Hippocampus and Spatial Memory—New Neurons Fit In, Old Ideas Are Challenged
Quick Links
A new kid on the block often injects vim and vigor into stale routines. New neurons in old brains might do the same. In the February 4 Nature Neuroscience online, Paul Frankland and colleagues at the Hospital for Sick Children and the University of Toronto, Canada, report that in adult mice, new neurons not only insinuate themselves into functional memory circuits in the hippocampus, but that they are much more likely to do so than older neurons. The finding hints that there may be more to adult neurogenesis than meets the eye. A different perspective on the hippocampus is given in this week’s PNAS online. Larry Squire, University of San Diego, California, and colleagues refute an old idea that the hippocampus is specifically required for remembering objects seen from a different viewpoint than before. That study suggests that degeneration of the hippocampus, such as might occur in Alzheimer disease (AD), does not impair viewpoint-independent memory any more than memory in general.
The hippocampus becomes dysfunctional early in AD, and this probably accounts for some of the earliest memory losses associated with the disease. The realization that adult neurogenesis occurs in mammalian hippocampus, including human (see Eriksson et al., 1998 and ARF related news story), raised the tantalizing hope that a way could be found to persuade new neurons to take over from their embattled brethren. But a fundamental question to be answered first is whether these new neurons actually integrate themselves into the neural circuitry? The science on that is conflicting. Some experiments designed to block adult neurogenesis suggest new neurons do have a functional role, while other, similarly designed studies suggest they do not. Frankland and colleagues approached this problem in a different way. Rather than block neurogenesis, they gave it free reign and then asked if the new neurons express immediate-early genes that are turned on when new memories are formed. The answer in this study seems to be a resounding yes.
Joint first authors Nohjin Kee, Cátia Teixeira and colleagues turned to the immediate-early gene c-fos to see if new neurons are functionally active. C-fos is turned on in the dentate gyrus of the hippocampus during the formation of new memories, such as when mice are trained in the Morris water maze, a commonly used tool to test rodent learning and memory. Kee and colleagues administered the proliferation marker bromodeoxyuridine (BrdU) to mice and then trained groups of animals in the water maze beginning between 1 to 8 weeks after BrdU treatment. C-fos and BrdU were then measured in all mice at 10 weeks. The researchers found not only that BrdU-positive cells do express c-fos, suggesting that they are part of the circuits induced by the training, but also that the recruitment depends on the age of the new cells. As the new cells matured, up to about 6 weeks, they were much more likely to express c-fos and become integrated into the circuitry. “There are probably two stages to the integration,” Frankland told ARF. “First the cells must become part of the hippocampus—that occurs in the 2-3 week range—then they become part of the memory.”
Since the new neurons make up only about one percent of the total number of neurons in the dentate gyrus, the physiological significance of their integration into memory circuits is unclear. Frankland suggests that the new neurons may be involved in temporal coding, or in transient expression of memory, or that they may even increase the computation power by protecting memories from interference, but he admits that this is all speculative at present.
What is clear is that the new neurons seem to respond better to the challenges of spatial learning than old neurons. Normalizing for cell count, the researchers found that new neurons were three times more likely to express c-fos after the training period ended. Furthermore, in mice deficient in spatial learning owing to a calmodulin kinase II mutation, adult neurogenesis proceeded apace even though c-fos was not turned on in the new neurons after water maze training. That finding adds weight to the idea that induction of c-fos in new neurons is directly related to the animals’ experience, not an inherent consequence of neurogenesis. “Learning seems to be a prerequisite for neurons to be integrated into the circuit,” said Frankland.
The potential of exploiting adult neurogenesis to help compensate for neurodegeneration in diseases like Alzheimer’s or Parkinson’s is generating a lot of interest in the field, though whether that will ever pan out remains to be seen. “These results are relevant in broad terms because one of the fundamental steps in developing neuron replacement treatments is understanding how new neurons become integrated into circuits, and particularly from an Alzheimer’s perspective, how they become integrated into memory circuits,” said Frankland. He cautions, however, that old memories that have been lost may never be retrievable.
