When it comes to spatial learning, the hippocampus and cortex may be less crucial than previously thought, a new preprint suggests. Some mice born without a hippocampus and most of the neocortex can still learn and remember their way through a complex maze.
It remains unclear whether typical mice rely on the same circuitry to learn the maze. But a separate preprint by an independent team of researchers shows that typical mice carry spatial information throughout the brain, not just in the hippocampus. Both preprints were posted on bioRxiv in September.
The findings “challenge our preconceptions about, on the one hand, the importance of the hippocampus and the cortex, and on the other, how specialized different parts of the brain are generally,” says Kevin Mitchell, associate professor of genetics and neuroscience at Trinity College Dublin, who was not involved in either study. But the hippocampus-free mice results do not necessarily unseat the hippocampus as the brain’s primary center for navigation because the mice developed without it, and their brains might have compensated for its absence, Mitchell says. “Clearly, there’s enough plasticity in the developing brain to accommodate.”
The results in typical mice add to mounting evidence that signals once assigned to specialized regions turn up across the brain. Decision-making engages neurons brain-wide, and the visual and olfactory cortices contain spatial signals, according to previous studies.
“Now navigation is everywhere; decision-making is everywhere,” says Enny van Beest, a senior research fellow in Matteo Carandini’s lab at University College London who worked on one of the new studies. This raises several questions, she says: “Which of these brain-wide networks are used in different memory requirements? Are they overlapping? Are they different subpopulations but just brain-wide?”
I
n the other study, researchers used mice that lack the PALS1 gene in the cells that give rise to the hippocampus and most of the cortex, which prevents these two brain structures from forming. These acortical mice struggled at first to navigate the Manhattan Maze, a grid-like arena in which they receive a small sip of water at certain locations. They appeared anxious, scanned their surroundings repetitively and were about two to three times slower than control mice to obtain the first few rewards, says study investigator Jieyu Zheng, a graduate student in Markus Meister’s lab at the California Institute of Technology.Past that stage, however, their learning, retention and generalization were largely similar to those of control mice, the preprint shows. Both sets of mice found a highly efficient route to the reward within 10 successful attempts and remembered a nine-turn route three to four weeks later. When the maze was reconfigured, both groups got off to a much faster start than they had when learning their first maze.
“Obviously there is a system for learning and generalization that exists outside of cortex and hippocampus,” says Meister, professor of biological sciences. “It is possible to perform navigation without a six-layered cortex.”
Whether wildtype mice also rely on subcortical circuits to solve the maze is an open question. But “it would be surprising if two entirely different brain systems were able to solve the same problem in the same way,” Meister says.
The result seems at odds with decades of work showing that removing or lesioning the hippocampus in adult rodents impairs spatial navigation. But those lesions tend to delay learning rather than abolish it, Zheng and Meister say. “If you really look at that literature, it’s not as clear that these mice can’t navigate as it seems,” Zheng says.
And growing up without a cortex or hippocampus could severely rewire the brain throughout development, meaning some areas might take on hippocampus-like traits they wouldn’t normally have, Mitchell says.
The piriform cortex could help the animals compensate, says Lisa Giocomo, professor of neurobiology at Stanford University, who was not involved in the work. It has “the same laminar structure” as the hippocampus and is known for olfactory place coding, she says. “The anatomy is there to perform the same types of computations.”
She says she is skeptical that the hippocampus isn’t the primary locus of spatial navigation, especially during challenging navigation tasks. She adds that she would like to see acortical mice attempt tasks that are thought to rely “very critically” on the hippocampus, such as contextual fear conditioning.
The mice were raised in standard cages and tested within hours of first encountering the maze, Zheng and Meister say, leaving little opportunity for the brain to adapt to the task itself. And subcortical circuits were guiding animals through the world long before the cortex expanded, Mitchell says. “Look at a lamprey. It doesn’t have much of a pallium, but it gets around, right? It can learn.”
Some mice born without a cortex or hippocampus never learned the task. Most of those that failed never explored the full maze, suggesting that the hippocampus may help kick off exploration, Zheng says.
F
or the other preprint, van Beest and her colleagues recorded from more than 20,000 neurons in the visual, motor and somatosensory cortices, hippocampal formation, striatum, thalamus and midbrain as head-fixed mice ran on a wheel through a linear virtual corridor. The corridor’s two halves looked exactly alike; a mouse couldn’t rely on sensory cues to tell which half it was in. The mice received a small sip of water in only one part of the corridor on most trials.Neurons that responded differently in the two halves, even after the team accounted for running speed and reward, were deemed spatial.
Spatial neurons appeared in every region recorded, and a computational model could predict the mouse’s position from each of them. The hippocampal formation had roughly the same proportion of cells that carried spatial information as each other area did, but it covered the corridor slightly more uniformly. It encoded position less precisely than the visual cortex did, however.
The setup may shortchange the hippocampus, says Carandini, professor of visual neuroscience at University College London. The track is one-dimensional, and the animal’s head doesn’t move, so there are no vestibular signals. “If we had a two-dimensional environment, it’s very likely that the hippocampus would have been special,” he says.
Decoding a signal does not necessarily mean that the brain uses it. “What they’ve shown is that as an experimenter, you could decode location from different parts of the brain,” Mitchell says. But one could argue that such activity isn’t a representation of space unless another brain area reads it out as one, he adds. The hippocampus could be broadcasting the information to other regions of the brain.
The acortical mice, however, have no hippocampus to broadcast spatial information, so it must come from somewhere else, and recording from those animals could reveal where that is, Meister says. The superior colliculus, basal ganglia and cerebellum are potential candidates, he adds.
Investigators on both preprints say the next step is to silence the hippocampus transiently, to test whether typical mice rely on other brain regions to remember spatial information, and whether these brain regions are simply receiving input from the hippocampus.
