Alterations of activity can result either promote or inhibit the survival of ABNs in the OB (Cummings and Brunjes, 1997
; Mandairon et al., 2003
; Yamaguchi and Mori, 2005
; Kelsch et al., 2009
; Moreno et al., 2009
; Mouret et al., 2009
). Here we used a viral vector encoding an siRNA to knock-down sodium channel expression and naris occlusion to reduce the activity of ABNs that migrate to the OB (which we note to reduce survival of ABNs by 44 and 24% respectively, data not shown). Both methods used to reduce neuronal activity resulted in significant decreases in the total number of spines, overall spine density, and several measures of branching complexity. In siRNA knock-down cells we observed that reductions in spine number, spine density, and branching complexity incurred primarily in proximal and intermediate regions of the dendritic tree. However naris-occluded cells did not exhibit significant changes in any specific region along the dendrite. Rather changes in dendritic length and branching were distributed along the length of the dendrite.
Previously published results by Kelsch et al. (2009
) reported that naris occlusion resulted in highly localized changes in the number of formed connections of ABNs labeled by a lentiviral vector expressing GFP tagged PSD-95. Specifically, changes seen in ABNs experiencing sensory deprivation by naris occlusion resulted in significant decreases in the number of PSD-95 labeled synapses at distal regions of the dendrite, and significant increases in the number of synapses at proximal regions of the dendrite (Kelsch et al., 2009
). This effect does not appear in our results. However, it is important to note that PSD-95-GFP likely labels glutamatergic inputs onto ABNs rather than GABAergic outputs (Sheng, 2001
). While Kelsch et al. (2009
) did observe output synapses of ABNs via labeling of synaptophysin (SypG), which localizes to presynaptic neurotransmitter vesicles (Sudhof and Jahn, 1991
), they reported significant changes in the number of SypG+ clusters at distal regions of the dendrite only. Because our experimental paradigm aimed to look at the effect of naris occlusion on the functional
integration of ABNs into the OB, we chose looked at changes in spines, which serve as the primary sites of GABAergic lateral inhibition onto mitral/tufted cells of the OB (Panzanelli et al., 2009
). While our dataset appeared to exhibit trends consistent with analysis of SypG clustering by Kelsch et al. (2009
) it is unclear why our results did not reach significance.
Conversely, Lin et al. (2010
) observed the effects of genetically manipulating excitability on both survival and morphology of ABNs. They noted that neither genetically decreasing excitability (via expression of ESKir2.1
) nor genetically increasing excitability (via expression of NaChBac
) had any effect on morphology of ABNs (Lin et al., 2010
). While these results appear to contrast our own, it is important to note that Lin et al. (2010
) altered the excitability of ABNs by inserting ion channels rather than altering expression of endogenous channels. Thus, the degree to which activity was changed may be different in our studies vs. the previous work. Alternatively, homeostatic mechanisms may more effectively compensate for reductions in activity not caused by knock-down of endogenous channels. These subtle and specific manipulations may provide insight into how differences in activity across unmanipulated ABNs can influence their connectivity.
Our results suggest that reduced levels of activity in ABNs adversely affect their ability to integrate into existing OB circuitry. Specifically, fewer spines formed by ABNs experiencing reduced activity indicates fewer connections made with surrounding circuitry, and thus reduced participation in olfactory processing. Further it appears that these two different manipulations of activity differentially affect spine development of ABN dendrites. While it is believed that naris occlusion results in sensory deprivation (and thus reduction of activity) in the OB (Frazier-Cierpial and Brunjes, 1989
), the effect of knocking down expression of voltage gated sodium channels (NaV
1.1–1.3) in ABNs is less clear.
In mature olfactory granule cells, lateral inhibition onto mitral cells can be evoked by global action potentials (Petreanu and Avarez-Buylla, 2002
; Carleton et al., 2003
) and low-threshold spiking (Egger, 2008
). Global action potentials result in long-lasting depolarizations, mediated by calcium activated non-specific cation currents (Egger, 2008
), and NMDAR-dependent calcium influx. This results in prolonged release of GABA at dendrodendritic synapses, and thus more efficient lateral inhibition onto mitral and tufted cell dendrites when compared to inhibition due to low-threshold spiking (Hall and Delaney, 2002
). By knocking down NaV
expression in ABNs, we hoped to reduce the number of global action potentials and their resulting activity.
Given the role of NaV
channels in the physiology of olfactory granule cells, we hypothesize that global spiking of ABNs serves as the primary form of activity in proximal regions of the dendrite (Egger et al., 2003
). However at distal regions of the dendrite, highly localized calcium transients due to dense populations of T-type calcium channels in spines facilitate NMDA receptor activation (Egger et al., 2005
) and allow for dendrodendritic inhibition independent of sodium channel activation (Isaacson and Strowbridge, 1998
; Schoppa et al., 1998
). That is, activity in more proximal regions of the dendrite may be more vulnerable to fluctuations of NaV
channel expression, where activity in distal synapses may be more electrically isolated from the soma and thus depend only weakly on voltage gated sodium channels. This selective change in activity may be relevant to the differences between our results and those of Kelsch et al. (2009
One question is whether siRNA knock-down changes overall activity of affected OBs, including uninfected cells. Since our lentiviral injections selectively labeled neural progenitors, only a small fraction of cells in a given OB are infected by the virus; consequently the activity reduction in the infected granule cells is likely to have little effect on overall OB activity. Thus the overall level of input received by new granule cells in the hemisphere injected with siRNA virus is likely to be comparable to cells in the control hemisphere. By contrast, cells in naris-occluded animals experience a reduction of activity via a reduction of synaptic input from sensory neurons to mitral cells. This will directly reduce input to distal dendrites of granule cells and also reduce somatic spiking. Thus changes in the spatial profile of formed spines may reflect differences in the profile of activity in cells under these two different conditions.
We have focused on differences observed in mature
ABNs. Therefore we chose a single time point (35 days post-injection), which allowed ample time for ABN maturation following birth. As reported previously, ABNs reach “class 5” morphology (full maturation) 15–30 days after birth (Petreanu and Avarez-Buylla, 2002
). Cells in this morphological class have been shown to have adult-like membrane properties and sodium channel expression (Carleton et al., 2003
). Given the typical maturation time of ABNs, we do not expect any time points beyond 35 days post-injection to show differences from those reported here, but examining earlier time points could identify when activity reduction begins to affect synapse formation.
We have shown that ABNs can develop nearly normal morphological features, despite manipulations in their activity levels. That is, these neurons adapt to their altered states or environments by altering specific aspects of their morphologies, while maintaining their gross structural features. Thus, major morphological features of these neurons develop in a manner that is largely insensitive to changes in activity, and observed changes are linked to more subtle and possibly highly specific changes in input number and location.