WTF! No neurogenesis in humans??

The latest study in the saga that is adult neurogenesis was just published today: Sorrells et al, (2018) Human hippocampal neurogenesis drops sharply in children to undetectable levels in adults. Nature, doi:10.1038/nature25975

I wrote a brief piece for Nature News and Views but since they have very strict space limitations I thought it would be worth elaborating just a bit more since I think this paper is sure to generate a lot of excitement. And by excitement I mean fear and dismay because the ultimate goal of neurogenesis research is to identify how new neurons can be used to promote human health. If we don’t have them, how can we use them? And so this new paper challenges us but also guides us in terms of optimizing our research questions.

SUMMARY

Briefly, Sorrells et al. looked at post mortem human hippocampal tissue from pre-term infants through old age. They immunostained for markers of stem cells, dividing cells and immature neurons and found plenty of signs of neurogenesis in fetal brains but it was much lower by one year of life, and the oldest brain that contained immature neurons was in a 13 year old (the next oldest sample was 18 years old). This contrasts markedly with previous studies that have reported neurogenesis throughout adult life in the human hippocampus. What are some of the significant points of this paper?

• the quality of the histology is excellent, which is critical for interpreting any study, especially studies of human tissue
• the use of young samples is a plus, since this demonstrates that they are capable of identifying neurogenesis using the same techniques that are used in the older brains
• their definition of immature cells are those that express both of two markers that are commonly used in animal studies: DCX and PSA-NCAM. This is more stringent criteria than is common but it is warranted because they show that either marker on their own can be non-specific (and identify mature neurons or glia)
  • the downside is that there could be legitimate immature cells that only express one of the markers, that would be missed in the analyses
• given species differences (between well-understood animals and poorly-understood humans) and methodological differences between studies, more research is needed to reconcile these negative findings with previous positive evidence for adult hippocampal neurogenesis in humans
• it suggests a number of directions for future research: better comparisons between rodents and longer lived mammals, the possibility that neurogenesis is more relevant at earlier developmental stages, developing neurogenic strategies for repaired the damaged brain…

CONTEXT

Adult neurogenesis is a great case study in how science moves forward (and then backward, but then forward again, with more forward happening overall, if you’re patient) and there are a number of articles that have already highlighted the progress of the field, as well as the challenges and controversies along the way. I will not rehash them here but suggest you read them because they describe in depth the factors that lead to such polarized interpretations of a phenomenon, whether between investigators or in terms of a changing consensus over time:

Gross CG (2000) Neurogenesis in the adult brain: death of a dogma. Nat Rev Neurosci, 1: 67–73. An historical overview of the factors that led to the persistence and death of the idea that the adult brain cannot produce new neurons.

Kaplan MS (2001) Environment complexity stimulates visual cortex neurogenesis: death of a dogma and a research career. Trends Neurosci, 24: 617-20. Michael Kaplan discusses his early experiments on adult neurogenesis and the opposition which led him to change careers.

Rakic P (2002) Neurogenesis in adult primate neocortex: an evaluation of the evidence. Nat Rev Neurosci, 3: 65-71. Pasko Rakic argues that findings of neocortical neurogenesis are methodologically flawed.

Gould E (2007) How widespread is adult neurogenesis in mammals? Nat Rev Neurosci, 8: 481-8. Elizabeth Gould examines the histological and microscopy methods involved in detecting new neurons, in particular the criteria that scientists use to make conclusions about their images. Covers hippocampal neurogenesis and controversies related to species (humans and primates) and brain regions (eg neocortex).

Cameron HA, Dayer AG (2008) New Interneurons in the Adult Neocortex: Small, Sparse, but Significant? Biological Psychiatry, 63: 550-5. If people are looking for pyramidal neurons they may not find interneurons, it is argued.

Altman J (2008) Memoir: The Discovery of Adult Mammalian Neurogenesis. Joseph Altman (1925-2016) discovered adult neurogenesis in the 1960s and reflects (no-holds-barred) on the early excitement and subsequent blowback.

Does adult neurogenesis exist? No, yes, no, yes, yes, no, yes, yes, yes, YES!

Does adult neurogenesis exist in primates? No, no, yes, maybe, yes, yes, YES!

Does neurogenesis occur in the neocortex? No friggin way, yes, no way, yes, no way Jose, yes, DEPENDS WHO YOU ASK.

Does adult neurogenesis happen in humans? Yes, yes, yes, no, WAIT WTF* DID YOU SAY NO???

