Hippocampal Synaptic Expansion Induced by Spatial Experience in Rats Correlates with Improved Information Processing in the Hippocampus

Abstract

Spatial water maze (WM) overtraining induces hippocampal mossy fiber (MF) expansion, and it has been suggested that spatial pattern separation depends on the MF pathway. We hypothesized that WM experience inducing MF expansion in rats would improve spatial pattern separation in the hippocampal network. We first tested this by using the the delayed non-matching to place task (DNMP), in animals that had been previously trained on the water maze (WM) and found that these animals, as well as animals treated as swim controls (SC), performed better than home cage control animals the DNMP task. The "catFISH" imaging method provided neurophysiological evidence that hippocampal pattern separation improved in animals treated as SC, and this improvement was even clearer in animals that experienced the WM training. Moreover, these behavioral treatments also enhance network reliability and improve partial pattern separation in CA1 and pattern completion in CA3. By measuring the area occupied by synaptophysin staining in both the stratum oriens and the stratun lucidum of the distal CA3, we found evidence of structural synaptic plasticity that likely includes MF expansion. Finally, the measures of hippocampal network coding obtained with catFISH correlate significantly with the increased density of synaptophysin staining, strongly suggesting that structural synaptic plasticity in the hippocampus induced by the WM and SC experience is related to the improvement of spatial information processing in the hippocampus.

Fig 1. Spatial pattern separation in the DNMP task improves after water maze and swimming treatment.

(A) Performance of animals during training in the Morris Water Maze task is expressed as the latency to reach the target platform; each point represents the average latency to reach the target from each pair of trials (5 pairs) from a total of 10 trials each animal underwent each daily session. The animals were trained during 5 sessions that occurred during 5 consecutive days. The trained animals showed a significant decrease in their latency to reach the target between trial pairs each day, particularly during the first 3 days (*** p<0.001, *p<0.05, repeated measures ANOVA). (B) The DNMP task consists of placing the animal in the middle of the start arm and then allowing it to find food at the end of the sample arm (SA), this occurs during the sample phase; during the next phase the animal is release in the same place but it now needs to find the choice arm (CA), which is the only place where food is now available and is open only in this choice phase. To solve the task the animal needs to distinguish the location of the newly available corridor from that one visited in the previous trial and the difficulty comes with the proximity of the 2 arms. (C) Four different separations between the SA and CA were tested (upper diagram); in Sp1 the SA and CA were adjacent to each other; in Sp2 one arm separates the SA and CA; in Sp3 two arms separate the SA and CA; and finally in Sp4, the SA and CA were opposite from each other; note that the start arm (in black) was 90° from either the SA or the CA (lower diagram), and the position of these baited arms alternated between left choice (LC) and right choice (RC) throughout trials. (D) The DNMP results are expressed as the average percentage of errors in all trials for each group (±SEM); note that WM and SC animals present significantly fewer errors than the IC group (***p<0.001, **p<0.01, *p<0.05) except in Sp4 were all groups perform well.

Fig 2. Exploration procedures and catFISH.

(A) The animals explored the first environment for 5 min, and 25 min later they explored the second environment for 5 min in either the AA (same box same room), AA’ (different box same room), or AB (different box different room) condition. Animals were sacrificed immediately after the second 5 min exploration and their brains extracted, frozen and later process for FISH to detect Arc mRNA. Examples of CA1 (B) and CA3 (C) confocal images (1 filtered plane from the stack) stained for Arc mRNA detection and used to perform the catFISH analysis; in green is the nuclear counterstaining “Sytox” and in red is the Arc signal. Note that neuronal nuclei have texture, and their Sytox signal is dim, while glial cells look solid and bright. The Arc signal can be observed as 2 transcription foci inside the nuclei (solid arrows), or as cytoplasmic Arc staining surrounding the neuronal nuclei (open arrows), or as double Arc staining (arrowhead). The bar represents 50 μm.

Fig 3. catFISH images from CA1 and CA3 from the different exploration procedures.

Dorsal hippocampal tissue stained with fluorescent in situ hybridization (FISH) for Arc mRNA (in Red) revealed with Cy3 (Promega kit see Methods). Images were taken with the 40x/1.3 NA objective using a Zeiss LSM 510 or Zeiss LSM 710 confocal system with 2 different lasers either set to excite Cy3 (Arc) or SITOX-Green (Nuclear counterstaining in Green). All images were process with Image J and 40–55% of the stack was collapsed in 1 single image (using Z Project). Images then went through the median filter, using the exact same parameters for all (0.8 pixels). From A to D, the images were taken from the CA1 hippocampal network and from E to G, from the CA3 network. Images A and E were taken from a Cage control animal; B and F from an animal that underwent the AA double exploration; C and G from an animal that underwent the AA’ double exploration condition; and D and H from an animal that underwent the AB double exploration condition. Calibration bar (lower right) represents 100 μm.

Fig 4. Raw catFISH results.

