Skip to main content
. 2017 Feb 24;6:e20147. doi: 10.7554/eLife.20147

Figure 2. Elevated LTD before training impairs LTD-dependent learning.

(A) Saturation hypothesis to explain impaired learning with enhanced synaptic plasticity. Top, In naïve WT mice, at the start of training, synapses are presumably available (white synaptic spines) to selectively undergo associative synaptic plasticity (long-term depression, LTD; blue spines) during training, thereby supporting normal learning. Bottom, In DKO mice, the lower induction threshold for LTD could enable spontaneous activity in the circuit to aberrantly recruit LTD at a random subset of spontaneously active synapses before training, thereby depleting the pool of synapses eligible to undergo LTD, and preventing normal learning. Behavioral pre-training (orange arrow) restores the capacity for LTD-dependent learning in the DKO mice (Figure 3). We tested whether LTD saturation and impairment of LTD-dependent learning can be induced in WT mice with climbing fiber stimulation (cyan arrow; Figure 2B). (B) Climbing fiber stimulation in WT mice before VOR training recapitulates the learning impairment in the DKO mice. Optogenetic stimulation of climbing fibers for 30 min, to induce pf-Pk LTD in the flocculus of WT mice, blocked subsequent VOR-increase learning (solid cyan trace; *p=0.03, F(1,10) = 5.912, two-factor repeated measures ANOVA, CF stim n = 6, Sham n = 6) but had no effect on VOR-decrease learning (dashed cyan trace; n.s. p=0.68, F(1,5) = 0.20) relative to sham stimulation controls in animals that did not express ChR2 in the climbing fibers (black). Mean ± s.e.m.

DOI: http://dx.doi.org/10.7554/eLife.20147.008

Figure 2—source data 1. Elevated LTD before training impairs LTD-dependent learning.
Climbing fiber stimulation in WT mice before VOR training recapitulates the learning impairment in the DKO mice. Data show time course of VOR learning in WT mice during 30 min of normal visual-vestibular VOR training following 30 min of optical stimulation to either optogenetically stimulate climbing fibers (CFs) or provide sham controls (light only in WT without ChR2 expressed in the CFs). Learning was calculated as the percentage change in VOR gain measured after each 10 min block of visual-vestibular training relative to the baseline VOR gain measured immediately before visual-vestibular training. Positive values indicate increase VOR learning and negative values indicate decrease VOR learning corresponding to the respective training stimuli. Each row of 4 numbers within the condition columns corresponds to the time course of learning in an individual animal.
DOI: http://dx.doi.org/10.7554/eLife.20147.009
elife-20147-fig2-data1.xlsx (44.3KB, xlsx)
DOI: 10.7554/eLife.20147.009

Figure 2.

Figure 2—figure supplement 1. Climbing fiber stimulation did not permanently impair VOR-increase learning.

Figure 2—figure supplement 1.

(A) Circuit diagram of the VOR, showing optogenetic stimulation of the climbing fiber inputs to floccular Purkinje cells in the cerebellar flocculus. Expression of ChR2 in climbing fibers was achieved with injections of adeno-associated virus (AAV) carrying ChR2 under the CaMKIIα promoter (CaMKIIα-ChR2(H134R)-EYFP) into the inferior olive (IO). Blue light stimulation was delivered directly to the cerebellar flocculus to activate climbing fibers relevant to VOR learning. CF: climbing fiber, IO: inferior olive, PF: parallel fiber, GC: granule cell, Pk: Purkinje cell, VN: vestibular nuclei. (B) Thirty minutes of pre-training with climbing fiber stimulation did not significantly affect the VOR at the start of VOR-increase (circles/solid traces) or VOR-decrease (triangles/dashed traces) training. Blue: mice with ChR2 expression in the climbing fibers (n = 6). Black: Sham stimulation control mice experiencing the same illumination of the cerebellar flocculus during pre-training but not expressing ChR2 in the climbing fibers (n = 6). During pre-training with optogenetic climbing fiber stimulation, animals were restrained with their head stationary in the dark. At 10 min intervals, the optogenetic stimulation was briefly interrupted to test the VOR. All groups exhibited a temporary decrease in the VOR during the pre-training period, but this effect did not depend on climbing fiber stimulation (blue vs. black traces). Mean ± s.e.m. (C) Climbing fiber stimulation immediately before training impaired VOR-increase learning (Figure 2B), but the same animals exhibited normal learning in response to the same visual-vestibular VOR-increase training stimuli several days later when trained without additional climbing-fiber stimulation pre-training (blue circles/solid trace) compared to sham stimulation control animals not expressing ChR2 in the climbing fibers (black circles/solid trace; ANOVA, p=0.79, F(1,70) = 0.074). Thus, climbing-fiber stimulation had no long-lasting adverse effects, but rather created a temporary state of the circuit that was unresponsive to VOR-increase training (Figure 2B). VOR-decrease learning was also normal when tested several days after climbing fiber stimulation (blue triangles/dashed trace; ANOVA, p=0.83, F(1,88) = 0.048 compared to sham controls without ChR2 shown in black triangles/dashed trace). Mean ± s.e.m.
DOI: http://dx.doi.org/10.7554/eLife.20147.010
Figure 2—figure supplement 2. Non-specific LTD may have no immediate effect on behavior, yet deplete the pool of synapses available to support LTD-dependent learning.

Figure 2—figure supplement 2.

For the VOR, non-specific LTD induced by climbing fiber stimulation or the lower threshold for LTD in DKO mice has no effect on the VOR amplitude (Figure 1—figure supplement 2A and Figure 2—figure supplement 1B), yet could saturate LTD in the synapses that support VOR increase learning. This would occur if the LTD is induced both in synapses whose depression contributes to VOR-increase learning, and in additional synapses whose depression could negate the effects of LTD in the first subset. Two specific possibilities are illustrated. (A) There could be some pf-Pk synapses whose depression causes an increase in VOR amplitude (up arrow) and others whose depression causes a decrease in VOR amplitude (down arrow). The latter may not normally undergo LTD in response to the VOR-decrease training paradigms used in our study (since there is no apparent effect of LTD enhancement or impairment on the learning induced by such training; Figure 1E, solid bars, Figure 1—figure supplement 1, bottom, and Boyden et al., 2006). Nevertheless, when they do undergo LTD through a non-specific induction process such as climbing fiber stimulation or a higher spontaneous LTD rate in DKO mice, this would oppose the effects of LTD in the synapses whose depression tends to increase the gain of the VOR, to yield no net change in the gain of the VOR (top). (B) There could be some pf-Pk synapses whose depression causes an increase in VOR amplitude (up arrow) and others whose depression is irrelevant for the VOR, followed by homeostatic normalization. Top: If too many synapses undergo LTD, there could be a homeostatic process, such as synaptic rescaling or an increase in dendritic excitability, that resets all synaptic weights to values close to their original value, without resetting the LTD mechanism or the capacity for additional LTD, so that these synapses remain LTD-ineligible (blue shading). Thus, the VOR would not be altered, but the LTD mechanism would not be available to support learning. Bottom: If LTD is induced selectively in the synapses that contribute to VOR-increase learning, the weight of these specific synapses remain depressed, and hence the gain of the VOR increased, even after homeostatic normalization restores the summed weight across all synapses to the original value.
DOI: http://dx.doi.org/10.7554/eLife.20147.011