Examining memory linking and generalization using scFLARE2, a temporally precise neuronal activity tagging system

Abstract

How memories are organized in the brain influences whether they are remembered discretely versus linked with other experiences or whether generalized information is applied to entirely novel situations. Here, we used scFLARE2 (single-chain fast light- and activity-regulated expression 2), a temporally precise tagging system, to manipulate mouse lateral amygdala neurons active during one of two 3 min threat experiences occurring close (3 h) or further apart (27 h) in time. Silencing scFLARE2-tagged neurons showed that two threat experiences occurring at distal times are dis-allocated to orthogonal engram ensembles and remembered discretely, whereas the same two threat experiences occurring in close temporal proximity are linked via co-allocation to overlapping engram ensembles. Moreover, we found that co-allocation mediates memory generalization applied to a completely novel stimulus. These results indicate that endogenous temporal evolution of engram ensemble neuronal excitability determines how memories are organized and remembered and that this would not be possible using conventional immediate-early gene-based tagging methods.

Keywords: CP: Neuroscience; amygdala; coding schemes; engram ensembles; memory; memory generalization; memory linking; mouse; neuronal-activity-dependent tagging systems; optogenetics; scFLARE2.

Figure 1.. Using single-chain fast light- and activity-regulated expression 2 (scFLARE2) to tag lateral amygdala neurons active during a single threat-conditioning experience

(A) scFLARE2 tagging system. A transcription factor (TF; orange) induces transcription of a tetracycline (Tet)-inducible response element (TRE)-dependent target gene only in the presence of high Ca²⁺ and blue light (BL). Left: transmembrane domain (TM; turquoise) bound to intracellular components, a Ca²⁺-sensitive tobacco etch virus (TEV) protease (CaTEV; purple), a BL-sensitive light-oxygen-voltage (LOV) domain (blue), a protease cleavage site (TEVcs; green), and a TF. Under basal Ca²⁺ conditions without BL, the protease is inactive, and the cleavage site is masked by the LOV domain, such that the TF is tethered to the TM (no target gene transcription). Right: high Ca²⁺ activates CaTEV, and BL induces conformational change of the LOV domain, uncaging TEVcs. Proteolytic cleavage releases TF, allowing nuclear translocation and transcription of TRE-dependent target genes. (B) Identifying optimal BL parameters for scFLARE2 tagging of active neurons during threat conditioning. Mice expressing scFLARE2 and TRE-mCherry (mCherry) in lateral amygdala (LA) receiving auditory threat conditioning (tone CS + footshock). Mice received either BL (10 min, 0.25 Hz; 10 min, 20 Hz; 3 min, 20 Hz) or no BL (control) during training. An additional control group (non-mnemonic) received BL (3 min, 20 Hz) during contextual threat training (without tone). Memory tested in novel context and freezing before and during tone assessed. Neurons active in test (cFos⁺) in the mCherry⁺ (active during training) population compared. (C) During memory test, all mice showed high freezing during CS except control non-mnemonic mice (group [non-mnemonic, no BL, 10 min, 0.25 Hz; 10 min, 20 Hz; 3 min, 20 Hz] 3 test period [pre-CS, CS] interaction, F_4,35 = 6.42, p < 0.001, followed by post hoc). All groups (n = 8). (D) scFLARE2-tagged LA neurons (mCherry, red) and neurons active during memory retrieval (cFos, gray) in mice from control (non-mnemonic, no BL) and experimental (10 min, 0.25 Hz; 10 min, 20 Hz; or 3 min, 20 Hz BL) groups. DAPI nuclei (blue). Scale bar: 200 μm. (E) No-BL mice showed low number of scFLARE2-tagged neurons (mCherry⁺, left); all other groups showed similar numbers of scFLARE2-tagged neurons regardless of BL condition (p < 0.001). There was no difference in the number of active neurons during the test (cFos⁺, middle left) (p > 0.05). The number of double-positive neurons (cFos⁺ and mCherry⁺/DAPI⁺, middle right) (p < 0.001) and cFos⁺ neurons in the mCherry⁺ population was higher in the experimental (10 min, 0.25 Hz; 10 min, 20 Hz; 3 min, 20 Hz) than in the control (non-mnemonic, no BL) groups (right) (p < 0.001), but there was no difference between experimental groups, indicating similar reactivation of scFLARE2-tagged neurons during memory retrieval regardless of BL parameter. (F) Examining specificity of scFLARE2 tagging in discriminative threat conditioning. Mice expressing scFLARE2 and TRE-mCherry were identically trained. In novel contexts, CS⁻ (7.5 kHz, pips) was presented alone, and CS⁺ (2.8 kHz, tone) was paired with footshock. BL-tagged neurons active during CS⁻ (CS⁻ BL) or CS⁺+shock (CS⁺ BL). Time spent freezing before and during CS⁺ was tested, and neurons active during the test were identified with cFos immunohistochemistry. (G) During test, both groups, regardless of when active neurons were tagged, froze to CS⁺ at similarly high levels (group [CS⁻ BL, CS⁺ BL] × test period [pre-CS⁺, CS⁺] ANOVA; no significant interaction, F_1,15 = 1.96, p > 0.05, or main effect of group, F_1,15 = 0.0094, p > 0.05, but significant main effect of test period, F_1,15 =173.8, p < 0.001). CS⁺ BL (n = 9); CS⁻ BL (n = 8). (H) LA neurons active during memory test (cFos⁺, gray) and scFLARE2-tagged neurons (mCherry⁺, red) active during either CS⁻ or CS⁺ training. DAPI nuclei (blue). Scale bar: 200 μm. (I) Higher mCherry⁺ (left) (p < 0.05, unpaired t test), cFos⁺ and mCherry⁺/DAPI⁺ (middle right) (p < 0.001), and cFos in mCherry⁺ population in CS⁺-tagged mice (right) (p < 0.001) but similar overall number of cFos⁺ neurons (middle left) (p > 0.05) after testing in both CS⁺- and CS⁻ -tagged groups, indicating specific reactivation of scFLARE2-tagged neurons. (J) Examining efficiency of scFLARE tagging in LA with pentylenetetrazole (PTZ). Mice expressing scFLARE2 and TRE-mCherry were treated with PTZ (50 mg/kg, s.c.) or saline (Sal) 60 min before BL stimulation. Neurons active (cFos⁺) in the mCherry⁺ population were calculated. (K) mCherry (red) and cFos (gray) in Sal and PTZ groups. DAPI (blue). Scale bar: 200 μm. (L) PTZ increased mCherry⁺ (p < 0.001, unpaired t test) and cFos⁺ (p < 0.001) neurons, cFos⁺ and mCherry⁺/DAPI⁺ (middle right) (p < 0.001), and cFos in mCherry⁺ population over Sal (right) (p < 0.001). n = 8 per group. All data are mean ± SEM. n.s., not statistically different, *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 2.. Manipulating scFLARE2-tagged neurons in LA during auditory threat conditioning to examine critical importance to memory

