I am a senior undergraduate in Biological Sciences at Fudan University, interested in understanding and controlling distributed neural dynamics.
I have studied selective brain vulnerability and behavior in He Lab at Fudan University, and autonomic control of stress physiology in Oka Lab at Caltech. These experiences led me from studying individual brain regions and pathways toward a broader question: how does information emerge, transform, and ultimately drive behavior and physiology across distributed neural networks?
Goal: To help build a future in which neurological disease no longer determines what a person can perceive, express, or do—and in which human–machine symbiosis expands, rather than diminishes, our possibilities.
Research Interest: The intersection of neuroengineering, physiology, and systems neuroscience.
Research Question: Do selected cortical populations contain enough stable and generalizable information to estimate—and eventually intervene in—brain–body states?
Neocortex: I see the neocortex as both a practical substrate for scalable BMIs and a window into distributed neural computation: it is accessible, while also sitting at the intersection of sensory input, motor output, memory, prediction, and internal-state representation.
Stress alters gastrointestinal motility, but the neural pathways mediating these changes remain incompletely understood. Using acute and repeated restraint stress in mice, we combined gastrointestinal assays, c-Fos mapping, sympathetic denervation, and chemogenetic manipulation to investigate brain-gut signaling. Acute stress slowed upper gastrointestinal transit, increased fecal output, and delayed feeding initiation, accompanied by marked c-Fos expression in the celiac-superior mesenteric ganglion complex (CG-SMG) and medial amygdala (MeA). Peripheral sympathetic denervation with 6-hydroxydopamine partially attenuated these gastrointestinal changes, with a stronger effect on fecal output. Broad MeA activation did not significantly alter upper gastrointestinal transit or feeding latency under the tested conditions. After repeated restraint stress ended, fecal output was reduced during a three-hour observation period without restraint. These findings identify distinct acute and post-stress gastrointestinal phenotypes and support a contribution of peripheral sympathetic signaling to acute stress-induced dysmotility, while the central pathways and mechanisms underlying post-stress changes remain unresolved.
Thiamine deficiency causes severe neurological dysfunction with regional selectivity, yet how functional deterioration unfolds over time and why particular brain regions are preferentially affected remain incompletely understood. Here, we characterized pyrithiamine-induced thiamine deficiency (PTD) in mice using longitudinal behavioral monitoring, gait analysis, metabolic measurements, multiscale histopathological analysis, and brain-wide metabolic and vascular datasets. Across pyrithiamine doses, disease onset shifted in time while retaining a common trajectory consisting of a phase of relatively preserved functions followed by rapid behavioral and neurological decompensation. Higher-resolution home-cage monitoring detected abnormalities only modestly earlier than this transition, supporting early functional compensation. Brain pathology was spatially selective, with the mammillary body (MBO) among the earliest and most consistently affected regions. Vulnerable regions tended to exhibit high baseline glucose utilization and dense vascular organization, although these features did not fully account for selective vulnerability. Together, our findings frame PTD as a transition from functional compensation to rapid decompensation and suggest that regional vulnerability reflects the interaction between systemic metabolic failure and local tissue properties.
The medial mammillary body (MM) contains molecularly distinct neuronal populations whose anatomical organization and circuit connectivity remain incompletely understood. This project aims to characterize the spatial distribution and circuit organization of Penk⁺ and Nts⁺ neurons in the MM. Using Allen Brain Atlas in situ hybridization data together with Penk-Cre;Ai14 and Nts-Cre;Ai14 reporter mice, we found that Nts⁺ neurons are relatively abundant and concentrated in the middle-to-posterior MM, whereas Penk⁺ neurons are sparse and largely restricted to a narrow region in the anterior MM. We next attempted to identify upstream inputs to Penk⁺ MM neurons using a Cre-dependent monosynaptic rabies tracing strategy, with the interanteromedial thalamic nucleus (IAM) or anteromedial thalamic nucleus (AM) serving as downstream targets. However, reliable starter-cell labeling could not be achieved because of the highly restricted distribution of Penk⁺ neurons, the high precision required for viral targeting, and substantial off-target infection in surrounding Penk-rich regions. These results establish the distinct spatial organization of Penk⁺ and Nts⁺ neuronal populations in the MM and identify key technical constraints for projection-defined circuit tracing in this small hypothalamic structure. Future work will focus on improving viral targeting accuracy and extending circuit mapping to the more abundant Nts⁺ population.
Gene therapy holds promise for durable correction of genetic defects, with recombinant adeno-associated virus (rAAV) as a leading vector, though limited transduction efficiency often necessitates high doses. Using L1000, the transcriptomic data of FDA-approved compounds, we identified 2,138 AAVR (KIAA0319L)-linked data entries and selected two promising candidates for following verifications. Entinostat (ENT), a benzamide HDAC inhibitor, significantly enhanced AAV2 and AAV8 transduction in Huh7 and HEK293 cells, whereas Lapatinib (LAP) was ineffective. Mechanistically, ENT upregulated AAVR and GPR108 in Huh7 but not in HEK293, where enhancement persisted even with AAVR-independent AAVrh32.33, indicating additional pathways. In vivo hepatic studies in normal mouse liver showed no significant change in AAV transduction or AAVR expression, consistent with the possibility that AAVR upregulation is restricted to cancer-derived cell lines. These findings support ENT as a potential adjuvant for rAAV delivery in specific cellular contexts and highlight the value of our L1000-based pipeline for identifying small-molecule enhancers of AAV transduction.