The bar graphs (in A and C) show the percentage of Arc-expressing cells in the CA1 (A) and CA3 (C) networks in each epoch. The solid bars represent the proportion of active cells for epoch 1, and the lighter, textured bars represent the proportion of active cells for epoch 2 (*p<0.01 Bonferroni posthoc analysis for all exploration groups vs. the cage control). The percentage of Arc-expressing cells for each classification is shown in B and D. Each bar represents the proportion of nuclear (Diagonal line pattern), cytoplasmic (Dots), or double (Solid) Arc mRNA staining. Note that the proportion of double activated cells in both CA1 and CA3 after the AA exploration condition is larger than other classifications (i.e., nuclear and cytoplasmic), which is particularly clear in the WM-treated animals (*p<0.01 Bonferroni posthoc analysis vs. the cage control).

Fig 5. The similarity score (SiSc) results from catFISH.

This measure takes into account the 3 Arc staining classifications and expresses it in a single value. This value (±SEM) represents the degree of overlap between the two recruited ensemble in the CA1 (A) and CA3 (B) networks from animals exposed to the AA, AA’, and AB double exploration conditions after being pre-treated as either intact controls (IC), swimming controls (SC), or water maze trained (WM). Bonferroni post hoc analysis of the WM group vs. its respective IC ***p<0.001, **p<0.01, *p<0.05; or vs. its respective SC (###p<0.001, #p<0.05), and the important intragroup differences are shown with lines (***p<0.001, **p<0.01).

Fig 6. Synaptophysin/Map2 segmentation and Map2 staining area analysis.

(A) A representative synaptophysin (Red) / Map2 (Green) stained image is shown with the drawings that defined the different hippocampal dendritic segments regions of interest (hippocampal segments ROIs). These hippocampal segments ROIs included the CA3 stratum oriens, divided into 3 regions based on their proximity to the dentate gyrus (DG): stratum oriens distal (SOd), stratum oriens medial (SOm), and stratum oriens proximal (SOp); the CA3 stratum lucidum was also divided into 3 regions, the stratum lucidum distal (SLd), stratum lucidum medial (SLm), and stratum lucidum proximal (SLp). The CA3 stratum radiatum was divided into 6 regions, depending on their proximity to the DG and to the pyramidal cell soma: stratum radiatum distal medial (SRdm), stratum radiatum medial medial (SRmm), stratum radiatum proximal medial (SRpm), stratum radiatum distal distal (SRdd), stratum radiatum medial distal (SRmd), and stratum radiatum proximal distal (SRpd). Finally, the last 2 ROIs correspond to the CA1 stratum oriens (CA1 SO) and the CA1 stratum radiatum (CA1 SR). In the bar graph (B) the Map2-stained area expressed in pixels is shown for each hippocampal segment ROI. It is important to emphasise that No significant differences were found among groups (IC, SC, and WM) in the Map2-stained area used for the synaptophysin analysis.

Fig 7. Synaptophysin/Map2 Split Channel and Higher Magnification Images.

Immonostained tissue for synaptophysin and Map2 imaged with the Zeiss ApoTome™ system, equipped with a motorized stage which allows the acquisition of montage or mosaic images with the MosaiX software. Images were taken with the 25X/0.8 NA objective. One Zeiss LSM file containing the MosaiX montage was open using imageJ software and the color channels were separated (split channels). Then each channel went through the median filter and assigned its representative color. The nuclear counterstaining DAPI is shown in blue (A for CA1 and E for CA3). Map 2 is shown in Green (B for CA1 and F for CA3) and synaptophysin is shown in RED (C for CA1 and G for CA3). The merge colors image is shown in D for CA1 and H for CA3. The magnification shown here was done by only trimming the montage image and by applying a regular zoom-in using adobe photoshop. This is a proper example of the image resolution the experimenter had available for analysis (He can perform the same simple zoom-in magnification with the same results), and was obtained from a similar montage image as those shown in Figs 6 and 8. Calibration bar (lower right) represents 200 μm.

Fig 8. Synaptophysin-staining analysis in the hippocampus.

Representative MosaiX montage images that covers all the CA3 and part (after trimming) of the CA1 hippocampal regions: in green is the Map-2 immunostaining and in red the synaptophysin staining in animals from the IC (A), SC (B), and WM (C) groups; the white bar represents 500 μm. (D) The bars represent the average synaptophysin/Map2 staining (±SEM) in each group (IC, SC, and WM) for each ROI (see Methods). Note that only in the CA3 SOd (stratum oriens distal), the CA3 SLd (stratum lucidum distal), and the CA1 SO (stratum oriens) were significant differences found between groups. Fisher post hoc analysis of the WM vs. the IC ***p<0.001, **p<0.01, *p<0.05; WM vs. the SC ##p<0.01.

Fig 9. Synaptophysin/Map2 area and Similarity score correlation analysis.

Correlation graphs between: the area of synaptophysin staining in the CA3 SOd and the similarity scores (SiSc) obtained in the CA3 network (first row); the area of synaptophysin staining in the CA3 SOd and the SiSc obtained in the CA1 network (second row); the area of synaptophysin staining in the CA3 SLd and the SiSc obtained in the CA3 network (Third row); and the area of synaptophysin staining in the CA3 SLd and the SiSc obtained in the CA1 network (Fourth row). The first column presents the correlation graphs for the AA double exploration condition, where pattern completion is observed in both the CA3 and CA1 networks; the second column presents the correlation graphs for the AA’ condition, where pattern completion occurred in the CA3 network, and partial pattern separation occurred in CA1; and the third column presents the graphs for the AB condition where pattern separation occurred in both CA1 and CA3. ***p<0.001, **p<0.01, *p<0.05.