Keck Graduate Institute (KGI) has awarded seed grants to three pioneering research projects that span disciplines and schools within the Institute, reflecting KGI’s mission to unite science and innovation and enabling faculty to forge new collaborations, pilot ambitious ideas, and build the foundation for high-impact science. These internal awards spark early-stage collaboration, providing faculty the means to generate preliminary data, explore novel questions, and pursue larger external funding.
The recipients represent the intersection of fields as diverse as machine learning, immunology, gene editing, and neurodegeneration — exemplifying how cross-disciplinary thinking can uncover hidden connections and accelerate progress in biotechnology and healthcare.
Predicting Stem Cell Fate with Machine Learning and Spectroscopy
Gargi Ghosh, Associate Professor of Bioprocessing
Animesh Ray, W. M. Keck Foundation Professor of Systems, Computational, and Molecular Biology
Mesenchymal stem cells (MSCs) hold enormous promise in regenerative medicine, but one critical challenge stands in the way: scaling up production without compromising quality. While current methods rely on antibody-based staining to check the cells' status, these techniques are slow, invasive, and impractical for industrial-scale use.
To solve this, Gargi Ghosh and Animesh Ray are building a faster, label-free way to determine stem cell fate using infrared (IR) spectroscopy and machine learning (ML).
"The idea is to scan the cells and use their biochemical signatures to predict whether they’re maintaining stemness or differentiating toward bone, fat, or cartilage lineage," Ghosh, who's extensively studied MSCs and their role in regenerative medicine, said. "And to do that in just 10 to 15 minutes."
Rather than using stains or antibodies, the new approach analyzes how cells absorb infrared light — producing a unique chemical fingerprint of their internal state. Ghosh's team collects this spectral data, while Ray, whose work blends biology and computer science, designs machine learning models to interpret the results.
"IR spectroscopy generates signals across many wavelengths — far too much for the human eye to parse," Ray said. "Machine learning helps us extract meaningful patterns and classify the cells based on their biochemical state."
The need for real-time quality checks is urgent. Industrial-scale therapeutic applications often require producing more than a billion cells at a time. But if some cells begin to differentiate into unintended types — like fat instead of bone — the therapy can lose its effectiveness.
"You can’t afford to wait hours for results,” Ghosh said. "You need to catch those changes early and intervene."
Their method would make that possible, enabling researchers to monitor and guide stem cell differentiation in real time — improving reproducibility and efficiency. It could also extend beyond natural MSCs, supporting quality control in gene therapy and engineered cell treatments.
"There are already gene therapies and clinical trials using genetically modified stem cells to treat disorders like sickle cell anemia and thalassemia," Ray said. "But the cost is very high. If we could streamline the quality control process, we might start making these therapies more accessible."
Backed by a one-year seed grant, the team aims to generate early data to support NIH and other external funding proposals.
"This type of interdisciplinary collaboration wouldn’t be possible without seed funding," Ghosh said. "It’s allowing us to combine engineering, biology, and AI in a novel way — and potentially open up a new frontier in stem cell manufacturing."
Unraveling the Mechanisms of Alzheimer’s Disease
Subhrajit Bhattacharya, Assistant Professor of Pharmacology
Derick Han, Keck Professor of Biopharmaceutical Sciences, Associate Professor of Biochemistry and Pharmacology
Early intervention in Alzheimer’s disease remains a major scientific hurdle — largely because there are still no reliable biomarkers that signal the first stages of decline. Subhrajit Bhattacharya and Derick Han are working to change that by studying a molecular interaction that may hold the key to both early diagnosis and intervention.
Their project investigates the interaction between a neural adhesion molecule called PSA-NCAM and a subtype of glutamate receptor (GluN2B) located outside the synapse. These extrasynaptic receptors are increasingly linked to pro-dementia signaling.
"Normally, PSA-NCAM helps suppress that activity, but in Alzheimer’s, that protective mechanism breaks down," Bhattacharya said. "We’re trying to check that downslide before it gains momentum."
In a healthy brain, PSA-NCAM acts like a natural brake — buffering the overactivation of GluN2B receptors and protecting against neuronal damage. But with aging and Alzheimer's, this protective function weakens.
"If we can intervene early, we have a better chance of preserving memory," Bhattacharya said.
By mapping how PSA-NCAM and GluN2B interactions shift during aging and disease progression, the researchers aim to identify new therapeutic targets. Bhattacharya brings expertise in synaptic signaling and molecular neurobiology, while Han contributes deep experience in Alzheimer’s pathology and model development.
"What excites me most is that this marker appears to shift significantly between normal aging and Alzheimer’s," Han said. "If we can understand that mechanism, we may be able to modulate it."
The team will use their seed grant to expand into more physiological models, including cortical neuron cultures and potentially behavioral studies in animals. They also plan to incorporate stem cell–derived organoids — 3D tissue cultures that better mimic the complexity of the brain.
At the same time, they're developing therapeutic strategies aimed at blocking the GluN2B disruptions seen in Alzheimer’s. These include testing polysialic acid and creating novel compounds that restore PSA-NCAM's regulatory role.
“Alzheimer’s is so complex,” Han said. “Finding even one regulatory switch between normal aging and disease progression is a big deal.”
Targeting a Novel Receptor in Multiple Sclerosis
Jeniffer Hernandez, Associate Professor of Immunology
Barbara Bailus, Assistant Professor of Genetics
Autoimmune diseases like multiple sclerosis (MS) remain among the most complex conditions to treat. In MS, the body’s own immune system mistakenly attacks the protective myelin sheath around neurons — slowing nerve signals and triggering a range of symptoms from muscle weakness to cognitive decline.
“You can think of it like insulation on a wire,” Barbara Bailus said. “When that insulation gets damaged, the signal can’t travel efficiently.”
To address this, Bailus and Jeniffer Hernandez are exploring a potential breakthrough: targeting a receptor called GPR65 that appears to drive inflammation in autoimmune disease. The collaboration emerged naturally — sharing lab space led them to recognize a unique opportunity to combine their complementary expertise when the seed grant program launched.
Hernandez, who has studied GPR65 for nearly a decade, recently had a manuscript accepted for publication confirming its elevated expression in TH17 cells — a subset of immune cells heavily involved in MS-related inflammation. Her previous work in mice models showed that blocking this receptor reduced inflammatory activity.
“We’ve shown that blocking GPR65 in mice reduces TH17 cell activity and lowers inflammation,” Hernandez said. “Now we want to see if that holds true in human cells.”
The team will use CRISPR-Cas9 gene editing to disable GPR65 in a human T-cell line and observe how its absence affects cytokine production, metabolism, and cell survival.
“T-cells need a lot of energy to carry out their functions,” Hernandez said. “If we can inhibit their metabolism, we can potentially slow their destructive activity.”
Bailus, who specializes in gene-editing tools and previously worked on diseases like Huntington's, will lead the CRISPR component. Her expertise in delivering treatments across the blood-brain barrier adds a critical translational dimension.
“Our role is to precisely remove the GPR65 gene so Jeniffer’s team can observe how the cells respond,” Bailus said.
If successful, the research could lay the groundwork for targeted therapies such as antibodies or small molecules that block GPR65 activity without compromising healthy immune function.
“This is a proof-of-concept project,” Bailus said. “We’re starting with cells in a dish, but if the data is promising, we could eventually move into animal models and apply for larger grants from NIH or the National MS Society."
Interest in GPR65 is growing rapidly. At a recent conference, Hernandez's students drew attention from major biotech companies for their research on the receptor.
“It’s a hot target right now,” she said. “We’re excited to see where this work could lead.”