Past events

In vivo and in vitro, spontaneous and evoked neuronal activity are sparsely distributed across neocortical networks, where only a small subset of cells show firing rates greater than 1 Hz. Understanding the stability, network connectivity, and functional properties of this active subpopulation has been hampered by an inability to identify and characterize these neurons in vitro. Here we use expression of a fosGFP transgene to identify and characterize the properties of cells with a recent history of elevated activity. Neurons that had induced fosGFP expression in vivo maintained elevated firing rates in vitro over the course of many hours. Paired-cell recordings indicated that fosGFP+ neurons have a greater likelihood of being connected to each other, both directly and indirectly. These findings indicate that highly active neuronal ensembles are maintained over long time periods and suggest that specific, identifiable groups of neurons may dominate the way information is represented in the neocortex.

Carboxypeptidase E (CPE) is a prohormone processing enzyme that cleaves C-terminal basic residues from peptide hormone intermediates to yield active hormones, within secretory granules of neuroendocrine cells. A transmembrane form of the enzyme has been shown to be a sorting receptor that sorts prohormones and BDNF at the trans Golgi network and targets them to the regulated secretory pathway. Recently, live cell imaging studies have demonstrated that transport of peptidergic/BDNF secretory vesicles to the release site is dependent upon CPE. The cytoplasmic tail of CPE on the vesicles binds to microtubule motors, KIF1A/KIF3A and dynein via dynactin to effect transport of prohormone/BDNF vesicles in a bidirectional manner from the soma to the process terminals and return. In addition, CPE has been found to play a neuroprotective role in adult brain. In CPE-knockout (KO) mice, degeneration of pyramidal neurons was observed in the hippocampal CA3 region of animals equal or greater than 4 weeks of age, whereas the hippocampus was intact at 3 weeks and younger. Calbindin staining indicated early termination of the mossy fibers before reaching the CA1 region, and a lack of staining of the pyramidal neurons and apical dendritic arborizations in the CA1 region of CPE-KO mice. Ex vivo studies showed that cultured hippocampal neurons transfected with an enzymatically inactive form of CPE were protected against H2O2 oxidative-stress-induced cell death but not in non-transfected or LacZ transfected neurons. Thus CPE has an anti-apoptotic role in the maintenance of survival of adult hippocampal CA3 neurons, although the mechanism of action is unknown. In hepatocellular carcinoma (HCC) cells, overexpression of CPE resulted in enhanced proliferation and migration. SiRNA knockdown of CPE expression in highly metastatic HCC cells inhibited their growth and metastasis in nude mice. These results indicate that CPE is a new mediator of tumor growth and metastasis. Thus CPE is a multi-functional protein which actions include both enzymatic and non-enzymatic to mediate various physiological functions.

Recent advances in human neuroimaging have shown that it is possible to accurately read out a person's conscious experience based only on non-invasive fMRI measurements of their brain activity. This "brain reading" is possible because each thought is associated with a unique pattern of brain activity that can serve as a "fingerprint" of this thought in the brain. By training a computer to recognize these fMRI "thought patterns" it is possible to read out what someone is currently thinking with high accuracy. Here several studies will be presented that also directly address the relationship between neural encoding of information (as measured with fMRI) and its availability for awareness. These studies include comparisons of neural and perceptual information, unconscious information processing, decoding of time courses of perception, as well as decoding of high-level mental states. This will show that it is possible to read out a person's concealed intentions and even to predict how someone is going to decide a few seconds later. Finally, the talk will discuss fundamental challenges and limitations of the field, along with the ethical question if such methods might one day be a danger to our mental privacy.

Hierarchy provides a major conceptual framework for understanding structure-function relationships of the cortex (Felleman and Van Essen, Cerb Cortex 1991). Feedforward (rostral directed) projections link areas in an ascending series and have a driving influence; feedback (caudal directed) projections link areas in a descending series and have a modulatory influence. This has led to the suggestion that feedforward projections are uniquely reciprocated by feedback projections i.e no strong loops (Crick and Koch, Nature 1998). We have re-examined this issue by making retrograde tracer injections in 22 areas spanning the occipital, parietal, temporal and frontal lobes. Injections were placed in areas V1, V2, V4 TEO, STPa, STPm, STPp, AudPba, AudPbp, 5, 7a, 7b, F1, 2, 8a, 45b, 9/46d, 9/46v, 46d, F5, ProM, 24c. High frequency sampling allows determination of indices of laminar distribution (SLN) and the relative strength (FLN) of connections (Vezoli et al., The Neuroscientist 2004). Analysis shows an inverse relationship between strength of connection and distance and revealed many (30%) hitherto unknown long-distance connections. Elsewhere we have shown that cortico-cortical projections form a smooth gradient: long-distance ascending connections are strongly feedforward (high SLN XX 100%) and on approaching the injection site have progressively lower SLN values (reaching 51%); likewise long-distance descending connections are strongly feedback (low SLN XX 0) and approaching the injection site reduce SLN 49% (Barone and Kennedy, J. Neurosci. 2000). The Felleman and Van Essen data is strictly hierarchical (no strong loops). A topological model of our data shows small world features (high cluster index and short average path distances) and five strong loops. Strong loops link frontal areas with occipital (areas 45-V4, 8A-V4), temporal (areas 45-TEO, 46-TEO) and parietal (areas 8A-7A, 46-7A) areas. The areas participating in strong loops exhibit high degrees of connectivity and constitute the hubs promoting small world attributes in the cortical architecture. The strong loops make it possible to go from V4 to all higher areas and back to V4 by uniqely feedforward pathways in an average of 3 and a maximum of 8 steps. One consequence of these anti-hierarchical connections is that the computations carried out in the supragranular layers of the cortex (Douglas and Martin, Annual Rev Neurosci. 2004) can be widely distributed in large-scale cortical networks mediating top-down control.