1. Neuroscience
Download icon

Local projections of layer Vb-to-Va are more prominent in lateral than in medial entorhinal cortex

  1. Shinya Ohara  Is a corresponding author
  2. Stefan Blankvoort
  3. Rajeevkumar Raveendran Nair
  4. Maximiliano J Nigro
  5. Eirik S Nilssen
  6. Clifford Kentros
  7. Menno P Witter  Is a corresponding author
  1. Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Norway
  2. Laboratory of Systems Neuroscience, Tohoku University Graduate School of Life Sciences, Japan
Research Article
  • Cited 0
  • Views 860
  • Annotations
Cite this article as: eLife 2021;10:e67262 doi: 10.7554/eLife.67262

Abstract

The entorhinal cortex, in particular neurons in layer V, allegedly mediate transfer of information from the hippocampus to the neocortex, underlying long-term memory. Recently, this circuit has been shown to comprise a hippocampal output recipient layer Vb and a cortical projecting layer Va. With the use of in vitro electrophysiology in transgenic mice specific for layer Vb, we assessed the presence of the thus necessary connection from layer Vb-to-Va in the functionally distinct medial (MEC) and lateral (LEC) subdivisions; MEC, particularly its dorsal part, processes allocentric spatial information, whereas the corresponding part of LEC processes information representing elements of episodes. Using identical experimental approaches, we show that connections from layer Vb-to-Va neurons are stronger in dorsal LEC compared with dorsal MEC, suggesting different operating principles in these two regions. Although further in vivo experiments are needed, our findings imply a potential difference in how LEC and MEC mediate episodic systems consolidation.

Introduction

Everyday memories, which include information of place, time, and content of episodes, gradually mature from an initially labile state to a more stable and long-lasting state. This memory maturation process, called memory consolidation, involves gradual reorganization of interconnected brain regions: memories that are initially depending on hippocampus become increasingly dependent on cortical networks over time (Frankland and Bontempi, 2005). Although various models have been hypothesized for this systems-level consolidation, such as the standard consolidation model and multiple trace theory (Nadel and Moscovitch, 1997; Squire and Alvarez, 1995), they all share a canonical hippocampal-cortical output circuit via the entorhinal cortex (EC), which is crucial to mediate long-term memory storage and recall (Buzsáki, 1996; Eichenbaum et al., 2012). The existence of this circuit was originally proposed based on the ground-breaking report of a non-fornical hippocampal-cortical output route mediated by layer V (LV) of the EC in monkeys (Rosene and Van Hoesen, 1977), which was later confirmed also in rodents (Köhler, 1985; Kosel et al., 1982).

The EC is composed of two functionally distinct subdivisions, the lateral and medial EC (LEC and MEC, respectively). MEC processes allocentric, mainly spatial information, whereas LEC represents the time and content of episodes (Deshmukh and Knierim, 2011; Hafting et al., 2005; Montchal et al., 2019; Tsao et al., 2018; Tsao et al., 2013; Xu and Wilson, 2012). Despite these evident functional differences, both subdivisions are assumed to share the same cortical output system mediated by LV neurons. Recently, we and others have shown that LV in both MEC and LEC can be genetically and connectionally divided into two sublayers: a deep layer Vb (LVb), which contains neurons receiving projections from the hippocampus, and a superficial layer Va (LVa), which originates the main projections out to forebrain cortical and subcortical structures (Ohara et al., 2018; Ramsden et al., 2015; Sürmeli et al., 2015; Wozny et al., 2018). These results indicate that for the hippocampal-cortical dialogue to function we need to postulate a projection from LVb to LVa neurons. Although the existence of such a LVb-LVa circuit is supported by our previous study using transsynaptic viral tracing in rats (Ohara et al., 2018), experimental evidence for functional connectivity from LVb-to-Va in LEC and MEC is still lacking.

