Homosynaptic LTP stimulus minutes before, after, or 24 hrs after a weak associative conditioning produces lasting memory.

a) Diagram showing the experimental timeline. b) Left: High frequency stimulation (HFS) of the thalamic inputs (Th) to the lateral amygdala (LA) applied either 24 hours (WTh+24hHFSTh, corresponding to panel a, top branch), or immediately after a weak thalamic associative conditioning (WTh+HFSTh, corresponding to panel a, middle branch), significantly increased the CS-evoked freezing (n=9; One-way ANOVA, F (2, 23) = 8.202, p-value=0.0020) with Tukey test correction). Right: HFS of the thalamic input immediately before (HFSTh+WTh, corresponding to panel a, bottom branch) (n=6) or after a weak associative conditioning (n=8) (WTh+HFSTh, corresponding to panel a, middle branch) is equally effective in increasing the CS-evoked freezing. Colors of the bar graphs represent the experimental protocols for each group of mice (colored boxes in panel a). Subscripts with blue font indicate stimulation of the blue-shifted channelrhodopsin oChIEF using the selective procedure. c) Diagram showing the experimental setup of the in vivo electrophysiology recording (Rec) in anesthetized mice. Evoked field EPSP was produced by blue light stimulation (450nm) of the thalamic inputs expressing oChIEF. d) Plot of average in vivo field EPSP slope (normalized to baseline period) in the LA before and after HFS (n=5). Right inset: Superimposed traces of in vivo field responses to single optical stimulus before (dashed line) and after (solid line) HFS. Scale bar, 0.1 mV, 5ms. Results are reported as mean ±S.E.M. **, p<0.01. Ctx: Cortical input; Th: Thalamic input; LA: lateral amygdala; HFS: High Frequency Stimulation; EPSP: excitatory postsynaptic potential; WTh: Recall session after a weak thalamic associative conditioning.

Submilliwatt yellow light renders a red-shifted channelrhodopsin insensitive to blue light.

a) Diagram showing ex vivo electro-physiology recordings in slices where ChrimsonR-expressing thalamic inputs to the lateral amygdala (red lines) were optically activated. Synaptic responses were evoked by pulses of 450nm blue light (450nm), or pulses of blue light co-illuminated with a 561nm yellow light pulse (co-illumination). b, c) Bar graph (normalized to blue light (450nm) stimuli) (b), and example recording (scale bar, 50 pA, 20ms) (c), of optically driven synaptic responses to pulses of blue light (450nm) or pulses of blue light co-il-luminated with yellow light (450nm with yellow underline) (2 animals, 3 slices, 7 cells). d) Left: Diagram showing the experimental set up of electrophysiology recordings in freely moving mice where ChrimsonR-expressing thalamic inputs (Th) to the lateral amygdala (LA) were optically activated. Right: Comparison of a representative waveform average of the response to pulses of red light (638nm), pulses of red light co-illuminated with a 500ms yellow light pulse (638nm with yellow underline), pulses of blue light (450nm), and pulses of blue light co-illuminated with a 500ms yellow light pulse (450nm with yellow underline) (n=3). e) Left: Diagram showing the experimental set up of electrophysiology recordings in freely moving mice where oChIEF-expressing thalamic inputs (Th) to the lateral amygdala (LA) were optically activated. Right: Comparison of a representative waveform average of the response to pulses of red light (638nm), pulses of blue light (450nm), and pulses of blue light co-illuminated with a 500ms yellow light pulse (450nm with yellow underline) (n=3). f) Left: Diagram showing the experimental setup of electrophysiology recordings in anesthetized mice where ChrimsonR-expressing thalamic inputs (Th) to the lateral amygdala (LA) were optically activated. Middle: Comparison of a representative waveform average of the response to pulses of blue light (450nm), pulses of blue light co-illuminated with a 500ms yellow light pulse (A), pulses of blue light following the yellow light pulse by 50 ms (B), or 500 ms (C) (n=4). Right: Comparison of the waveform average responses to pulses of blue light (450nm), pulses of blue light co-illuminated with a 500ms yellow light pulse (450nm with yellow underline), and pulses of blue light after sequentially applying NBQX and TTX, and later in a euthanized mouse (Dead), (n=4). Scale bar, 0.2 mV, 10ms. g, h) Representative traces for 10 Hz (g) and 85 Hz (h) stimulation of Chrim-sonR-expressing thalamic inputs to the lateral amygdala, which were activated with blue light (450 nm, in blue). Yellow traces are the represantative evoked responses of the inputs to 10 Hz (g) and 85 Hz (h) blue light stimulation (450 nm) co-illuminated with a 561nm yellow light pulse (n=3).

