Rapid and continuous activity-dependent plasticity of olfactory sensory input

Abstract

Incorporation of new neurons enables plasticity and repair of circuits in the adult brain. Adult neurogenesis is a key feature of the mammalian olfactory system, with new olfactory sensory neurons (OSNs) wiring into highly organized olfactory bulb (OB) circuits throughout life. However, neither when new postnatally generated OSNs first form synapses nor whether OSNs retain the capacity for synaptogenesis once mature, is known. Therefore, how integration of adult-born OSNs may contribute to lifelong OB plasticity is unclear. Here, we use a combination of electron microscopy, optogenetic activation and in vivo time-lapse imaging to show that newly generated OSNs form highly dynamic synapses and are capable of eliciting robust stimulus-locked firing of neurons in the mouse OB. Furthermore, we demonstrate that mature OSN axons undergo continuous activity-dependent synaptic remodelling that persists into adulthood. OSN synaptogenesis, therefore, provides a sustained potential for OB plasticity and repair that is much faster than OSN replacement alone.

Figure 1. Specific labelling of immature and mature OSN axons in the OB.

(a) Schematic representation illustrating generation of transgenic lines expressing tdTom and sypGFP specifically in either immature or mature OSNs using the tTA system. (b,c) Widefield fluorescence images of coronal sections through OBs of 3-week-old OMP-sypGFP-tdTom (b) and Gγ8-sypGFP-tdTom (c) mice. Inset: maximum intensity projections of confocal z-stacks for boxed regions. (d,e) As b,c but for 8-week-old mice. Note difference in scale bar for main images. (f,g) Two-photon images of immature OSN axons in glomerular (f) and outer nerve (g) layers of a Gγ8-sypGFP-tdTom mouse. (h,i) As f,g for an OMP-sypGFP-tdTom mouse. (j,k) Confocal images of anti-OMP staining in the GL of the OB of Gγ8-sypGFP-tdTom mice, showing little co-expression of OMP in Gγ8+ OSN axons.

Figure 2. Immature OSN axons form synapses in the OB.

(a) Low-magnification EM images from Gγ8-sypGFP-tdTom (left) and OMP-sypGFP-tdTom (right) mice demonstrating specific labelling of OSN presynaptic terminals with immunogold anti-GFP immunostaining. (b,c) Anti-GFP immunogold transmission electron micrographs of synapses formed by immature (b) or mature (c) OSN axons in the OB of 8-week-old mice. (d,e) No difference in maximum bouton diameter (d; P=0.36) or PSD length (e; P=0.55) between synapses formed by immature (n=18) or mature (n=16) OSNs or adjacent unlabelled synapses (‘other', n=16; one-way ANOVAs). Bar charts show mean±s.d.

Figure 3. Optogenetic photoactivation of immature OSN axons evokes robust firing of OB neurons.

(a) Schematic representation illustrating generation of transgenic mice expressing ChIEF-Citrine in immature OSNs using the tTA system. (b) Confocal image of coronal section of olfactory epithelium from P16 Gγ8-ChIEF-Citrine mouse showing colocalization with GAP43 (98.5%) and OMP (6.2%; n=1,106 Gγ8⁺ OSNs from three mice). 49.7% (1,106/2,225, n=3 mice) of GAP43⁺ neurons expressed Citrine. (c) Widefield fluorescence image of coronal OB section showing ChIEF-Citrine expression in olfactory nerve layer and GL of the same mouse as in b. Inset: maximum intensity projections of confocal z-stacks for boxed regions. (d) Schematic representation of multi-channel recording experiment. Optrode consisted of 16 electrodes (pink) with 100 μm spacing and an optical fibre (blue) terminating 200 μm above the most superficial electrode. D, dorsal; and M, medial. (e) Example traces from deepest four electrodes in a Gγ8-ChIEF-Citrine mouse showing robust multi-unit firing during photoactivation (blue band) in the GL, EPL (superficial and deep) and MCL. Scale bars, 10 μV, 1 s. (f–i) Raster plots and peri-stimulus time histograms (20 ms bins) for trials using an excitation power of 10 mW for the same mouse shown in e; traces correspond to trial 1. (j) Relationship between excitation power (473 nm) and mean change in firing rate (ΔFR) for neurons in GL, EPL and MCL in Gγ8-ChIEF-Citrine mice (n=5, mean values shown). GL P<0.001; EPLs P<0.001; EPLd P=0.003; MCL P=0.006; and Gγ8-ChIEF-Citrine versus tetO-ChIEF-Citrine, two-way ANOVA. tetO-ChIEF-Citrine data shown as mean values across all four electrodes (n=3).