In the present study, we examined the presence of this hypothetical intrinsic EC circuit by using a newly generated LVb-specific transgenic (TG) mouse line obtained with an enhancer-driven gene expression (EDGE) approach (Blankvoort et al., 2018). To compare the LVb intrinsic circuit between LEC and MEC, we ran identical in vitro electrophysiological and optogenetical experiments in comparable dorsal portions of LEC and MEC. To our surprise, we found differences in the postulated intrinsic LVb-LVa pathway between the two entorhinal subdivisions: the connectivity is prominent in dorsal LEC but is apparently sparse in dorsal parts of MEC. In contrast, other intrinsic circuits from LVb to layers II and III (LII and LIII), which constitute hippocampal-entorhinal re-entry circuits, are very similar in both entorhinal subdivisions. Our data seem to suggest that the current view of the canonical hippocampal-cortical output circuit that allegedly is crucial for systems consolidation might need revision, though the functional impact of our findings awaits further in vivo studies.

Results

Characterization of LVb TG mouse line

Entorhinal LV can be divided into superficial LVa and deep LVb based on differences in cytoarchitectonics, connectivity and genetic markers such as Purkinje cell protein 4 (PCP4) and chicken ovalbumin upstream promoter transcription factor interacting protein 2 (Ctip2) (Ohara et al., 2018; Sürmeli et al., 2015; Figure 1—figure supplement 1, see Materials and methods for details). To target the entorhinal LVb neurons, we used a TG mouse line (MEC-13-53D) that was obtained with the EDGE approach (Blankvoort et al., 2018). In this TG line, the tetracycline-controlled transactivator (tTA, Tet-Off) is expressed under the control of a specific enhancer and a downstream minimal promoter. To visualize the expression patterns of tTA, this line was crossed to a reporter mouse line, which expresses mCherry together with GCaMP6 in a tTA-dependent manner.

In both LEC and MEC, mCherry-positive neurons were observed mainly in LVb (93.2% in LEC and 82.9% in MEC) and some in layer VI (LVI; 5.1% in LEC and 16.8% in MEC) but hardly in LVa (1.7% in LEC and 0.3% in MEC), and none in superficial layers (Figure 1A–D). The proportion of PCP4-positive LVb neurons that show tTA-driven labeling was 45.9% in LEC and 30.9% in MEC (Figure 1E). The tTA-driven labeling colocalized well with the PCP4 labeling (percentage of tTA-expressing neurons that were PCP4-positive was 91.7% in LEC and 99.3% in MEC; Figure 1F), highlighting the specificity of the line. In another experiment using a GAD67 TG line expressing green fluorescent protein (GFP), we showed that the percentage of double-labeled (PCP4+, GAD67+) neurons among total GAD67-positive neurons is very low in both LEC and MEC (4.3% and 2.3%, respectively, Figure 1—figure supplement 2). This percentage of double-labeled neurons was significantly lower than in the Ctip2-stained sample in both regions (18.1% for LEC and 7.2% for MEC). This result shows that PCP4 can be used as a marker for excitatory entorhinal LVb neurons. Occasionally PCP4-positive neurons were observed in what seems to be layer Va, where there is a lack of continuity in the cell layer as indicated by retrograde tracing (Figure 1—figure supplement 1). Although sparse, these ‘misplaced’ LVb neurons were also targeted in our TG mouse line. The MEC-13-53D is thus an attractive TG mouse line to target excitatory LVb neurons in both LEC and MEC.

Figure 1 with 4 supplements see all
Lateral entorhinal cortex (LEC) and medial entorhinal cortex (MEC) layer Vb (LVb) neurons show distinct morphological features.

