Optogenetic inhibition of either the anterior or posterior retrosplenial cortex disrupts retrieval of a trace, but not delay, fear memory

Abstract

Previous work investigating the role of the retrosplenial cortex (RSC) in memory formation has demonstrated that its contributions are not uniform throughout the rostro-caudal axis. While the anterior region was necessary for encoding CS information in a trace conditioning procedure, the posterior retrosplenial cortex was needed to encode contextual information. Using the same behavioral procedure, we tested if there was a similar dissociation during memory retrieval. First, we found that memory retrieval following trace conditioning results in increased neural activity in both the anterior and posterior retrosplenial cortex, measured using the immediate early gene zif268. Similar increases were not found in either RSC subregion using a delay conditioning task. We then found that optogenetic inhibition of neural activity in either subregion impairs retrieval of a trace, but not delay, memory. Together these results add to a growing literature showing a role for the retrosplenial cortex in memory formation and retention. Further, they suggest that following formation, memory storage becomes distributed to a wider network than is needed for its initial consolidation.

The retrosplenial cortex (RSC) has received an increasing amount of attention for its role in associative learning and memory (Fournier, et al., 2019; Keene & Bucci, 2008a; Kwapis, et al., 2014, 2015; Todd, et al., 2015; 2016; Trask, et al., 2021; see Smith, et al., 2018; Todd & Bucci, 2015, for reviews), after being extensively studied for its role in spatial learning and navigation (Neave, et al., 1994; Miller, et al., 2020; Vann & Aggleton, 2004; Vann, et al., 2003). An overwhelming amount of evidence demonstrates that the RSC is needed for complex forms of associative learning, like learning that a context predicts an unconditional stimulus (Keene & Bucci, 2008b; Robinson, et al., 2012; Todd, et al., 2017) or that a conditional stimulus (CS) and unconditional stimulus (UCS) separated in time by a trace interval are associated (e.g., trace conditioning; Kwapis, et al., 2014; Trask, et al., 2021). Simpler forms of conditioning in which a CS coterminates with a UCS (i.e., delay fear conditioning) do not rely on the RSC for memory formation (Kwapis, et al., 2014; 2017).

Previous work shows that spatial learning relies preferentially on the posterior, but not anterior, portion of the RSC (Neave, et al., 1994; Vann & Aggleton, 2004; Vann, et al., 2003). Conversely, the anterior region seems to be especially important for object memory (de Landeta, et al., 2020). However, the role of each RSC subregion in other types of associative learning is less clear. While a majority of the research examining the contributions of this region to associative learning has examined the effects of complete RSC lesions, emerging evidence suggests that anterior and posterior regions of the RSC can have differential roles in memory acquisition. While infusions of a protein synthesis inhibitor (i.e., anisomycin) into the RSC immediately before trace fear conditioning can disrupt later freezing to both the CS and the context (Kwapis, et al., 2015), optogenetic inhibition of the anterior RSC during training disrupted CS-elicited responding during testing whereas inhibition of the posterior RSC disrupted context-elicited responding (Trask, et al., 2021). Together, these results suggest that neural activity in the anterior and posterior RSC is critical during training trials and depends on the content of the learning for the acquisition of a trace fear response. The pattern of results in our optogenetic studies suggest that the anterior region of the RSC is needed for the encoding of event-related information and the posterior RSC is necessary for encoding context-related information.

The dissociation between event- and context-related information during acquisition is likely a result of anatomical and functional differences between the anterior and posterior RSC (aRSC and pRSC, respectively). While the aRSC is highly connected with regions necessary for the acquisition of cue-elicited responding like the ACC and mPFC (Gilmartin, et al., 2013; Milad, et al., 2007a; 2007b; Tang, et al., 2005), the pRSC shares dense reciprocal connections with areas important for spatial and context learning, like the hippocampus and the entorhinal cortex (Czajkowski, et al., 2013; Kononenko & Witter, 2012; Sugar, et al., 2011). Despite differential roles for the anterior and posterior RSC during memory acquisition, it has been hypothesized that information becomes integrated in the RSC following consolidation, and recall may therefore require both the aRSC and pRSC. Activity in the RSC is necessary for expression of memory following multiple types of complex associative learning, including contextual conditioning (Fournier et al., 2019) and discrimination learning (Todd et al., 2015; see Todd & Bucci, 2015, for a review). In line with the rodent literature, the RSC is important for episodic memory retrieval in both humans (Foster et al., 2013; Rosenbaum et al., 2008; Valenstein et al., 1987) and non-human primates (Buckely & Mitchell, 2015; Hussin et al., 2021).

