Volume 92, Issue 2, Pages 518-529 (October 2016) Laminar Module Cascade from Layer 5 to 6 Implementing Cue-to-Target Conversion for Object Memory Retrieval in the Primate Temporal Cortex  Kenji W. Koyano, Masaki Takeda, Teppei Matsui, Toshiyuki Hirabayashi, Yohei Ohashi, Yasushi Miyashita  Neuron  Volume 92, Issue 2, Pages 518-529 (October 2016) DOI: 10.1016/j.neuron.2016.09.024 Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 1 Experimental Design (A) Lateral and coronal views of a monkey brain. Six cortical layers (L1–L6) are shown in area 36 (A36). A, anterior; P, posterior; D, dorsal; V, ventral; L, lateral; M, medial. (B) Two models for laminar organization of neurons relevant to the cued recall of visual objects. In a “laminar module model,” different response types of neurons coding for the presented cue and/or the to-be-recalled target are located in distinct cortical layers, thus forming functional laminar modules for mnemonic processing. In a “non-laminar functional model,” each type of functional neuron is distributed throughout cortical layers, leading to a non-laminar (salt-and-pepper) functional cluster. Parameters for the behavioral task are also shown in the bottom row. (C) Localization procedure of recorded neurons at the resolution of six cortical layers. In “MRI-based in vivo localization” of recorded neurons in each recording track (left column), the microelectrode tip position was localized in MRI scan sessions and the locations of recorded neurons were reconstructed in the MRI volume by referring to the locations of the microelectrode tip (Figure S1C). After completion of all the recordings, the neurons in the MRI volume were registered onto the postmortem histological volume by aligning both of these volumes with the aid of metal deposit marks that served as common references visible in both the MRI volume and the histological volume (right column; Figure S1D). Black dot, recorded neuron; red cross and arrowhead, electrode tip; yellow dot and arrowhead, elgiloy-deposit fiducial mark. (D) Representative reconstruction of laminar locations of recorded neurons. The electrode tip on the MR images (top panel), as well as location of recorded neurons, is registered to a Nissl-stained histological section (bottom panel). Elgiloy-deposit marks are visible both in the MR image (top panel) and a histological section (bottom panel). The framed area in the top panel indicates the position of the bottom panel. Scale bars, 5 (top) and 1 mm (bottom). See also Figures S1–S3 and Table S1. Neuron 2016 92, 518-529DOI: (10.1016/j.neuron.2016.09.024) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 2 Responses and Laminar Locations of Representative Neurons (A) An example of a recording track reconstruction with MRI scan sessions at two different cortical depths. Shown are the coronal view of the brain (left column), MR images of the electrode tip at two different depths (column second from the left), the corresponding histological section (top, middle column), and line drawing of laminar positions of recorded neurons (bottom, middle column). Spike density functions (column second from the right) and stimulus coding properties for cue/target (cue-holding index [CHI] and pair-recall index [PRI]; see Experimental Procedures) (right column) are depicted for two representative neurons (B1577 and B1582) that are shown in the line drawing panel (light blue circle). An area framed by the dotted rectangle in the MR image and histological section indicates the position of the panel showing the laminar position of the neurons. (B) Another example with MRI scan sessions at two different depths. The conventions are the same as in (A). (C–E) Other examples of localized neurons in L6 (C), L5 (D), and L3 (E), with a single MR image of an electrode tip. See also Figure S3. Neuron 2016 92, 518-529DOI: (10.1016/j.neuron.2016.09.024) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 3 Distinct Dynamics of Cue- and Target-Stimulus Coding across Layers (A) A Nissl-stained section of A36 showing the six-layered structure. (B and C) Time courses of the cue-holding index (CHI) (B) and pair-recall index (PRI) (C) for all the stimulus-selective neurons in each layer. Each row represents a single neuron, sorted according to its depth location (Supplemental Experimental Procedures). n = 20 (L2), 98 (L3), 33 (L4), 135 (L5), and 102 (L6). (D) Stimulus coding indices (z transformed) of each neuron as a function of normalized cortical depth. Color of each circle denotes statistical significance of the index for each neuron. (E and F) Stimulus coding indices (mean ± SEM) (E) and relative number of neurons with statistically significant indices (F) in each layer. ∗p < 0.05; ∗∗p < 0.01; two-tailed t test against zero with Bonferroni’s correction. †p < 0.05; ‡p < 0.01; Tukey’s post hoc test after one-way ANOVA. §p < 0.05; §§p < 0.01; significant increase from ratio of all cells (horizontal dotted line) with post hoc residual analysis after χ2 test. See also Figure S4. Neuron 2016 92, 518-529DOI: (10.1016/j.neuron.2016.09.024) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 4 Functional Dissociation of L5 and L6 (A) Cyto- and chemoarchitectonic differences between L5 and L6. Scale bars, 200 and 500 μm in high- and low-magnification images, respectively. (B) Population time courses (mean ± SEM) of the pair-recall index (PRI, left column) and the cue-holding index (CHI, right column) in L5 and L6. (C) Cumulative plots of PRI latency (left column) and CHI fall time (right column). n = 70 (L5) and 56 (L6) for PRI latency and 135 (L5) and 102 (L6) for CHI fall time, respectively. Histograms (inset) depict distribution of neurons with PRI latency <300 ms. See Supplemental Experimental Procedures for measurements of PRI latencies and CHI fall time. See also Figure S5B. ∗p < 0.05; ∗∗p < 0.005; Kolmogorov-Smirnov test. Neuron 2016 92, 518-529DOI: (10.1016/j.neuron.2016.09.024) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 5 L6 Contains Two Cell Assemblies (A) Relationship between CHI fall time and PRI latency in L5 (left column) and L6 (right column). L6 neurons were further subdivided into early and late-recall groups based on a cluster analysis (Experimental Procedures). n = 25 and 31 neurons for the early and late group, respectively. A quadratic discriminant function for L6 was shown in black line. (B) Temporal relations between CHI fall time and PRI latency. ∗∗p < 0.01; Wilcoxon signed-rank test against zero with Bonferroni’s correction. ‡p < 0.001; Mann-Whitney U test with Bonferroni’s correction for multiple comparisons after Kruskal-Walis test. (C–E) Spike phase-locking to the local field potential (LFP). (C) Spike-LFP phase-locking (bottom row) and time courses of response indices (top row) of a representative neuron in each group. (D), Population phase-locking (mean ± SEM) in L5 (n = 57), L6 early group (n = 17), and L6 late group (n = 21) during the fixation (dotted line) and delay period (solid line). Note that neurons with fewer than 150 spikes were excluded from the analysis of phase-locking to increase the reliability of the results. (E) Difference in phase-locking between the delay and fixation period (mean ± SEM) in the theta frequency range (5–8 Hz). See Supplemental Experimental Procedures for calculation of the spike-LFP phase-locking. ∗∗p < 0.005; two-tailed t test against zero with Bonferroni’s correction. †p < 0.05; ‡p < 0.01; two-tailed t test with Holm’s correction. (F) A proposed model of the laminar module cascade for cued recall of visual objects. See also Figure S5. Neuron 2016 92, 518-529DOI: (10.1016/j.neuron.2016.09.024) Copyright © 2016 Elsevier Inc. Terms and Conditions