Li-Huei Tsai

Our laboratory is interested in elucidating the pathogenic mechanisms underlying neurological disorders that impact learning and memory. We are taking a multidisciplinary approach to investigate the molecular, cellular, and circuit basis of neurodegenerative disorders.

The Role of Cdk5 in Neurodegeneration and Normal Brain Function

Alzheimer’s disease (AD) is a devastating and irreversible brain disorder that eventually leads to dementia and death. Cyclin-dependent kinase 5 (Cdk5) is a brain-specific protein serine/threonine kinase essential for brain development, synaptic plasticity, learning, and memory. We have shown that the hyperactivation of Cdk5 occurs when its regulatory protein p35 is cleaved by the Ca²⁺-activated protease calpain, under neurotoxic conditions, to liberate the carboxyl-terminal fragment p25. We hypothesized that p25 generation and accumulation play important roles in AD-like neurodegeneration. We created an inducible mouse model of p25 accumulation (the CK-p25 mouse) that displays key pathological hallmarks of AD, including profound neuronal loss in the forebrain, increased β-amyloid (Aβ) peptide production, Tau pathology, and severe cognitive impairment. In this model, increased Aβ levels are observed prior to neuronal loss; furthermore, reducing Aβ production ameliorates memory deficits in the CK-p25 mouse model, suggesting that this event operates synergistically with p25 to lead to the manifestation of neurodegeneration and memory impairment.

To further decipher the role of p25 generation in neurodegeneration, we created a mouse “knock-in” model whereby the endogenous p35 gene is replaced by a mutant p35 that is resistant to calpain cleavage (Δp35KI). Δp35KI mice show normal hippocampus-dependent learning and memory and LTP. However, they exhibit impaired hippocampal long-term depression (LTD). Moreover, Δp35KI hippocampal neurons are resistant to Aβ peptide-induced synaptic depression. The 5XFAD model is a well-established AD mouse model exhibiting abundant amyloid plaque pathology, inflammation, and memory deficits by 6 months of age. Interestingly, memory deficits are not observed in Δp35KI/5XFAD compound mice, and the animals also show ameliorated inflammatory response and reduced Aβ levels. These results strongly suggest that p25 mediates Aβ peptide-associated pathology.

Currently, we are evaluating whether p25 generation also plays a role in Tau-mediated pathology. In addition, we are using CRISPR/Cas9-mediated genome editing to create calpain-resistant p35 alleles in human induced pluripotent stem cells (iPSCs) derived from AD patients (see below) to evaluate the role of p25 generation in the degeneration of human neurons and to create human cellular models of AD for therapeutic investigation.

Chromatin Remodeling and Learning and Memory

Among the available AD mouse models, the CK-p25 mouse uniquely recapitulates the profound neuronal loss observed in advanced stages of AD. Thus, we have used this model to explore novel therapeutic approaches that may be beneficial for cognition even after profound synaptic loss and neuronal death have occurred. We showed that promoting chromatin remodeling in CK-p25 mice with chemical histone deacetylase (HDAC) inhibitors induces robust synaptogenesis and dendritic growth, restores learning and recovers long-term memory—even after massive neuronal loss has occurred. Our findings demonstrated that an epigenetic mechanism involving increased histone acetylation and chromatin remodeling can be beneficial for learning and memory, and that these manipulations are effective even in the face of neuronal loss and neurodegeneration. These observations suggest that memory is not completely erased after neurodegeneration, and provide compelling evidence for developing HDAC inhibitors to reverse cognitive impairment in Alzheimer’s disease.

As our lab delved into the question of what leads to neuronal loss, we found that histone deacytylases regulate neuronal DNA damage and that compromised genome integrity, particularly DNA double-strand breaks (DSBs), is a common and early hallmark of neurodegeneration. Intriguingly, we recently discovered that DSBs are induced by neuronal activity and are actually required for normal learning and memory to occur. We also found that DSBs occur non-randomly in the neuronal genome, but are associated with specific genes. Now that we know the location and purpose of activity-induced DSBs in the neuronal genome, we can begin to determine how these physiological breaks are impacted during disease, and identify potential approaches to rectify DSBs production and enhance DNA damage repair in neurons of neurodegenerative diseases.

To better understand the changes in the epigenome during neurodegeneration, we collaborated with computer scientist, Manolis Kellis, in a comprehensive study that identified new master regulatory elements of AD that are conserved between the inducible CK-p25 mouse model and human AD brains. The combined power of the mouse model with advanced bioinformatics enabled us to identify novel enhancers, rather than coding regions, as significant factors with substantial contribution to the genetic risk for AD. In particular, we found a marked upregulation of genes and enhancers associated with innate immune responses in both AD mouse models and human AD patients. These results have dramatically altered our understanding of the genetic elements contributing to AD risk, and highlight important roles for epigenetic regulation and immune responses inAD pathogenesis. Currently, we are performing cell type specific epigenomic analysis, including ATAC-seq and ChIP-seq, and single cell RNA-sequencing in mouse and human brains to further understand the roles of different neural cell types including neurons, astrocytes, and microglia in AD-related neurodegeneration.

