Previous work suggests that activation patterns of neurons in superficial layers of the neocortex are more sensitive to spatial context than activation patterns in deep cortical layers. A possible source of this laminar difference is the distribution of contextual information to the superficial cortical layers carried by hippocampal efferents that travel through the entorhinal cortex and subiculum. To evaluate the role that the hippocampus plays in determining context sensitivity in superficial cortical layers, behavior-induced expression of the immediate early gene Arc was examined in hippocampus-lesioned and control rats after exposing them to 2 distinct contexts. Contrary to expectations, hippocampal lesions had no observable effect on Arc expression in any neocortical layer relative to controls. Furthermore, another group of intact animals was exposed to the same environment twice, to determine the reliability of Arc-expression patterns across identical contextual and behavioral episodes. Although this condition included no difference in external input between 2 epochs, the significant layer differences in Arc expression still remained. Thus, laminar differences in activation or plasticity patterns are not likely to arise from hippocampal sources or differences in external inputs, but are more likely to be an intrinsic property of the neocortex.
Monitoring the cellular distribution of behavior-induced immediate early gene (IEG) expression has been productive in at least 2 ways: First, it has become clear that some IEGs are crucially involved in the plasticity processes necessary for durable memory (e.g., Guzowski et al. 2000; Ploski et al. 2008). In addition, the expression of some of these genes can be used as an indicator of patterns of activated cells that reflect active information processing in the system. For example, the IEG Arc (Lyford et al. 1995), also known as Arg3.1 (Link et al. 1995), can monitor the degree to which cellular activity in specific regions of the brain differentiates between 2 separate episodes over 2 distinct points in time. For Arc mRNA, the analysis of the cellular compartment that this IEG resides in after behavior can be used to infer patterns of activated neurons during 2 distinct episodes of experience (Guzowski et al. 1999, 2001). This method of single cell imaging has been called compartment analysis of temporal activity by fluorescence in situ hybridization (catFISH). Using the catFISH method, Burke et al. (2005) showed that when a rat performs the same behavior in 2 different rooms, activation patterns of neuron populations in the CA1 region of the hippocampus and in superficial layers of posterior parietal (PPC) and granular insular (GI) cortices differ between the 2 episodes, while activation patterns of neuron populations in deep layers of these cortical regions do not. Episode differentiation by neuron populations in the hippocampus was expected (as shown by Guzowski et al. 1999), given the known firing patterns of hippocampal pyramidal neurons in specific spatial locations (O'Keefe and Nadel 1978; Wilson and McNaughton 1993). In contrast, mechanisms underlying the difference in episode differentiation between deep and superficial layers remain unclear.
One explanation for the difference in activity patterns observed across cortical layers by Burke et al. (2005) is that cells in superficial neocortical layers are more sensitive to the spatial context compared with cells in deeper layers. This different sensitivity to context may originate from afferent signals from the hippocampus, in which its output reaches rhinal cortices (including the entorhinal, perirhinal, and postorhinal cortices in rats) and subsequently projects differentially to superficial layers throughout the neocortex (Swanson and Kohler 1986; Amaral and Witter 1995; Insausti et al. 1997; Lavenex et al. 2002). This hypothesis predicts that damage to the hippocampus will eliminate laminar differences in Arc expression. Alternatively, it is possible that the laminar differences originate from an intrinsic property of neocortical networks that is independent of external input. In this case, the laminar difference will still exist even when animals are exposed to an identical environmental and behavioral context twice. To test these predictions, we used the catFISH method to examine the pattern of activated neurons during 2 distinct episodes of behavior in rats with hippocampal damage as well as during 2 identical episodes in intact rats. The present result suggests that laminar differences in activity patterns and/or the possible engagement of plasticity mechanisms are not likely to arise from hippocampal sources, but are more likely to be an intrinsic property of the neocortex.
Materials and Methods
Thirty-four, 5-month-old, male Brown Norway/Fisher 344 hybrid rats obtained from Harlan Sprague Dawley were handled for 1 week prior to behavioral training. During handling, all rats were food deprived to 80–85% of ad libitum weight. Weights were maintained for the remainder of the experiment, with the exception of a post-surgery recovery period. All animal handling and surgical procedures complied with National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee of the University of Arizona.