Less work has thoroughly examined how subregions along the anterior-posterior axis of the rodent RSC might differentially contribute to retrieval, rather than acquisition, of conditional fear memory. First, we examined how trace and delay fear retrieval impacted neural activity in the anterior and posterior regions of the RSC using expression of the immediate early gene zif268. We then tested if each subregion was necessary for memory retrieval of either a trace or delay fear memory by inhibiting activity in either region when memory was tested during a retrieval session where the CS was presented alone. We hypothesized that optogenetic inhibition of the anterior, but not posterior, RSC during retrieval would result in decreased freezing to the CS in animals trained with trace fear conditioning, but similar inhibition would not impact behavior after training with delay fear conditioning.

Methods

Subjects.

Subjects were male Long-Evans rats (300–400g at the time of receipt; Envigo, IN). Animals were housed individually in plastic cages with chip bedding and free access to food and water. The room where animals were housed was maintained on a 14:10 light/dark cycle.

Surgical Procedures.

The AAV9-CAG-ArchT-GFP or AAV9-CAG-GFP recombinant virus (obtained from the University of North Caroline Vector Core; titer: 2×10¹² molecules/ml) was infused into either the aRSC or pRSC. Rats were anesthetized with isoflurane (induction: 4%; maintenance: 2–2.5%) and placed in a stereotaxic frame. Two 0.5-mm diameter holes were drilled in the skull above either the anterior or posterior retrosplenial cortex (one on each hemisphere). Coordinates for the anterior RSC infusions were 0.5mm lateral, 1.6mm ventral, and 2.6 posterior with respect to bregma. Coordinates for the posterior RSC infusions were 1.0mm lateral, 1.8mm ventral, and 5.6 posterior with respect to bregma. Using a 10-ul syringe and a 33-gauge needle (World Precisions Instruments, Sarasota, FL), 0.5 ul of either the ArchT or control virus was injected at a rate of 0.05nl/min. After injection, the needle was left in place for an additional 10 minutes to allow for virus diffusion away from the injector. Following virus infusion, the incision was sutured and each animal was given six weeks to allow for optimal virus expression.

Groups received a second surgery approximately 6 weeks following virus infusions to implant optical fibers into the site where virus was previously infused. Fibers were secured to the skull with four skull screws surrounded by acrylic cement. Rats were allowed 5–7 days of recovery following fiber implantation prior to behavioral training and testing.

Animals in the immunofluorescence experiments (Figure 1) did not receive either surgery and only received the behavioral procedure outlined below without light delivery or a second retrieval test.

Figure 1.

Trace, but not delay, fear retrieval results in an increase in zif268 expression in both the anterior and posterior retrosplenial cortices. A) Experimental design. Animals were trained in a trace fear conditioning task then given CS retrieval the next day. B) Mean freezing scores during the trace fear training session. C) Mean freezing scores during the trace fear retrieval session. D) Quantified zif268 (red) expression in the anterior RSC calculated as a percentage of the total amount of DAPI staining (blue) present on the same slice and normalized to the no retrieval condition. Representative images from both the No Retrieval and Retrieval groups in the aRSC. DAPI staining is in blue and zif268 is in red. E) Quantified zif268 expression and representative images in the posterior RSC. F) Experimental design. Animals were trained in a delay fear conditioning task then given CS retrieval the next day. G) Mean freezing scores during the delay fear training session. H) Mean freezing scores during the trace fear retrieval session. I) Quantified zif268 expression and representative images in the anterior RSC. J) Quantified zif268 expression and representative images in the posterior RSC. * indicates p < 0.05 from No Reactivation control.

Behavioral Procedure.