Heterosynaptic LTP stimulus produces lasting memory if delivered within minutes after a weak associative conditioning.

a,b) Diagram showing the experimental timeline of the heterosynaptic LTP protocol manipulation following a weak thalamic associative conditioning. HFS with yellow upperline indicates that the delivery of high frequency stimulation with blue light overlapped with long pulses of yellow light. This co-illumination prevents the activation of ChrimsonR-expressing thalamic inputs (Th) by blue light, while the oChIEF-expressing cortical inputs remain unaffected. Note that yellow light specifically renders ChrimsonR, and not oChIEF, insensitive to blue light. c) Left: High frequency stimulation (HFS) of the thalamic input expressing ChrimsonR immediately following a weak associative conditioning on the same input (WTh+HFSTh, corresponding to panel b) (n=9) was ineffective in producing the CS-evoked freezing. HFS with yellow upperline indicates that HFS with blue light overlapped with long pulses of yellow light. This was to prevent the activation of ChrimsonR-expressing thalamic inputs by blue light, as described above and detailed in figure 2. The same HFS protocol in mice that additionally expressed oChIEF in the cortical inputs (WTh+HFSCtx, corresponding to panel a, bottom branch) (n=10), significantly increased the CS-evoked freezing (heterosynaptic LTP) (Unpaired t-test, p-value=0.0100). Middle: HFS on the cortical input, induced 24 hours after a weak associative conditioning (WTh+24hHFSCtx, corresponding to panel a, top branch) was ineffective in producing the CS-evoked freezing (n=9; Paired t-test, p-value=0.2193). Right: Comparison of the effect of homo-synaptic LTP protocol (WTh+HFSTh) (same dataset from figure 1b) and heterosynaptic LTP protocol (WTh+HFSCtx) (same dataset from panel c, left) (Unpaired t-test, p-value=0.9740). Results are reported as mean ±S.E.M. **, p<0.01. Subscripts with red font and blue font indicate stimulation of the red-shifted channelrhodopsin ChrimsonR and the blue-shifted channelrhodopsin oChIEF, respectively.

Homosynaptic and heterosynaptic LTP protocols produce lasting memory when applied within minutes after an unpaird conditioning.

