Selective serotonin reuptake inhibitors, the enormously successful group of antidepressants that includes Prozac (fluoxetine, from Indianapolis-based Eli Lilly and Co.), increase brain serotonin levels. But they do more than that.

SSRIs also lead to the formation of new neurons in the brain. Moreover, if the new neurons do not form, the antidepressants would have no behavioral effects. That shows "neurogenesis doesn't just accompany the actions of antidepressants, but is necessary for their actions," Grigori Enikopolov of the Cold Spring Harbor Laboratory told BioWorld Today.

But the devil is in the details, as the devil is wont to be: Which part of the neurogenesis process were SSRIs influencing? Enikopolov rattled off half a dozen possibilities, including fewer neurons dying, postmitotic neurons going back into cell division mode, mitotic cells dividing once more for the road, or a stem cell replicating itself or giving rise to more daughter cells.

In a study in the May 23, 2006, issue of the Proceedings of the National Academy of Sciences, senior author Enikopolov and his colleagues pinned down the effects of Prozac to a single step in the road to neurogenesis. "We saw changes in one and only one class: the daughters of stem cells," also known as amplifying neural progenitors. "Whatever [increases in mature neurons] we see at the end is the consequence of that change."

SSRIs increase the number of new mature neurons in the hippocampus by about 50 percent. But because there are not that many new neurons in the adult hippocampus to begin with, that effect is fairly small in absolute terms. Also, Enikopolov said, "there is always production of new neurons in the hippocampus." So neurogenesis in response to antidepressants is "an augmentation" of what goes on under normal conditions, rather than the induction of a process that usually doesn't exist in the adult brain, which also makes its effects.

That subtlety intersected with the fact that the available labeling technique for stem cells (technically speaking, early progenitor cells, since there is no proof that they self-renew in the adult hippocampus) made them hard to count. In mature neurons, labeling usually is confined to the nuclei — "in the end, one is counting dots," Enikopolov said. In contrast, stem cell labeling techniques labeled the whole cell, making the cells easy to see "but very difficult to count."

Enikopolov and his colleagues at Cold Spring Harbor Laboratory in Cold Spring Harbor, N.Y., generated a mouse whose fluorescent signal in the stem cells was once again restricted to the nucleus, turning the signal back into a dot. That reporter mouse, plus the division of the process of hippocampal neuron formation into steps that were easy to distinguish by immunostaining, allowed researchers to count even small changes in cell numbers. That increased accuracy, in turn, allowed them to zero in on the increased division of an early progenitor cell as the root cause of the increase in mature neurons.

For all of the clinical successes of selective serotonin reuptake inhibitors, little actually is known about their mechanism. So little, in fact, that a recent issue of PloS Medicine devoted a whole article to the difference between what's known scientifically about SSRI's mechanisms, what's claimed or implied in promotional materials, and whether the twain shall ever meet.

Most marketing materials use an increase in serotonin levels as the mechanistic explanation for why SSRIs work.

Enikopolov said his work "doesn't refute the serotonin hypothesis," and added that work by other researchers using serotonin receptor knockout mice shows that antidepressants do indeed work in part by increasing serotonin levels.

What the PNAS study does, he said, is to shift the balance a bit toward neurogenesis in terms of the relative importance of the two mechanisms.

For their next challenge, Enikopolov and his colleagues are planning to see how general their findings are.

"One question is whether other SSRIs work the same way, but a more interesting question is whether other families, such as SNRIs, work the same way," he said. "And the even more interesting question is whether other treatments - procedures such as electroconvulsive therapy or deep brain stimulation - involve neurogenesis, and whether they target the same cell population. So that is what we are trying to determine now."