So, some things are still not clear, which is funny given such recent technological advances with lasers and all, and with neurogenesis we often can’t even agree on the interpretation of neuroanatomical and immunohistochemical images. And the question about humans is important because we know from rodent work that adult neurogenesis plays a significant role in hippocampal physiology and behavior. There is a lot of evidence in favor of adult human hippocampal neurogenesis. But human studies are challenging. The tissue is rare, you don’t have control over how it was prepared and preserved, the subjects may be ill, histological techniques that work in nicely-prepared animal tissue may not work the same in humans, the brain is different so it may be hard to even know what to look for etc. But this shouldn’t stop us (in fact it should do just the opposite and motivate us to crack this nut). In 2011 I made a list of all the human studies of adult hippocampal neurogenesis (see here, may be incomplete). The pioneering study was by Eriksson et al in 1998, where they found BrdU+ cells in cancer patients. Studies such as Knoth et al and Epp et al have used endogenous markers of immature neurons (such as DCX) to identify adult neurogenesis. Lastly, Spalding et al used another complementary and creative approach to quantify adult neurogenesis: radiocarbon dating. Individually, none of these studies are entirely convincing of adult neurogenesis in humans but, collectively, they provide strong converging evidence. They convinced me. And while the new Sorrells paper raises concerns about these previous papers, it cannot definitively refute them either.

MOVING FORWARD

We need more studies of neurogenesis in humans

The more data we have the more the truth will become apparent. But in addition to more studies we also need better studies, better sharing, better objectivity, better transparency. The Sorrells et al study is strong because the immunohistochemistry is clear, there is a large sample size which includes young subjects as positive controls, and there are all sorts of pictures from zoomed-in profiles of individual cells to overview images of the entire dentate gyrus. There’s nothing more frustrating (to me) than picking up an interesting paper and seeing images that are low resolution, poorly-stained, or too “representative” or zoomed in to know what the tissue really looks like. You have to cut human studies a bit of slack here but Sorrells raises the bar, which we need.

I would go a step further and suggest that future studies of human or primate (or anything, really) tissue provide more images. Why not provide images of all the samples, of many hippocampi, of many other brain regions? These won’t fit in a standard peer-reviewed article but this is 2018 and we don’t have to limit ourselves to a 7 page pdf. Maybe this was one of the problems of the past: limited availability of images and data that the research community can use to make conclusions. In the above articles you will read about investigators visiting other labs to look at their slides. Today, we can easily share these online.

Sorrells makes the point that DCX can label glia, PSA-NCAM can label mature cells, and BrdU-like staining can be observed even when antibodies are omitted. These are important caveats to appreciate and if you read the above articles you will see that these are not new issues. Thus, it is unlikely that there are false-positives in Sorrells work, but there could be false negatives since some DCX-only or PSA-NCAM-only cells might really be legitimate adult-born neurons, because they could downregulate expression of one marker before another. Previous studies have also used DCX as a marker of adult neurogenesis in the human hippocampus. For example, the images by Epp et al are some of the nicest I’ve seen. Are they a bit odd-looking? Sure, maybe. But no one has ever described the morphology of newborn human hippocampal neurons so I don’t know what is normal. Are their DCX neurons a bit more in the hilus than I would expect from rodent studies? Perhaps, but again, might it be that if there is a germinal zone in adult humans, it looks different than rodents? Some previous human studies (eg Boldrini 2009) have examined a band that extends more into the molecular layer and hilus. Which is correct? I don’t know.

It is also worth considering that both DCX and PSA-NCAM fluctuate based on experience. And probably the people that were examined had some “experiences” before their brain tissue was collected. For example, DCX disappears in bats within 30 minutes of capture (they argue due to stress). Post-mortem delay quickly causes DCX+ dendrites to disappear, which can make valid DCX+ neurons appear non-neuronal. PSA-NCAM in the dentate gyrus can also increase (and therefore maybe decrease) independent of adult neurogenesis (see Pham 2003 and Lopez-Fernandez 2007). But, you say, Sorrells’ Ki67 data indicated that cell division occurred in infants but not adults, right? Yes, but everything is harder to stain for in older tissue and children are stress hyporesponsive so…I bring up all of these points and perspectives not to negate the work of Sorrells et al, because I do think it is an excellent study. But because we have been through these ambiguities before and come to conclusions prematurely in the past (again, read the above articles). We are still learning here.

