Overview
Neurons are the basic functional unit of the brain; the brain contains about 200 billion such neurons. The average neuron has over 1,000 connections ("synapses") to other neurons, but some have over 100,000. The connections between neurons are largely made onto structures that are unique to neurons, an array of branches called dendrites (margin figure). Even though the total extent of the dendrites is often on the order of hundreds of microns (tenths of a millimeter), individual dendrite diameters fall in the one micron (one millionth of a meter) regime. Furthermore, connections are often made on dendritic specializations called spines, which are even smaller than dendrites.
Within these thousands of inter-connections between neurons, one finds a large variety of temporal activation patterns. Some inputs are random, others regular. Even more complex patterns have been observed, such as a pattern in which a burst of activity is followed by a longer quiescent interval, followed by another burst, and so on (margin figure). The time interval between activation of any single input can be as short as 1 millisecond (one thousandth of a second). Each of the thousands of inputs is potentially indendent of the others.
It is this spatio-temporal patterning of activation - the spatial distribution of time-varying synapses across the elaborate dendrites - which possibly underlies single neuron function. But an important question remains: what does a neuron actually do? Does it merely collect inputs and report the sum? Or can it be more complex? What features of a neuron allow it to accomplish this task? These are the fundamental scientific questions of our lab.
But studying these problems is experimentally difficult. As previously shown, neurons are small and fast. To characterize neuron physiology, one requires high speed, to faithfully capture events, and high resolution, to obtain detailed information from small structures. Here one encounters the well-known engineering principle of making performance compromises: one can have one capability, but not the other. Traditionally, neuroscientists have had high temporal resolution, or high spatial resolution, but not both. Our lab's engineering goal is to break this barrier: acquiring data from neurons at high speed and high resolution simultaneously. We do so by leveraging the advantages provided by optical imaging techniques and extending them in novel ways to create neurophysiology intruments with minimal compromises. A key component of this plan is the use of special optical probes: fluorscent molecules or photo-sensitive chemicals. Combining these optical probes with technologies that allow fast, precise positioning of laser illumination achieves this goal.
Physiology of Neuronal Dendrites
We are currently studying the principles of spatio-temporal synaptic integration. We are focusing on the underlying biophysical mechanisms that influence the interactions between two to thousands of synaptic inputs. Since this is experimentally difficult, we are taking a combined approach utilizing advanced optical techniqes and theoretical modeling. With the former, we employ the fast laser scanner developed in our lab to release precisely-defined spatio-temporal patterns of synaptic activation (margin figure), independent of the neuron's endogenous inputs. Since we dictate the input and can measure the output electrically or optically, we can begin to elucidate the transform between the input and output. Modeling these systems on a digital computer allows us to pre-test hypotheses, explore parameter spaces, confirm experimental data, and in general stimulate thinking about the problem.
Recently, our lab has been investigating the role of calcium in synaptic transmission, both pre-synaptically and post-synaptically.
Brad Losavio, Zanwen Chen, Neel Parikshak
Fast, Multi-Site Optical Imaging
project description/details/images...
Duemani Reddy, Vivek Bansal, Vijay Iyer, Rudy Fink, Brad Losavio
Super-resolution Imaging
project description/details/images...
Olga Gliko, Abbas Yaseen
collaborators: Bill Brownell, Bahman Anvari
Automated Reconstruction of Neuronal Dendrites ("ORION")
Current technology for extracting detailed morphological information about neuronal dendrites is severely limited. It requires a human operator to manually trace each and every branch, a process that takes several hours per neuron. In addition to this excessive time requirement, the accuracy of this information suffers from individual subjectivity, incomplete dye loading, and fixed tissue shrinkage.
In collaboration with two labs from the University of Houston, we are developing techniques to resolve these issues (margin figure). The morphology extraction is performed by a computer from a set of fluorescent images. The information is more accurate, due to the use of an objective computer algorithm and the fact that the tissue is not fixed. The process is also faster, requiring only a fraction of the time of a manual tracing by a human. A side benefit is that the neuron can remain alive while the reconstruction is performed. We can thus perform experiments guided by knowledge gleaned from the reconstruction.
Brad Losavio, Sean O'Malley, Alberto Santamaria
collaborators: Costa Colbert, Ioannis A. Kakadiaris
funding sources ? NIH, NSF...