(A) Examining whether LA neurons tagged by scFLARE2 during auditory threat training are necessary for subsequent memory retrieval. Mice expressing scFLARE2 and TRE-NpHR3.0-mCherry (a red-light-sensitive inhibitory optogenetic construct and mCherry) in LA were given BL during auditory threat training (to tag active neurons). Subsequent freezing to tone CS either with or without red light (RL)-induced silencing of scFLARE2-tagged neurons was assessed in different groups. Neurons active during the test were identified with cFos immunohistochemistry. (B) During memory test, silencing tagged neurons disrupted freezing to tone CS (group [RL⁻, RL⁺] × test period [pre-CS, CS] ANOVA; interaction, F_1,15 = 19.17, p < 0.001, post hoc). RL⁻ (n = 8); RL⁺ (n = 9). (C) scFLARE2-tagged neurons during training (red) and neurons active during memory retrieval (cFos, gray) in RL⁺ and RL⁻ groups. DAPI nuclei (blue). Scale bar: 200 μm. (D) Similar mCherry levels (mCherry⁺/DAPI⁺, left) (p > 0.05), but number of neurons active during test (cFos⁺/DAPI⁺, middle) (p < 0.001) and cFos⁺ neurons in mCherry⁺ population (engram reactivation, right) (p < 0.001) decreased in RL⁺ inhibition group. (E) Examining necessity of scFLARE2-tagged neurons in memory retrieval using a within-subjects design. Mice expressing TRE-eNpHR3.0-YFP and scFLARE2 (scFLARE2⁺) or no scFLARE2 (scFLARE2⁻) were given BL during auditory threat training. Freezing to tone CS was assessed with or without RL. (F) scFLARE2-tagged LA neurons (NpHR3.0-YFP, green). DAPI nuclei (blue). Scale bar: 200 μm. (G) RL disrupted freezing in scFLARE2⁺, but not scFLARE2⁻, mice (group × test period ANOVA interaction, F_3,60 = 7.98, p < 0.001, followed by post hoc). scFLARE2⁻ (n = 10); scFLARE2⁺ (n = 12). (H) Examining whether artificial activation of LA neurons tagged by scFLARE2 during auditory threat training induces behavior similar to memory retrieval even in the absence of retrieval cues. Mice expressing scFLARE2⁺ (or not [scFLARE2⁻]) and TRE-ChRmine-mScarlet (RL-sensitive excitatory optogenetic construct) in LA were given BL during auditory threat training. Memory tested in novel context and freezing to RL alone and CS was assessed. (I) scFLARE2-tagged LA neurons (ChRmine-mScarlet, red) in scFLARE2⁻ and scFLARE2⁺ groups. DAPI nuclei (blue). Scale bar: 200 μm. (J) During test, scFLARE2⁻ mice showed no freezing to RL but froze to tone CS. In contrast, scFLARE2⁺ mice showed similar freezing to RL stimulation of neurons tagged during training and CS tone (group × test period ANOVA, F_3,48 = 5.51, p < 0.01, followed by post hoc). scFLARE2⁻ (n = 8); scFLARE2⁺ (n = 10). All data are mean ± SEM. n.s., not statistically different, *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 3.. Examining neural coding strategies mediating memory linking versus separating in mice with endogenous neuronal excitability