(A–C) Expression of tetracycline-controlled transactivator (tTA) in the enhancer-driven gene expression (EDGE) mouse line (MEC-13-53D), which is visualized with mCherry (green) by crossing to a tTA-dependent mCherry line. A horizontal section was immunostained with an anti-Purkinje cell protein 4 (PCP4) antibody (magenta) to label entorhinal LVb neurons. Images of LEC (B) and MEC (C) correspond with the boxed areas in (A) and show from left to right PCP4 expression, mCherry expression, and a merged image. (D) Percentage of tTA-expressing neurons among layers in LEC and MEC. (E) Percentage of tTA-expressing neurons among the total PCP4-positive neurons in LEC and MEC. (F) Percentage of PCP4-positive neurons among the total tTA-expressing neurons in LEC and MEC. Error bars: mean ± standard errors. The tTA-expressing neurons mainly distributed in LVb of EC and colocalized with PCP4. (G–I) Morphology of LVb neurons targeted in MEC-13-53D in MEC (G, H) and LEC (I). tTA-expressing LVb neurons were first labeled with green fluorescent protein (GFP) (green) by injecting AAV2/1-TRE-Tight-EGFP in MEC-13-53D, and then intracellularly filled with biocytin (magenta, G) Images of MEC (H) and LEC (I) show biocytin labeling, which correspond with the boxed area in each inset. The four neurons shown in (G) correspond to the neurons in (H). Double arrowheads show the cell bodies, the single arrowheads show their dendrites, and different neurons are marked in different colors (green, blue, red, and yellow). The distribution of apical dendrites largely differs between MEC-LVb and LEC-LVb neurons. (J) Proportion of morphologically identified cell types of LVb neurons in LEC and MEC. These data were obtained in 10 animals and 22 slices. Scale bars represent 500 μm for (A) and inset of (H) and (I), 100 μm for (H) and (I), 50 μm for (B) and (C), and 20 μm for (G). Figure 1—source data 1. See also Figure 1—figure supplement 1, Figure 1—figure supplement 2, Figure 1—figure supplement 3, and Figure 1—figure supplement 4

Figure 1—source data 1

Specificity of tetracycline-controlled transactivator expression in MEC-13-53D.

https://cdn.elifesciences.org/articles/67262/elife-67262-fig1-data1-v2.xlsx

Morphological properties of LVa/LVb neurons in LEC and MEC

We next examined the morphological and electrophysiological properties of the LVb neurons in LEC and MEC in this TG mouse line. Targeted LVb neurons were labeled by injecting tTA-dependent adeno-associated virus (AAV) encoding GFP (AAV2/1-TRE-Tight-GFP) into either LEC or MEC and filled with biocytin during whole-cell patch-clamp recordings in acute slices (Figure 1G). Consistent with our histological result showing that this line targets excitatory cells, all recorded cells showed morphological and electrophysiological properties of excitatory neurons (Figure 1, Figure 2). In line with previous studies, many MEC-LVb neurons were pyramidal cells with apical dendrites that ascended straight toward layer I (LI; Figure 1H, J; Canto and Witter, 2012a; Hamam et al., 2000). In contrast, more than 40% of the targeted LEC-LVb neurons were tilted pyramidal neurons (Canto and Witter, 2012b; Hamam et al., 2002) with apical dendrites not extending superficially beyond LIII (Figure 1I, J). Since this latter result may result from severing of dendrites by the slicing procedure, we also examined the distribution of LVb apical dendrites in vivo. After injecting AAV2/1-TRE-Tight-GFP in the deep layer of LEC in the TG line, the distribution of labeled dendrites of LEC-LVb neurons was examined throughout all sections (Figure 1—figure supplement 3). Even with this approach, the labeled dendrites mainly terminated in LIII and only sparsely reached layer IIb. These morphological differences indicate that MEC-LVb neurons sample inputs from different layers than LEC-LVb neurons: MEC-LVb neurons receive inputs throughout all layers, whereas LEC-LVb neurons only receive inputs innervating layer IIb–VI. In contrast to LVb neurons, the morphology of LVa neurons was relatively similar in both regions: the basal dendrites extended horizontally mostly within LVa, whereas the apical dendrites reached LI (Figure 1—figure supplement 4). These morphological features of LVa neurons are in line with previous studies (Canto and Witter, 2012a; Canto and Witter, 2012b; Hamam et al., 2000; Hamam et al., 2002; Sürmeli et al., 2015).

Figure 2 with 1 supplement see all
Electrophysiological properties distinguish layer Va/layer Vb (LVa/LVb) neurons in both lateral entorhinal cortex (LEC) and medial entorhinal cortex (MEC).