a) Diagram showing the experimental timelines for fiber photometry from the thalamic inputs (Th) expressing GCaMP7s. b) Averaged trace of the thalamic input activity in response to footshock (onset indicated by the dotted line), n=5. c) Diagram showing the experimental timelines for fiber photometry from the lateral amygdala (LA) neurons expressing GCaMP8m with intact or lesioned thalamic inputs. d) Averaged trace of the LA neurons activity in response to footshock (onset indicated by the dotted line) in mice with lesion (dash line) or no lesion (solid line) in the lateral thalamus (Th), n=6 per group. e) Diagram showing the experimental timelines of the homosynaptic LTP protocol manipulation following an unpaired thalamic conditioning. f) Unpaired conditioning on the thalamic inputs (UTh, corresponding to panel e, top branch) produced no CS-evoked freezing, while if unpaired conditioning was immediately followed by high frequency stimulation (HFS) on the same inputs (UTh+HFSTh, corresponding to panel e, bottom branch) it significantly increased the CS-evoked freezing (homosynaptic LTP), (n=11 per group; Mann-Whitney test, p-value=0.0002). Subscripts with blue font indicate stimulation of the blue-shifted channelrhodopsin oChIEF using the selective procedure. g) Diagram showing the experimental timelines of the heterosynaptic LTP protocol manipulation following an unpaired thalamic conditioning. h) High frequency stimulation (HFS) of the thalamic input expressing red-shifted channelrhodopsin ChrimsonR immediately following an unpaired conditioning on the same input (UTh+HFSTh, corresponding to panel g, top branch) was ineffective in producing the CS-evoked freezing, while the same protocol in mice that, in addition, expressed oChIEF in the cortical inputs (UTh+HFSCtx, corresponding to panel g, bottom branch), significantly increased the CS-evoked freezing (heterosynaptic LTP), (n=11 per group; Mann-Whitney test, p-value=0.0002). During HFS, blue light pulses overlapped with long pulses of yellow light. This co-illumination prevents the activation of ChrimsonR-expressing thalamic inputs (Th) by blue light, while the oChIEF-expressing cortical inputs remain unaffected. Note that yellow light specifically renders ChrimsonR, and not oChIEF, insensitive to blue light. Subscripts with red font and blue font indicate stimulation of the red-shifted channelrhodopsin ChrimsonR and the blue-shifted channelrhodopsin oChIEF, respectively. i) Diagram showing the experimental setup of the in vivo electrophysiology recordings (Rec) in anesthetized mice where the thalamic input expressing ChrimsonR and/or the cortical input expressing oChIEF were optically activated independently. j) Left: Plot of average in vivo field EPSP slope (normalized to baseline period) in the LA evoked by optical activation of the thalamic inputs, before and after footshock delivery (n=5; Paired t-test, p-value=0.2916). Middle: Plot of average in vivo field EPSP slope (normalized to baseline period) in the LA evoked by optical activation of the cortical inputs (Ctx), before and after high frequency stimulation (HFS) of these inputs (n=6; Paired t-test, p-value=0.0031). Right: Plot of average in vivo field EPSP slope (normalized to baseline period) in the LA evoked by optical activation of the thalamic inputs (Th), before and after HFS delivery on the cortical inputs (heterosynaptic LTP) (n=5; Paired t-test, p-value=0.0074). HFS with yellow upperline indicates that the delivery of high frequency stimulation with blue light overlapped with long pulses of yellow light. This co-illumination prevents the activation of ChrimsonR-expressing thalamic inputs (Th) by blue light, while the oChIEF-expressing cortical inputs remain unaffected. Superimposed traces of in vivo field response to single optical stimulus before (dash line) and after (solid line) the induced protocols. Results are reported as mean ±S.E.M. **, p<0.01; ***, p<0.001.

Heterosynaptic LTP protocol when applied within minutes after a weak associative conditioning produces a long-lasting memory accompanied by the synaptic potentiation of the conditioned inputs.

a) Diagram showing the experimental setup of the in vivo electrophysiology recordings (Rec) in freely moving mice where the thalamic inputs expressing ChrimsonR and cortical inputs expressing oChIEF were optically activated independently. b) Left: Plot of average in vivo field EPSP slope (normalized to baseline period) in the LA evoked by optical activation of the cortical inputs, before and 24 hours after a weak thalamic conditioning followed immediately by HFS delivery on the same cortical inputs. Right: as left, except field EPSP was evoked by activation of the thalamic inputs. The potentiation of the field EPSP of the cortical (homosynaptic LTP) (n=4; Paired t-test, p=0.0082) as well as the thalamic inputs (heterosynaptic LTP), (n=4; Paired t-test, p=0.0336) is evident 24 hours after the delivery of HFS. Superimposed traces of in vivo field response to single optical stimulus before (dash line) and after (solid line) HFS. Scale bar, 0.5mV, 5ms. c) The behavioral responses of the mice tested for their homo- and hetero-synaptic plasticity in panel b. Note a significant CS-evoked freezing 24 hours after a weak thalamic conditioning followed immediately by HFS delivery on the cortical inputs (heterosynaptic LTP). These mice did not show a CS-evoked freezing prior to the protocol (BL) (n=4; Paired t-test, p=0.0478). d) Positioning of the stimulating electrodes (Th. Stim and Ctx. Stim.) and the recording electrode. e) (left) Weak stimulation of the thalamic inputs (purple circle) results in a transient LTP. No change was observed in the basal synaptic transmission of the cortical inputs (control pathway, yellow circle). Strong stimulation of the cortical inputs following the weak stimulation of the thalamic inputs stabilized synaptic potentiation of the thalamic inputs (right). Dash line indicates the onset of HFS induction. f) A paired-comparison of the decay of synaptic potentiation of the thalamic inputs with (WTh+SCtx) or without (WTh) the strong stimulation of the cortical inputs. (n=4; Welch’s t test, p=0.0062). (10 animals, 15 slices; one cell per slice). Results are reported as mean ±S.E.M. *, p<0.05; **, p<0.01.

