Differential Laminar Activation Dissociates Encoding and Retrieval in the Human Medial and Lateral Entorhinal Cortex

Abstract

The hierarchically organized structures of the medial temporal lobe are critically important for episodic memory function. Accumulating evidence suggests dissociable information processing pathways are maintained throughout these structures including in the medial and lateral entorhinal cortex. Cortical layers provide an additional dimension of dissociation as the primary input to the hippocampus derives from layer 2 neurons in the entorhinal cortex, whereas the deeper layers primarily receive output from the hippocampus. Here, novel high-resolution T2-prepared functional MRI methods were successfully used to mitigate susceptibility artifacts typically affecting MRI signals in this region providing uniform sensitivity across the medial and lateral entorhinal cortex. During the performance of a memory task, healthy human subjects (age 25-33 years, mean age 28.2 ± 3.3 years, 4 female) showed differential functional activation in the superficial and deep layers of the entorhinal cortex associated with task-related encoding and retrieval conditions, respectively. The methods provided here offer an approach to probe layer-specific activation in normal cognition and conditions contributing to memory impairment.SIGNIFICANCE STATEMENT This study provides new evidence for differential neuronal activation in the superficial versus deep layers of the entorhinal cortex associated with encoding and retrieval memory processes, respectively, in cognitively normal adults. The study further shows that this dissociation can be observed in both the medial and the lateral entorhinal cortex. The study was achieved by using a novel functional MRI method allowing us to measure robust functional MRI signals in both the medial and lateral entorhinal cortex that was not possible in previous studies. The methodology established here in healthy human subjects lays a solid foundation for subsequent studies investigating layer-specific and region-specific changes in the entorhinal cortex associated with memory impairment in various conditions such as Alzheimer's disease.

Keywords: 7T; T2prep; cortical layer; encoding and retrieval; high field functional MRI; medial temporal lobe.

Figure 1.

The memory task is designed to tax medial temporal lobe–dependent pattern separation and completion (see below, Materials and Methods). The stimuli in this task consisted of 384 pairs of color photographs of common namable objects. Ninety-six pairs consisted of identical pictures (referred to as repeated pairs), 96 pairs consisted of highly similar but not identical pictures of the same object (referred to as similar pairs), and 192 pairs consisted of unrelated single pictures of objects (referred to as foils). Stimuli were presented for 2500 ms with a 500 ms interval during which a dark screen was displayed. Illustration of the functional paradigm with examples of two pairs of repeated stimuli, two pairs of similar stimuli, and one foil stimulus are shown here. For each item, participants were asked to judge whether the item was new, old, or merely similar to an item they have seen before.

Figure 2.

Segmentation of the medial and lateral components of the ERC. A–C, Coronal views of (A) T1-weighted (1 mm isotropic voxel, whole brain), (B) T2-weighted (0.5 mm isotropic voxel, partial brain), and (C) T2prep BOLD fMRI images (0.9 mm isotropic voxel, partial brain) from one representative subject are shown. The location of the medial and lateral ERC is illustrated on T1-weighted whole-brain images and the manual segmentation for this subject is shown on the T2-weighted and T2prep BOLD images. R, Right; L, left; S, superficial layer; D, deep layer; M, medial; L, lateral. The regions close to the CSF and WM are the superficial and deep layers, respectively. The green box indicates the coverage of the T2-weighted and T2prep BOLD fMRI images. The yellow box indicates the area magnified in the T2prep BOLD fMRI image.

Figure 3.

Representative images acquired using GRE EPI BOLD and T2prep BOLD fMRI (coronal view). In most subjects, susceptibility artifacts (signal dropout and geometric distortion caused by the nearby cavity in the ear canal) in the entorhinal cortex were significantly reduced in T2prep BOLD (middle) compared with GRE EPI BOLD (bottom). The high resolution T2-weighted image is shown as a dropout and distortion free anatomic reference (top).

Figure 4.

Average laminar tSNR profile in the lateral and medial entorhinal cortex across all subjects (n = 7). The x-axis shows the layer number with layers close to the CSF reflecting the superficial layers and layers close to the WM reflecting the deep layers of the entorhinal cortex. The tSNR values in the shaded layers were averaged to provide mean tSNR values in the superficial (green) and deep (orange) layers (Table 2). Error bars indicate the intersubject SEs. Data from the 1st, 2nd, 19th, and 20th layers are not shown because of the known limitation in the LayNii software for the segmentation of layers close to the rim.

Figure 5.

A, Top row, Representative high-resolution T2-weighted images acquired from one subject used for layer segmentation. Bottom row, The segmented superficial and deep layers in the lateral and medial entorhinal cortex are overlaid on the T2-weighted images with different colors; purple, superficial lateral; blue, deep lateral; yellow, superficial medial; red, deep medial. B, Mean laminar activation profile in the lateral and medial entorhinal cortex across all subjects showing activation differences between successful encoding and successful retrieval trials. Layers close to CSF and WM are superficial and deep layers, respectively. The beta values in the shaded layers were averaged to give mean activation in the superficial (green) and deep (orange) layers. Error bars indicate the intersubject SEs.

Figure 6.

A, Left, Individual high-resolution T2-weighted images from each subject used for layer segmentation. Right, The segmented superficial and deep layers in the lateral and medial entorhinal cortex are overlaid on the T2-weighted images with different colors; purple, superficial lateral; blue, deep lateral; yellow, superficial medial; red, deep medial. B, Individual fMRI results and selection of layers in analysis method 3 with peak layers. The shaded areas indicate the selection of three consecutive layers with the most positive and most negative beta values among the 1st to 10th layers and the 11th to 20th layers for the superficial (green) and deep (orange) layers, respectively. The beta values in the shaded layers were averaged to give mean beta values in the superficial and deep layers, respectively. Error bars indicate the intrasubject SEs. In analysis methods 1 and 2, selection of layers is the same for all subjects (data not shown).

Figure 7.

Laminar activation contrasting successful retrieval and successful foil trials in the lateral and medial entorhinal cortex. A, Mean group results. Beta values in the shaded layers were averaged to provide mean activation values in the superficial (green) and deep (orange) layers. Error bars indicate the intersubject SEs. Data from the 1st, 2nd, 19th, and 20th layers are not shown because of the known limitation in the LayNii software for the segmentation of layers close to the rim. B, Individual laminar beta profiles from all subjects.