That new neurons are born in the adult mammalian brain is by now a foregone conclusion. But whether this phenomenon is important for the brain’s function is still hotly debated. A paper in yesterday’s Nature suggests that it is. Ronald Evans and colleagues at The Salk Institute for Biological Studies, La Jolla, California, report that relieving mice of the neural stem cell protein TLX not only eliminates neurogenesis but also weakens the rodents’ spatial memory skills. Oddly enough, the mice had no problem learning to associate an audible tone with a mild foot shock. The results suggest that while neurogenesis can contribute to learning and memory, its effects are not broad but may be limited to specific neural circuits or learning paradigms.

The potential for exploiting neurogenesis to help maintain or improve declining learning and memory skills makes this phenomenon of particular interest to AD researchers. In fact, exercise, which may be protective for AD (see Teri et al., 2003), is known to both induce adult neurogenesis in mice and reduce amyloid burden in models of the disease (see ARF related news story). But whether exercise-induced neurogenesis, or neurogenesis in general, contributes to learning and memory has been controversial. Some studies suggest neurogenesis has no effect on spatial learning but boosts contextual fear conditioning (see Saxe et al., 2006), while others suggest that it has no effect on learning at all (see ARF related news story).

One of the problems with teasing out a role of neurogenesis in adult animals is finding a mild means to prevent the process from happening that does not otherwise impair the mice. Researchers have alternatively used irradiation to ablate neural stem cells, or mitotic inhibitors to prevent them from dividing. Neither is very gentle to mice and may lead to collateral damage. The approach taken here by Evans and colleagues is more direct. First author Chun-Li Zhang and colleagues used genetic recombination to conditionally knock out the TLX gene in adult forebrain.

TLX is an orphan nuclear receptor. In other words, its biological ligand is unknown and its exact function is a bit of a mystery. But the protein is essential for normal neural development because deleting the gene leads to a small brain, blindness, and aggressive behavior. Neural stem cells express the protein, which led Evans and colleagues to explore its function in neurogenesis.

Zhang and colleagues made a “floxable” TLX mouse by flanking the TLX gene with two loxP motifs, which are sites where the Cre recombinase can attack the DNA and remove the intervening sequences. These Tlxf mice also express a Cre recombinase regulated by an attached, though modified, estrogen receptor (ER), which is sensitive to the ER ligand tamoxifen. Giving tamoxifen to adult mice induces Cre activity and excision of the Tlx gene. Zhang and colleagues found that in cell culture, floxing the Tlx gene reduced proliferation of neural stem cells (NSCs) by 80 percent.

In living mice, tamoxifen administration reduced cell proliferation by 65 percent, as judged by analysis of bromodeoxyuridine (BrdU), a thymidine analog that is incorporated into DNA. Exercise-induced increases in dividing cells were also down by about half, suggesting perhaps that the non-TLX NSC pool can still respond to exercise but not fully compensate for the missing stem cells. As for learning and memory, the tamoxifen-treated mice performed as well as controls in a test of contextual fear conditioning. However, they floundered in the Morris water maze test of spatial memory. Over 8 days of training, the floxed mice matched controls on days 1 to 4, but on days 5 and 6 they were taking significantly longer to find the hidden platform. “Indeed, probe trials carried out 12 h after the fifth day session demonstrated a major deficiency in short-term memory for these mutant mice, measured by time in the target zone or platform crossings,” write the authors.

The effects do not seem to be due to lack of motivation, visual or locomotor problems since the floxed mice performed just as well as controls in a version of the test where the platform was not hidden. But there is a possibility that the effect might be due to loss of TLX in mature neurons. To address this, the researchers determined that TLX is expressed primarily in excitatory neurons. Then they repeated the floxing, but this time they expressed Cre under the control of the CaM kinase II promoter, which is turned on in mature excitatory neurons. These animals performed as well as controls in the Morris water maze and their NSC complement was unaffected. “These data indicated that most, if not all, neuronal ‘non-neurogenic’ TLX is not required for spatial learning and memory,” write the authors.

