Optogenetic stimulation of the basolateral amygdala accelerates acquisition of object-context associations

Abstract

The basolateral complex of the amygdala (BLA) is capable of modulating memory and is thought to do so via projections to regions such as the hippocampus. The present study used optogenetic stimulation of glutamatergic projection neurons in the BLA as rats learned object-context associations during a well-studied hippocampus-dependent memory task. Relative to a control condition, optogenetic BLA stimulation resulted in accelerated acquisition of object-context associations when stimulation was delivered following correct choices but not when it was delivered during the intertrial interval. These results extend prior examples of amygdala-mediated memory enhancement to a canonical example of hippocampus-dependent memory and provide opportunity for future dissection of amygdalar modulation of object-context associative memory.

Declarative memory for stimuli with affective salience or for emotionally-arousing events tends to be boosted relative to neutral stimuli and events (Hamann, 2001). The hippocampus and adjacent cortical areas are essential for all declarative memories (Squire, Stark, & Clark, 2004). The boost for emotional memories additionally depends on the basolateral complex of the amygdala (BLA), including the lateral, basal, and accessory basal nuclei (LaBar & Cabeza, 2006; Pitkänen, Pikkarainen, Nurminen, & Ylinen, 2000). The BLA sends prominent glutamatergic projections to the temporal two-thirds of the hippocampus as well as to the entorhinal and perirhinal cortices (Pitkänen et al., 2000), and a widely-held view is that the BLA directly influences declarative memory by modulating activity in these regions (McGaugh, 2004; Paré, 2003). The BLA also innervates many other brain regions that could indirectly influence declarative memory, for example, frontal and striatal regions import for attention and motivation (LaBar & Cabeza, 2006). An important goal is to understand how activation of amygdala outputs to distributed brain regions can engage a coordinated collection of memory-enhancing responses.

In line with this goal, several previous studies in rats have found that direct electrical stimulation of the BLA following exploration of novel objects led to improved recognition memory relative to control objects when memory was tested the next day (Manns & Bass, 2016). A subsequent study in humans found a similar benefit to object recognition memory resulting from direct electrical stimulation of the amygdala, despite participants reporting no awareness of the stimulation (Inman et al., 2018). Several other studies in rodents have found that optogenetic stimulation of BLA neurons improved spatial memory, for example, for buried food (Kanta, Pare, & Headly, 2019) and escape hole (Wahlstrom et al., 2018) locations as well as during inhibitory avoidance learning (Huff, Miller, Deisseroth, Moorman, & LaLumiere, 2013). These studies using electrical and optogenetic BLA stimulation built on prior research demonstrating that post-training pharmacological BLA interventions could modulate subsequent performance on a variety of memory tasks (McGaugh, 2004; Paré, 2003; Roozendaal, Barsegyan, & Chen, 2018). The correspondence between the memory enhancement resulting from emotional arousal and direct interventions of the BLA suggests that outputs from the BLA play a central role in modulating declarative memory.

A fundamental feature of the mammalian hippocampal memory system is the capacity to associate spatial and nonspatial information funneled from widespread neocortical inputs to the hippocampus via the perirhinal, parahippocampal (postrhinal in rodents), and entorhinal cortices (Witter, Wouterlood, Naber, & Van Haeften, 2000). Indeed, a common view is that hippocampus-dependent declarative memory is typified by the formation of associations between nonspatial items and spatial contexts (Davachi, 2006; Knierim, Lee, & Hargreaves, 2006; Manns & Eichenbaum, 2006). One particularly well-studied test of this type of hippocampus-dependent memory is an object-context association (OCA) task (Komorowski, Manns, & Eichenbaum, 2009; Komorowski et al., 2013), in which rats learn to associate object X with context A and object Y with context B. Single-unit recording studies in rats performing the task have identified complementary contributions of many different brain regions to the task, including the hippocampus and entorhinal, perirhinal and orbitofrontal cortices (see McKenzie et al., 2016 for a combined analysis). Notably, the BLA strongly innervates many of these areas, and an important question is how activation of glutamatergic BLA neurons might enhance performance on this test of canonical hippocampus-dependent declarative memory. The present study addressed this question by using optogenetic stimulation of glutamatergic neurons in the BLA during acquisition of object-context associations in the OCA task.