Apparatus and Training
Before rats received lesions of the hippocampus or sham lesions, they were trained to traverse a circular track (20 cm outer diameter, 10 cm width, 10 cm elevation) for a food reward for 7 days. This ensured that the animals formed the association between the track and food reward. On each day, the first training session began at 9 AM, and the second training session began at 4 PM. Both sessions took place in the same room, and the rats were required to run 15 laps each session. To control visual inputs during training, 2 patterns of room light conditions were used; in the Light-ON condition, the room light was on while rats were running on the track, and in the Light-OFF condition, the room light was on for 10 s to allow rats to look around the room they were in, and for the remainder of the traversals of the track, the rats ran in darkness. The Light-OFF condition was used to minimize the variance between rats with respect to visual stimulation. The Light-ON condition was used during the first 4 days of pre-training; the Light-OFF condition was used during the latter 3 days of pre-training.
During the first 3 days of Training, all rats were trained on the circular track as in pre-training. On day 1 and 2, the Light-ON condition was used, and on day 3, the Light-OFF condition was used. During the rest of 6 days of training, the rats were exposed to 2, 5-min epochs of track running on a rectangular track (60 × 20 cm, 10 cm elevation) in a clockwise direction for food reward (Fig. 1A). The 2 epochs were separated by a 20-min period in the home cage and took place in different rooms that had different sets of distal cues and brightness (room C and D). A new set of rooms (room A and B) was used from day 8 to keep the memory for spatial context “new” to maximize the dependence of the memory on the hippocampus (Bontempi et al. 1999; Maviel et al. 2004). The Light-ON condition was used on day 4, 5, and 8, and the Light-OFF condition was used on day 6, 7, and 9. On day 9, the rats were quickly anesthetized in a container of isoflurane immediately after the second 5-min track running epoch and decapitated with a rodent guillotine. Their brains were extracted, hemisected, and quickly frozen in a cold solution of isopentane (approximately −50 °C).
Thirty-four rats were divided into 5 groups (Fig. 1B). Two groups of rats received pre-training and then sham or hippocampal-lesion surgeries. After recovery, these 2 groups began training. On the last training day, they traversed the rectangular track in 2 different rooms immediately before they were sacrificed (AB-CONTROL, n = 6 and AB-HPC, n = 8, respectively). The third and fourth groups did not receive pre-training or surgery, but they received training. On the last training day, the third group traversed the rectangular track in the first room twice (AA, n = 6), and the fourth group did not traverse the track (CAGED, n = 6). The fifth group did not receive any behavioral training or surgery, but received maximal electoconvulsive shock treatment (MECS; 85 mA, 100 Hz, for 1 s) to activate all IEG expressing neurons 5 min before sampling of the brains (MECS, n = 6). This group was used as a positive control for FISH.
Ibotenic acid (Sigma-Aldrich, St. Louis, MO, United States of America) was dissolved in artificial cerebrospinal fluid to provide a solution with a concentration of 10 mg/mL and a pH of 7.4. Each rat was anesthetized with isoflurane (1–1.5% by volume in oxygen at a flow rate of 1.5 L/min), placed in a stereotaxic holder (David Kopf Instruments, Tujunga, CA, United States of America) with the skull surface in the horizontal plane. The skull was exposed by incision along the midline. Ten craniotomies (0.8 mm diameter) were drilled on each side of the skull (stereotaxic coordinates of the injection sites described by Jarrard (1989)). The most dorsolateral coordinate was −2.4 mm anteroposterior (AP), ±1.0 mm mediolateral (ML), and −3.0 mm dorsoventral (DV) to bregma, and the most ventrolateral coordinate was −5.4 mm AP, ±5.0 mm ML, and −6.1 mm DV. Ibotenic acid solution was injected via 10-µL Hamilton syringe mounted on a stereotaxic frame and held with a microinjector (World Precision Instruments, Sarasota, FL, United States of America). An injection needle was inserted into each of 10 craniotomies and volumes of 0.05 or 0.10 µL ibotenic acid solution were infused at either 1 or 2 different depths (Jarrard 1989). Infusion was made at a rate of 0.1 µL/min, and the needle was left in situ for an additional 1 min after completing each infusion. Ten rats received infusion of ibotenic acid solution (AB-HPC). Six rats (AB-CONTROL) underwent a similar surgical procedure, in which the skin was incised, the cortex exposed, and the dura perforated with a 25-gauge needle, but injection was not made. The rats were allowed to have a 2-week period for recovery. During the second week, all rats were handled again, and food deprived to 80–85% of ad libitum weight.
