Showing papers by "Xiaohui Xie published in 2004"

Errα and Gabpa/b specify PGC-1α-dependent oxidative phosphorylation gene expression that is altered in diabetic muscle

Abstract: Recent studies have shown that genes involved in oxidative phosphorylation (OXPHOS) exhibit reduced expression in skeletal muscle of diabetic and prediabetic humans. Moreover, these changes may be mediated by the transcriptional coactivator peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α). By combining PGC-1α-induced genome-wide transcriptional profiles with a computational strategy to detect cis-regulatory motifs, we identified estrogen-related receptor α (Errα) and GA repeat-binding protein α as key transcription factors regulating the OXPHOS pathway. Interestingly, the genes encoding these two transcription factors are themselves PGC-1α-inducible and contain variants of both motifs near their promoters. Cellular assays confirmed that Errα and GA-binding protein a partner with PGC-1α in muscle to form a double-positive-feedback loop that drives the expression of many OXPHOS genes. By using a synthetic inhibitor of Errα, we demonstrated its key role in PGC-1α-mediated effects on gene regulation and cellular respiration. These results illustrate the dissection of gene regulatory networks in a complex mammalian system, elucidate the mechanism of PGC-1α action in the OXPHOS pathway, and suggest that Errα agonists may ameliorate insulin-resistance in individuals with type 2 diabetes mellitus.

Learning in neural networks by reinforcement of irregular spiking.

Abstract: Artificial neural networks are often trained by using the back propagation algorithm to compute the gradient of an objective function with respect to the synaptic strengths. For a biological neural network, such a gradient computation would be difficult to implement, because of the complex dynamics of intrinsic and synaptic conductances in neurons. Here we show that irregular spiking similar to that observed in biological neurons could be used as the basis for a learning rule that calculates a stochastic approximation to the gradient. The learning rule is derived based on a special class of model networks in which neurons fire spike trains with Poisson statistics. The learning is compatible with forms of synaptic dynamics such as short-term facilitation and depression. By correlating the fluctuations in irregular spiking with a reward signal, the learning rule performs stochastic gradient ascent on the expected reward. It is applied to two examples, learning the XOR computation and learning direction selectivity using depressing synapses. We also show in simulation that the learning rule is applicable to a network of noisy integrate-and-fire neurons.