Method

Subjects

Twelve 2-3 month-old male Long Evans rats weighing between 300-450 g were used. Rats were housed individually on a 12:12 light-dark cycle with testing conducted during the light period. Rats had free access to water but restricted access to food such that they were motivated for food rewards but never fell below 90% of their free-feeding weight. All procedures involving rats were approved by the Emory University Institutional Animal Care and Use Committee.

Surgery and histology

Rats underwent a sterile-tip surgery to infuse a viral vector (AAV₅-CaMKII-hCH2(H134R)-EYFP; UNC Vector Core) and implant optical fibers (200/230 nm, 0.66 NA; Plexon, Inc.) bilaterally into the BLA using a procedure similar to that used in Ahlgrim and Manns (2019). Rats were anesthetized with isoflurane and received pre- and post-operative buprenorphine (0.03 and 0.05mg/kg, respectively) and meloxicam (1mg/kg). A 33-gauge needle was used to inject 350nL of the virus in each BLA (−3.5AP, ±5.1ML, −8.9DV) at 150nL/min and then left in place for 10min. Optical fibers were implanted 0.5mm dorsal to the infusion site. Rats were monitored for 3 days and allowed to recover for a week before behavioral testing. Following the experiment, all brains were perfused with fixative, extracted, and sectioned at 40 μm for staining sections with cresyl violet or for acetylcholinesterase. Virus expression was confirmed by either endogenous fluorescence (n=7) or immunohistochemical labeling (n=4) of the EYFP virus reporter. Expression was not apparent in a twelfth rat, whose data were excluded.

Procedure

Figure 1 shows a schematic of the hippocampus-dependent OCA task procedure, which was adapted from Komorowski et al. (2009; 2013). On each acquisition testing day, rats learned a new set of X-A and Y-B object-context associations across 80 trials. The objects consisted of glass jars filled with a digging medium (e.g., plastic beads) and scented with a fragrant oil (e.g., lavender), and the contexts were two visually distinct boxes connected by a tunnel. Prior to the experiment, rats were pre-trained on one X-A/Y-B object-context association to habituate them to the general procedure. Each session of object-context association learning in the actual experiment involved new objects (novel odors and media) and altered contexts (i.e., changed black-and-white patterns on the enclosure walls). In the initial phase of the experiment, stimulation was delivered to the BLA immediately following each correct response, after rats retrieved the buried food reward. All rats completed two (experimental and control) counterbalanced acquisition days separated by at least 8 days using different objects and altered contexts. During the experimental condition, BLA stimulation consisted of blue (465 nm) light capable of exciting the transfected ChR2 opsins (Mattis et al., 2012). During the control condition, stimulation consisted of near-infrared (850 nm) light outside the excitation spectrum for ChR2 opsins (Mattis et al., 2012). Power at the optical fiber tip was approximately 11 mW for the blue LED and 7 mW for the near-infrared LED. In both conditions, the 1-s stimulation included eight on-off cycles in which the on period lasted 62.5 ms and consisted of 50-Hz light pulses, which was based on a prior study that observed theta-modulated gamma oscillations in the hippocampus in response to blue-light but not near-infrared BLA stimulation using a similar optogenetic approach (Ahlgrim & Manns, 2019). Longer (5 s) BLA stimulation in the prior study resulted in decreasing hippocampal gamma power across the stimulation period, and therefore the current study focused on shorter (1 s) stimulation. In another prior study, theta-modulated hippocampal gamma oscillations were previously observed to correlate with OCA task performance (Tort, Komorowski, Manns, Kopell, & Eichenbaum, 2009). In both experimental and control conditions, the acquisition day was followed one week later by a retention test in which no stimulation was delivered. In the second phase of the experiment, rats again completed two (experimental and control) counterbalanced acquisition days separated by at least 8 days using different objects (novel odors and media) and altered contexts. In the second phase, the procedure was the same as in the initial phase except that stimulation was delivered during the intertrial interval (ITI), when rats were in the tunnel connecting the contexts.

Figure 1. Experimental approach.

A. Object-context association (OCA) task. Context A and context B differed in the black-and-white paneling on the walls and their fixed location in the testing room. Objects were 250 mL glass jars filled with unique scented digging media. In context A, object X contained a buried piece of sweetened cereal. In context B, object Y contained a buried piece of sweetened cereal. B. Optogenetic stimulation. A viral vector (AAV₅-CaMKII-hCH2(H134R)-EYFP) was infused and optical fibers were implanted bilaterally in the BLA. One second of blue light (or near-infrared light in the control condition) was delivered for each stimulation. Virus expression was confirmed by either endogenous fluorescence or immunohistochemical labeling of the EYFP reporter. See Method for details. Expression was visible in BLA cell bodies as well as in presumptive BLA axons in major target regions (e.g., prefrontal and entorhinal cortices), particularly in the ventral hippocampus.