Fluorescence in Situ Hybridization
Activated cells during behavior were visualized by using high sensitivity FISH, which detects transcription of the IEG, Arc (Guzowski et al. 1999, 2001). Expression of Arc mRNA is very low at baseline, but Arc mRNA is expressed within ∼2 min of behaviorally induced activation of neurons and detected as 1 or 2 intensely staining intranuclear foci (Guzowski et al. 1999, 2001). The processed Arc mRNA subsequently accumulates in the cytoplasm, where it can be detected ∼20–45 min after induction (Guzowski et al. 1999, 2001). Furthermore, a second round of Arc transcription can be initiated in the same neurons when 2 experiences are separated by as little as 20 min (e.g., Guzowski et al. 1999; Vazdarjanova and Guzowski 2004). Coronal sections 20-μm thick were cut and arranged so that tissues from all groups were represented on a single slide. FISH was performed as previously described (Guzowski et al. 1999; Burke et al. 2005). Briefly, Arc antisense riboprobes were generated from the full length Arc cDNA (Lyford et al. 1995) using a commercial transcription kit (Maxiscript; Ambion, Austin, TX, United States of America) and RNA-labeling nucleotide mixes containing digoxigenin-labeled uridine-5’-triphosphate (UTP, Roche Molecular Biochemicals, Nutley, NJ, United States of America). The riboprobe was hybridized with the tissue overnight. The slides were treated with 2% H2O2 to quench any residual horseradish peroxidase (HRP) activity. Subsequently, the digoxigenin-labeled Arc riboprobe was detected with anti-digoxigenin-HRP conjugate (Roche Applied Science, Indianapolis, IN, United States of America) and revealed with cyanine-3 substrate kit (CY3 DirectFISH; PerkinElmer Life Sciences, Emeryville, CA, United States of America). The nuclei were counterstained with TOPRO (Molecular Probes, Eugene, OR, United States of America).
Image Acquisition and Manual Analysis
Images were collected from 20-μm thick sections of tissue using a Zeiss 510 Metaseries laser scanning confocal microscope with a ×40 oil objective. Photomultiplier tube assignments, pinhole size, and contrast values were kept constant for each brain region within a slide. Areas of analysis consisted of a 0.45 × 0.45-mm region brain on the XY scale and were Z-sectioned in 1.0-μm optical sections. As in the Burke et al. (2005) study, the superficial and deep layers of the PPC and GI were examined. These cortical regions were chosen because they receive inputs from predominantly different regions (with the exception of the hippocampus), while they are involved in memory encoding in a spatial task with food reward (McNaughton et al. 1994; Katz et al. 2001). Six images were collected from the deep and superficial layers of PPC between coordinates 4.6 and 5.6 mm posterior to bregma and 2.0 and 3.0 mm lateral to bregma. All GI images were collected between coordinates 0.0 and 2.8 mm posterior to bregma and ∼6.0 mm below the dural surface, dorsal to the rhinal fissure (Paxinos and Watson 1998). Images from superficial layers contained layers 2 and 3, while images from deep layers contained layers 5 and 6. Because there was no apparent difference in Arc-expression patterns between layers 2 and 3 or between layers 5 and 6 in each image, we did not separately analyze Arc-expression patterns in each layer. The mean number of cells sampled in each layer of the PPC and GI is summarized in Table 1. There was no significant difference in total number of counted cells, which included both cells with and without Arc-labeling, between the superficial and deep layers [2-way repeated-measures analysis of variance (ANOVA); PPC, F1,22 = 0.141, P > 0.05; GI, F1,22 = 0.003, P > 0.05] or across groups (PPC, F3,22 = 0.773, P > 0.05; GI, F3,22 = 0.524, P > 0.05).
