Glioblastoma (GBM) is a malignant and aggressive brain tumor, with a five-year survival rate of 6.9 percent. Immunotherapy offers opportunities for personalized treatment strategies, leveraging the patient’s immune system to fight against cancer cells. While these approaches have seen success in other cancer types, the unique immunosuppressive microenvironment of GBM has rendered such therapies as unsuccessful. Currently, there lacks an in vitro model that recapitulates the interactions between tumor, immune, and normal cells. Our goal is to develop an immunocompetent model to directly study the tumor-immune microenvironment in our established cortical organoid tumor transplant system. Human immune cells in the form of peripheral blood mononuclear cells (PBMCs) are first isolated from whole blood. PBMCs are then fluorescently stained and introduced onto stem cell-derived cortical organoids containing GBM cells. To optimize the survival and composition of both PBMC and tumor cells, multiple culturing methods, PBMC cell densities, and transplant strategies were tested. After three days, transplanted organoids are dissociated and PBMC cell yield is analyzed using flow cytometry. We are optimizing both PBMC cell number recovered and composition to find which condition reflects that of the tumor microenvironment. Out of the tested conditions, an equal mix of organoid and immune cell media yielded the best PBMC recovery after co-culture. Collaborating with UCLA Neurosurgery, we will utilize our immunocompetent model to transplant dissociated primary GBM cells and patient PBMCs, to study patient-specific tumor-immune cell interactions, allowing improvements in the efficacy of existing immunotherapies.

Stroke is the leading cause of adult disability in the United States and ischemic stroke, characterized by loss of blood flow to an area in the brain, is responsible for 87% of all strokes. Damage to the stroke cavity, or the infarct, results in functional disability for affected individuals and is characterized by dying tissue devoid of neurons and with disordered vasculature. Following stroke, two mechanisms of repair associated with functional recovery– axonal sprouting and angiogenesis– have been observed in the healthy tissue surrounding the infarct, the peri-infarct, across model organisms. Current treatments, such as physical therapy, target functional recovery by leveraging plasticity within the peri-infarct but there are no reparative therapies targeting the infarct. The delivery of biomaterials, such as the porous microporous annealed particles (MAP) scaffold, has shown promise in remodeling the infarct through cell infiltration and decreased cerebral atrophy. A prior study with a nonporous hydrogel containing clustered vascular endothelial growth factor (VEGF) nanoparticles (CLUVENA) has demonstrated angiogenesis and axonal sprouting in the infarct and peri-infarct. In this study, cortical photothrombotic strokes were induced in mice and a hydrogel consisting of a MAP scaffold and heparin nanoparticles with or without CLUVENA was injected at 5 days post-stroke (dps). The mice were sacrificed at 30 or 60 dps and the slide-mounted brain tissue was stained with immunohistochemical assays to visualize vessels and axons within the cerebral cortex and hydrogel regions. We hypothesize increased axonal sprouting and angiogenesis in the MAP hydrogel containing CLUVENA relative to the MAP hydrogel.

The basal ganglia (BG) serve complex functions including motor control, reinforcement learning, and emotional regulation. Among BG’s interconnected nuclei (striatum, globus pallidus, subthalamic nucleus, substantia nigra), the subthalamic nucleus (STN) projects extensively to all the other structures and is considered the main modulator of BG output. Treatment of Parkinson’s disease by deep brain stimulation takes advantage of STN’s strategic role, but the procedure suffers from suboptimal electrode placement due to lack of understanding in STN connectivity. We address this knowledge gap by mapping STN afferents with the TRIO method. We labeled structures that are connected to STN neurons with efferents to substantia nigra pars reticulata (SNr), a major downstream target of STN. Our results confirmed previous literature on the major inputs STN receives from the entire primary somatosensory cortex, primary motor cortex, Raphe nuclei, and also the less established input from the superior colliculus. In the Globus Pallidus external segment (GPe), we found that efferents to STN concentrate in the ventral medial GPe, suggesting an internal topological organization exists in this structure. Interestingly, we also observed labeling in the caudoputamen (CP) never reported in previous literature. We followed up this finding with a cTRIO labeling experiment in which Cre recombinase is expressed by the parvalbumin (PV) promoter in STN. To our surprise, the cTRIO result didn’t show any labeling in CP. Due to the low probability of Cre-AAV injection leak in the TRIO construct, we hypothesize that the target STN neurons projecting to SNr don’t coincide with PV-expressing STN neurons.

The medial prefrontal cortex and anterior striatum have been implicated in decision-making. Previous studies have reported choice-related neural activity in them. Another major function of these regions is movement control. Although the coexistence of decision-making and movement control signals is well-established, it remains unknown whether they are represented within different neurons or the same neurons. In this study, we recorded videos of freely moving rats making perceptual decisions. We first employed Keypoint-MoSeq, a behavior analysis software built on unsupervised machine learning, to identify sub-trial level movement dynamics. We found that each rat’s movements can be divided into a set of explainable movement units ("syllables"). These syllables included turning sideways and staying in the center. We then combined syllable labels with electrophysiological data collected in medial prefrontal cortex and anterior striatum. We found that these syllables corresponded to distinguishable neuronal activity at population level. Finally, we tested how the timing of decision signals relates to syllables. We discovered that trials with varying syllable dynamics did not show significant variations in SVM (support vector machine) accuracy when decoding the decision. Also, we found no significant difference in syllable dynamics between trials where SVM accurately predicted decisions early versus those predicted later in the trial. These results implied that there might not be a straightforward relationship between the signals of movement and those of decision-making. Taken together, our findings shed new light on the complex interplay between decision-making and movement control. Future research may further expand our understanding of the mechanism underneath.