It’s fun to zoom out and get the big picture sometimes. This is one such picture I took long ago when I wanted to see if staining for zinc transporter 3 effectively labels the mossy fiber axons of the dentate gyrus. You can see by the perfect overlap with calbindin that it does the job, though the staining wasn’t quite as bright and obvious as calbindin. The abundance of zinc in mossy fiber axons is one of the peculiarities of the DG and it underlies numerous synaptic properties of DG neurons.

I think the goal was to build on previous work by Lipp, Ramirez-Amaya, and Routtenberg showing that spatial learning causes “sprouting” of mossy fibers, though when I found out that this phenomenon does not occur in mice the project was aborted.

But what else can you see in this picture?

• clear differential expression of calbindin: DG (lots) > CA1 > CA3 (none), and a scattering of strongly-positive interneurons (e.g. 5 cells where CA3 and CA1 meet)
  • in CA1 you can see calbindin is expressed only in the lower band of cells (see hi res photo if needed; there is a ref for this, somewhere)
• a thin band of calbindin-positive fibers crossing the corpus callosum (CC)
• A small group of cells that are not contacted by the calbindin-positive mossy fiber axons (i.e. beyond CA3) yet do not express somatic calbindin (as seen in CA1). I’m guessing this may be mysterious and ambiguous field CA2.

A fundamental property of the hippocampus is its ability to rapidly encode memories while simultaneously keeping them distinct. Recording from hippocampal neurons one can clearly see that different populations of neurons are active as a rat explores two environments. This is thought to be one mechanism by which information is kept distinct in the brain.

For the last 15-20 years it has been thought that the dentate gyrus (DG), a major subfield of the hippocampus, serves to take small changes in incoming sensory information and orthogonalize them (i.e. make them more different). This idea was built in part on the fact that there are many more DG neurons than upstream cortical neurons. Thus, the DG could use completely different populations of neurons to represent different sets of incoming information and then pass on these representations to CA3, which may bind them into coherent events/memories (the interconnectedness of CA3 neurons, via “recurrent collatorals”, is thought to be a mechanism by which the different components of a memory are bound together).

However, a “problem” arose when Leutgeb et al. found that it is always the same population of dentate granule neurons (~1% of the total population) that are active as an animal explores different environments, even very different ones. This was a bit of a surprise. Still consistent with the proposed role of the DG in orthogonalizing information, however, was the fact that the DG neurons fired (i.e. generated action potentials, which transmit information from neuron to neuron) at different rates/frequencies in the different environments. Thus, changes in sensory information were represented by changes in patterns of activity within the same population of cells, not by recruiting different populations of cells. This is but one study – the question of how the DG encodes and extracts information is far from settled (e.g. what are the other 99% of granule neurons doing? Surely there is a situation in which they are active, no?). But the findings were robust and raise many questions, namely: How does the same population of DG neurons activate different populations of downstream CA3 neurons, during different experiences?

Until now I had been in denial, fixated on trying to understand what types of behavioral experiences might activate different populations of dentate gyrus neurons. But maybe now it’s time to face the data.

The consensus, both in vitro (e.g. here and here) and in vivo (here), seems to be that if DG neurons are sufficiently active they can reliably activate CA3 neurons. Can a single population of DG neurons account for the amount of CA3 activity seen in the behaving animal? Well, 1% activation of the total DG population (1 million neurons) is 10 000 DG neurons. Each DG neuron contacts about 10 CA3 neurons. So if all active DG neurons activated all their downstream targets, then you’d expect about 100 000 active CA3 neurons – a third of the population. Indeed, about 20% of CA3 neurons are active when a rat explores a novel environment. So it’s possible. But it’s probably unlikely.

One reason it’s unlikely is that it doesn’t explain how different populations of CA3 neurons are activated by different experiences if it is the same population of DG neurons that are always driving them. In other words, since DG neurons are relatively hard-wired to CA3 neurons, how could a given DG neuron activate a CA3 neuron under some conditions and not others? One answer is that maybe it doesn’t – quite a while ago, McNaughton et al. showed that, even when the DG is lesioned, CA3 neurons are still able to selectively encode spatial locations as a rat traverses the environment, probably due to direct inputs from the cortex. And so perhaps the primary function of the DG is not to selectively activate different CA3 populations. However, the DG could certainly shape activity within CA3 or insert unique information into the CA3 network. How?

One possible mechanism, which may be dead obvious to electrophysiologists, is frequency itself. Leutgeb et al. found that frequency of activity is how DG neurons encode information and so frequency of activity may also be the way DG neurons transmit information to CA3 during different experiences.

It has been known for some time now that the output of DG neurons, the mossy fiber axons, show extraordinary frequency-dependent synaptic facilitation. Basically, as a DG neuron fires more action potentials over shorter periods of time, the amount of neurotransmitter it releases onto CA3 neurons increases (thereby increasing the likelihood a CA3 neuron will in turn fire action potentials and be recruited to participate in memory encoding). This means that at low firing rates, a DG neuron will activate some CA3 neurons and, at higher firing rates, it will recruit different or at least additional CA3 neurons.