|Superficial||559.7 ± 29.7||589.0 ± 39.8||483.1 ± 44.0||667.8 ± 50.8|
|Deep||610.0 ± 38.3||580.8 ± 37.4||452.9 ± 37.5||687.2 ± 47.4|
|Superficial||440.2 ± 33.5||453.5 ± 19.0||454.9 ± 39.8||482.3 ± 26.2|
|Deep||467.8 ± 24.1||482.3 ± 56.7||410.4 ± 26.4||467.2 ± 34.1|
|Superficial||559.7 ± 29.7||589.0 ± 39.8||483.1 ± 44.0||667.8 ± 50.8|
|Deep||610.0 ± 38.3||580.8 ± 37.4||452.9 ± 37.5||687.2 ± 47.4|
|Superficial||440.2 ± 33.5||453.5 ± 19.0||454.9 ± 39.8||482.3 ± 26.2|
|Deep||467.8 ± 24.1||482.3 ± 56.7||410.4 ± 26.4||467.2 ± 34.1|
Experimenters blind to experimental conditions counted cellular regions positive for Arc FISH labeling. Manual cell counts were obtained with the help of Metamorph software (version 7.1, Universal Imaging Corp, West Chester, PA, United States of America). First, nuclei present in the median 20% of the planes in the Z-stack were identified and outlined. Nuclei out of this range were excluded, taking into consideration stereological concerns, and an increased likelihood of false negatives (West et al. 1999). Glial cell nuclei could be clearly identified by their small size (∼5 μm in diameter) and their bright, uniform counterstaining. Furthermore, because glial cells do not express Arc (Vazdarjanova et al. 2006), they were not inadvertently included in the mRNA expression counts. Cells were categorized into 4 categories based on the intracellular distribution of Arc-labeling: Neurons without Arc-labeling, neurons with intranuclear Arc-labeling, neurons with cytoplasmic Arc-labeling, and neurons with both intranuclear and cytoplasmic Arc-labeling (Fig. 1C). The odds ratio was used to quantify the degree of overlap in a population of activated cells during the first and second epochs. Logarithm of the odds ratio (ODD) was calculated as
Qualitative Analysis of Hippocampal Damage
Using in-house software written in Matlab, a 3D image of the hippocampus was reconstructed from the slides of the brain atlas (Paxinos and Watson 1998) by interpolating the boundaries of the hippocampus between the slides. This produced a new set of standard slides with 80-μm intervals. The location of the damaged area was identified in Nissl-stained serial sections of brains (80-μm intervals) and was drawn onto the corresponding, standard slide from the reconstructed 3D images (Fig. 1D). The proportion of damaged volume was defined as weighted average of the proportion of damaged area in each slide.
Two-way repeated-measures ANOVA was used to evaluate the main effects of experimental condition (AB-CONTROL, AB-HPC, AA, or CAGED) as a between-subject factor, the neocortical layer (superficial or deep), epoch (first or second epoch), or intracellular distribution of Arc-labeling (intranuclear Arc-labeling, cytoplasmic-labeling Arc, and both) as a within-subject factor, and the interaction between the 2 factors. If the interaction was significant at the P < 0.05 level, 1-way ANOVA was conducted with a corrected significance level to control for type I error. When the main effect of group was significant at the P < 0.05 level, additional comparisons were conducted with Tukey's honestly significant difference (HSD) post hoc tests. All statistical tests were performed with SPSS software (SPSS Inc., Chicago, IL, United States of America).
When rats performed behaviors in 2 different environments, different, but partially overlapping, sets of neurons were activated in the cortex (Burke et al. 2005). In the present study, the size of the activated population and the degree of overlap in the activated population were compared among 4 groups: The AB-CONTROL and AB-HPC groups were exposed to 2 different rooms after they received sham operation or removal of the entire hippocampus, respectively. The AA group was exposed to 2 running epochs in an identical room twice (i.e., presumably identical experiences). The CAGED condition was sacrificed immediately after removed from the home cage. The AB-CONTROL condition was used to replicate the previous finding (Burke et al. 2005), while the AB-HPC group was used to examine the impact of hippocampal damage on the pattern of activated neurons during 2 epochs. The AA condition was used to estimate the maximum degree of overlap of activated neurons between 2 epochs, and the CAGED condition was used to estimate the expression level of Arc at baseline.
Amount of Hippocampal Damage
The amount of damage to the hippocampus was estimated by using a series of Nissl-stained sections (see Materials and Methods). The estimated volume of damaged tissue is summarized in Table 2. In 8 of 10 rats, nearly complete damage of the dorsal hippocampus and at least 65% damage to the ventral hippocampus were observed. In the remaining 2 rats, damage to the hippocampus was relatively limited, and the data were excluded from further analysis.
|Rat#||Right hemisphere||Left hemisphere||Total (%)|
|Dorsal (%)||Ventral (%)||Dorsal (%)||Ventral (%)|
|Rat#||Right hemisphere||Left hemisphere||Total (%)|
|Dorsal (%)||Ventral (%)||Dorsal (%)||Ventral (%)|
The location of damaged area was identified in Nissl-stained serial sections of brains. The proportion of damaged volume was estimated as a weighted average of the proportion of damaged area in each slide.