Following recovery from fiber implantation, animals were placed in a Med Associates (St. Albans, VT) conditioning chamber (30.5 × 24.1 × 29.2 cm) housed in individual sound attenuating chambers. Chambers were illuminated with an incandescent house light and exhaust fans provided a 65-dB background noise. A scent was presented by cleaning each chamber with a 5% ammonium hydroxide solution immediately before the animal was placed in the chamber. For the trace conditioning procedure, rats received 6 CS–UCS pairings following a six-minute baseline period. The CS was a 10-s 72dB white noise stimulus played from a speaker mounted to the wall of the sound-attenuating chamber. The UCS was a 1.0 mA footshock. A 20-s trace interval period separated each CS and UCS and the ITI between these pairings was an average of 240 s. Animals remained in the chamber for four minutes following the final footshock. The delay procedure was similar, except that now the 10-s CS co-terminated with the footshock and only 4 CS-UCS pairings were given, based on previous work that demonstrated equivalent freezing levels following delay and trace fear conditioning using these parameters (Kwapis, et al., 2017). The ITI was an average of 110 s.

To assess auditory fear retention, each animal was tested 24 hrs later in a novel context (20.5 × 26.5 × 21 cm) for conditional freezing to the white noise CS. These conditioning boxes were housed in a sound attenuating chamber in a separate room in the laboratory. Each chamber was illuminated and had a black plexiglass flooring. Chambers were cleaned with 100% ethanol immediately before the animals were placed in the lab. Following a one-minute baseline period, animals received four 30-s CS presentations (with an average ITI of 60s). Animals remained in the chamber for one minute following the final CS presentation. In optogenetics experiments, light delivery was delivered as described below. 24 hrs following this test, animals were given a second retrieval session that was identical with the exception that no light was delivered.

Light Delivery.

The LEDs were controlled via TTL pulses from a Med Associates computer (Med Associates, St. Albans, VT). Fibers were plugged in to the patch cord and placed in the chambers at the beginning of the training session. Light exposure (540 nm; 10mw) began 1 s before each CS presentation and lasted until 1 s following the cessation of each CS.

Immunofluorescence.

A subset of animals was deeply anesthetized with isoflurane 60 minutes following memory retrieval to examine changes in activity in the aRSC and pRSC following retrieval. This was measured by examining zif268 expression and comparing it to animals who received the same behavioral training the day before but no subsequent retrieval test. Animals in the No Retrieval condition were matched to animals in the Retrieval condition for time of sacrifice and tissue collection. Other animals were sacrificed 60 minutes following TFC with either the light on or off during CS-UCS trials to examine the efficacy of the optogenetic inhibition using fibers using procedures similar to those used to examine optogenetic inhibition using surface LEDs in our previous work (Trask, et al., 2021). Brains were immediately removed and stored at −80°C until sliced in 20-micron sections and mounted onto charged slides. Slides were rehydrated in wash buffer (PBS + 0.05% Tween-20) and permeabilized (PBS + 0.3% Triton X) for 15-min and incubated in blocking solution (PBS + 0.7% NGS). Slides were then incubated in zif268/EGR1 antibody (Cell Signaling, 1:500, #4153) solution (PBS + 0.3% Triton X + 5% NGS) overnight at 4°C. The next day, slides were incubated in secondary antibody solution for 2 hours and rinsed with wash buffer, a DAPI counterstain was applied, and slides were cover slipped. Images were captured on the Olympus Fluoview FV1200 confocal microscope using a 20x objective lens. Serial z-stack images covered a depth of 4.55μm through five consecutive sections (0.91μm per section) and were acquired using Fluoview software (Olympus). For verification of the optogenetic inhibition, eight slides were collected to represent 1mm of tissue along the A/P axis (e.g., −1, −2, etc.), covering the entire infusion area for both anterior and posterior groups. 12 images (6 from the left hemisphere, 6 from the right hemisphere) were taken for each slide for each animal (with exceptions being made for damaged tissue), giving a detailed representation of activity along the anterior/posterior axis. For retrieval-induced IEG expression, four sagittal sections (two from each hemisphere) were taken from each animal and one image was taken in the anterior (−2.6 bregma) and posterior (−5.6 bregma) RSC on each section. For all immunofluorescent experiments, zif268 activity was normalized as a proportion of DAPI present on the same section.