Neuronal Networks, Circuits, and Neurodegeneration

Altered cellular processes in neurodegenerative diseases ultimately lead to disruptions in neural circuits and network connectivity to result in cognitive deficits. Thus, to better understand how disease genes and pathology impact cognitive function in disorders such as AD, we have aspired to alter the activity of specific neural circuits and to evaluate the consequences on pathology, network activity, and behavior. We use targeted genetic approaches such as optogenetics and Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) to manipulate the activity of specific populations of neural cells in the brain. In combination with various AD mouse models, we are examining the effect of stimulating or inhibiting specific cells within hippocampal, basal forebrain, and amygdalar circuits to gain insights into circuit dysfunction underlying neurodegeneration and cognitive deficits.

In particular, our work has revealed network-level disruptions in gamma oscillations in multiple mouse models of AD. Conversely, we are also interested in applying circuit manipulations to ameliorate cognitive deficits in AD. We previously showed that we could induce population-wide gamma oscillations in mice using optogenetics to entrain large groups of neurons to synchronize their firing patterns. Most recently, we developed non-invasive methods of inducing gamma oscillations by presenting sensory stimulation at specific frequencies to the mice. In collaboration with Ed Boyden’s lab at MIT, we used optogenetics or non-invasive sensory stimulation to induce gamma oscillations, and surprisingly found a marked reduction in Aß levels in several AD mouse models. The therapeutic effects of gamma oscillations appear to involve the concerted actions of neurons and microglia to reduce the production and enhance clearance of Aβ, respectively. We are currently testing various parameters of inducing gamma oscillations, such as using other sensory modalities, and examining the potential of gamma stimulation to improve cognition in AD model mice. We are also working on developing other non-invasive methods to affect neural oscillations with spatial and temporal specificity.

Through collaboration with MIT’s Clinical Research Center and faculty members across multiple disciplines at MIT, we are gearing up to apply this technology to human subjects. Initial tests will establish the safety and feasibility of gamma entrainment in humans through sensory stimulation. State-of-the-art technology, such as magnetoencephalography (MEG), electroencephalography (EEG), and functional magnetic resonance imaging (fMRI) will be used to investigate and visualize potential brain oscillation changes. Eventually, our goal is to apply our non-invasive technology to subjects with AD, and determine whether gamma stimulation can help rescue cognitive deficits.

In addition to the manipulation of specific circuits, we aim to map out the sequential temporal and spatial disruptions of neural circuits by the deposition of Aß peptides and aggregated Tau protein in AD, and how Aß and Tau pathology propagates throughout the brain. Current work involves mapping various AD-related pathology in a 3-D manner in mouse and human brains. We are collaborating with Kwanghun Chung at the MIT Picower Institute to apply enhanced CLARITY techniques to interrogate the relationship of ß-amyloid plaques, Tau tangles, neural inflammation, and vasculature pathology in different mouse models of AD and in postmortem human brain samples of AD subjects. We will also use this approach to assess how environmental enrichment and epigenetic drugs, previously shown to improve cognitive function of AD mouse models, influence neuronal circuitry necessary for memory formation and retrieval.

Modeling neurological disorders using induced pluripotent stem cells (iPSCs) and tissue bioengineering

Our lab has collected human skin fibroblast lines from healthy individuals as well as late onset sporadic AD (LOAD), early onset familial AD (fAD), Autism spectrum disorder (ASD), schizophrenia, bipolar disease, and Down syndrome patients and reprogrammed them into induced pluripotent stem cells (iPSCs). We use genome-editing techniques such as CRISPR/Cas9 to create isogenic cell lines to facilitate the assessment of phenotypic consequences of disease associated genetic variants. These iPSCs are then differentiated into major brain cell types including excitatory neurons, astrocytes, and microglia that can be used for a number of basic and applied research purposes. For example, we can use this system to both examine how specific gene perturbations affect AD-like pathology directly in human neurons and glia, while at the same time screening libraries of drug-like chemicals in a high-throughput fashion to determine potential therapeutic candidates. Increasingly sophisticated culture techniques also allow us to evaluate how disease pathology and genetic variants affect each of the different cell types populating the brain. Using techniques of bioengineering combined with multiphoton deep imaging, optogenetics, and electrophysiology, we can recapitulate and study complex human brain tissue. In these “mini-brain” or organoid cultures, we can examine neuronal and glial activity and examine relevant disease phenotypes such as protein aggregation, neuronal connectivity, and synapse loss. Current efforts aim to better recapitulate the in vivo brain environment by adding engineered vasculature to our 3D organoids to mimic the blood brain barrier. Ultimately, we hope that these techniques will facilitate and expedite drug screening and discovery by allowing us to directly screen engineered human brain organoids for compounds and therapies likely to work in the in vivo human brain.