Wouldn’t this cause a problem where, as DG firing rates increase, it is not different populations of CA3 neurons that become activated, but more populations? Well, it is known that some DG neurons increase their activity, and others decrease their activity, as an animal has different experiences, so the net activity in CA3 could remain constant, while still activating different CA3 populations. However, the DG-CA3 circuitry is certainly complicated enough to allow for other mechanisms. For example, while the dentate gyrus projects to CA3, and it is connections between these hippocampal regions that are thought to encode memories, DG neurons actually contact more inhibitory interneurons than CA3 neurons. Furthermore, there is a wide variety of synaptic connections between DG neurons and interneurons and these connections can be made weaker or stronger in a state- and frequency-dependent manner. Suffice it to say, by firing at different frequencies, it is plausible that a given DG neuron could activate different populations of interneurons, which in turn could inhibit different populations of downstream CA3 neurons, making them less likely to participate in memory encoding.

This ties in loosely to a peculiarity of the dentate gyrus that, until now, has just been a source of pretty histological images (to me) – the variability of calbindin expression in dentate gyrus neurons. Calbindin is a protein that binds calcium, it acts as a buffer, and gives DG neurons their property of facilitation (briefly: A single action potential in a DG neuron will travel down the axon and trigger the opening of calcium channels in the synaptic terminal at a CA3 neuron. Calcium is necessary for neurotransmitter release and subsequent activation of the CA3 neuron. Calbindin will bind this small amount of calcium, thereby preventing neurotransmitter release and CA3 activation. However, as the number and frequency of action potentials increases, calbindin will fail to effectively “mop up” the extra calcium and neurotransmission will proceed.). If you look at the picture at the top of this post, you can see that the amount of calbindin varies greatly in DG neurons. Immature DG neurons, identified by PSA-NCAM expression, are devoid of calbindin (arrows point to clear examples) and even when they are quite mature (10w of age) 40% will still be devoid of calbindin (see my data in this montage). Lastly, calbindin expression can be modified by experience. So the variable and modifiable expression of calbindin might be yet another mechanism by which DG neurons are capable of shaping activity in CA3 neurons. Indeed, at least one study, from Robert Sapolsky’s lab, has shown that genetically altering calbindin expression in the dentate gyrus dramatically influences DG-CA3 physiology and impairs memory.

Thanks to A.P. for posing the question.

Reference

Leutgeb JK, Leutgeb S, Moser MB, & Moser EI (2007). Pattern separation in the dentate gyrus and CA3 of the hippocampus. Science, 315 (5814), 961-6 PMID: 17303747

Neurons are not born with dendrites and spines – they are acquired during a developmental process that takes many weeks (see here & here). During early development, the pattern of formation of dendrites and spines are sculpted by experience, as might be expected if dendrites and spines are anatomical structures involved in processing and storing sensory information. While a body of work has emerged suggesting adult-born neurons are involved in memory and behavior, no one has yet investigated whether experience is capable of altering the dendritic development of these new neurons. This paper by Tronel et al. is therefore very important because it is the first to look at this phenomenon. They show a dramatic acceleration of dendritic development in response to learning, suggesting a potentially powerful role for new neurons in storing and processing information.

It has been 10+ years since Gould et al. and Kempermann et al. showed that learning and enriched environments can enhance the survival of new neurons. These findings are logical precursors to the current study since, if these new neurons have all the necessary components,  they suggest experience could add to the mnemonic functions of the hippocampus. But subsequent studies indicated that experience could also decrease the survival of new neurons. So perhaps structural changes to new neurons that are more relevant to learning might be worth investigating. For example, in many of my own experiments, I have failed to observe learning-induced changes in the number of new neurons but, if the number of dendrites or spines is increased, then there could still be an enhanced ability to process information. Or there could be the removal of some spines and the formation of others, suggesting a transformation in the type of information processed by new neurons. To get at these possibilities, Tronel et al. used doublecortin (DCX) staining and retroviral-GFP labeling to visualize the dendritic structure of newborn neurons in rats that had either remained in their cage (non-learners) or had learned a spatial memory task, the Morris water maze.

Since the authors had previously shown that water maze learning enhances the survival of 1-week-old cells, they first examined whether water maze learning would also alter the dendritic structure of this same population of neurons. Training rats for 6 days and examining new BrdU+/DCX+ neurons the following day (i.e. when new neurons were 14-days-old) they found that the dendritic length and the number of dendritic branches was doubled compared to rats that sat in their home cage.

More remarkable is the duration that the increased dendritic complexity persisted. To get at this question a GFP retrovirus was used to label new neurons born 1 week before learning, since DCX is eventually downregulated and cannot be used to examine dendritic morphology in neurons more than ~2 weeks old.  They found that even 3 months after learning, maze-trained rats had longer dendrites, more branch points, and more dendritic ends. The differences were not trivial either – maze-trained rats had ~70% increases for all of these measures. The number of spines (and therefore putative synapses) was also elevated, 3-fold, and the proportion of spines that showed a mature, mushroom-shaped morphology was 6-fold greater than naive, untrained rats. Since the dendritic morphology of developmentally-born hippocampal neurons can be altered by learning, physiological changes in hormones, and exercise, it is also worth noting that in this study learning did not affect the dendritic complexity of mature granule neurons (though spines were not analyzed in mature neurons and it is possible that learning caused retraction and formation of spines in mature neurons with no overall effect in spine numbers or morphology), suggesting adult-born neurons are particularly sensitive to learning-related activity.