Statistical analysis.

All results were analyzed using analyses of variance (ANOVAs) or t-tests (as appropriate) using SPSS 25 (Statistical Package for Social Sciences; IBM) software, with alpha set to 0.05. Outliers were screened according to the methods outlined in Field (2005). Figures were created using Graphpad Prism.

Verification of Optogenetic Inhibition Efficacy.

In order to verify that the light-induced inhibition effectively reduced neural activity in that region and was selective to the targeted region without influencing activity throughout the RSC, we quantified zif268 throughout the RSC using immunofluorescence following temporary inactivation of either the anterior or posterior RSC during the CS-UCS trials in trace fear learning paradigm (Supplementary Figure 1). As in Trask, et al. (2021), each millimeter of tissue was taken as a separate observation between groups. One-way ANOVAs revealed significant group differences at all points, smallest F_{(2, 93)} = 3.40, p = .04, with the exception of −8 bregma, F < 1. Bonferroni corrections were used to determine where inactivation groups differed from no light controls. This found that animals with aRSC inactivation showed less zif268 expression only at −1 (p = 0.01), with trends towards a decrease in −2 and −4 (ps = 0.08, 0.07, respectively). pRSC inactivation did not produce decreases in these areas (ps = 1.00). pRSC inactivation produced decreased zif268 expression at −5, 6, and 7 bregma (ps = 0.005, 0.001, < 0.001, respectively). aRSC inactivation did not produce decreases in these regions (smallest p = 0.55). These results confirm that inactivation of each brain region resulted in selective inhibition of the areas surrounding virus infusion without affecting neural activity throughout the entire RSC.

Results

Trace, but not delay, fear retrieval results in increased neural activity in both the anterior and the posterior retrosplenial cortex.

First, we examined whether or not the retrosplenial subregions showed elevated activity as measured by zif268 expression following trace fear memory retrieval. The experimental design is depicted in Figure 1A. Animals acquired freezing similarly during training (Figure 1B), confirmed by a 2 (Group: No Retrieval, Retrieval) x 3 (Trial Type: Pre-CS, CS-UCS, Post-CS) ANOVA which found a main effect of trial type, F_{(2, 20)} = 172.54, MSE = 114.54, p < 0.001, η_p² = .95, but no main effect of group nor an interaction, Fs < 1. During retrieval, animals increased freezing when the CS came on, confirmed by a main effect of trial type, F_{(3, 15)} = 25.50, MSE = 192.29, p < 0.001, η_p² = .84. Trace fear retrieval resulted in increases in zif268 expression in the aRSC (Figure 1D), t₍₅₁₎ = 3.67, p < 0.001, and the pRSC (Figure 1E), t₍₅₂₎ = 2.90, p = 0.006.

We then examined if retrieval of a delay fear memory influenced activity in the RSC (experimental design depicted in Figure 1F). As was observed during trace fear conditioning, animals acquired freezing similarly during delay fear training (Figure 1G). This was confirmed by a 2 (Group: No Retrieval, Retrieval) x 3 (Trial Type: Pre-CS, CS-UCS, Post-CS) ANOVA which found a main effect of trial type, F_{(2, 20)} = 214.20, MSE = 113.07, p < 0.001, η_p² = .96, but no main effect of group nor an interaction, largest F = 1.60, p = 0.23. During delay fear retrieval (Figure 1H), animals increased freezing to the CS relative to the pre-CS period, confirmed by a main effect of trial type, F_{(3, 15)} = 22.90, MSE = 216.63, p < 0.001, η_p² = .82. Following delay fear retrieval, there were no increases in zif268 expression in either the aRSC (Figure 1I), t₍₄₄₎ = 1.64, p = 0.11, or pRSC (Figure 1J), t₍₄₅₎ = 0.10, p = 0.92.

Optogenetic inhibition of the anterior or posterior retrosplenial cortex impairs trace memory retrieval.

Acquisition (Figure 2C).

Figure 2.