Effect of Hippocampal Damage and Environment on Intracellular Distribution of Arc-Labeling in Superficial and Deep Neocortical Layers
As previously characterized (Guzowski et al. 1999; Vazdarjanova et al. 2002; Burke et al. 2005), nuclei containing cytoplasmic Arc indicate that neurons were active ∼30 min before the animal's death (i.e., during the first epoch). Nuclei with intranuclear Arc indicate that neurons were active within ∼10 min of the animal's death (i.e., during the second epoch). Neurons containing both intranuclear and cytoplasimic Arc were likely active during both the first and second epochs.
As illustrated in Figure 2A, the 4 groups of rats showed different distributions of Arc-labeling across the 3 categories (nuclear, cytoplasmic, and both) in the superficial layers of the PPC (2-way repeated-measures ANOVA, distribution × group interaction, F6,44 = 5.638, P < 0.001; the main effect of distribution, F2,44 = 11.626, P < 0.001; the main effect of group, F3,22 = 20.426, P < 0.001). Arc expression was very low in the CAGED group (n = 6), which had spent the first and second epochs in the home cage, whereas Arc expression was relatively higher in groups that ran on the track, including the AB-CONTROL (n = 6), AB-HPC (n = 8), and AA groups (n = 6, post hoc Tukey's HSD, CAGED vs. 3 other groups, all comparison, P < 0.001). An Arc-expression level was not different among the AB-CONTROL, AB-HPC, and AA groups (post hoc Tukey's HSD, AB-CONTROL vs. AB-HPC or AA, P > 0.05). Consistent with a previous finding (Burke et al. 2005), in the AB-CONTROL group, which was exposed to 2 different rooms, the ratio of cells with both cytoplasmic and intranuclear Arc-labeling was comparable to the ratio of cells with either cytoplasmic or intranuclear Arc-labeling. In the AB-HPC group, which received hippocampal lesion before being exposed to the 2 rooms, the pattern of distribution of Arc-labeling was very similar to that in the AB-CONTROL group. The AA treatment group (with intact hippocampi and exposed to the identical room twice) showed a relatively higher percentage of neurons with both intranuclear and cytoplasmic Arc-labeling compared with the AB-CONTROL group. This difference, however, did not reach statistical significance (the simple main effect of groups for Arc-labeling in both nuclei and cytoplasm, F3,22 = 12.561, P < 0.001; post hoc Tukey's HSD, CAGED vs. AA, AB-CONTROL, or AB-HPC P < 0.01; AB-CONTROL vs. AB-HPC or AA, P > 0.05).
In deep layers of the PPC (Fig. 2B), the groups that ran on the track (AB-CONTROL, AB-HPC, and AA) had a significantly higher percentage of Arc-expressing neurons compared with the CAGED group (the main effect of group, F3,22 = 16.940, P < 0.001; post hoc Tukey's HSD, CAGED vs. the other groups, P < 0.001). In contrast to cells in the superficial PPC layers, the majority of neurons in the deep layers of PPC showed Arc-labeling in both nuclear and cytoplasmic compartments, and this pattern was consistent across all 3 treatment groups (the simple main effect of groups for Arc-labeling in both nuclei and cytoplasm, F3,22 = 12.109, P < 0.001; post hoc Tukey's HSD, CAGED vs. AB-CONTROL, AB-HPC or AA, P < 0.01; AB-CONTROL vs. AB-HPC or AA, P > 0.05).