They go on to show that these structural changes in adult-born neurons are even more pronounced when rats learn a more challenging version of the water maze task, where the spatial location of the escape platform moves on a daily basis. They also show these effects require NMDA receptors, which are required for many forms of hippocampal-dependent memory. These additional experiments are notable but it is the basic finding – the magnitude and duration of the structural changes – that is most interesting to me. Here are some of the reasons why:

• previous studies have suggested that adult-born neurons reach a plateau in their functional development by ~8+ weeks of age. These data suggest that new neurons still have a long way to go before they become fully mature.
• the 8w developmental plateau in earlier studies could be normal for animals that have not had any significant life experience (what does this mean when the majority of scientific studies of the brain use naive, deprived animals as models?)
• when experience accelerates the dendritic development of new neurons, are those neurons now less plastic and less likely to contribute to future behaviors? In trying to understand why some studies report behavior deficits after neurogenesis ablation whereas others do not, I’m imagining that 6 weeks of neurogenesis ablation could have major effects on behavior if older (>6w) adult-born neurons are less plastic, perhaps because experience (experimenter handling, group housing, previous learning) accelerated maturation in the way Tronel et al. report. In contrast, if animals have been deprived of learning experiences, 6 weeks of neurogenesis ablation might not have any effects on behavior, because older neurons are still relatively immature and able to compensate.
• depending on how you look at it, it is valid to wonder how a relatively small population of new neurons can be important for behavior. If you now consider the fact that 3-month-old cells still have significant amounts of untapped storage capacity, the cumulative numbers of new neurons generated over 3 months no longer seems so small and insignificant

Reference

Tronel S, Fabre A, Charrier V, Oliet SH, Gage FH, & Abrous DN (2010). Spatial learning sculpts the dendritic arbor of adult-born hippocampal neurons. Proceedings of the National Academy of Sciences of the United States of America, 107 (17), 7963-8 PMID: 20375283

One positive thing that came out of looking at these very immature neurons was that I got the chance to see several examples of pyknotic (dying) cells. Older, adult-born neurons also die, particularly after an experience (see here and here), but it’s infrequent and hard to visualize. However, a relatively large proportion of new neurons die within a few days of their birth making them easier to find – the cluster of cells shown below is an example that caught my attention.

You can clearly see two BrdU-labeled cells (in green; marked with arrowheads) that also express doublecortin (DCX; red). The blue stain, Hoechst, stains DNA allowing for the visualization of all cell nuclei. Collectively, these 3 stains tell us that the cells are 1-day-old (because BrdU was injected 1 day before brains were collected), that they’re neurons (because they express the immature marker DCX) and that they’re dying (because BrdU and Hoechst both label DNA and show that the DNA is condensed in a ball, as is typically seen when cells undergo pyknosis). The arrow points to a lucky, neighboring neuron that is not dying.

Why were these two cells born if they’re only going to die 24 hours later?

I can understand the speculation that neural activity influences the survival of more mature neurons in a “use it or lose it” manner – essentially, if a memory is stored in a young neuron there must be a mechanism to ensure that the neuron, and therefore the memory, survives. But is it possible that a similar mechanism also influences the survival of very immature neurons? It’s hard to imagine, since very young neurons do not have synapses and cannot participate in memory processing/storage. Consistent with this idea, Tashiro has shown that NMDA receptors (a synaptic ingredient essential for many forms of memory) regulate the survival of 2-3 week-old neurons, which are just beginning to form synapses, but not younger neurons that have not yet formed synapses. However, the possibility remains that learning could do something to these 1-day-old neurons – e.g. epigenetically imprint them – so that they have some sort of cellular memory that causes them to subsequently participate in certain behaviors but not others. Since information is typically thought to be stored at synapses, I can’t imagine that these memories could be terribly specific but they could bias a young neuron to be more involved in a general class of behavior (e.g. spatial memory vs. stress) that is associated with certain broad differences in activity (e.g. firing patterns, neuromodulators, hormones). It would be really cool if someone shows this.

Two recent papers have attracted a lot of media attention because they draw direct links between adult neurogenesis and behavioral disorders: Noonan et al. showed that rats lacking adult neurogenesis (stopped with irradiation) are more susceptible to cocaine addiction. Jin et al. showed that mice lacking adult neurogenesis (using a transgenic model) suffer greater infarct size and have more severe motor deficits after stroke.

While the papers themselves have important implications, what caught my attention was the angle taken by press releases: both articles studied the effects of reducing neurogenesis but the media focused on potential benefits of increasing neurogenesis. See speculation that antidepressants, by increasing neurogenesis, might be stroke-protective here. And, from Science Daily:

While the research specifically focused on what happens when neurogenesis is blocked, the scientists said the results suggest that increasing adult neurogenesis might be a potential way to combat drug addiction and relapse.

It may very well be the case that increasing neurogenesis is good in the same way decreasing neurogenesis is bad but it shouldn’t be assumed – maybe we have all the neurogenesis we need and, while completely arresting neurogenesis could be harmful, increasing neurogenesis beyond normal levels is just redundant.