Results

Figure 2 shows performance (n=11) across 8 blocks of 10 trials each on the OCA task for the acquisition and one-week retention test days. Rats performed significantly better during acquisition when receiving blue light BLA stimulation as compared to the near-infrared light control. Repeated measures general linear models were conducted on the 8 blocks of averaged performance data with block and light color as within-subject factors. Analysis of the acquisition results indicated that both block (F(7,70)=16.985; p<0.001; partial η²=0.629) and stimulation condition (F(1,10)=6.222; p=0.032; partial η²=0.384) had a significant effect. Moreover, there was a significant interaction of linear trends (F(1,10)=6.541; p=0.028; partial η²=.395), reflecting the faster learning in the experimental condition. A follow-up analysis to assess potential complications of testing order indicated that the testing order (counterbalanced experimental first or control first) was not a statistically-significant factor (F(1,9)=0.134; p=0.723; partial η²=0.015) and the effects of block, stimulation condition, and their linear interaction all remained statistically-significant (p<.05) when testing order was included in the model, a lack of carryover between experimental and control conditions that likely resulted from the use of new objects (novel odors and media) and altered contexts in each condition.

Figure 2. Effect of optogenetic (black) and control (gray) BLA stimulation on performance.

A. Acquisition and one-week retention performance on the OCA task. Stimulation was delivered to the BLA for 1 s after each correct digging choice during acquisition (n=11). B. Performance on OCA task when BLA stimulation was delivered at the end of the intertrial interval (ITI; in the tunnel connecting contexts) as rats learned new object-context associations (n=10). Across all graphs, circles show means, and error bars show SEM. Stepped lines with an asterisk denote a statistically significant (p<.05) effect of trial block. Crossed lines with an asterisk denote a statistically significant (p<.05) interaction of linear trends (see Results).

In comparison to their acquisition performance, rats performed similarly on the one-week retention test between experimental and control stimulation conditions (F(1,10)=0.233; p=0.640; partial η²=0.023), and there was no significant linear interaction between stimulation condition and block (F(1,10)=0.058; p=0.815; partial η²=0.006). Performance did increase over blocks during the retention test for both stimulation conditions (effect of block: F(7,70)=13.861; p<0.001; partial η²=0.581). Thus, optogenetic stimulation of the BLA during acquisition of object-context associations improved acquisition but not retention performance.

To next ask if the benefit of optogenetic stimulation during acquisition could be explained by generic or context-independent factors, optogenetic BLA stimulation (or control stimulation) was delivered to rats during the intertrial interval (ITI; in the tunnel connecting the A and B contexts) as rats were trained to acquire new object-context associations. Similar to the initial phase of the study, all rats received optogenetic and control stimulation conditions in a counterbalanced order, and both conditions used new objects (novel odors and media) and altered contexts (i.e., changed black-and-white patterns on the enclosure walls). The number of optogenetic and control stimulations were yoked to each rat’s performance in the first phase of the experiment. Figure 2 (panel B) shows performance (n=10; one rat became sick) across 8 blocks of 10 trials each on the OCA task for the acquisition test day (a one-week retention test was not administered). Analysis of the context-independent stimulation results indicated that block had a significant effect (F(7,63)=15.526; p<0.001; partial η²=0.633). However, stimulation condition was not a statistically-significant effect (F(1,9)=0.453; p=0.518; partial η²=0.048), and there was no significant linear interaction between stimulation condition and block (F(1,9)=0.274; p=0.614; partial η²=0.030). Thus, BLA stimulation delivered outside of the A and B contexts did not significantly improve performance relative to the control condition.

Discussion

The current study demonstrated that 1-s optogenetic stimulation of the BLA immediately following correct choices during the OCA task accelerated acquisition of object-context associations. Good performance on the OCA task depends on the hippocampus (Komorowski et al., 2013), and the broader category of item-context associations has been frequently argued to represent a prototypical example of memory supported by the hippocampus and associated structures (Davachi, 2006; Knierim et al., 2006; Manns & Eichenbaum, 2006). Moreover, a recent study found that the same optogenetic BLA stimulation used here triggered theta-modulated gamma oscillations in the hippocampus (Ahlgrim & Manns, 2019), an oscillatory pattern previously associated with good performance on the OCA task (Tort et al., 2009). Thus, taken together with past findings, the present results indicate that the BLA can enhance object-context associative memory at least in part by briefly modulating memory states in downstream regions such as the hippocampus. One possibility is that increased levels of theta-modulated gamma oscillations in the hippocampus might benefit spike-timing dependent plasticity at recently active synapses (Bass and Manns, 2015).