a) Representative image of a coronal section of mice expressing AAV-oChIEF-tdTomato in the lateral thalamus (Left). Scale bars, 1 mm. Axonal expression of AAV-oChIEF-tdTomato in the lateral amygdala (Right). Scale bars, 500 μm. b) The CR is significantly higher after strong conditioning (S, n=8) compared to weak conditioning (W, n=9), and unpaired conditioning (U, n=7). Optical CS alone did not elicit any CR (N, n=4; Ordinary one-way ANOVA, F (3, 28 = 60.79, p-value<0.0001 with Tukey test correction). c) Optic fiber placement of individual mice from figure 1. Results are reported as mean ±S.E.M. ****, p<0.0001.

a) Representative image of a coronal section of mice expressing AAV-oChIEF-tdTomato in the LA-projecting cortical region and AAVChrimsonR-tdTomato in the LA-projecting thalamic region. Scale bars, 1 mm. Axonal expression of AAV-oChIEF-tdTomato and AAVChrimsonR-tdTomato in the lateral amygdala. Scale bars, 500 μm. b) Optic fiber placement of individual mice from figure 2.

a) Representative image of a coronal section of mice expressing AAV-GCaMP7s in the lateral thalamus (Left). Scale bars, 1 mm. Axonal expression of AAV-GCaMP7s in the lateral amygdala. Scale bars, 500 μm. b) Optic fiber placement of individual mice from figure 3. c) Overlay of the maximum extent of the lesion in the thalamic-lesioned group (n=6). d) Optic fiber placement of individual mice from figure 3 f,h.

a) Comparison of the freezing levels evoked by high frequency stimulation of the thalamic inputs in naive mice (HFSTh, n=4) compared to mice subjected to unpaired conditioning (U; n=11) and to unpaired conditioning followed by HFS (U + HFSTh, n=11; Ordinary one-way ANOVA, F (2,23)= 10.09, p-value=0.0007 with Tukey test correction). b) A paired comparison of the freezing levels evoked by thalamic axons activation when thalamic LTP is applied 24 hours after the unpaired conditioning (n=7, Wilcox test, p-value=0.0625). c) Optic fiber placement of individual mice from figure 3. d) Plot for average in vivo field EPSP slope (normalized to baseline period) in the LA. The response was evoked by optical activation of the thalamic inputs (Th), before and after HFS delivery to the cortical inputs (heterosynaptic LTP) without delivering a foot shock (n=4; Paired t-test, p-value=0.9437). Results are reported as mean ±S.E.M.

a,b) Diagram and histology of the brain sections showing the AAVs injection sites, and the optetrode implantation sites. c) Representative examples of the EPSP amplitude recorded in the LA by stimulation of cortical axons (left, repeated-measures Two-way RM ANOVA for group by light intensity, F (1.752, 3.505) = 4.302 p-value=0.1147 with Sìdak test correction), and thalamic axons (right, repeated-measures Two-way RM ANOVA for group by light intensity, F (1.936, 3.873) = 4.438 p-value=0.0999 with Sìdak test correction) before and after weak conditioning and cortical LTP. d) Plot for average in vivo field EPSP slope (normalized to baseline period) in the LA. Response was evoked by optical activation of the thalamic inputs before, 2hrs after, and 24hrs after a weak thalamic conditioning protocol. In the absence of HFS of the cortical inputs, there is no change in the field EPSP of the thalamic inputs (n=3; repeated-measures Two-way RM ANOVA for group by time interaction, F (1.644, 3.288) = 0.8228 p-value=0.4900 with Tukey correction). Superimposed traces of in vivo field response to single pulse optical stimulation before (black line), 2hrs after (dash orange line) and 24hrs after (solid orange line) HFS. e) Representative example of the EPSP amplitude recorded in the LA by stimulation of thalamic axons (repeated-measures Two-way RM ANOVA for group by light intensity, F (10, 10) = 1.235 p-value= 0.3727). Results are reported as mean ±S.E.M. #, p<0.1; **, p<0.01; ***, p<0.001.