Or, maybe the key is where you’re increasing or decreasing neurogenesis from:

In this model, assuming most of us have normal levels of neurogenesis, further increases will provide no benefit to behavioral performance, whether we’re talking resistance to addiction or recovery from stroke or anything else. However, decreases in the number of young neurons below a critical level would impair storing/processing of information, protection against stroke etc.

This speculation is under the assumption that we all have “normal” or sufficiently high levels of constitutive neurogenesis. However, in the case of addiction, covariables like stress, poor nutrition and narcotics themselves might all serve to reduce neurogenesis below healthy levels (see “unhealthy levels” in figure). In this case, increasing neurogenesis would cause dramatic improvements in behavior and is consistent with the authors’ speculation. For similar reasons, increasing neurogenesis might provide cognitive benefits to people in other situations where neurogenesis is known to be compromised: the aged, those experiencing chronic stress, patients undergoing cranial irradiation or chemotherapy.

It may be surprising that this idea is actually very hard to test, even in the laboratory. While many tools exist for reducing neurogenesis to unhealthy levels (irradiation, antimitotic drugs, genetics) there are no tools for selectively increasing neurogenesis beyond normal/healthy levels. Yet.

References

Noonan MA, Bulin SE, Fuller DC, & Eisch AJ (2010). Reduction of adult hippocampal neurogenesis confers vulnerability in an animal model of cocaine addiction. The Journal of neuroscience : the official journal of the Society for Neuroscience, 30 (1), 304-15 PMID: 20053911

Jin K, Wang X, Xie L, Mao XO, & Greenberg DA (2010). Transgenic ablation of doublecortin-expressing cells suppresses adult neurogenesis and worsens stroke outcome in mice. Proceedings of the National Academy of Sciences of the United States of America PMID: 20385829

I’ve been interested in new neuron function since 1999 and so I’m actually quite surprised I missed this study until so recently. In 1999 the neurogenesis literature was so scant that it was easy to know ALL of the studies, even the early Altman, Kaplan and Nottebohm studies from the 1960s through 1980s. Even studies that were not interesting were interesting, because there was nothing else to read! So, had I known about it back then, I would have been pretty interested in this study by Okano et al. if only for its focus on cell cycle markers. But I really would have been interested in it because it has a small functional experiment that was way ahead of it’s time:

(Figure Legend) E, Thymidine-labeled/Fos-negative cells detected in the subgranular region 3 d after receiving ³H-Thy and 3 hr after pentylenetetrazolinduced seizure activity. Note the presence of many Fos-positive cells in the adjacent granule cell layer. F, Thymidine-labeled/Fos-positive cells detected in the granule cell layer 4 weeks after receiving ³H-Thy and 3 hr after pentylenetetrazol-induced seizure activity. g, granule cell layer.

(Results) To determine if the newly differentiated neurons were functional, we examined the expression of Fos-IR in response to pentylenetetrazole-induced seizure activity. In the absence of induced seizures, very few (< 1%) ³H-Thy-labeled cells in the dentate gyrus were immunoreactive for Fos-IR at any of the six time points examined. However, within 3 hr following seizure activity, an induction of Fos-IR within 21 .O% (at 1 week post-thymidine injection) and 81.3% (at 4 weeks post-thymidine injection) of the ³H-Thy-labeled cells detected in the granule cell layer was observed (Fig. 3E,F), suggesting that these cells had formed functional connections.

Fos is one of a number of immediate-early genes (IEGs) that are expressed following synaptic activity. IEGs allow short-term changes in synaptic function to turn into long-term changes (which is thought to be required for short-term memory to transform into long-term memory). IEGs can therefore be used to identify neurons that are contributing to memory formation. Okano et al. noted that < 1% of new cells expressed Fos when rats were not stimulated overtly. It could have been that the rats’ experience prior to death was not memorable or it could have reflected the fact that only a fraction of hippocampal neurons are activated by normal experiences. To get around this they activated all neurons that were possibly “activateable”, by using a convulsant. Therefore, all neurons that had synapses expressed Fos and a timecourse of neuronal maturation, one that fits nicely into the subsequent literature, was obtained. When I say subsequent literature I’m referring to examples like the next study which used IEGs to characterize new neuron maturation, 10 years later. And also the many electrophysiological studies that emerged beginning 10-15 years later. It’s funny that none of them referred to this study. Had I known about it I certainly would have cited it as potential evidence that new neurons in rats mature faster than in mice (21% of 7-day-old rat neurons expressed Fos – higher than I’ve observed here but consistent with a recent rat electrophysiology study?)

Now, this is a small experiment but how did it go unnoticed when so many people are curious about the function of adult neurogenesis? Well, the study does focus mainly on the expression of cell cycle markers and less on neuronal function – most of the 102 subsequent papers that cite it are cell cycle studies. But, a number of familiar papers do cite it – papers I thought I’d read thoroughly back when there was nothing else to read on the subject, so it’s just proof you can never totally be on top of things…hence the often-used “To the best of our knowledge, this is the first example of….”

Oh, and this is not the first example of functional neurogenesis. That title goes to Paton & Nottebohm who, to the best of my knowledge, demonstrated it for the first time in songbirds.