(A) Representative voltage responses to hyperpolarizing and depolarizing current injection of LEC-LVa (orange), LEC-LVb (green), MEC-LVa (blue), and MEC-LVb (magenta) neurons. (B) Voltage responses at rheobase current injections showing afterhyperpolarization (AHP) wave form and depolarizing afterpotentials (DAP). (C) Voltage responses to hyperpolarizing current injection with peaks at −90 ± 5 mV showing Sag. (D) Voltage responses to depolarizing current injection with 10 ± 1 action potentials (APs) showing adaptation. (E) Voltage responses to +200 pA of 1-s-long current injection showing maximal AP number. (F–K) Differences of sag ratio (F, one-way ANOVA, F3,117 = 36.88, ***p<0.0001, Bonferroni’s multiple comparison test, ***p<0.001), time constant (G, one-way ANOVA, F3,117 = 53.39, ***p<0.0001, Bonferroni’s multiple comparison test, **p<0.01, ***p<0.001), AP frequency after 200 pA injection (H, one-way ANOVA, F3,117 = 44.37, ***p<0.0001, Bonferroni’s multiple comparison test, ***p<0.001), fast AHP (I, one-way ANOVA, F3,117 = 7.536, ***p=0.0001, Bonferroni’s multiple comparison test, **p<0.01, ***p<0.001), medium AHP (J, one-way ANOVA, F3,117 = 21.99, ***p<0.0001, Bonferroni’s multiple comparison test, **p<0.01, ***p<0.001), and adaptation (K, one-way ANOVA, F3,117 = 21.6, ***p<0.0001, Bonferroni’s multiple comparison test, **p<0.01, ***p<0.001) between LEC-LVa (N = 31), LEC-LVb (N = 45), MEC-LVa (N = 20), and MEC-LVb (N = 25) neurons (error bars: mean ± standard errors). (L) Principal component analysis based on the 12 electrophysiological parameters shown in Figure 2—source data 1 show a separation between LVa and LVb neurons as well as a moderate separation between LEC-LVb and MEC-LVb neurons. Data representing 121 neurons from 27 animals (also holds for M and N). (M) Separation of LEC-LVa (orange), LEC-LVb (green), MEC-LVa (blue), and MEC-LVb (magenta) neurons using sag ratio, AP frequency at 200 pA injection, and time constant as distinction criteria. (N) Separation of LEC-LVb (green) and MEC-LVb (magenta) neurons using fast AHP, AP frequency at 200 pA injection, and time constant as distinction criteria.

Figure 2—source data 1

Electrophysiological properties of entorhinal layer V neurons.

https://cdn.elifesciences.org/articles/67262/elife-67262-fig2-data1-v2.xlsx

Electrophysiological properties of LVa/LVb neurons in LEC and MEC

Previous studies have reported that the electrophysiological profiles of LV neurons are diverse both in LEC and MEC (Canto and Witter, 2012a; Canto and Witter, 2012b; Hamam et al., 2000; Hamam et al., 2002), but whether these different electrophysiological properties of entorhinal LV neurons relate to the two sublayers, LVa and LVb, was unclear. Here, we examined this by analyzing a total of 121 neurons recorded from the TG mouse line (MEC-13-53D): 31 LEC-LVa, 45 LEC-LVb, 20 MEC-LVa, and 25 MEC-LVb neurons (Figure 2A). As previously reported (Canto and Witter, 2012a; Canto and Witter, 2012b; Hamam et al., 2000; Hamam et al., 2002), only a few LV neurons showed weak depolarizing afterpotentials (DAP; Figure 2B), with a higher incidence in MEC than in LEC. Among the 12 examined electrophysiological properties (Figure 2—source data 1), differences were observed between the LVa and LVb neurons in most parameters except for resting potential, input resistance, and action potential (AP) threshold (Figure 2F–K, Figure 2—figure supplement 1). Principal component analysis based on the 12 parameters resulted in a clear separation between LVa and LVb neurons, and also in a moderate separation between LEC-LVb and MEC-LVb (Figure 2L). Sag ratio (Figure 2C, F), time constant (Figure 2G), and AP frequency after 200 pA injection (Figure 2E, H) were the three prominent parameters that separated LVa and LVb neurons (Figure 2M) with the sag ratio and AP frequency after 200 pA injection being smaller in LVa than LVb, and the opposite was true for the time constant. The difference in sag ratio may indicate that LVb neurons show more prominent subthreshold oscillations, which have been reported to occur in LV, although details on differences between the two sublayers have not been studied (Egorov et al., 2002b; Schmitz et al., 1998). The clearest features aiding in separating LEC-LVb and MEC-LVb were time constant (Figure 2G), AP frequency after 200 pA injection (Figure 2E, H), and fast afterhyperpolarization (AHP; Figure 2I, N). Neurons in MEC-LVb showed a smaller time constant, higher AP frequency, and smaller fast AHP than LEC-LVb neurons. Although LVb neurons in LEC and MEC thus differed in some of their electrophysiological characteristics, as well as morphologically (described above; Figure 1G–J), it remains to be determined how these two features influence neuronal and network activity.

