Yellow arrow indicates a stable spine that lasted over eight days of imaging; blue and red arrows indicate transient spines. Modified, with permission, from Trachtenberg et al. © Nature Publishing Group 2002.

Dendritic spines in primary sensory cortex are surprisingly stable — or surprisingly dynamic, depending on who you ask. Two papers published in Nature use the same technique to image the dendrites of cortical pyramidal cells over time, and come up with very different results on the stability of spines — tiny protrusions on dendrites that receive most of the excitatory synapses in the brain.

To be able to image identified dendrites and spines repeatedly in the living brain over days or even weeks is a remarkable feat. The two groups — Trachtenberg et al. and Grutzendler et al. — both used transgenic mice in which a small subset of cortical pyramidal neurons expressed fluorescent marker proteins. The fluorescence was then visualized, through thin 'windows' in the skull, using two-photon microscopy. This technique allowed the researchers to image the dendrites of individual layer 5 pyramidal cells with a contrast and resolution that was high enough to see individual spines, and to identify the same dendrites and spines repeatedly by using landmarks such as blood vessels.

Grutzendler et al. used this technique to observe neurons in the primary visual cortex. They compared spine stability in young (1 month) and adult (∼4 months) mice. Even in young mice, most spines (∼73%) remained stable over a 1-month period, and most changes represented elimination of spines (consistent with earlier work showing that the number of synapses in the visual cortex decreases during development). In the adult, around 96% of spines were stable over one month, although their shapes (length and head diameter) often changed. Such morphological changes could, in theory, mediate changes in synaptic efficacy.

Trachtenberg et al. looked at a different area of cortex, the barrel cortex. This is the primary sensory cortex that receives inputs from the whiskers. They also used mice of a different age (6–10 weeks, corresponding to young adulthood). In these neurons, around 20% of spines disappeared from one day to the next, and only about 60% of spines persisted for more than a week. Despite the high turnover of spines, the total spine density remained constant, presumably reflecting some homeostatic mechanism.

Ultrastructural analysis showed that new spines, generated during the course of the experiment, formed synapses. Taking the study a step further, the authors looked at the effect of sensory deprivation on spine dynamics. When a whisker is removed from a mouse, the corresponding area of cortex is rapidly remodelled so that the neighbouring intact whiskers are more strongly represented. Trachtenberg et al. found that whisker removal led to an increase in the number of 'transient' spines — those that were present for a day or less — and a corresponding decrease in the number of 'stable' spines, although spine density did not change. This finding supports the idea that sensory deprivation induces cortical plasticity that is mediated by synaptic reorganization.

What could be responsible for the differences between the findings of these two studies? In a News and Views article in the same issue of Nature, Otterson and Helm suggest that it is unlikely to result solely from the differences in the ages of the animals used. Intrinsic differences in spine stability between visual and barrel cortex might hold the answer, but much more work will be needed before this can be confirmed or refuted. In the meantime, the question of whether spines represent stable connection sites or dynamic sources of plasticity (or both) remains open.