Janelia Research Campus
Dr. Spruston is also scientific program director at the Janelia Research Campus.
How Do Neurons Compute Output from Their Inputs?
After taking a neurophysiology class as a college senior, it struck Nelson Spruston that the brain is "what makes us who we are." While writing a research paper for that class, he was drawn to Daniel Johnston's work on the hippocampus, a brain structure that creates and houses memories.
"Memory is absolutely fascinating," says Spruston, "you can see something that will remind you of an event and a flood of memories come back to you even though you haven't thought of this event in 10 or 20 years. How does that happen?"
Spruston would go on to do his doctoral work with Johnston at Baylor College of Medicine, studying the neurons of the hippocampus. And Spruston has continued to investigate the hippocampus ever since, including his 16 years on the faculty at Northwestern University.
Now a laboratory head at the Janelia Research Campus, Spruston deciphers how individual neurons in the hippocampus, in particular a common type called pyramidal neurons, integrate and compute the inputs they receive to produce an outgoing signal. A typical pyramidal cell has a complex dendritic tree with numerous branches. Those dendrites receive roughly 30,000 excitatory synaptic inputs—connections from other cells—and about 5,000 inhibitory synaptic inputs. These inputs to the dendritic tree determine whether a neuron fires off an action potential, the electrical signal that travels down the cell's axon to relay the signal on to other neurons or release neurotransmitter chemicals in the brain.
Spruston's group and others have shown that it's not a simple calculation of summing up the excitatory and inhibitory signals to dendrites. Rather, it's the biophysics underlying those integrated signals that determines the outcome.
"The simplest possible model is that the current flowing at any synapse is funneled passively down the dendrites to the cell body, but the reality is much more complicated than that," he says. For instance, he and others in the field have shown that the distribution of voltage-gated ion channels in dendrites as well as the branching pattern of the dendritic tree influence the way the neuron integrates synaptic inputs.
Spruston uses an analogy to the electoral college model of elections, in which each of the 35,000 synapses has a vote and each branch of the tree can be thought of as a local election. The local outcome is then reported to a central voting office (the axon), like a state precinct. The tally of local elections, but not the individual votes, determines whether the state will cast its vote in favor of firing the neuron. In the case of neurons, he explains, "the branches might not all have equal weight; the votes of some branches might be weighted more heavily than others."
Spruston's lab is beginning to study other types of hippocampal neurons, including inhibitory interneurons. Working with those cells, his group discovered a new way for action potentials to fire by directly arising from the cell's axon—the "wire" that propagates action potentials—without any inputs from the dendrites. By continuously stimulating an axon to fire for about a minute, the scientists caused it to "pop into persistent firing mode" for about another minute without further stimulation.
Although it's unclear what such axon-initiated firing means for an animal's memory, Spruston notes that nature rarely provides cellular mechanisms for no reason. "We're going to discover new types of signaling going on in these neurons because we've identified a completely new form of action potential firing. If we do the right experiments, we'll figure it out, and there's going to be some cool new biology there."