Figure 4. In vivo two-photon time-lapse imaging of OSN presynaptic terminals.

(a,b) Maximum intensity projections of in vivo two-photon z-stacks showing the dorsal surface of the OB in 3-week-old Gγ8-sypGFP-tdTom (a) and OMP-sypGFP-tdTom (b) mice. (c) Maximum intensity projections of time-lapse images of an immature OSN axon. (d) Single optical sections of time-lapse images from an OMP-sypGFP-tdTom mouse. (c,d; left) Fluorescence images and (right) detected sypGFP puncta. Yellow arrowheads: puncta destined to be lost; cyan arrowheads: newly formed puncta.

Figure 5. Rapid and continuous turnover of OSN presynaptic terminals.

(a) Percentage turnover of sypGFP puncta over 3 h. P=0.002 Gγ8 versus OMP; P=0.23, 3-week old versus 8-week old, two-way ANOVA. (b) Mean percentage gain of new sypGFP puncta over 30 min. P=0.002 Gγ8 versus OMP, P=0.32, 3-week old versus 8-week old, two-way ANOVA. (c) Mean percentage loss of existing sypGFP puncta over 30 min. P=0.001 Gγ8 versus OMP; P=0.56, 3-week old versus 8-week old, two-way ANOVA. Magenta dashed lines in b,c indicate potential contribution of movement noise (Methods and Supplementary Fig. 4C). In a–c, mean±s.d. is shown. (d) Survival time of newly-formed sypGFP puncta. (e) Time-lapse image series of single optical sections from two-photon z-stack showing formation of new sypGFP punctum (arrowhead) at t60, which persists throughout the remainder of the imaging session (survival time >120 min).

Figure 6. Turnover of OSN presynaptic terminals is strongly reduced by naris occlusion.

(a) Timeline for naris occlusion. (b) Widefield image of OB coronal section from 8-week-old OMP-sypGFP-tdTom mouse following 3 weeks of naris occlusion, showing decreased OB size on the occluded (right) side. (c) Comparison of size of occluded (right) and open (left) OBs (P<0.001, paired t-test, n=6 mice). (d) Confocal images of left and right sides of septal olfactory epithelium in coronal sections. (e) No difference in density of Gγ8⁺ (P=0.87) or OMP⁺ (P=0.84) OSNs between open (O) and closed (C) nares. Two-way ANOVA; three mice/group. (f,g) Time-lapse images of Gγ8-sypGFP (f) and OMP-sypGFP (g) puncta showing stability over 3 h. (left) Acquired images and (right) detected sypGFP puncta. (h) Turnover of immature and mature OSN presynaptic terminals is strongly reduced in the ipsilateral OB following naris occlusion (P<0.001, effect of naris occlusion; P<0.001, effect of OSN maturity; P=0.046, interaction; two-way ANOVA). Ctrl: unoccluded mice. (i) Naris occlusion reduces gain and loss of immature OSN presynaptic terminals in the ipsilateral OB (P=0.005, effect of occlusion; P=0.98, gain versus loss; P=0.87, interaction; two-way ANOVA). (j) Naris occlusion reduces gain and loss of mature OSN presynaptic terminals in the ipsilateral OB (P<0.001, effect of occlusion; P=0.50, gain versus loss; P=0.48, interaction; two-way ANOVA). (k) Survival time of newly formed sypGFP puncta. Magenta dashed lines as Fig. 5b,c. Bar charts show mean±s.d.