As in the PPC, the behavior groups exhibited a significantly higher percentage of Arc-expressing neurons in the superficial (Fig. 2C) and deep (Fig. 2D) layers of the GI than did the no behavior group (GI; distribution × group interaction, F6,44 = 3.173, P< 0.05; the main effect of group, F3,22 = 16.707, P< 0.001; the main effect of distribution, F2,44= 8.560, P< 0.01 in the superficial GI layers; distribution × group interaction, F6,44 = 5.951, P< 0.001; the main effect of group, F3,22= 13.606, P< 0.001; the main effect of distribution, F2,44= 31.262, P< 0.001 for deep GI layers; post hoc comparisons of CAGED to the other groups, P< 0.001 in the deep and superficial layers). In the superficial layers, the percentage of cells with both intranuclear and cytoplasmic Arc-labeling was comparable between the AB-HPC and AB-CONTROL groups. The AA group showed a relatively higher percentage of cells with both intranuclear and cytoplasmic Arc-labeling compared with the AB-CONTROL and AB-HPC groups; however, this difference did not reach statistical significance (the simple main effect of groups for Arc-labeling in both nuclei and cytoplasm, F3,22 = 11.210, P < 0.001; post hoc Tukey's HSD, CAGED vs. AA, AB-CONTROL, or AB-HPC P < 0.01; AB-CONTROL vs. AB-HPC or AA, P > 0.05). In the deep layers of GI, the pattern of cellular distribution of Arc-labeling was comparable across 3 groups (the main effect of group, F3,22 = 10.318, P < 0.001; post hoc Tukey's HSD, CAGED vs. AA, AB-CONTROL, or AB-HPC P < 0.01; AB-CONTROL vs. AB-HPC or AA, P > 0.05). Together, these patterns suggest that the degree of overlap of the activated population was comparable among the AB-CONTROL, AB-HPC, and AA groups in these cortical regions.
Another noticeable pattern was that the ratio of cells with cytoplasmic Arc-labeling was smaller than that of cells with intranuclear Arc-labeling in some groups. This raises a possibility that the room used for the second exposure induced a greater amount of Arc expression than the room for the first exposure. This possibility is unlikely because a similar trend was observed in the previous study (Burke et al. 2005), which used a different set of rooms from the rooms used in the present study.
Effect of Hippocampal Damage and Environment on the Degree of Overlap of Activated Neuron Populations in Deep and Superficial Cortical Layers
To compare the degree of overlap of active neurons between layers and across groups, percentages of the intracellular distribution of Arc-labeling were converted into logarithm of an ODD (see Materials and Methods). ODD becomes infinite if 2 populations are perfectly overlapping and becomes 0 if a degree of overlap is comparable with the level that is expected by chance. Consistent with a previous study (Burke et al. 2005), in the AB-CONTROL group (n = 6), superficial layers in the PPC showed lower overlap of Arc-labeled neurons during the first and second epochs compared with deep layers (Fig. 3A). In the AB-HPC group (n = 8), superficial layers also exhibited lower overlap than did deep layers, and the degree of overlap in superficial layers was comparable with that observed in the AB-CONTROL condition. Finally, in the AA group (n = 6), the degree of overlap was higher than that in the AB-CONTROL group; however, the overlap in superficial layers was still lower than that in deep layers. Statistical significance was established by using a 2-way repeated-measures ANOVA that revealed a significant effect of layer (F1,17 = 53.798, P < 0.001) and group (F2,17 = 5.275, P < 0.05), but no layer-by-group interaction (F2,17 = 1.233, P > 0.05). Post hoc comparison with Tukey's HSD showed that the ODD scores of the AA group were significantly different from the AB-CONTROL and AB-HPC groups (P < 0.05), whereas the AB-HPC group was not different from the AB-CONTROL group (P > 0.05).
In the GI (Fig. 3B), all 3 groups showed a lower degree of overlap of Arc-labeled neurons in the superficial layers compared with the deep layers (the main effect of layer, F1,17 = 23.015, P < 0.001). Unlike the superficial layer of the PPC, the degree of overlap in the AA group did not significantly differ from that in the AB-CONTROL or AB-HPC group (the main effect of group, F2,17 = 0.712, P > 0.05; layer × group interaction, F2,17 = 1.280, P > 0.05).
Correlation Between the Extent of Hippocampal Damage and the Amount of Neocortical Arc Expression
To further examine the effect of hippocampal damage on Arc expression in the neocortex, we calculated Pearson's correlation coefficient between the size of damage to the dorsal or ventral hippocampus and the percentage of cells in 1 of 6 categories: Cells with cytoplasmic Arc-labeling (Cyto), cells with intranuclear Arc-labeling (Foci), cells with both cytoplasmic and intranuclear Arc-labeling (Double), cells either cytoplasmic or intranuclear with Arc-labeling (All labeled), cells activated during the first epoch (1st epoch), and cells activated during the second epoch (2nd epoch, Fig. 4). None of the comparisons in either superficial or deep layers of the 2 cortices reached statistical significance.
