The brain pumps out fresh neurons throughout its lifetime. Inspired by this decade-old discovery (Eriksson et al., 1998 and see ARF related news story), scientists have sought ways to harness these new cells to fill gaps left by their dying kin in the wake of neurodegenerative disease. Such efforts seemed to get the go-ahead from rodent studies suggesting that newborn neurons do appear important for cognition. In those investigations, suppression of neurogenesis brought impairment in various hippocampus-dependent memory tasks (Winocur et al., 2006), including long-term spatial memory (Snyder et al., 2005).

More recent studies, such as one led by Paul Frankland at the Hospital for Sick Children in Toronto, Canada, have showed that the young neurons integrate themselves into functional neural circuits when new memories are formed and that they seem even more likely than older neurons to do so (Kee et al., 2007 and see ARF related news story). In that analysis, the scientists trained mice in the Morris water maze and saw that some of their new neurons expressed c-fos, an immediate-early gene that gets switched on during memory formation. What the current study does, Frankland said, “is try to characterize in more detail the particular role that these [new] neurons have.”

Using an experimental approach similar to Frankland’s, Rampon and colleagues set out to test whether new neurons that get recruited into functional networks during memory acquisition support retrieval of that same memory a month later. “There has never been direct evidence that specific neurons present at the time of training would be activated at the time of recall,” Rampon told ARF.

To address this, first author Stephanie Trouche and colleagues injected mice with the proliferation marker bromodeoxyuridine (BrdU) and, nine days later, trained them to find a hidden platform in a circular pool. They chose the day 9 timepoint because prior studies had suggested that young neurons are most sensitive to surrounding network activity at this immature stage, Rampon said. One month after this training, the researchers exposed the mice to the identical spatial task and checked to see whether any of the BrdU-labeled neurons got recruited to neural circuits encoding the learned behavior. As readout for integration into functional circuits, her team looked by immunofluorescence for cells triply labeled with BrdU, the neuron-specific nuclear protein NeuN, and an activity-dependent protein (either c-fos or another marker, Zif268) during various behavioral conditions.

The researchers found that 4.1 percent of the BrdU-positive neurons on hand for the initial water maze training expressed Zif268 when the mice faced the same spatial challenge one month later. In a separate analysis, they determined that the proportion of Zif268-expressing cells in the dentate gyrus (a prime site for neurogenesis in the brain) was at least four times lower in control mice that sat in their cages during the water maze training. Animals given a more rigorous retraining regimen—consisting of nine additional trials instead of just one—had a much larger proportion (11.2 percent) of BrdU-positive neurons expressing Zif268. This suggests that new hippocampal cells may “contribute to the processes underlying updating and strengthening of remote spatial memory,” the authors write.

When the mice faced an adjusted water maze task, in which the hidden platform was in the quadrant opposite to that used in their training, only 3.6 percent of the new neurons turned on Zif268. In addition, in mice that were not trained until day 39, only 3.9 percent of the BrdU new neurons expressed Zif268. All told, the findings suggest that “if they are present at the time of training, they are more prone to be recruited later when the same situation is encountered and the same memory is recalled,” said Rampon.

Interestingly, when the researchers used the Morris water maze protocol in which the hidden platform present for initial training was removed for the probe test a month later, a mere 0.9 percent of the BrdU-positive neurons expressed Zif268 (compared to 4.1 percent when the platform remained in the same place for both initial training and retesting). Rampon suggests that this discrepancy may result from activation of different neural circuits in the mismatch condition. “You’re not looking at the same networks,” she said. “The neurons that become activated are not the ones storing the [initial memory]. They are the ones starting to encode a new memory (i.e., the platform is not there anymore, let’s look for it somewhere else).”

Frankland’s 2007 study, which used the water maze for spatial learning in the mismatch situation, suggested that the integration of new neurons into functional memory circuits depends on their maturation status. “It’s not until four or five weeks of age before they start doing something,” Frankland told ARF. On the other hand, he said Rampon’s data “suggests that something important happens when the neurons are relatively immature. Some general activation of this pool of immature neurons, even though they haven't fully established connections, does lead to their integration into circuits supporting spatial memory.”