Local projections of LVb neurons in LEC and MEC are different

Subsequently, we examined the local entorhinal LVb circuits by injecting a tTA-dependent AAV carrying both the channelrhodopsin variant oChIEF and the yellow fluorescent protein (citrine, AAV2/1-TRE-Tight-oChIEF-citrine) into the deep layers of either LEC or MEC in mouse line MEC-13-53D (Figure 3A). This enabled specific expression of the fused oChIEF-citrine protein in either LEC-LVb (Figure 3B–F) or MEC-LVb (Figure 3G–K). Not only the dendrites and the soma but also the axons of these LVb neurons were clearly labeled. As shown in the horizontal sections taken at different dorsoventral level (Figure 3B–D, G–I), citrine-labeled axons were observed mainly within the EC, and only very sparse labeling was observed in other regions, including the angular bundle, a major efferent pathway of EC. This result supports our previous study (Ohara et al., 2018) showing that the main targets of the entorhinal LVb neurons are neurons in superficially positioned layers. Within EC, the distribution of labeled axons differed between LEC and MEC (Figure 3L). Although in both LEC and MEC, labeled axons were densely present in LIII rather than in layers II and I, we report a striking difference between LEC and MEC in LVa, as is easily appreciated from Figure 3L, M: many labeled axons of LEC-LVb neurons were present in LVa, whereas in case of MEC-LVb, the number of labeled axons was very low in LVa. Such entorhinal labeling patterns were not affected by the unintended labeled neurons in the deep perirhinal cortex (PER; Figure 3D) or postrhinal cortex (POR; Figure 3I) since these neurons hardly project to LEC and MEC (Figure 3—figure supplement 1A–C). It is also very unlikely that the sparse labeling patterns in MEC-LVa is a false negative result due to the selective targeting of a LVb subpopulation that modestly project to LVa for two reasons. First, the PCP4-labeling patterns referred to above also differ in LVa between MEC and LEC: PCP4-labeled fibers are hardly present in MEC LVa, whereas the axonal density is much higher in LEC LVa (Figure 1—figure supplement 1). Second, a strikingly similar labeling pattern was observed in LVa of MEC following an anterograde tracer injection into MEC-LVb in wild-type mice (Figure 3—figure supplement 1D). Note that we also confirmed this labeling pattern in rat MEC (Figure 3—figure supplement 1E), which is in line with a previous study (Köhler, 1986). Based on these anatomical observations, we predicted that LVb neurons in both LEC and MEC innervate LIII neurons rather than LII neurons. Importantly, our findings further indicate that LVb-to-LVa connections, which mediate the hippocampal-cortical output circuit, are much more prominent in LEC than in MEC. To test these predicted connectivity patterns, we used optogenetic stimulation of the oChIEF-labeled axons together with patch-clamp recordings of neurons in the different layers of EC.

Figure 3 with 1 supplement see all
Layer Vb (LVb) neurons project locally, and their projections differ between lateral entorhinal cortex (LEC) and medial entorhinal cortex (MEC).