Effect of Hippocampal Damage on Behavior During 2 Epochs
Finally, to confirm that the rats' behavior did not differ between the 2 epochs, we examined the number of laps each rat ran during 5 min of track running on the final day of training (i.e., on the day when the rat was sacrificed). The number of laps did not differ between the first and second epochs in any group (Fig. 5A, 2-way repeated-measures ANOVA; the main effect of epoch, F1,17= 4.424). The rats with hippocampal damage (AB-HPC, n = 8) as a group ran significantly more laps than did the rats with an intact hippocampus (AB-CONTROL and AA, each n = 6, P > 0.05; the main effect of group, F2,17= 36.494, P < 0.001; epoch × group interaction, F2,17= 1.987, P > 0.05; post hoc Tukey's HSD, AB-HPC vs. AB-CONTROL or AA, P < 0.001). Furthermore, within the AB-HPC group, the amount of hippocampal damage was positively correlated with the number of laps traversed (Pearson's correlation coefficient: r2 = 0.864, P< 0.001, Fig. 5B). These results suggest that while rats' behavior did not differ between the 2 epochs within a group, the amount of hippocampal damage did proportionally change the behavior, confirming that the amount of hippocampal damage produced in this experiment was sufficient to induce detectable alterations in behavior.
The present study reveals fundamental differences between neuron activation patterns in superficial compared with deep layers of the neocortex. The experiments first replicated the results by Burke et al. (2005), who examined expression patterns of the activity-regulated, IEG Arc in the PPC and GI to show that superficial and deep layers of neocortex differ in the degree of overlap in activated cells across 2 episodes. Previously, it had been hypothesized that the factor determining episode discrimination was the contextual (or “episodic”) information carried by the hippocampus, which projects (by way of the entorhinal cortex) differentially to superficial layers of the neocortex. The results from the present study are not consistent with this hypothesis, as hippocampal lesions had no effect on episode discrimination in either the PPC or the GI. Furthermore, the present data are not consistent with the idea that episode discrimination in superficial cortical layers takes place because the animal experiences different spatial environments between the 2 epochs. Space may have partially contributed to Arc expression patterns in the superficial layers of PPC, because exposing rats twice to the same spatial environment induced a higher degree of overlap in Arc-expressing neurons than exposing rats to 2 different environments. The effect of space, however, was not observed in the GI. Thus, the present data suggest that space plays some role in the PPC, but it is not the critical determinant of the episode discrimination differences between superficial and deep neocortical layers.
One technical concern should be discussed concerning the sensitivity of our method to detect the Arc signal. Specifically, because Arc signals are generally weaker in superficial layers, what effect might such biases have on accurately counting double-labeled cells in superficial and deep layers when the animals were exposed to the same environment twice? It is unlikely that such a measurement artifact could account for the results obtained here, because the method used to quantify the degree of overlap in activated cells (i.e., logarithm of ODD) will become higher if the percentage of cells with Arc-signals is mistakenly underestimated, which was not observed.
Yet another possibility is that compensation occurs following permanent hippocampal damage, and it is this rearrangement that is responsible for the absence of significant effect on neocortical Arc expression. Because this possibility is inconsistent with all known effects of permanent hippocampal damage on the formation of accurate contextual representations (e.g., Morris et al. 1990; Frankland et al. 1998; Winocur et al. 2010), it is unlikely to explain the effects observed here. If the hippocampus is not responsible for the observed activation changes between epochs, it is tempting to suggest that inputs from another region cause superficial layers to differ between the 2 experiences. Such a region might carry information that changes between the 2 epochs, such as temporal order of episodes or states of hunger or fatigue. Although this possibility cannot be eliminated, it seems unlikely for several reasons. First, hippocampal damage is well known to disrupt memory for events that take place as little as 10–30 s prior to test, or as long as several minutes in behavioral paradigms that reward animals for maintaining the memory (Nemanic et al. 2004). This means that it is unlikely that any information based on recent experiences, such as temporal order, will be present in the brains of hippocampal damaged animals. Secondly, although states of hunger or fatigue can fluctuate with time, the time period between epochs is brief relative to circadian cycles, and similar running speeds on the track between the 2 epochs did not suggest motivation differences. One additional control for the participation of motivational state in laminar differences observed in the present study is unpublished data from our laboratory, showing that the laminar differences remain even when medial forebrain bundle stimulation was used for reward. Another possible explanation for the effect presented here is that it may originate from subcortical regions. Because the PPC and GI receive inputs from predominantly different sets of subcortical regions (Chandler et al. 1992; Lundy and Norgren 2004), this does not provide a strong explanation of the laminar differentiation observed here.
