Chronic activation of fear engrams induces extinction-like behavior in ethanol-exposed mice

Alcohol withdrawal directly impacts the brain’s stress and memory systems, which may underlie individual susceptibility to persistent drug and alcohol-seeking behaviors. Numerous studies demonstrate that forced alcohol abstinence, which may lead to withdrawal, can impair fear-related memory processes in rodents such as extinction learning, however the underlying neural circuits mediating these impairments remain elusive. Here, we tested an optogenetic strategy aimed at mitigating fear extinction impairments in male c57BL/6 mice following exposure to alcohol (i.e., ethanol) and forced abstinence. In the first experiment, extensive behavioral extinction training in a fear-conditioned context was impaired in ethanol-exposed mice compared to controls. In the second experiment, neuronal ensembles processing a contextual fear memory in the dorsal hippocampus were tagged and optogenetically reactivated repeatedly in a distinct context in ethanol-exposed and control mice. Chronic activation of these cells resulted in a context-specific, extinction-like reduction in fear responses in both control and ethanol-exposed mice. These findings suggest that while ethanol can impair fear extinction learning, optogenetic manipulation of a fear engram is sufficient to induce an extinction-like reduction in fear responses.

These data suggest that aspects of ethanol withdrawal can promote ethanol-seeking behavior 62 potentially through mechanism of enhanced stress (Koob, 2008) where disruption of extinction 63 occurs, thereby potentially hampering clinical attempts to utilize extinction processes to treat 64 AUDs. 65 66 In rodents withdrawn from ethanol, preclinical studies also show impairments in memory 67 processes unrelated to the ethanol itself (e.g. fear learning). These deficits include impaired 68 extinction learning, heightened, stress-induced reinstatement, and increased fear generalization 69 impairments in ethanol-withdrawn rodents reflect retrieval deficits related to the extinction 74 memory (e.g., enhanced fear generalization) this may be manifested as increased freezing in a 75 neutral context (Jasnow, et al. 2017). In this case, targeting retrieval processes may prove 76 effective at mitigating the extinction impairments observed following withdrawal from ethanol. 77 Here, we employ a strategy that drives context-specific associative learning that may overcome 78 these retrieval errors by targeting the original fear memory (Chen et al, 2019). 79 80 Fear memory retrieval is a dynamic process that can be measured at the neuronal, circuit, and 81 behavioral level. Moreover, a behavioral strategy to suppress fear and induce extinction learning 82 involves returning a rodent to the fear-conditioning context in the absence of shock (Goode &  83 Maren, 2019). Repeated sessions where retrieval of a fear memory in the absence of the 84 unconditioned stimulus leads to an overall reduction in conditioned response (i.e., freezing 85 (Goode & Maren, 2019). An alternative method involves an artificial strategy, which attempts to 86 optogenetically activate a fear engram -the neuronal ensemble active during acquisition which 87 undergoes plasticity, and which reactivation of facilitate retrieval (Liu, et Figure 1A). 107

108
We observed no group differences in fear acquisition in Ctx A ( Figure 1B) however, there was a 109 main effect of trial demonstrating increases in freezing in the latter trials compared to baseline 110 and earlier trials (two-way RM ANOVA; F (4,72) = 54.98, p <0.0001). This was also the case in Ctx 111 B ( Figure 1C) (two-way RM ANOVA; F (4,72) = 15.54, p <0.0001). Following fear acquisition, mice 112 underwent 10-min extinction sessions in Ctx A (2 sessions per day for 5 days). Across extinction, we observed a gradual decrease in freezing levels ( Figure 1D) (two-way RM ANOVA, 114 session) (F (9,234) = 65.47, p <0.0001). Given our effects were stronger in the morning extinction 115 sessions, we followed up this analysis by looking at these sessions specifically and found a 116 similar effect (two-way RM ANOVA, session) (F (4,104) = 84.65, p <0.0001). Post-hoc analyses 117 revealed that EtOH-exposed mice were impaired in extinction learning. On the third day (S5), 118 EtOH mice froze significantly more than Sal mice (Tukey's HSD, p = 0.049). As the extinction 119 trials progressed these mice eventually started to reach normative levels (S7: Tukey's HSD, p = 120 0.082) ( Figure 1D). 121

122
Following extinction, mice were tested for contextual fear memory recall in Ctx A ( Figure 1E). As 123 expected, freezing levels of EtOH-exposed mice were significantly higher relative to Sal-control 124 mice (unpaired t-test, one-tailed, t (26) = 2.020, p = 0.0269). In contrast, no group differences 125 were found in freezing during the recall test in Ctx B (unpaired t-test, two-tailed, t (26) = 0.968, p = 126 0.3418) ( Figure 1F). Following behavior, cFos+ cells in the dDG were quantified as a proxy of 127 activation during recall in Ctx A ( Figure 1G-H). We observed no significant difference in this 128 cellular marker of activity during recall (unpaired t-test, two-tailed, t (5) = 0.629, p = 0.5567) 129 ( Figure 1G). Together, these data show that EtOH-exposed mice take longer to extinguish a 130 fear-conditioned freezing response, and a higher degree of freezing during recall which was 131 context-specific despite comparable levels of cFos activation in the dDG. group (unpaired t-test on difference score, two-tailed, t (18) =0.3030, p=0.7654; Figure 2E). 156