(A) Tetracycline-controlled transactivator (tTA)-expressing entorhinal LVb neurons were visualized by injecting a tTA-dependent adeno-associated virus (AAV) expressing oChIEF-citrine into either LEC or MEC of MEC-13–53G. (B–F) Horizontal sections showing distribution of labeled neurites originating from LEC-LVb at different dorsoventral levels (B–D). Images of entorhinal cortex (EC) (E, F) correspond to the boxed area in (C). Note that the cell bodies of labeled neurons are located in LVb of LEC (citrine label in yellow, E), and that the labeled neurites mainly distribute within EC (citrine immunolabeling in green, F). The labeling observed in perirhinal cortex (PER; D) originates from the sparse infection of PER neurons due to the leakage of the virus along the injection tract. (G–K) Horizontal sections showing distribution of labeled fibers originating from MEC-LVb at different dorsoventral levels (G–I). Images of EC (J, K) correspond to the boxed area in (H). Note that the cell bodies of labeled neurons are located in LVb of MEC (J), and that the labeled neurites mainly distribute within EC (K). The labeling observed in postrhinal cortex (POR; I) originates from the sparse infection of POR neurons due to the leakage of the virus along the injection tract. (L) Comparison of labeled neurites originating from LEC-LVb and MEC-LVb neurons (green), of which the cell bodies are visualized with citrine (yellow). The left panel corresponds to the boxed area in (F) and is 90° rotated to match the orientation of the right panel, which represents the boxed area in (K). The distribution of the labeled fibers is strikingly different between LEC and MEC in LVa (black arrowhead) with a strong terminal projection in LEC and almost absent projections in LVa of MEC. (M) Intensity of citrine immunolabeling in LVa and LII of LEC and MEC, normalized against the LIII labeling (error bars: mean ± standard errors, N = 4). The normalized labeling was significantly higher in LEC-LVa than in MEC-LVa (two-tailed paired t-test for LEC-LVa vs. MEC-LVa: t6 = 7.68, ***p=0.0003, LEC-LII vs. MEC-LII: t6 = 0.24, p=0.82). Scale bars represent 1000 μm for (B) and (G) (also apply to C, D, H, I), 500 μm for (E) and (J) (also apply to F and K), and 100 μm for (L). Figure 3—source data 1. See also Figure 3—figure supplement 1.

Figure 3—source data 1

Distribution of labeled fibers of layer Vb neurons in entorhinal layer V.

https://cdn.elifesciences.org/articles/67262/elife-67262-fig3-data1-v2.xlsx

Translaminar local connections of MEC-LVb neurons

We first examined the LVb circuits in MEC by performing patch-clamp recording from principal neurons in layers II (n = 20 for stellate cells, n = 18 for pyramidal cells), III (n = 30), and Va (n = 18), while optically stimulating LVb fibers in acute horizontal entorhinal slices (Figure 4, Figure 4—figure supplement 1). Recorded neurons were labeled with biocytin, and the neurons were subsequently defined from the location of their cell bodies, morphological characteristics, and electrophysiological properties. In line with previous studies, LIII principal neurons were pyramidal cells, while neurons in LII were either stellate cells or pyramidal neurons (Figure 4A; Canto and Witter, 2012a; Fuchs et al., 2016; Winterer et al., 2017). LII stellate cells were not only identified by the morphological features but also from their unique physiological properties, characterized by the pronounced sag potential and DAP (Figure 4B; Alonso and Klink, 1993).

Figure 4 with 2 supplements see all
Medial entorhinal cortex-layer Vb (MEC-LVb) neurons preferentially target LII/III pyramidal neurons.

