Heterogeneity in Neuronal Dynamics Is Learned by Gradient Descent for Temporal Processing Tasks.

Individual neurons in the brain have complex intrinsic dynamics that are highly diverse. We hypothesize that the complex dynamics produced by networks of complex and heterogeneous neurons may contribute to the brain's ability to process and respond to temporally complex data. To study the role of complex and heterogeneous neuronal dynamics in network computation, we develop a rate-based neuronal model, the generalized-leaky-integrate-and-fire-rate (GLIFR) model, which is a rate equivalent of the generalized-leaky-integrate-and-fire model. The GLIFR model has multiple dynamical mechanisms, which add to the complexity of its activity while maintaining differentiability. We focus on the role of after-spike currents, currents induced or modulated by neuronal spikes, in producing rich temporal dynamics. We use machine learning techniques to learn both synaptic weights and parameters underlying intrinsic dynamics to solve temporal tasks. The GLIFR model allows the use of standard gradient descent techniques rather than surrogate gradient descent, which has been used in spiking neural networks. After establishing the ability to optimize parameters using gradient descent in single neurons, we ask how networks of GLIFR neurons learn and perform on temporally challenging tasks, such as sequential MNIST. We find that these networks learn diverse parameters, which gives rise to diversity in neuronal dynamics, as demonstrated by clustering of neuronal parameters. GLIFR networks have mixed performance when compared to vanilla recurrent neural networks, with higher performance in pixel-by-pixel MNIST but lower in line-by-line MNIST. However, they appear to be more robust to random silencing. We find that the ability to learn heterogeneity and the presence of after-spike currents contribute to these gains in performance. Our work demonstrates both the computational robustness of neuronal complexity and diversity in networks and a feasible method of training such models using exact gradients.

A Subpopulation of Microglia Generated in the Adult Mouse Brain Originates from Prominin-1-Expressing Progenitors.

Microglia maintain brain health and play important roles in disease and injury. Despite the known ability of microglia to proliferate, the precise nature of the population or populations capable of generating new microglia in the adult brain remains controversial. We identified Prominin-1 (Prom1; also known as CD133) as a putative cell surface marker of committed brain myeloid progenitor cells. We demonstrate that Prom1-expressing cells isolated from mixed cortical cultures will generate new microglia in vitro To determine whether Prom1-expressing cells generate new microglia in vivo, we used tamoxifen inducible fate mapping in male and female mice. Induction of Cre recombinase activity at 10 weeks in Prom1-expressing cells leads to the expression of TdTomato in all Prom1-expressing progenitors and newly generated daughter cells. We observed a population of new TdTomato-expressing microglia at 6 months of age that increased in size at 9 months. When microglia proliferation was induced using a transient ischemia/reperfusion paradigm, little proliferation from the Prom1-expressing progenitors was observed with the majority of new microglia derived from Prom1-negative cells. Together, these findings reveal that Prom1-expressing myeloid progenitor cells contribute to the generation of new microglia both in vitro and in vivo Furthermore, these findings demonstrate the existence of an undifferentiated myeloid progenitor population in the adult mouse brain that expresses Prom1. We conclude that Prom1-expressing myeloid progenitors contribute to new microglia genesis in the uninjured brain but not in response to ischemia/reperfusion.SIGNIFICANCE STATEMENT Microglia, the innate immune cells of the CNS, can divide to slowly generate new microglia throughout life. Newly generated microglia may influence inflammatory responses to injury or neurodegeneration. However, the origins of the new microglia in the brain have been controversial. Our research demonstrates that some newly born microglia in a healthy brain are derived from cells that express the stem cell marker Prominin-1. This is the first time Prominin-1 cells are shown to generate microglia.