If the differences between Arc-expression patterns in superficial and deep layers cannot be explained by changes in input from the hippocampus or as argued in the previous paragraph changes in input from other brain regions, what other explanations exist? One possibility is that network activity in superficial layers is more unstable and more likely to drift over time than activity in deep layers. “Drift” in this context refers to gradual transitions in the set of neurons that are active (i.e., firing action potentials) independent of the input that the network is receiving from an afferent region. To further unpack this concept, in recurrently connected networks, the set of neurons that are active at a given time depends not only on the stimulation that the neurons receive from afferent regions, but also on the stimulation they receive from other neurons in the local network. In the absence of afferent stimulation, the set of neurons that are firing in the local network will drift as a result of factors such as heterogeneities in the strength of local synaptic connections (Zhang 1996; Itskov et al. 2011), or heterogeneities in neuron excitability levels (Renart et al. 2003). Superficial layers of cortex may be more likely to drift over time than deep layers, either because they better resemble a recurrent network without afferent stimulation (meaning they contain more or stronger synapses between neurons within the local network, relative to synapses from presynaptic regions) or because of ongoing synaptic or excitability changes.
One example mechanism that describes how an intrinsic network variable could influence that the set of neurons activated by a behavioral experience is offered by investigations of cAMP response element-binding protein (CREB). The level of phosphorylated CREB in a neuron can influence its excitability (nucleus accumbens: Dong et al. 2006; hippocampus: Lopez de Armentia et al. 2007; amygdala: Zhou et al. 2009). Manipulating CREB levels in a set of neurons in the amygdala prior to fear learning can also determine whether or not that same set will be subsequently activated during retrieval (Han et al. 2007; Zhou et al. 2009). It is possible therefore that any baseline fluctuation in the levels of phosphorylated CREB among superficial layer neurons may be enough to induce a drift in those that are activated between epochs.
One other variable that must be taken into account in the interpretation of these experiments is the specific marker used for determining network activity; in fact, it may be the case that neuron firing rates in superficial layers do not differentiate between similar experiences, and that changes take place between epochs only with respect to the subpopulation of stimulated neurons that express Arc. The IEG Arc plays an important role in synaptic plasticity (reviewed by Bramham et al. 2008 and Korb and Finkbeiner 2011). Although Arc is expressed in hippocampal pyramidal neurons in similar proportions to those neurons expected to have place fields, there are certain conditions in which neuron firing activity is not coupled with Arc expression. For example, Arc expression is reduced following multiple consecutive exposures to an environment, while neuron firing rates do not substantially change (Guzowski et al. 2006). Behaviorally induced Arc expression in the hippocampus is also attenuated following fornix lesions (Fletcher et al. 2006), but fornix lesions do not eliminate hippocampal neuron firing selectivity (Shapiro et al. 1989). Similarly, in the primary auditory cortex, an auditory stimulus induces Arc expression in the absence of sensory-evoked action potential responses (Carpenter-Hyland et al. 2010). Thus, while evidence suggests that many forms of synaptic plasticity are contingent upon Arc expression, Arc expression is not contingent upon action potential signaling alone. It therefore may be the case that signaling patterns between superficial layer neurons do not change between epochs, but that changes do take place in the neuron subpopulations undergoing the cellular signaling cascades that lead to Arc transcription. In summary, whether the episode differentiation observed in the present study takes place due to network activity changes, or due to network plasticity changes, the Arc expression patterns probably reflect something internal to the cortical circuit.
In conclusion, the present study revealed that laminar differences in the degree to which Arc-expression patterns differ between 2 behavioral episodes are determined by intrinsic differences in cellular or network dynamics within the cortex itself, rather than by inputs from the hippocampus directed to superficial neocortical layers. Future studies on differences in receptor distributions or gene expression levels between superficial and deep layers will be useful to identify the source of the difference in episodic encoding between neocortical layers, thereby providing further insight into the way that episodic memory is represented in neocortical networks.
This work was supported by National Institutes on Aging (grant number AG012609) to C.A.B., the McKnight Brain Research Foundation to C.A.B., and the Human Frontier Science Program (LT-00282/2006) to K.T.
We thank Bruce L. McNaughton for discussions on the experiments and Karl Duston for help with training of animals. Conflict of Interest: None declared.