Reference:
Okano HJ, Pfaff DW, & Gibbs RB (1993). RB and Cdc2 expression in brain: correlations with 3H-thymidine incorporation and neurogenesis. The Journal of neuroscience : the official journal of the Society for Neuroscience, 13 (7), 2930-8 PMID: 8331381

Below are what I hope to be comprehensive visual collages of all published timecourse experiments, where a certain property of new neurons is examined at multiple (≥ 3) different ages. They are grouped by studies of: 1) cell survival, 2) marker expression, 3) functionality, and 4) miscellaneous studies that do not quite fit into the first 3 categories. I’ve ordered the data roughly chronologically and have included the first author’s name and publication year so you can read deeper, if needed. Indeed, if you know these studies already, a brief look at the graphs will bring back the take home message. However, since the data is stripped of text, if the studies are unfamiliar, you’ll have to go to the original source to figure out what the heck they mean (use Pubmed to at least obtain abstracts for the original studies if I didn’t provide a direct link).

Personally, I like timecourse studies for the same reason I like to have all my music albums or books visible at the same time: at a single glance they provide a lot of information – each individual stage of maturation can be interpreted within a bigger picture. The result of these many hours of work will either be a) that the purpose of adult neurogenesis will become immediately clear, or b) that we’ll all have some fancy collages to pin on our bulletin boards and look intelligent.

The survival timecourse

New neurons are born and then many die. The survival timecourse answers the questions: How many new neurons are born? Where are they born and where do they end up, anatomically? How many of them survive and can their survival be altered? Survival timecourses are typically performed by injecting animals with a mitotic marker that will label new neurons as they’re being born, e.g. ³H-thymidine (old school), BrdU (tried and true – example), or a GFP-expressing retrovirus (new school). At a later date one can then detect these birthdated new neurons and count them, see where they’re located etc.

What do these survival timecourses tell us?

• many newborn neurons die between 1w and 4w of age but after that they all survive
  • neurons born during infancy are an exception as they DO die off many months after their birth (Dayer 2003), lending support to the sexy-but-underexplored idea that neuronal turnover might underlie memory turnover in the hippocampus
• the number of new cells labeled with a birthdating marker (e.g. BrdU) grows between 2 hours and several days after the birthdating marker is administered
  • this is caused by continued division of the stem cell or precursor cell that took up the marker in the first place (see expression timecourse, below). After a few cell divisions the marker gets diluted to undetectable levels.
• the general timecourse of cell death is similar in young and aged animals (McDonald 2005) and in mice and rats, although more cells die in mice (Snyder 2009)
• the addition and culling of newborn neurons in the monkey hippocampus (Gould 2001) follows a delayed timecourse compared to rodents
• CREB signalling is critical for neurons to survive between 5-7 days old (Jagasia 2009), NMDA receptors are critical for survival from 14-21 days (Tashiro 2006)
  • thus, CREB signalling would appear to regulate survival before new neurons have formed excitatory connections and are functional (see below) and NMDA receptors regulate survival during the early phase of excitatory synapse formation, when new neurons are just beginning to be able to contribute to behavior. Knowing how to regulate neuronal survival has obvious implications for disorders where reduced neurogenesis might be a causative factor.
• learning increases the survival of new neurons (Leuner 2004)
• learning does not increase the survival of new neurons (Snyder 2005)

The expression timecourse

All cell types within the body express different genes/proteins that serve the cell’s function. Since a muscle cell has a completely different function than a skin cell, it will naturally express different proteins. In like manner, a 1 week-old neuron is functionally distinct from a 4 week-old neuron and the two will also express different proteins (to some extent). Many people have taken advantage of this, using these different proteins as markers that identify a new cell as a neuron vs. a glial cell or, more specifically, an immature neuron vs. a mature neuron. By simultaneously visualizing (via immunohistochemistry) both the birthdating marker (e.g. BrdU) and these phenotypic markers, one can know both the exact age of the neuron and its general degree of maturity. For a 10 sec guide to cell labeling with BrdU and phenotypic markers, see here.

What do these expression timecourses tell us?

• some markers (proteins) are increasingly expressed as new neurons mature over 4 weeks (NSE, NeuN, calbindin)
• other markers are mainly expressed when new neurons are < 4 weeks-old (DCX, PSA-NCAM, calretinin)
• most studies have used the same markers (e.g. DCX, NeuN) to simply demonstrate that new cells are neurons, but some have examined expression of markers that are associated with a more specific function, such as glucocorticoid receptors (Cameron 1993, Garcia 2004) or vascular markers (Palmer 2000)
• BrdU (or other birthdating markers) labeled cells express cell division markers (e.g. Ki67) several days after BrdU is administered. This does not mean newborn neurons are dividing – what it represents is the continued division of the stem cell, or precursor cell, that was originally labelled. (therefore you can never know the exact age of a new cell, but pretty close)

The functional timecourse

The previous timecourses are all well and good but sheer numbers of cells say nothing about whether the new neurons actually work. And markers like DCX or NeuN may give a general hint at the maturity of a neuron but not much more. Functional timecourses address these gaps. A direct measure of function would be whether a new neuron has electrophysiological properties that enable it to process information (e.g. input and output synapses, action potentials). Less direct signs of function can be inferred from the morphology of a new neuron and whether a new neuron is capable of expressing activity-dependent immediate early genes.