Optogenetic BLA stimulation accelerated acquisition but did not improve retention of object-context associations when memory was tested without stimulation one week later (Fig. 2). Past research has highlighted the BLA as an important modulator of memory consolidation (McGaugh, 2004), and several previous studies have specifically found that pharmacological activation of the BLA following a study session improved subsequent object (Roozendaal et al., 2018) and object-context recognition memory (Barsegyan, McGaugh, & Roozendaal, 2014). The present results complement these prior findings to the extent that the present study manipulated BLA activity throughout acquisition rather than after encoding opportunities had ended. Thus, the benefit of BLA stimulation to acquisition but not retention performance may in part be explained by the current study’s intervention during the stage of memory formation rather than memory consolidation. Indeed, to the extent that brief BLA stimulation can trigger transient pro-memory oscillatory states in the hippocampus (Ahlgrim & Manns, 2019; Bass and Manns, 2015), it represents one of likely several BLA-mediated mechanisms of memory modulation.

Nevertheless, given that BLA stimulation accelerated acquisition in the present study, it is unclear why that benefit was not visible a week later. Indeed, based on the idea that immediate small synaptic changes might cascade into more robust changes over time (e.g., Frey and Morris, 1997; Richter-Levin & Akirav, 2003), it was hypothetically possible that the experimental-control difference would have been even bigger on the retention test as compared to the end of the acquisition test. One possible explanation for the lack of an effect at retention is that the acquisition-retention interval was too long and that forgetting erased the benefit. The one-week interval was chosen because pilot testing suggested ceiling effects at intervals as long as three days and because the potential to observe savings (faster relearning) during the retention test was hypothesized to mitigate floor effects. However, it is possible that using acquisition-retention intervals between three and seven days might have revealed an residual effect of optogenetic BLA stimulation.

In a second experiment in the present study, optogenetic BLA stimulation delivered during the ITI, when rats were outside the training contexts, did not result in statistically significant improvement in acquisition performance relative to the control condition. The absence of a significant benefit from ITI BLA stimulation indicates that a general boost to arousal or circulating hormone levels (e.g., via polysynaptic activation of the sympathetic nervous system) was unlikely to account for the benefit observed when BLA stimulation was delivered immediately following correct responses. A caveat is that this second experiment involving ITI BLA stimulation was conducted after all rats had already undergone testing with BLA stimulation following correct choices. New objects and altered contexts were used in the ITI stimulation experiment, and any complications arising from testing order effects would have applied equally to optogenetic and control BLA ITI stimulation conditions. Nevertheless, non-specific factors such as increasing fluency with the general procedure or increasing memory interference across different object-context associations cannot be ruled out as part of the explanation for the contrasting efficacy (relative to control) of BLA stimulation delivered after correct choices versus during the ITI.

The present study capitalized on past findings regarding the fundamental role of the hippocampus in acquiring object-context associations and the capacity of the BLA in modulating hippocampus-dependent memories. In particular, an analysis of spiking data from multiple brain regions found that firing rates of neurons in orbitofrontal cortex, lateral entorhinal cortex, medial entorhinal cortex, and hippocampus related most specifically to reward, object identity, object location, and object-context combinations, respectively (McKenzie et al., 2016). Thus, a good deal is known about how multiple brain regions contribute to the OCA task. Moreover, the BLA sends substantial projections to these regions in addition to the hippocampus (Pitkänen et al., 2006; Kita and Kitai, 1990). These BLA projections are unlikely to be essential for normal learning on the standard OCA task as used here, as indicated by the relatively good performance in the present control condition. However, the sufficiency of stimulating these projections for enhancing performance offers an opportunity to dissect subsets of BLA-mediated memory mechanisms, strands of what the BLA ordinarily offers during emotional arousal. It is also an opportunity to ask how target regions might each contribute to the overall performance enhancement observed in the present study. Thus, the current study highlighted the capacity of BLA glutamatergic projections to enhance canonical hippocampus-dependent object-context memories and opened a path to charting how the constellation of BLA targets each contribute to the enhancement.