157
We measured freezing in Ctx A and Ctx B for ChR2 EtOH-exposed and Sal-control mice as well 158 as eYFP controls. Overall, we found a main effect of context, a main effect of virus, and no main 159 effect of treatment. We found a significant context by treatment interaction (three-way RM 160 ANOVA) (F (1,40) = 4.274, p = 0.0452). This effect is driven by the observation that across 161 treatment, there is a significantly lower degree of freezing in Ctx A for ChR2 animals (Tukey's 162 HSD, EtOH-ChR2: p < 0.001; Sal-ChR2: p = 0.032), suggesting that chronic optogenetic 163 activation induced a context-specific decrease in freezing. This effect was not observed in eYFP 164 mice. Additionally, there was a significant context x virus interaction (three-way RM ANOVA) (F 165 (1,40) = 7.721, p = 0.0083) and this effect was due to the difference in freezing within Ctx A, 166 between the EtOH ChR2 group and the Sal eYFP group (Tukey's HSD, p < 0.001) and between 167 ChR2 and eYFP mice overall (Tukey's HSD, ChR2 vs. eYFP: p < 0.001) ( Figure 2F). 168 169 Following behavior, cFos+ cells in the dDG were quantified as a proxy of activation during recall 170 in Ctx A ( Figure 2G-H). We observed no significant difference in cellular activity during recall 171 (unpaired t-test, two-tailed, t (6) = 1.6, p = 0.1606) ( Figure 2F). Together, these data demonstrate 172 that chronic activation of a tagged dDG fear memory leads to a context-specific reduction in 173 freezing in both ChR2 EtOH and Sal mice. 174

175
We utilized a behavioral and optogenetic approach to respectively evaluate and reduce aberrant 176 fear extinction responses in mice that experienced ethanol exposure and forced abstinence. 177 Behaviorally, repeated exposure to a fear-associated context led to fear retrieval. Although both 178 groups of mice acquired fear to a similar extent to both contexts, EtOH-exposed mice 179 demonstrated a slower rate of context extinction and an overall moderate impairment in 180 extinction memory relative to Sal-control mice. Next, chronic optogenetic activation of tagged 181 dDG cells induced a context-specific, extinction-like reduction of freezing behavior in both Sal-182 control and EtOH-exposed mice, but not in eYFP controls. Moreover, freezing levels in the 183 tagged context did not differ between ChR2 EtOH and Sal mice. These data lend credence to 184 the idea that artificially reactivating a fear memory over multiple sessions, is sufficient to reduce 185 fear memories in an extinction-like manner in both Sal-control and EtOH-exposed mice, and 186 points to dDG-mediated engrams as a key node, sufficient to alleviate maladaptive behavioral 187 responses. 188

189
We hypothesize that forced abstinence of ethanol may have caused withdrawal-induced 190 changes to stress systems in our mice, which led to a general enhancement in fear learning. It is possible that this increased fear learning was therefore more resistant to behavioral 192 extinction, but susceptible to optogenetic perturbations. Fear extinction impairments in the 193 EtOH-exposed mice reported here dovetails with previous reports of ethanol-withdrawal 194 memory impairments, including heightened fear responses, increased context generalization, 195 impaired extinction, and stress/cue-induced reinstatement in rodents (Holmes, et al. 2012;196 Bertotto Additionally, we found no difference in cFos+ cell quantifications between EtOH-exposed and 198 Sal-control groups in both the normative and optogenetic experiments. Though future studies 199 can increase sample sizes to potentially detect subtle differences across groups, it is likely an 200 upstream brain region mediates the behavioral impairment observed in normative EtOH- both EtOH-exposed and Sal-control mice. In line with this view, and despite our negative cFos 209 data in Figure 2, we propose that our chronic stimulation protocol bypassed numerous brain-210 wide systems previously implicated in mediating addiction-related behaviors, such as the insular 211 cortex, which governs interoceptive feelings of drug craving and withdrawal (Naqvi & Bechara 212 2009). Measuring immediate early genes in the insula may reveal differential activity levels 213 between EtOH-exposed and Sal-control mice, which potentially underlie the impaired ability of 214 EtOH-exposed mice to mitigate fear memories. promising value in better understanding the underlying mechanisms of learning and memory. In 241 particular, modulating addiction-related engrams permits a brain-wide cataloguing of the maladaptive structural and functional changes while pointing to key cellular mechanisms that 243 may be sites of future intervention (Whitaker & Hope 2018). By monitoring and manipulating the 244 cellular, circuit, and systems-wide changes that occur when a brain transitions into a state of 245 drug or alcohol-dependence, we believe it may be possible to artificially and enduringly restore 246 healthy neuronal functioning and corresponding behavioral outputs. 247

Acknowledgements 249
We thank Dr. Joshua Sanes and his lab at the Center for Brain Science, Harvard University, for 250 providing laboratory space within which the initial experiments were conducted, the Center for 251 Brain Science Neuroengineering core for providing technical support, and the Society of Fellows 252 at Harvard University for their support. We also thank Dr. Susumu Tonegawa and his lab for 253 providing the activity-dependent virus cocktail. This work was supported by an NIH Early 254