What do these functional timecourses tell us?

• electrophysiology (Esposito 2005; Ge 2006), anatomy/morphology (Hastings 1999; Zhao 2006; Toni 2007; Toni 2008), and activity-dependent gene expression (Jessberger 2003; Snyder 2009) all point to new neurons forming their first synapses at 2-4 weeks of age
• around 4 weeks of age, new neurons go through a phase where they have enhanced synaptic plasticity (Ge et al 2007) and enhanced activation during behavior (Snyder 2009)
  • thus, at a young age, new neurons are more modifiable by experience than mature neurons. This may enable them to make a greater impact on behavior, either at this age or by shaping their further integration into the circuitry so they can alter brain function when fully mature and functional
• young neurons have distinct neurotransmitter profiles: initially they receive GABAergic inputs and later receive excitatory glutamatergic inputs (Esposito 2005). Notably, GABA depolarizes immature neurons (Ge 2006), unlike its typically-inhibitory effects on mature neurons. Also, immature neurons have a unique form of the NMDA receptor (NR2B), which endows them with their enhanced plasticity (Ge 2007).
• blocking CREB signalling (Jagasia 2009) or the depolarizing effects of GABA (Ge 2006) inhibits the functional maturation of new neurons
• by 8 weeks of age, new neurons are pretty much fully developed, though Zhao 2006, Toni 2007 and Toni 2008 show that 8-10 week-old neurons are still slightly underdeveloped, presynaptically and postsynaptically
  • does this mean that 8-10 week-old neurons, despite no longer having enhanced synaptic plasticity or enhanced activation during behavior, might still function differently than fully mature neurons?

Other timecourses

There are some timecourse-ish studies that, instead of examining new neurons of different ages, have examined the final fate of same-aged neurons that had been manipulated at different stages of their development. From the first 3 figures we can see that specific stages during a new neuron’s development are associated with enhanced plasticity, unique neurotransmitter profiles and increased likelihood of cell death. Therefore, it is very possible that the ultimate fate of an adult-born neuron depends on when experiences occur, relative to these different stages.

What do these timecourses tell us?

Several show us that experience can modify the number of new neurons, but that the magnitude and direction of the change depends on how old the neurons are when the animal undergoes the experience. For example:

• environmental enrichment enhances survival of new neurons mainly when the neurons are 1-2 weeks old (Tashiro 2007)
• spatial learning in the water maze enhances survival when the neurons are 5-10 days old (Epp 2007) or 1-2 weeks old (Kee 2007)
• the stress hormone corticosterone decreases new neuron survival when administered for 18-day, but not 9-day, stints during new neuron development (Wong 2006)
• Dobrossy 2003 and Olariu 2005 show the interesting but difficult-to-interpret findings that neuron addition can be increased or decreased depending on the extent of learning and the age of the neuron relative to the learning experience
• The Kee 2007 data suggests that fully mature, 10-week-old neurons are activated during memory retrieval only if they were old enough, at the time of  learning, to be involved in forming the original memory

———————————————————————

Hopefully it’s now clear that a 3 day-old neuron differs from a 3-week-old neuron from a 3-month-old neuron. We have seen examples where the developmental stage of a new neuron influences it’s functional maturation and survival, clues that could someday be used to manipulate adult neurogenesis for therapeutic purposes.

Lastly, comparing many of these timecourses side by side, I’m reminded why I like them so much: I trust them. By examining the same thing at multiple time points, each study inherently has a lot of controls. If you take a single time point out of some of these studies, say the % of new neurons that have excitatory glutamatergic inputs at 14 days-old in Esposito 2005 and Ge 2006, you might wonder, who’s right here? One finds 5%, the other finds 70%. I am curious as to why they differ but I still trust both of these studies and refer to them often because, within their respective timecourses, both 14-day data points make sense and, across studies, the 2 timecourses themselves do generally agree, even if they are slightly shifted.

Since the original Eriksson study, a number of studies have attempted to characterize adult neurogenesis in the human hippocampus, by immunostaining for endogenous markers of proliferating precursors and immature neurons, thereby getting around the inconvenient fact that most human brains do not contain BrdU (similar techniques have been used to characterize neurogenesis in wild animals). The problem is that the histology in human studies often looks markedly different than in rodent studies – antibodies don’t recognize human antigens the same as in rodents, human brains may not be preserved as well as in controlled animal studies, and human brain tissue is obtained when an unhealthy person dies. Thus, it can be very difficult to interpret human data based on what is known about adult neurogenesis in rodents.

DCX-expressing cells in a 65 year-old hippocampus

However, the recent study by Knoth et al. in PLoS One does a pretty good job of getting around at least some of these problems and is perhaps the most informative study of adult neurogenesis in humans since the original Eriksson study. The authors focus their study on doublecortin (DCX), a protein involved in cell migration and the extension of neuronal processes. DCX can be expressed in mature neurons but, within the dentate gyrus, it is a reliable and specific marker of immature neurons…at least in rodents. Whether or not DCX can be applied to human studies of adult neurogenesis is a legitimate concern. Previous studies have found putative DCX-expressing cells in human tissue but these cells often lacked dendrites and axons and so it was questionable whether these cells were in fact neuronal (or sometimes even cellular).

