Neurogenesis

Cajal's dogma

Santiago Ramón y Cajal, whose drawings of stained neurons founded modern neuroanatomy and who shared the Nobel Prize in 1906, wrote one of the most consequential sentences in the history of the field. In the adult centres, he concluded, the nerve paths are something fixed and immutable: everything may die, nothing may be regenerated.

It is worth noticing what kind of statement that was. Cajal was not reporting an experiment showing that no new neurons appear. He was reporting a failure to find them with the tools he had, and drawing a general conclusion. In fairness to him, he explicitly framed it as something future science might overturn. But the qualification was forgotten and the doctrine stuck. For most of the twentieth century, the fixed adult brain was simply what everyone knew.

Altman and the first cracks

In the 1960s, Joseph Altman, working with radioactive thymidine to label dividing cells, reported evidence of newly generated neurons in the adult rat brain, including the hippocampus. The work was careful and it was, in retrospect, correct. It was also largely ignored for two decades. The tools of the day made it difficult to prove beyond doubt that a labelled cell was a neuron rather than a glial cell, and the weight of the dogma did the rest.

The finding was revived and extended in the 1980s and 1990s. Fernando Nottebohm's work on songbirds showed striking seasonal neuron addition in the song nuclei of adult canaries, which was hard to dismiss. Better markers, better microscopy, and the development of BrdU labelling then made the rodent case increasingly solid. By the late 1990s, adult neurogenesis in the rodent dentate gyrus and in the subventricular zone (feeding the olfactory bulb) was becoming mainstream. The question that remained was whether it happened in us.

Eriksson 1998: new neurons in human brains

The human question was answered, in a way that seemed decisive at the time, by an unusual opportunity. Peter Eriksson, Fred Gage, and colleagues studied post-mortem brain tissue from cancer patients who had, while alive, received bromodeoxyuridine (BrdU) as part of a clinical protocol to assess tumour cell proliferation. BrdU is incorporated into the DNA of any cell that divides while it is present, so it labels cells born during the treatment window.

In the hippocampal dentate gyrus of these patients, the team found BrdU-labelled cells that also carried neuronal markers. The 1998 paper in Nature Medicine concluded that new neurons are generated in the adult human hippocampus, and that the human brain retains the potential for self-renewal into old age. The oldest patient in the sample was in their seventies.

It was a landmark, and it was reported as the death of Cajal's dogma. The sample was small, as any study depending on such an unusual clinical circumstance must be, but the finding was clear and it was not seriously challenged for years.

Spalding 2013: dating cells with the Cold War

The most elegant piece of evidence came from an unlikely source: atmospheric nuclear weapons testing. Above-ground tests in the 1950s and early 1960s dramatically raised the level of carbon-14 in the atmosphere, and the ban on such testing in 1963 was followed by a steady, well-characterised decline. Plants take up atmospheric carbon, we eat plants, and a cell that divides incorporates the carbon-14 concentration of its year into its DNA, where it stays.

Kirsty Spalding, Jonas Frisén, and colleagues exploited this to date human cells. By measuring the carbon-14 content of neuronal DNA from post-mortem hippocampi, they could estimate when those neurons were born. Their 2013 paper in Cell concluded that a substantial subpopulation of hippocampal neurons undergoes turnover in adults, and put the figure at something on the order of hundreds of new neurons per day in the dentate gyrus, with only a modest decline across adult life.

The method's great virtue is that it does not depend on staining for the right marker or catching a cell in the act of dividing. Its limitation is that it infers a population-level turnover rate from an isotope measurement, which requires a model of how the tissue is composed. That model has itself been contested.

2018: the collision

And then, in the space of a few weeks in 2018, two prominent papers reached opposite conclusions about the same tissue.

Two well-equipped laboratories, publishing in two of the most selective journals in biology, using post-mortem human hippocampal tissue, and arriving at conclusions that cannot both be right. The disagreement has not been resolved by a subsequent decisive study, and later work has continued to land on both sides.

Where the evidence actually stands

Rather than pick a winner the data do not support, here is what can be said with each degree of confidence.

Why this is so hard to settle

The obvious question is why a field that can image single synapses cannot count new neurons in a human hippocampus. The answer is that almost every tool that makes the rodent case airtight is unavailable in humans.

None of this is a scandal. It is what a genuinely hard empirical question looks like from the inside. The appropriate response is neither to declare the matter closed nor to throw up one's hands, but to hold the uncertainty precisely.

The marketing problem

Unfortunately, "neurogenesis" escaped the laboratory and became a sales word. It has the ring of scientific authority, most people have a vague sense that it is good, and almost nobody knows how contested the human evidence is. That combination is commercially irresistible.

A useful test when you meet a neurogenesis claim: ask what species the underlying study was in, and whether anyone measured a new neuron in a living person. Almost invariably, the answer to the second question is no.

Related topics