(A) Image of a representative horizontal slice showing expression of oChIEF-citrine in LVb neurons (green) and recorded neurons labeled with biocytin (magenta) in MEC. Inset shows a low-power image of the section indicating the position of the higher power image. Scale bars represent 500 μm (inset) and 100 μm. (B) Voltage responses to injected current steps recorded from neurons shown in (A): i, pyramidal cell in layer Va (LVa); ii, pyramidal cell in LIII; iii, pyramidal cell in LII; iv, stellate cell in LII. Inset in (iii) and (iv) shows the depolarizing afterpotential (DAP) in expanded voltage- and time scale. Note that LII stellate cells (iv) show a clear sag potential and DAP compared to LII pyramidal cells (iii). (C) Voltage responses to light stimulation (light blue line) recorded from neurons shown in (A). Average traces (blue) are superimposed on the individual traces (gray). (D–G) The proportion of responding cells (D), excitatory postsynaptic potential (EPSP) amplitude (E), the normalized EPSP based on LIII response (F, one-way ANOVA, F3,47 = 33.29, ***p<0.0001, Bonferroni’s multiple comparison test, **p<0.01, ***p<0.001), and the input resistance (G, one-way ANOVA, F3,82 = 21.99, ***p<0.0001, Bonferroni’s multiple comparison test, ***p<0.001) was examined for each cell type (error bars: mean ± standard errors). (H) Latency of EPSP onset for MEC neurons to optical activation (F, one-way ANOVA, F3,47 = 11.65, ***p<0.0001, Bonferroni’s multiple comparison test, *p<0.05, ***p<0.001). LIIs: LII stellate cell; LIIp: LII pyramidal cell. Figure 4—source data 1. See also Figure 4—figure supplement 1, Figure 4—figure supplement 2.

Figure 4—source data 1

Patch-clamp recording data in medial entorhinal cortex.

https://cdn.elifesciences.org/articles/67262/elife-67262-fig4-data1-v2.xlsx

There was a densely labeled axonal plexus in LIII, which is the layer where LIII pyramidal neurons mainly distribute their basal dendrites. In line with this anatomical observation, all LIII neurons (30 out of 30 cells) responded to the optical stimulation (Figure 4C, D). In contrast, the axonal labeling was sparse in LII, and this distribution was reflected in the observed sparser connectivity. The percentage of pyramidal neurons in LII responding to optical stimulation was 61.1% (11 out of 18 cells), and this percentage was especially low in stellate cells (25.0%; 5 out of 20 cells). Even in the five stellate cells that responded to the light stimulation, evoked responses were relatively small as measured by the amplitude of the synaptic event (Figure 4C, E, Figure 4—figure supplement 1C, E). In order to compare the differences of excitatory postsynaptic potential (EPSP) amplitudes across different layers/cell types, the voltage responses of each neuron were normalized to the response of LIII cells recorded in the same slice (Figure 4F). The normalized EPSP amplitude of LII cells was significantly smaller than those of LIII pyramidal cells, and within LII cells, the normalized responses of stellate cells were significantly smaller than those of pyramidal cells (LII stellate cells, 0.24 ± 0.06; LII pyramidal cells, 0.63 ± 0.08; LIII pyramidal cells, 1.0 ± 0.03; p<0.001 for LIIs vs. LIII and LIIp vs. LIII, p<0.01 for LIIs vs. LIIp, one-way ANOVA followed by Bonferroni’s multiple comparison test). This difference in responses between LII and III neurons is likely due to the difference of the LVb fiber distribution within these layers. In contrast, the difference of responses between the two cell types in LII might be explained by the difference of the input resistance between these neurons, the input resistances of stellate cells being significantly lower than that of pyramidal cells (101.2 ± 5.9 vs. 278.9 ± 29.1 MΩ; p<0.001, one-way ANOVA followed by Bonferroni’s multiple comparison test; Figure 4G), although we cannot exclude a possible difference in total synaptic input between the stellate and pyramidal neurons in LII. Note that the latency of the EPSP onset for LII stellate cells was significantly longer than that for LIII cells (4.3 ± 0.7 vs. 2.9 ± 0.1 ms; p<0.5, one-way ANOVA followed by Bonferroni’s multiple comparison test; Figure 4H). In contrast, the inputs to LII pyramidal cells showed latencies similar to those of LIII cells (3.2 ± 0.2 vs. 2.9 ± 0.1 ms). These differences in latencies indicate that projections from LVb to LIII and LII pyramidals may be monosynaptic and those to LII neurons may be disynaptic. Since we did not assess this pharmacologically, it is hard to conclude, particularly since in a parallel study in mouse LEC LII, using similar viral and laser stimulation protocols, we measured latencies ranging from 4.5 to 6.8 ms, which consistently were shown to be monosynaptic when tested using pharmacological measures (Nilssen, 2019).