To get at these issues, Knoth et al. examined a large number of subjects spanning a huge age range (0 to 100 years!). Since neurogenesis declines with age in rodents and primates, if DCX is truly labeling immature neurons, and if neurogenesis in humans parallels that in rodents, you’d expect the numbers of DCX+ cells to decrease with age. And this is exactly what Knoth et al. report: the density of DCX+ cells decreases roughly tenfold from puberty to old age (their Fig. 9), a drop that is pretty similar to what is seen in rodents. It is also nice to see, in their Fig. 4, some pictures of DCX+ cells that actually resemble neurons. Some cells look like those I’ve seen before (e.g. their Fig. 4e,f) but the cells in a 65 year-old subject and those in the 9 day-old subject have dendritic processes and give the reader confidence that this DCX staining is real. Interestingly, Boekhoorn et al. have shown that, as the interval between death and brain fixation increases (which is quite variable in preservation of human tissue), DCX-expressing cells lose their dendritic arbor but do not altogether disappear. Therefore, it is likely that any potential morphologic variability caused by delayed fixation did not prevent Knoth et al. from accurately quantifying DCX+ immature neurons. It is also worth noting that their estimate of the magnitude of neurogenesis, ~1-10 DCX+ cells/mm² in adulthood is in the same ballpark as reported by Eriksson (humans) and me (rats) for BrdU-labeled cells during middle to old age (~100s of cells/mm³).

Using DCX as a marker of young neurons, Knoth et al. then go on to co-label DCX with other markers of proliferating cells and immature neurons, to better characterize the types of cells present in the adult human dentate gyrus, and to further validate DCX as a true marker of immature cells in humans. They found DCX+ cells also expressed various maturational markers previously characterized in rodents (e.g. proliferation: Ki67, PCNA; neuronal lineage: NeuN, calretinin; and many other markers I won’t pretend to understand). It would be nice to see the proportion of DCX+ cells that express the various markers, to see if the DCX population resembles that of rodents (e.g. is the proliferating proportion of a similar size? What about the “mature” proportion that co-expresses NeuN?). Regardless, the number of markers examined is very extensive and is certainly consistent with the idea that these DCX+ cells are adult-generated and go through distinct phases of maturation.

Morphology of DCX+ cells observed by Knoth et al., ranging from immature (left) to mature (right)

Likely due to the limitations of working with human tissue, many findings are qualitative and will have to be definitively answered in the future. For example, the oldest age at which DCX+ cells were still found to be proliferating depended on which endogenous marker of proliferation was used (Ki67 – 38yr, Mcm2 – 65yr, PCNA – oldest age). Depending on the validity of these markers, it is possible that proliferation ends in middle age and that DCX+ cells found in the oldest subjects are the result of a very protracted cellular maturation process. (Species differences in maturation have been found and there are hints that the neurogenesis process may be slower in primates than in rodents – e.g. cell addition and death are delayed, maturation markers may be delayed). Or does neuronal proliferation occur in old age but is simply too low to reliably detect without sufficient sampling? I also I would love to see low magnification pictures to get a better sense of how immunostaining for the various neurogenesis markers in humans compares to that of rodents, but I’m guessing that’s easier said than done.

All in all, this paper is a welcome addition to neurogenesis literature. And any wishes for more pictures or experiments has to be qualified with the disclosure that, until BrdU in injected into humans, I will always be begging for more. While on this topic – is there any way to get another human study that uses BrdU (or similar) to birthdate new neurons, and provide concrete evidence of the magnitude and maturational profile of adult neurogenesis? Once upon a time BrdU was used extensively to both treat and evaluate cancer. Might there be some BrdU-containing tissue hidden away in lab storage somewhere? Alternatively, many great discoveries have been made, and Nobels awarded, in the name of self-experimentation. Is there any good reason to suspect that a single glass of BrdU water is going to harm any of us? Probably could get a great paper out of it. Then again, we might not be alive to read it…or might not have a hippocampus with which to remember reading it, but we’re not doing science for ourselves anyway, are we?

Knoth, R., Singec, I., Ditter, M., Pantazis, G., Capetian, P., Meyer, R., Horvat, V., Volk, B., & Kempermann, G. (2010). Murine Features of Neurogenesis in the Human Hippocampus across the Lifespan from 0 to 100 Years PLoS ONE, 5 (1) DOI: 10.1371/journal.pone.0008809

At 0.6% of the way into the decade, we’re well beyond the timeframe when most “things of the decade” articles appear. Now that “decade hype” has settled down I thought it would be fun to write a series of posts that discuss some of the major themes in adult neurogenesis over the last decade. A lot has happened in this time; depending on how you birthdate the field (i.e. not counting the work of Joseph Altman), the last decade represents over half the lifetime of the field. BDHXV8966V35