(A) Examining the importance of neurons tagged during one event to the retrieval of a second event with different intertraining intervals. Mice expressing scFLARE2 and TRE-eNpHR3.0-mCherry trained on two threat events (train 1 [CS1; 2.8 kHz tone + footshock] and train 2 [CS2; 7.5 kHz pips + footshock]) occurring 3 or 27 h apart. Neurons active during train 1 were tagged via BL. Memory for CS2 and CS1 was assessed with and without RL inhibition of scFLARE2-tagged neurons. (B) RL inhibition of scFLARE2-tagged neurons (train 1) disrupted CS2 freezing in 3 h mice, showing co-allocation to overlapping engram ensembles (group × test period ANOVA, F_3,51 = 8.53, p < 0.001, followed by post hoc). 3 h (n = 10); 27 h (n = 9). (C) RL inhibition of scFLARE2-tagged neurons (train 1) disrupted CS1 freezing in both groups (group × test period ANOVA, no significant interaction, F_3,51 = 0.74, p > 0.05, or main effect of group, F_1,17 = 0.50, p > 0.05, but significant main effect of test period, F_3,51 = 52.47, p < 0.001, followed by post hoc). (D) RL-induced suppression index ((CS_RL− CS_RL+)/(CS_RL−+ CS_RL+)) high in all conditions except 27 h group tested on CS2, indicating co-allocation of engram ensembles at 3 h but dis-allocation of engram ensembles at 27 h (group × test period ANOVA; F_1,34 = 7.69, p < 0.01). All data are mean ± SEM. n.s., not statistically different, *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 4.. Examine neural coding strategy mediating memory generalization

(A) Examining training parameters to induce memory generalization. Mice trained on two events (train 1 [CS1; 2.8 kHz tone + footshock] and train 2 [CS2; 7.5 kHz pips + footshock]) 3 or 27 h apart. Time spent freezing during novel intermediate CS3 (5.6 kHz tone) was assessed. (B) Freezing to novel CS3 higher in 3 than in 27 h group (group × test period ANOVA; no significant interaction, F_3,42 = 2.08, p > 0.05; significant main effect of group, F_1,14 = 8.20, p < 0.05, and test period, F_3,42 = 30.51, p < 0.001). 3 h (n = 8); 27 h (n = 8). (C) Examining importance of neurons tagged during one event to memory generalization in mice trained with two events at different intertraining intervals. Mice expressing scFLARE2 and TRE-eNpHR3.0-mCherry (RL-sensitive inhibitory opsin with mCherry) trained on two events (train 1 [CS1; 2.8 kHz tone + footshock] and train 2 [CS2; 7.5 kHz pips + footshock]) 3 or 27 h apart. Mice were given BL during train 2. Memory for CS1 and a novel tone (CS3; 5.6 kHz tone) was assessed with and without RL inhibition of scFLARE2-tagged neurons during train 2. (D) RL inhibition of scFLARE2-tagged neurons (train 2) disrupted CS3 freezing in 3 h mice, showing co-allocation to overlapping neurons important in memory generalization (group × test period ANOVA; F_3,57 = 3.33, p < 0.05). 3 h (n = 11); 27 h (n = 9). (E) RL inhibition of scFLARE2-tagged neurons (train 2) disrupted CS1 freezing in 3 h mice, showing co-allocation to overlapping neurons (group × test period ANOVA; F_3,57 = 4.42, p < 0.01). (F) RL-induced suppression index ((CS_RL− CS_RL+)/(CS_RL− + CS_RL+)) for CS1 and CS3 higher in 3 than in 27 h group (group × tone test ANOVA; no significant interaction, F_1,38 = 1.43, p > 0.05, or main effect of tone test, F_1,38 = 0.44, p > 0.05, but significant main effect of group, F_1,38 = 20.19, p < 0.001). (G) Correlation between CS1 and CS3 freezing (without RL inhibition) in 3 (red circle, linear regression, p < 0.05) but not in 27 h (gray triangle, linear regression, p > 0.05) group. All data are mean ± SEM. n.s., not statistically different, *p < 0.05, **p < 0.01, ***p < 0.001.