Inactivation of either the anterior or posterior regions of the retrosplenial cortex reduces retrieval of a trace, but not delay, fear memory. A) Schematic of GFP expression showing the minimum (dark green) and maximum (light green) extent resulting from infusions into the anterior RSC (left panel) and the posterior RSC (right panel). B) Representative images of virus expression along the sagittal plane (4X magnification) and schematics of fiber placement. C) Animals did not differ during acquisition of a trace fear response. D) Optogenetic inhibition of either the anterior or posterior region of the retrosplenial cortex overlapping with CS presentations impaired responding to the CS. E) No persistent changes in fear responding were seen due to optogenetic inhibition. F) Animals did not differ in acquisition of a delay fear response. G) Optogenetic inhibition of either region had no impact on retrieval of a delay fear memory. H) No differences were observed in the light-free test following delay fear conditioning. * indicates p < 0.05 from GFP control.

All animals acquired responding throughout TFC. A 2 (Brain Region: aRSC, pRSC) x 2 (Virus: GFP, ArchT) x 3 (Trial Type: Pre-CS, CS-UCS, Post-CS) ANOVA found a main effect of trial type, F_{(2, 48)} = 159.39, MSE = 187.67, p < 0.001, η_p² = .87, but no other main effects or interactions, largest F = 1.03, p = .32.

Retrieval (Figure 2D).

A 2 (Brain Region: aRSC, pRSC) x 2 (Virus: GFP, ArchT) x 3 (Trial Type: Baseline, CS, ITI) found a main effect of trial, F_{(2, 48)} = 118.24, MSE = 239.58, p < 0.001, η_p² = .83, a main effect of virus, F_{(1, 24)} = 8.12, MSE = 941.58, p = 0.009, η_p² = .25, and a trial by virus interaction, F_{(2, 48)} = 3.83, MSE = 239.58, p = 0.03, η_p² = .14, but no other main effects or interactions, largest F = 1.59, p = 0.22. Follow-up comparisons found that inhibition of either the anterior (p = 0.06) or posterior (p = 0.04) RSC decreased freezing relative to controls only in the CS, but inhibition in either region did not impact freezing during the ITI (smallest p = 0.14). This was confirmed by follow-up comparisons that found that inhibition of the aRSC or the pRSC decreased freezing to the CS (p = 0.006). The disruption to freezing produced by light-induced inhibition was therefore restricted to the period in which the light was illuminated.

Light-free Test (Figure 2E).

A similar 2 (Brain Region: aRSC, pRSC) x 2 (Virus: GFP, ArchT) x 3 (Trial Type: Pre, CS, ITI) found a main effect of trial, F_{(2, 46)} = 83.99, MSE = 223.85, p < 0.001, η_p² = .79, but no other main effects or interactions, largest F = 2.22, p = 0.12, suggesting that the impairments from the light delivery on the preceding day did not have long-lasting impacts on behavior. Further, this demonstrated that the reduced freezing observed previously was due to the light-induced inhibition and not a consequence of pre-existing differences in memory acquisition.

Optogenetic inhibition of the anterior or posterior retrosplenial cortex has no impact on delay fear retrieval.

Acquisition (Figure 2F).

A 2 (Brain Region: aRSC, pRSC) x 2 (Virus: GFP, ArchT) x 3 (Trial Type: Pre-CS, CS-UCS, Post-CS) ANOVA found a main effect of trial type, F_{(2, 32)} = 277.61, MSE = 114.98, p < 0.001, η_p² = .95, but no other main effects or interactions, largest F = 1.93, p = 0.18, indicating that all groups acquired freezing and that this did not differ between groups.

Retrieval (Figure 2G).

Optogenetic inhibition during delay fear retrieval had no impact on responding. This was confirmed by a 2 (Brain Region: aRSC, pRSC) x 2 (Virus: GFP, ArchT) x 3 (Trial Type: Pre, CS, ITI) ANOVA that found a main effect of trial type, F_{(2, 32)} = 32.69, MSE = 330.38, p < 0.001, η_p² = .67, but no other main effects or interactions, largest F = 1.80, p = 0.20.

Light-free Test (2H).