We need more studies of neurogenesis in primates (and longer-lived mammals and other models)

A recent article by Michael Yartsev lamented that our knowledge of the human brain is limited because all we study are mice and rats. Rodents are just one model and they cannot model everything. Years back I made a comprehensive list of all the studies of adult hippocampal neurogenesis in primates, because they can provide a better approximation of our own neurodevelopment. People have looked at primates in the past and failed to find evidence for adult neurogenesis, but now that has been overturned. However, the extent is unclear. Sorrells looks at primates as well and concludes that neurogenesis may persist throughout adulthood (unlike humans) but numbers are still lower than in rodents. They find that after treating two seven year-old macaques with BrdU there are 0 and 2 new neurons when examined 10 and 15 weeks later, respectively. However, Gould et al (2001) found that newborn neurons in primates are short-lived, and decline between 5 and 9 weeks, meaning they could have been gone by the time Sorrells looked. In contrast, at 11 and 23 weeks Kohler et al found 1000s of newborn BrdU+ cells, of which hundreds expressed neuronal markers, and these cells were born in relatively older monkeys (they also used a different BrdU labelling paradigm). So, even in a controlled primate model there are discrepancies and it remains unclear exactly how many neurons are added in adulthood. More research may provide some insights into the processes that are happening in our own brains.

Irmgard Amrein and colleagues are some of the few looking at comparative hippocampal anatomy across mammals. Their approaches could be useful for relating findings across species, for example check out this paper where they report that dentate gyrus proliferation levels correlate with absolute age across mammalian species.

Recalibrating the function of neurogenesis throughout the lifespan

Adult neurogenesis is typically studied in rodents but rodents are born with a premature brain compared to humans. The dentate gyrus is pretty much fully formed in humans at birth but the peak of neurogenesis in rats is at about 5-7 days of age. Opinions vary but the human brain at birth may be developmentally equivalent to a 1-2 week old rodent. Since adult neurogenesis is routinely studied in 8 week-old rodents, and functional studies sometimes block or manipulate neurogenesis when rodents are as young as 4-5 weeks (because by manipulating larger numbers of new neurons, they are more likely to find functional effects), these are basically children/adolescents, i.e. not too different from the human subjects where Sorrells et al found immature cells. Granted, Sorrells found that immature neurons in children were likely born years earlier whereas in rodents new neurons continue to be born into adulthood (though drops off precipitously with age). But in terms of the big picture, the bulk of rodent “adult” neurogenesis really happens quite early in postnatal life, and could be equated to human hippocampal neurogenesis in childhood. Tools such as the translating time website may help with neurodevelopmental comparisons between animal models and humans,

An important additional factor to consider is the timecourse over which new neurons mature. Primate work (see Kohler 2011) and sheep work (Brus 2012) indicate that new neurons mature much slower in longer lived mammals (I’d say at least years) than in rodents. If we extrapolate to humans, childhood (or even prenatal) neurogenesis probably produces populations of new neurons that remain plastic and uniquely involved in learning and behavior through adolescence or even beyond. Even in rats it has been shown that 4-month-old adult-born neurons remain morphologically plastic, raising the question of when, or if, their critical period actually ends. We are just beginning to understand the surprisingly unique circuit connectivity in young neurons in rodents. For example, they connect with the lateral but not medial entorhinal cortex (see Vivar 2012 and Woods 2017), meaning they are performing completely different computations as compared to the older dentate gyrus granule neurons. Do they ever “mature” and become anatomically and physiologically equivalent to earlier-born cells? Or are functionally different cohorts of neurons born at different developmental stages? In sum, if Sorrells findings hold, new neurons born in childhood in humans could have enhanced plasticity for many years and they could have unique computational functions for life. By studying plasticity windows and circuit functions in rodents (even in adults) we can speculate upon the impact and functions of neurons born in childhood in humans. Even better if we can begin to identify functional properties, or provide something more than immunohistochemical characterization, in humans and phylogenetically-related animal models.

Maybe neurogenesis ends in childhood. But even if not, I don’t think people thought that we were bursting with new neurons in old age. There has long been an effort to understand stem cells and how we can harness them in old age, and Sorrells work highlights the importance of finding regenerative therapies. Look at all of the technological advances in neuroscience (again, I’m talking lasers). Do we think that we will not one day be able to figure out how to trick the brain into producing new neurons, or effectively transplant new neurons or their precursors? But we will need to keep studying the animal and human brain to learn how to get there…

-Jason

*speaking of, this was the original title I pitched to Nature News and Views.

PS – It is worth mentioning that a separate study in humans, published in 2016, came to similar conclusions as Sorrells: Dennis et al, (2016) Human adult neurogenesis across the ages: An immunohistochemical study. Neuropathology and Applied Neurobiology, 42: 621-38.