Huo Lu

Physiologically evoked neuronal current MRI in a bloodless turtle brain: detectable or not?

Contradictory reports regarding the detection of neuronal currents have left the feasibility of neuronal current MRI (ncMRI) an open question. Most previous ncMRI studies in human subjects are suspect due to their inability to separate or eliminate hemodynamic effects. In this study, we used a bloodless turtle brain preparation that eliminates hemodynamic effects, to explore the feasibility of detecting visually-evoked ncMRI signals at 9.4 T. Intact turtle brains, with eyes attached, were dissected from the cranium and placed in artificial cerebral spinal fluid. Light flashes were delivered to the eyes to evoke neuronal activity. Local field potential (LFP) and MRI signals were measured in an interleaved fashion. Robust visually-evoked LFP signals were observed in turtle brains, but no significant signal changes synchronized with neuronal currents were found in the ncMRI images. In this study, detection thresholds of 0.1% and 0.1 degrees were set for MRI magnitude and phase signal changes, respectively. The absence of significant signal changes in the MRI images suggests that visually-evoked ncMRI signals in the turtle brain are below these detectable levels.

Modeling oxygen effects in tissue-preparation neuronal-current MRI

Tissue‐preparation neuronal‐current MRI (ncMRI) was recently developed to directly detect neuronal activity without hemodynamic contamination. However, as a paramagnetic substance, the oxygen molecules present in the tissue may also alter the ncMRI signal through relaxivity and susceptibility effects. To study the effects of oxygen on the ncMRI signal and estimate their impact on tissue‐preparation experiments, oxygen‐induced MRI signal changes were formulated as a function of oxygen concentration (OC) of gas, oxygen consumption rate, and imaging parameters. Under favorable conditions of these parameters, the maximum oxygen‐induced signal magnitude and phase change were estimated to be 0.32% and 3.85°, respectively. Considering that the ncMRI signal changes obtained in previous tissue‐preparation experiments (3–5% in magnitude, 0.8–1.7° in phase) were tens or hundreds of times larger than the corresponding oxygen‐induced signal changes (0.03% in magnitude, 0.03–0.07° in phase), it is concluded that the oxygen had negligible effects in the previous experiments. Magn Reson Med 58:407–412, 2007.

The role of feedforward and feedback inhibition on frequency-dependent information processing in a cerebellar granule cell

Multi-modal sensory information entering the cerebellum via mossy fibers is processed through the granule cell (GC) network, the major cellular elements of which are the GC and an inhibitory interneuron, the Golgi cell. A GC model supporting both feedforward (FF) and feedback (FB) inhibition to its dendritic arbor was constructed. This model was used to examine the influence of Golgi cell inhibition on GC responses to mossy fiber inputs ranging from 10–100 Hz. Both FF and FB inhibitory signals reduced GC output. When both inhibitory loops accessed the same GC dendrite, the greatest decrement in GC output was produced by FB inhibition alone. However, if each inhibitory loop accessed a different GC dendrite, both inhibitory inputs were required to produce the greatest decrement in GC output.

Information processing in a cerebellar granule cell

A fifteen compartment, biologically realistic model of a cerebellar granule cell (GC) was developed to examine the signal processing capabilities of this most numerous element in the cerebellar cortical circuit. The model explicitly includes compartments for the soma, axon hillock, proximal axon, dendrites and terminal bulbs. All synaptic inputs were transduced via activation of glutamate receptor subtypes located on the dendritic bulb compartments, and were systematically varied in their number and frequency. An intriguing morphological feature, in which axonal location is shifted from the soma to a dendrite, was specifically examined to determine its impact on granule cell output. The GC was shown to be electrotonically compact, resulting in a lack of biasing of output based on axonal location. Biasing of output could be driven by changes in the passive parameters of the model, but required an unrealistically large change in resistive coupling between dendritic and somal compartments. Thus, axonal location does not induce physiologically relevant phase shifts between synaptic inputs located on multiple dendritic bulbs, suggesting that the GC relies heavily upon temporal aspects of its input signals for integrative processing.