My laboratory is interested in two questions: the mechanisms of memory and abnormalities that underlie schizophrenia. In both cases, we seek to determine how biochemical, neuronal and network processes can work as a system to perform physiological function.
What is the molecular basis of memory storage at synapses?This is a fundamental problem in neuroscience. The underlying question is non-trivial because memories can be very long-lasting (years), while the molecules of the synapse are short-lasting because of protein turnover.
We now have evidence that memory is stored by the abundant synaptic protein, CaMKII. During learning, this protein is activated, becomes phosphorylated, and binds to the NMDA channels at synapses. CaMKII holoenzymes contains 12 similar subunits (only four are shown in the figure); importantly, each subunit, if phosphorylated can phosphorylate neighboring subunits within the holoenzyme. This property of phosphorylation can explain the perpetuation of the activated state of the whole holoenzyme (and thus the memory). For instance, if a subunit becomes dephosphorylated by phosphatase, it can be rephosphorylated by a neighboring phosphorylated subunit. Similarly if in the process of protein turnover, a phosphorylated subunit is replaced by a non-phosphorylated subunit, the new subunit will be phosphorylated by a neighbor. Thus, CaMKII can stay activated for very long periods, thus producing long-term memory. We have tested the role of CaMKII in the maintenance of LTP and the maintenance of memory itself using the “erasure test”. We found that after LTP induction or after behavioral learning, interference with CaMKII function erases LTP and memory. Thus, CaMKII is essential for LTP and memory storage. We are currently trying to understand the mechanisms by which CaMKII triggers the structural enlargement of synapses that produces the actual potentiation of synaptic transmission at synapses that have learned. The first step in these structural changes is the binding of CaMKII to the NMDA channel. We using advanced optical methods (2-photon FRET) to study this binding in single dendritic spines.
What is the mechanism of short-term memory and why can we only remember about 7 items? The brain shows oscillations in electrical signals, as seen in the EEG. During wakefulness, dual oscillations in the theta (8Hz) and gamma (40Hz) frequency range occur together. We propose that these interacting oscillations form a framework by which multiple items can be held in short-term memory (the theta-gamma code accounts for why we can only about 7 items in working memory, as for instance a phone number). According to this hypothesis, the first item in a presented list is represented by the spatial pattern of cells that fire in the first gamma cycle within a theta cycle (see ovals in the diagram); the second items by the different cells that fire in the second gamma cycle, and the last item in the seventh gamma cycle (not shown). This process repeats on each theta cycle, thereby holding all 7 items in mind. In recent experiments using intracranial recording we have confirmed this hypothesis. Specifically, we found brain regions where activity is selective for particular letters. We found that for a site selective for G, the site is active in an early gamma cycle if G was given at the beginning of a list, but active in a late gamma cycle if G was at the end of the list. These results strongly support the hypothesis that short-term memory is organized by the theta-gamma code.
What is the abnormality in brain networks in schizophrenia? Our approach is based on two major observations. First, EEG measurements on subjects with schizophrenia show that delta frequency (1-4 Hz) oscillations are elevated in the disease in the awake state. Second, normal subjects, when given drugs that block (antagonize) NMDARs, acutely develop positive and negative symptoms of schizophrenia. We have brought these observations together by showing that NMDA antagonist can generate delta oscillations in the thalamus or rats and that these abnormal oscillations can interfere with functions known to show deficits in schizophrenia. We hope that by understanding the molecular and network mechanisms that generate these abnormal delta oscillations, new treatment strategies can be developed. One reason for particular optimism is that recent genetic studies have identified risk genes for schizophrenia, many of which are central to the thalamic processes that we have been studying.
Theory of Brain Function. How do different brain networks work together to produce mental function? I believe that there are general principles involved and that the rich data now being produced can reveal these principles. My current focus is on the way that oscillations organize information, pass it from one region to the other and organize action. The underlying coding mechanisms must obey general conventions about how to handle information so that different brain regions can work together. I think these and other ideas will yield relatively simple explanations of perception, memory, action-selection and consciousness.