to label microglia with an anti-Iba1 antibody, proliferating cells with an anti-BrdU antibody, and cell nuclei with DAPI. This allowed for the visualization of microglia using fluorescence microscopy and microglia were quantified using stereological principles. The aim of this research project is to use pharmacological DREADD receptor-mediated inhibition of microglia at different intervals post-injury to quantify the proliferation of microglia following CCI. We hypothesize that increased duration between CCI and CNO injection leads to more pronounced microglial activation represented by increased numbers and morphological changes. We also hypothesize that microglial activation can be observed as a gradient with the greatest proliferation closest to the CCI epicenter.

Manually analyzing neural histopathology is a tedious process, and while there are reference sources available such as the Allen Brain Atlas, one of the primary challenges of using image analysis software to compare neural pathology between animals is the alignment of the rostral-caudal axis. The current study aims to investigate the use of Visiopharm, an image analysis software, and supervised machine learning to automate the assignment of brain sections along the rostral-caudal axis.This research is a part of a larger study conducted to analyze potential neuropathological changes in male mice that have undergone blast-induced mild traumatic brain injuries. Glial fibrillary acidic protein (GFAP), Tyrosine hydroxylase (TH), Tryptophan hydroxylase (TPH), Kappa Opioid Receptor (KOR) and Ionized Calcium-Binding Adapter (IBA) were used to identify histopathology. First, we manually labeled sections in Visiopharm and trained a deep learning algorithm to outline the sections and compute section parameters (e.g., circumference, diameter, convexity). In parallel, we manually assigned each labeled section according to the Allen Brain Atlas (ABA) rostral-caudal axis. Finally, we used python to train a supervised machine learning regressor to predict the assigned ABA slide number based on the section parameters generated by Visiopharm. We have successfully completed workflow for TH labeled sections and are currently working to generalize the algorithms to the other stains. We expect that with additional labeling and training, we will be able to successfully develop an automated process for classifying sections on the rostral-caudal axis.The goal is that after training the app further through the labeling of sections, the app will be able to create parameters for future stains, thus creating a general classification framework.This framework will increase consistency across future labeling, which allows for more reliable and faster classification. This additional time can be devoted to analyzing specific stains and other research questions.