Whereas Rampon’s study suggests a role for developing hippocampal neurons in updating and strengthening memories, a recent paper from researchers led by Fred Gage at the Salk Institute for Biological Studies, La Jolla, also proposed a different function for such cells—one that ascribes time quality to memories (Aimone et al., 2009 and see ARF related news story). Like Rampon’s, that study, based on a computational model for adult neurogenesis, suggested that the function was restricted to a critical time window early in the development of these new cells. (For Gage’s recent review of neurogenesis mechanisms and functional implications, see Zhao et al., 2008.)

That newly formed neurons seem to play key roles in memory formation and retrieval suggests to Orly Lazarov, University of Illinois at Chicago, that our brains may be more adaptable than previously thought. “We should see neurogenesis in a much wider context than just addition of a few neurons to the system. It's much more than that,” she told ARF. “This process is highly responsive to stimulation, and the nature of those newly formed neurons and the nature of change depends on the context and the stimulus.” (See additional comment below.) Lazarov, together with Sangram Sisodia at the University of Chicago, Illinois, has shown that environmental enrichment, which can drive neurogenesis, also has beneficial effects in mouse models of Alzheimer disease (see ARF related news story). To make matters more complex, they have also shown that presenilin mutations, which drive familial forms of AD, impact neurogenesis as well (see ARF related news story). The links between neurogenesis, genetic and environmental factors, and AD remain somewhat elusive.

While the ins and outs of neurogenesis are being worked out, other scientists are plugging away at an alternative strategy for replenishing nerve cells lost to neurodegenerative disease—by devising better ways to grow new ones in the lab. The idea here is that if cells from a readily available tissue, such as skin, could be reprogrammed to a developmentally primitive state, they could be induced to differentiate into functional cells to replace dying neurons in people with Alzheimer’s, Parkinson’s, and related conditions. A wave of recent developments—using various methods to shuttle into target cells the same four pluripotency factors (Oct4, Sox2, c-Myc, and Klf4)—has made generation of these so-called induced pluripotent stem (iPS) cells easier and safer (see ARF related news story and ARF news story). The study by Thomson and colleagues that appears in this week’s ScienceXpress represents the latest contribution on this front.

First author Junying Yu and colleagues introduced six reprogramming genes (the four listed above, plus Nanog and Lin28) into human foreskin fibroblasts using a non-integrating episomal vector (oriP/EBNA1) derived from the Epstein-Barr virus. Though the reprogramming efficiency was low (~three to six colonies per million input cells), the technique holds promise because it leaves behind no trace of vector sequence in the host cell DNA. Capitalizing on a convenient feature of the oriP/EBNA1 episomal vectors—their disappearance from dividing cells grown without selection—the method seems to trump two recently published protocols. One of those removed integrated transgenes but left residual vector sequences in the host genome (Kaji et al., 2009). The second cleanly excised vector and transgene sequences but was only demonstrated in mouse cells (Woltjen et al., 2009).

Keisuke Kaji of the University of Edinburgh, U.K., who was lead author on one of those recent studies, has a few reservations about the new method (see full comment below), but nevertheless calls it “a great advance in reprogramming technology.” Andras Nagy of Mount Sinai Hospital in Toronto, who was lead investigator on the mouse cell study, sees the latest work as continued progress in the right direction. The methods “all have their own pros and cons and will go through further improvement in the near future,” he wrote in an e-mail to ARF (see full comment below). “At this point, there is no way to see which one is going to be superior over the others.”—Esther Landhuis.

References:
Trouche S, Bontempi B, Roullet P, Rampon C. Recruitment of adult-generated neurons into functional hippocampal networks contributes to updating and strengthening of spatial memory. PNAS Early Edition. March 2009. Abstract

Yu J, Hu K, Smuga-Otto K, Tian S, Stewart R, Slukvin II, Thomson JA. Human induced pluripotent stem cells free of vector and transgene sequences. 26 March 2009. ScienceXpress. Abstract