How do these findings fit in with other reports that neurogenesis has no effect on spatial learning and memory in mice (see Saxe et al., 2006 and Raber et al., 2004)? The discrepancy might be related to the experimental protocol. In other studies, researchers first trained the mice in a visible platform test before switching to the hidden platform. When Zhang and colleagues tried this approach, the Tlx-floxed mice no longer had spatial learning problems in the hidden platform test. This suggests that there are subtleties to studying the effects of neurogenesis that may not be fully appreciated. As for the contextual fear conditioning, which has been linked previously to neurogenesis, Zhang and colleagues suggest that those effects may be “attributable to differences in species or to the side effects of the methods used to knock down neurogenesis.”

How any of these studies relate to human neurogenesis and learning and memory is not at all clear. One thing that is clear, however, is that the debate over the role of neurogenesis is bound to continue. On that note, “our development of a new, inducible knockout mouse model provides a new and powerful tool to understand better the role of adult neurogenesis in normal behavior and disease, and should deepen our insight into which of the many facets of brain function are impacted by the progressive and dynamic cell population,” write the authors.—Tom Fagan


Make a Comment

To make a comment you must login or register.

Comments on News and Primary Papers

  1. This is a very elegant study supporting a role for neurogenesis in water maze performance. This study re-emphasizes that the design of the water maze test might modulate the ability to detect this role. For example, in this study the effects of reduced neurogenesis on water maze performance were not detected when the mice were first trained to locate a visible target (task learning). However, it might be more complex, and what might also be very important is whether there is only one probe trial (no target) at the end of the hidden water maze training or whether there are multiple probe trials during hidden water maze training. In addition, there might be a species difference. Training the animals first to locate a visible platform and using one probe trial only at the end of the hidden water maze training, we did not see impairments in water maze performance in 2-month-old C57Bl6/J wild-type mice X-irradiated at 10 Gy and tested 3 months later (Raber et al., 2004), although using a similar water maze paradigm (visible platform training; hidden platform training; probe trial), we did see impairments in water maze performance in gerbils: irradiated gerbils spent less time in the target quadrant in the probe trial than vehicle controls (Raber et al., 2004). This does not mean that there were no detrimental effects of cranial irradiation on spatial learning and memory in the mouse study, as such effects were detected in the Barnes maze test.

    Importantly, training mice first to locate a visible platform followed by training mice to locate a hidden platform and performing a probe trial at the end of each day of hidden platform training, we did see impairments in spatial memory retention in the probe trial in C57Bl6/J wild-type mice irradiated at 5 Gy at 3 weeks of age and behaviorally tested 3 months later (Rola et al., 2004) and in human ApoE transgenic female mice irradiated at 10 Gy at 2 months of age and behaviorally tested 3 months later (Villasana et al., 2006). Thus, the impairments in water maze performance might be overcome with extended training. Although not first training the mice to locate a visible platform or including additional probe trials during hidden water maze training might both make the water maze test more challenging and allow effects of reduced neurogenesis on water maze performance to be detected, there is an important difference. Training to locate a visible platform is not hippocampus-dependent and does not require hippocampal neurogenesis. In addition, without first training the mice to locate a visible platform, it is hard to distinguish potential differences in task learning (training the animals that they need to swim to the target) from differences in spatial learning and memory. Clearly, more work is needed to fully understand the exact role of neurogenesis in spatial learning and memory.

    View all comments by Jacob Raber


News Citations

  1. Sorrento: More Fun, Less Amyloid for Transgenic Mice
  2. Neurogenesis Not Needed for Environmental Enrichment Effects

Paper Citations

  1. . Exercise plus behavioral management in patients with Alzheimer disease: a randomized controlled trial. JAMA. 2003 Oct 15;290(15):2015-22. PubMed.
  2. . Ablation of hippocampal neurogenesis impairs contextual fear conditioning and synaptic plasticity in the dentate gyrus. Proc Natl Acad Sci U S A. 2006 Nov 14;103(46):17501-6. PubMed.
  3. . Radiation-induced cognitive impairments are associated with changes in indicators of hippocampal neurogenesis. Radiat Res. 2004 Jul;162(1):39-47. PubMed.

Further Reading

Primary Papers

  1. . A role for adult TLX-positive neural stem cells in learning and behaviour. Nature. 2008 Feb 21;451(7181):1004-7. PubMed.