One very influential theme that emerged, only to gain momentum, is the neurogenesis-depression hypothesis. Generally, the idea is that adult hippocampal neurogenesis is protective against depression. This idea was initially quite novel because, 10 years ago, most people were fixated on the hippocampus as a structure involved in learning and memory. Indeed, it’s not implausible that the ability to form rich, detailed memories (which the hippocampus is known for) could enable one to make associations and see perspectives that allow them to escape a depressive funk. But more direct evidence linking the hippocampus to mood has come from studies showing that manipulations to the hippocampus alter stress and anxiety-related behaviors. Searching “neurogenesis” and “depression” in Pubmed I can find papers by Daszuta, McEwen, and Duman discussing reduced adult neurogenesis as a potential factor in depression and mood disorders just before the decade, in 1999. But the study that really got the field going was the finding that antidepressants can increase neurogenesis, by Malberg et al., in 2000. And so, with the turn of the last decade, depression officially replaced epilepsy as the most popular disorder that is potentially related to neurogenesis (the neurogenesis-epilepsy hypothesis was killed in dramatic fashion briefly wounded in 1999*). Within the next few years it was clear that every type of chemical antidepressant increased neurogenesis, as did electroconvulsive shock, as did environmental influences such as running and social housing, which are known to have antidepressant effects. Furthermore, factors that precipitate depression, such as chronic stress, potently downregulate neurogenesis. But while these findings are all consistent with the possibility that neurogenesis plays a role in depression, antidepressants / ECS / exercise / stress also have many other effects on the brain that could explain their effects on mood. So more experiments were needed.

In 2003, Santarelli et al. delivered findings everybody was waiting for: without neurogenesis, antidepressants don’t work. Thinking about this study I am inspired to write a post on the “Top drool-producing papers in the field of adult neurogenesis.” Undoubtedly, this would be near the top, as it seemed to provide the killer evidence needed to link neurogenesis to depression. Moreover, merely reducing neurogenesis was not sufficient to induce anxiogenic or depressive behavior, it just blocked the benefits of antidepressants. This finding was welcome (to me) because it suggested neurogenesis is involved in regulating depressive behavior, but it is not the only factor, consistent with other studies that had failed to find neurogenesis-depression links. The remainder of the decade was rounded out with studies that, while partially confirming a role for neurogenesis in anxiety/depressive behavior, also illustrated the complexity of the situation. For example, neurogenesis does not mediate the effects of antidepressants in some strains of mice, and may mediate antidepressant effects in distinct behavior tests in different species (e.g. forced swim test in rats but not mice). Also confirming but complicating are the findings from Abrous’ group showing (with a transgenic model of reduced neurogenesis as opposed to irradiation in the previously mentioned studies) that merely reducing neurogenesis can be anxiogenic.

So we started the decade with a mere proposal that neurogenesis was relevant for depression/anxiety/mood disorders. We ended with hundreds of correlative studies and several studies spanning several groups showing that relatively specific reductions in adult neurogenesis can lead to depressive behavior if the conditions are right. What will 2010-2020 unveil? How about: 1) One hundred more studies of depressive behavior in neurogenesis-deficient animals, 2) the first test of the neurogenesis-depression hypothesis in humans, and 3) a switch to autism as the most popular disorder that is potentially related to adult neurogenesis.

*Lots of recent and interesting research points to a role for neurogenesis in the pathogenesis (also here) and also recovery from seizures

Malberg JE, Eisch AJ, Nestler EJ, & Duman RS (2000). Chronic antidepressant treatment increases neurogenesis in adult rat hippocampus. The Journal of neuroscience : the official journal of the Society for Neuroscience, 20 (24), 9104-10 PMID: 11124987

Santarelli, L. (2003). Requirement of Hippocampal Neurogenesis for the Behavioral Effects of Antidepressants Science, 301 (5634), 805-809 DOI: 10.1126/science.1083328

Current list in excel format | HTML | RSS feed of updates to the list

I’ve always enjoyed making lists. As a kid I can remember writing lists of rhyming words, lists of all the Ocean Pacific clothes I owned, lists of all the people I knew. Many years later, I hope I’ve now made a list that is actually useful.

Adult neurogenesis is now undisputed. Pretty much on a weekly basis there is a new paper that examines both levels of adult hippocampal neurogenesis and behavior, attempting to draw a functional connection. The good news is that the argument for a behavioral function for adult neurogenesis continues to get stronger. The bad news is that there’s a massive pileup of data, and it’s becoming hard to filter through the relevant studies – first you have to find them amongst the 1000+ studies of adult neurogenesis. Then you have to read them. What behaviors are examined? Is there an effect of reducing or enhancing neurogenesis? What method is used to manipulate neurogenesis? What do other studies find that performed a similar analysis?

In this spreadsheet I’ve tried to provide summary answers to these questions. The data can be sorted by the type of behavior examined (e.g. depressive behaviors, memory etc), how neurogenesis was manipulated (e.g. via irradiation, transgenic tools or exogenous factors like anti-mitotic drugs), and behavioral effect.

It should be noted that I essentially took authors’ claims at face value and nothing here should be blindly accepted as evidence for or against a behavioral function for neurogenesis – read the papers! Task, neuronal age, and other methods should all be considered. Also, at this time, I have only entered data for a fraction of the studies, namely those that have claimed to use a technique specific for reducing neurogenesis. In reality, no such technique exists and I’d like to enter the same data for all studies that correlate neurogenesis with behavior, even those that have manipulated neurogenesis using methods that have widespread effects in the nervous system (e.g. exercise, enriched environment). If you’d like to assist let me know. And if you have any suggestions about how to improve the list let me know – as far as databases go I’m quite a novice.