Again, groups did not differ during the light-free test as demonstrated by a 2 (Brain Region: aRSC, pRSC) x 2 (Virus: GFP, ArchT) x 3 (Trial Type: Pre, CS, ITI) ANOVA which found a main effect of trial type, F_{(2, 32)} = 56.27, MSE = 267.18, p < 0.001, η_p² = .78, but no other main effects or interactions, largest F = 1.31, p = 0.28.

Discussion

First, we found that trace fear retrieval increased expression of immediate early gene zif268 in both the anterior and posterior RSC. No changes in zif268 expression were seen following delay fear retrieval in either the anterior or posterior RSC. We then found that inhibition of either RSC subregion during the CS presentations of a trace retrieval session selectively reduced freezing to the CS and left ITI freezing unaffected. This suggests that inhibiting RSC subregions during retrieval affects learned fear expression during inhibition, but inhibition does not have a lasting impact on memory recall. Similar inhibition during retrieval of a delay fear memory had no impact on performance, in line with work in humans suggesting a specific role for the RSC in episodic memory (Foster, et al., 2013; Rosenbaum, et al., 2008; Valenstein, et al., 1987).

Based on prior results demonstrating increased zif268 expression following exposure to a novel context (e.g., Asok, et al., 2013), it is possible that the delay fear retrieval task which takes place in a novel context could have resulted in increased zif268 expression. However, novel context exposure has not yielded increases in zif268 or c-fos expression in our hands (Trask & Helmstetter, in revision). Several methodological differences might be responsible for the disparity in these results, included their use of adolescent rats and examining zif268 mRNA expression rather than zif268 protein expression as in our studies.

While we hypothesized that only the aRSC would be selectively important for trace fear recall, inhibition of either the anterior or posterior retrosplenial cortices had a similar detrimental impact on trace memory during the retention test. While our prior results (Trask, et al., 2021) demonstrated that the anterior and posterior RSC had dissociable roles during acquisition of a trace fear memory, with the aRSC encoding CS information and the pRSC encoding context information, the present results demonstrate that retrieval of a previously-consolidated memory relies on the entire RSC. Given that CS testing occurs in a shifted context, it might also be the case that the pRSC-dependent context memory functions differently during testing to support memory retrieval. It has been suggested that the RSC is important for information binding during memory acquisition (Fournier, et al., 2019; Todd & Bucci, 2015). Following memory consolidation, it seems likely that the memory relies on a more distributed network, where reducing activity in one region influences the memory circuit to disrupt performance. This might suggest that although distinct RSC regions have dissociable roles in encoding different aspects of memory, there are no longer dissociable once this information becomes integrated following consolidation.

It is unclear from the above results how inactivation of either subregion might affect memory for the context in which training occurred. While our prior results demonstrating a selective role for the pRSC in acquisition of the contextual memory associated with trace fear conditioning, the present results suggest a less well-defined role for the retrosplenial subregions in memory retrieval. Additional research will be needed to assess how each subregion might differentially affect retrieval of contextual memory. One hypothesis is that the aRSC might be important for retrieval of the contextual information acquired during trace fear conditioning, as it becomes linked with the CS memory during consolidation, but the aRSC may play less of a role in retrieval of a contextual fear memory that occurs in the absence of any conditional stimulus.

Together, our results demonstrate a role for both the anterior and posterior regions of the retrosplenial cortex in trace, but not delay, memory retrieval. Trace retrieval increases activity in both regions, and inhibition of these regions coinciding with CS presentations during retrieval reduces responding. This adds to a growing literature demonstrating the role of the RSC in memory retrieval as well as points to a possible role for the RSC in integrating multiple aspects of memory along its rostro-caudal axis.

Supplementary Material

1

Supplementary Figure 1. A) Sagittal view highlighting the RSC along the anterior/posterior axis. B) Percent quantified zif268 expression (expressed as a proportion of total DAPI staining; mean and SEM) across the retrosplenial cortex in animals that had their aRSC inhibited or pRSC inhibited. * indicates p < 0.05, # indicates p < 0.10 from no light control.

Highlights:

Trace memory retrieval increased activity throughout the retrosplenial cortex.

Inhibition of anterior or posterior RSC regions impairs trace memory retrieval.

Similar effects are not seen in delay memory retrieval.

These results suggest the entire RSC supports trace, but not delay, memory retrieval.