Dr Guy Major - DPhil
Imaging Dendritic Function and Neural Network Activity in Cerebral Cortex
There is a technological revolution under way in neuroscience. New ‘biophotonics’ techniques such as two-photon microscopy and patterned two-photon neurotransmitter uncaging give us the ability to simultaneously record from and stimulate multiple neurons or multiple branches of individual neurons. This can be done both in brain slices and in intact animals.
We are using these optical techniques in conjunction with electrical recordings and computer modelling to investigate two related questions:
a. What are the computational roles of dendrites in the cerebral cortex?
In brain slices, in response to stimulation by the neurotransmitter glutamate, 20 micrometer long segments of individual dendrites are capable of generating voltage pulses called ‘NMDA spike/plateaus’. These events have voltage and glutamate thresholds. Sections of individual dendrites can therefore be computationally equivalent to decision-making ‘units’ in neural network models. Because different sections can interact, individual dendrites could have very complex and versatile computational capabilities depending on the recent pattern of activity within the local neural network. NMDA spike/plateaus could allow dendrites to contribute to graded persistent firing, thought to underlie working memory. They could also endow individual dendrites with directional selectivity and enable them to decode information encoded as spike times across an array of axons. One of our main research aims is to determine whether dendritic NMDA spike/plateaus play an important role in the intact brain during normal function. We will do this by monitoring dendrites in vivo for the large calcium transients associated with NMDA spike/plateaus. We are also investigating interactions between NMDA spike/plateaus at multiple dendritic locations, using patterned two-photon glutamate uncaging in brain slices.
Focal iontophoresis of a brief glutamate pulse onto a single cortical pyramidal neuron basal dendrite can elicit a local dendritic spike/plateau potential and calcium transient.
View a PDF of an enlarged version of the above image.
(A). A layer 5 pyramidal neuron is filled with a low affinity calcium indicator via a whole-cell patch pipette. A double barreled sharp electrode with fluorescein in one barrel and glutamate in the other is aimed to within 1-2 μm of a basal dendrite.
(B) A 5 ms pulse of glutamate is iontophoresed onto the dendrite, eliciting a spike/plateau potential in the dendrite, an attenuated, smoothed version of which is recorded at the cell body.
(C) Corresponding 2-photon microscope image of the dendrite in vicinity of iontophoresis electrode. Average of baseline pre-stimulus frames.
(D) Image frame with peak fluorescence; the input dendrite becomes visibly brighter close to the iontophoresis electrode. The sister dendrite does not change in brightness.
(E) Higher time resolution rectangle scan, rotated so dendrite lies approximately along fast scan axis. Different regions of interest (ROIs) indicated by different colored rectangles; fainter areas require bigger ROIs to achieve comparable noise levels; background = white rectangle.
(F) Approximate time course of calcium transient at input site. Relative fluorescence transient from fast rectangle scan in panel E. Black: somatic voltage trace from panel B.
(G) Spatial profile of relative fluorescence transient shows ~20 μm long ‘hot-spot’ near stimulation electrode.
b. How is information encoded in the cerebral cortex?
Most somatosensory neurons only fire one or two (or fewer) action potentials during brief sensory stimuli such as whisker deflections, so it is possible that information about the the stimulus (such as location and direction) is encoded by relative action potential timing across a large array of neurons.
Dendritic calcium transients associated with back-propagating action potentials. Cells loaded with a high affinity indicator calcium indicator.
View a PDF of an enlarged version of the above image.
Action potentials can be monitored in neurons by means of calcium transients. In general, an action potential is associated with a calcium transient in the cell body and proximal dendrites with a rapid onset synchronised to the action potential. This allows the timing of action potentials to be determined to within a few milliseconds, if a sufficiently fast imaging technique is used. Calcium-sensing fluorescent dyes can be loaded into many neurons at once. Alternatively, strains of genetically modified mice are now available which express artificial calcium sensing fluorescent proteins in sub-sets of their neurons. We are in the process of combining these techniques with rapid two-photon imaging of the cerebral cortex in vivo, to measure relative action potential times across arrays of dozens to hundreds of neurons during sensory stimulation.
Royal Society Research Grant: "Imaging dendritic function in cerebral cortex"
BBSRC: "Developing imaging of action potential times in arrays of cerebral cortical neurons expressing transgenic calcium-sensing fluorescent proteins"
Wellcome Trust: "Imaging NMDA spike/plateaus in thin dendrites in brain slices using genetically-encoded calcium indicators"
David Tank, Princeton University , NJ, USA
Jackie Schiller, Technion, Haifa, Israel
Winfried Denk, Max-Planck-Institut für Medizinische Forschung, Heidelberg, Germany
Mark Goldman , Wellesley College, Worcester, MA, USA