The olfactory bulb contains several populations of disposable cells: primary sensory olfactory cells and periglomerular neurons undergo continuous turnover and replacement throughout adulthood. New cells are generated in the subventricular zone (SVZ), and migrate through the rostral migratory stream (RMS). Upon arriving at the granule cell layer of the olfactory bulb, most of the cells cease migrating, and differentiate into granule cells, then extend processes into the external plexiform layer and form synapses with mitral cells. Some cells continue migrating into the glomerular layer and become periglomerular neurons. (These are inter- neurons found in the glomerulus, the structure in which olfactory cells expressing receptors for odorant molecules form synapses with the fibres of the olfactory nerve.
About 1 month after each wave of migrating cells reaches the olfactory bulb, approximately 50% of them die, and the rest remain stable for at least several more months. Newly-generated olfactory neurons go through a number of developmental stages; these stages can be distinguished on the basis of morphology. Class 1 cells are neuroblasts migrating tangentially within the rostral migratory stream at 2-7 days after leaving the SVZ; class 2 are immature migrating neurons; class 3 cells are granule cells with a single, unbranched dendrite that extends no further than the mitral layer (days 9-13); class 4 are granule cells with a branched dendrite that lacks spines but projects into the external plexiform layer (days 11-12); and class 5 mature granule cells (days 15-30).
Adi Mizrahi, of the Hebrew University of Jerusalem, has used a novel technique to examine the dynamics of dendrites in new olfactory neurons as they become incorporated into the olfactory bulb during the later stages of their development (class 4 and 5 cells). Mizrahi injected a modified lentivirus containing the gene encoding green fluorescent protein (GFP) into the SVZ of adult mice. This resulted in the non-selective labelling of neurons and glial cells in the SVZ, as revealed by immunohistochemical staining for neuron- and glial cell-specific molecular markers; infected cells were not damaged.
Craniotomies were then performed on the animals – the partial removal of the skull overlying the olfactory bulbs created a “cranial window” through which time-lapse lase scanning two-photon microscopy could be performed. In this way, cells in the superficial layers of the olfactory bulb – the glomerular and external plexiform layers, which are no more than 250 µm (one quarter of a millimetre) beneath the surface of the bulb – were visualized. By collecting stacks of images from different focal planes, Mizrahi reconstructed the dendrites and spines of the cells during the latter stages of development. (Watch film clips of the time-lapse imaging in the Supplemental data.)
Mizrahi detected numerous GFP-labelled cells in the RMS and olfactory bulb within 2-3 days of injection of the modified lentivirus. All of these cells were neurons and not glia. The earliest granule cell dendrites appeared in the external plexiform layer between 10-14 days post-injection. At the start of this period, most of the neurons had smooth dendrites with few or no spines; the dendrites were sparsely distributed and had a short and stubby appearance. The dendrites were highly dynamic – within 24 hours of the onset of imaging, new dendritic branches formed, and spines appeared on existing dendrites.
Similar dendrite dynamics were observed in the periglomerular neurons imaged in the study. However, differences in the development of the two types of periglomerular neuron – the spiny and non-spiny neurons – were also observed. The dendrites of non-spiny cells were highly dynamic over time – the morphology of the dendritic trees in these neurons was continuously changing throughout development. By contrast, the dendrites of spiny neurons were more stable, but had highly dynamic dendritic spines. Thus, both spiny and non spiny olfactory neurons seemed to be constantly rewiring, although in non spiny cells this process apparently occurs over a larger region.
The time-lapse imaging also revealed that the structure of all the new cells remained highly dynamic even after incorporation into the olfactory bulb. Mizrahi imaged both granular cells and periglomerular neurons between 40-47 days after injecting the lentivirus-GFP construct. By this stage, the newborn cells had generated granule cells with diverse structures (see figure below), but the dendritic trees of all of them were densely covered with spines, and the dendritic dynamics were limited to partial elongations or retractions at the distal tips of the dendrites. Occasionally, however, new branches formed, or older ones retracted.
Morphology of randomly selected perigranular neurons, reconstructed from in vivo imaging experiments of green fluorescent protein (GFP) labelled cells, 45 and 90 days after infection with lentiviral vector. (Scale bar = 50 µm; from Mizrahi, 2007.)
Mizrahi also examined other aspects of cell migration in the RMS. Lentiviruses infect both mitotic and post-mitotic (i.e. dividing and non-dividing) neurons, so the ages of the fluorescent cells migrating into the olfactory bulb cells could not be accurately determined. By using bromodeoxyuridine (BrdU), which is incorporated into newly-synthesized DNA and therefore labels only dividing cells, and comparing the numbers of BrdU-positive cells with GFP-labelled cells, it was shown that, 3 months after lentivirus infection, only a small proportion of cells were labelled with both GFP and BrdU. Time-lapse imaging at 10, 45 and 90 after lentivirus injection further showed that the numbers of migrating cells decreased with time.
Mizrahi then examined the role of sensory deprivation on the development of newly-generated periglomerular neurons. One nostril was blocked to prevent stimuli reaching the corresponding olfactory bulb, and the effect of this on the morphology of new cells incorporated into the bulb was determined. It was found that there was a small increase in the numbers of dendritic branches on periglomerular neurons in the deprived olfactory bulb. There appeared to be no significant effect on the morphology, dynamics or spatial distribution of the dendrites, suggesting that sensory input has only a minor role in the development of newborn cells and their incorporation into the olfactory bulb nerve network. But immature neurons migrating in the RMS express GABAA, AMPA, and, later on, NMDA receptors; electrical activity induced by sensory stimuli is therefore likely to be involved, but exactly how remains to be seen.
The exact function of new olfactory neurons is unclear. Because the distal tip of the olfactory bulb is almost exposed to open air, neurons in that region of the bulb are prone to damage from, for example, toxic fumes. It seems logical that one function of the new cells coming into the olfactory bulb through the RMS would be to replace those cells in the distal tip of the bulb that get damaged. This study shows that neurons in the adult olfactory bulb remain highly plastic throughout adulthood; Mizrahi suggests that regeneration provides cells which act as a substrate for the plasticity of the wiring in the olfactory bulb.
The study also demonstrates a novel experimental system for studying the development and plasticity of new cells in the adult olfactory bulb in vivo. The system may also offer a new way of investigating developmental processes, such as synaptogenesis, in the adult brain. The spatio-temporal dynamics of development make such processes difficult to study in the growing brain; this system may provide a means by which developmental processes can be investigated in the less chaotic environment of the adult brain.
Petreanu, L. & Alvarez-Buylla, A. (2002). Maturation and death of adult-born olfactory bulb granule neurons: Role of olfaction. J. Neurosci. 22: 6106-6113. [Full text]