The exact role of the human hippocampus in memory remains a subject of debate. Advancements have turned in large part on the study of people who have sustained damage to this area of their brain. Squire and colleagues recruited the help of such patients to address the role played by the hippocampus in “allocentric” memory, which is invoked to remember the location of objects after the viewpoint has shifted. In this case, the viewpoint shift was achieved by rotating a virtual medieval town square.
In the PNAS paper, first author Yael Shrager and colleagues asked six amnesic patients to view the virtual town as images popped briefly into view. The patients tried to remember the locations of the images (one to five, depending on the test), and then the virtual town was rotated. Images were shown again and the patients had to decide whether those images appeared in the same place as before. Patients performed normally when they had to remember one or two images, but when asked to remember four images, their performance declined no matter by how much the viewpoint had shifted (not at all, 55, 85, or 140 degrees). “The results suggest that the hippocampus is not dedicated to, or especially important for, viewpoint-independent (or allocentric) memory. Rather, the hippocampus is generally important for declarative memory, and viewpoint-independent memory is one example of that broad category,” write the authors.—Tom Fagan
References
News Citations
Paper Citations
- Eriksson PS, Perfilieva E, Björk-Eriksson T, Alborn AM, Nordborg C, Peterson DA, Gage FH. Neurogenesis in the adult human hippocampus. Nat Med. 1998 Nov;4(11):1313-7. PubMed.
Further Reading
Primary Papers
- Kee N, Teixeira CM, Wang AH, Frankland PW. Preferential incorporation of adult-generated granule cells into spatial memory networks in the dentate gyrus. Nat Neurosci. 2007 Mar;10(3):355-62. PubMed.
- Shrager Y, Bayley PJ, Bontempi B, Hopkins RO, Squire LR. Spatial memory and the human hippocampus. Proc Natl Acad Sci U S A. 2007 Feb 20;104(8):2961-6. PubMed.
Annotate
To make an annotation you must Login or Register.
Comments
Baylor College of Medicine
This uses a clever experimental design to ask whether newborn hippocampal neurons are activated by learning and memory, and if so, how their involvement compares to that of older, established granule cells. The authors find that newborn hippocampal neurons are preferentially activated by spatial learning tasks compared to older granule cells, but that this preferential activation only occurs after the new neurons reach 4-6 weeks of age. This timing is remarkably consistent with previous studies from the Gage lab and others showing that newborn neurons take several weeks to reach their target field and develop synaptic spines, and require up to 5 weeks to become functionally mature.
The preferential activation of this maturing newborn population of neurons suggests that these cells may be especially suited for the acquisition and/or storage of new information. However, the water maze task used to induce learning and memory in this study has been the one test that has consistently shown little impact of treatments designed to inhibit neurogenesis. The lack of effect on water maze performance has been interpreted to suggest that newborn neurons are not required for spatial tasks. The data presented here indicates that, in fact, when this pool of newborn neurons is available, it is preferentially activated over older granule cells during the recall of spatial information.
The apparent contradiction between the blocking experiments and the endogenous activation is intriguing. It may suggest that either treatment used to block neurogenesis in past work (MAM, X-irradiation, and targeted ablation) does not completely suppress the production of new neurons, or that memories created from spatial tasks are especially resilient to degradation of the memory network. Kee et al. suggest yet another consideration based on their findings. Studies designed to inhibit neurogenesis generally test its effect within a few weeks of treatment. As shown by Kee et al., greatest incorporation of newborn neurons into memory networks occurs much later, roughly 4-8 weeks after division. However, this explanation does not account for treatments based on X-irradiation, which are usually delivered several months before behavioral testing, leaving the exact role of newborn neurons in spatial learning unclear.
The role of newborn hippocampal neurons in learning and memory is especially relevant for Alzheimer disease, as several groups have shown decreased neurogenesis in FAD transgenic APP and PS1 mice. Given that Kee et al. show these newborn neurons to be preferentially activated during learning and memory, the diminishment of this population in mouse models for AD may contribute to the cognitive deficits described in many APP and PS1 mice. However, directly linking deficits in neurogenesis with functional decline in Alzheimer disease is a tall order that will require future study.
View all comments by Joanna JankowskyMake a Comment
To make a comment you must login or register.