As shown in Figure 1—figure supplement 4, and also in line with previous studies (Canto and Witter, 2012a; Hamam et al., 2000; Sürmeli et al., 2015), LVa pyramidal cells have their basal dendrites mainly confined to LVa, which is the layer that MEC-LVb neurons avoid to project to (Figure 3L, Figure 4A, Figure 4—figure supplement 1D, G). In line with this anatomical observation, only 27.8% (5 out of 18 cells) responded to the light stimulation (Figure 4D). The EPSP amplitudes of the responding LVa neurons were relatively small (Figure 4C, E, Figure 4—figure supplement 1F, H), and the normalized EPSPs were significantly smaller than those of LIII neurons (0.47 ± 0.10 vs. 1.0 ± 0.03; p<0.001, one-way ANOVA followed by Bonferroni’s multiple comparison test; Figure 4F). In addition, the latency of the EPSP onset for LVa cells was significantly longer than that for LIII cells (5.6 ± 1.7 vs. 2.9 ± 0.1 ms; p<0.001, one-way ANOVA followed by Bonferroni’s multiple comparison test; Figure 4H), indicating that these responses are either the result of monosynatic inputs onto the apical dendrite in LIII or that they represent disynaptic responses.

We recorded in slices taken at different dorsoventral levels. Since functional differences along this axis have been reported (Steffenach et al., 2005; Stensola et al., 2012), we examined whether the LVb-LVa connectivity differs along the dorsoventral axis by grouping the recorded LVa responses in three distinct dorsoventral levels (Figure 4—figure supplement 2). The voltage responses of the more ventrally positioned LVa neurons were significantly higher than those measured more dorsally in MEC (p<0.01, one-way ANOVA followed by Bonferroni’s multiple comparison test). Since the EPSP amplitudes of LIII neurons did not differ at different dorsoventral levels, it is unlikely that the observed response differences are caused by different levels of oChIEF expression in LVb fibers along the dorsoventral axis. The observed difference may be caused by the difference in severing of apical dendrites at different dorsoventral levels since the axons of MEC-LVb neurons massively distribute in LIII and frequently reach into LI. This, however, does not seem to be the case since LVa cells in the dorsal MEC, irrespective of whether they have full (n = 3) or severed dendrites (n = 2), did not respond to light stimulation. Similarly, for the central dorsoventral level, four out of five neurons with intact apical dendrites in LIII did not respond. In the ventral MEC recordings, four out of five neurons with seemingly intact dendrites responded to the optical stimulation (Figure 4—figure supplement 2D).

Translaminar local connections of LEC-LVb neurons

We next examined the LVb local circuits in LEC with the similar method as applied in MEC (above). In LEC, LII can further be divided into two sublayers: a superficial layer IIa (LIIa) composed of fan cells and a deep layer IIb (LIIb) mainly composed of pyramidal neurons (Leitner et al., 2016). Fan cells mainly extend their apical dendrites in LI, where the density of LVb labeled fibers is extremely low (Figure 5A, Figure 5—figure supplement 1). This contrasts with LIIb, LIII, and LVa neurons, which distribute at least part of their dendrites in layers with a relatively high density of LVb axons. In line with these anatomical observations, only 26.9% of the fan cells (7 out of 26 neurons) responded to the light stimulation (Figure 5C, D). On the other hand, the response probabilities of LIIb, III, and Va were high, 76.9% (20 out of 26 cells), 100% (34 out of 34 cells), and 94.7% (18 out of 19 cells), respectively. The voltage responses of these neurons were also significantly larger than those of the LIIa neurons (LIIa neurons, 0.15 ± 0.03; LIIb neurons, 0.79 ± 0.15; LIII neurons, 1.0 ± 0.04; LVa neurons, 0.93 ± 0.10; p<0.01 for LIIa vs. LIIb, p<0.001 for LIIa vs. LIII and LIIa vs. LVa, one-way ANOVA followed by Bonferroni’s multiple comparison test; Figure 5E, F). In contrast to MEC-LII stellate cells, the input resistance of LIIa fan cells was significantly higher than that of LIIb and LIII neurons (LIIa neurons, 323.6 ± 16.1; LIIb neurons, 226.5 ± 17.2; LIII neurons, 194.3 ± 13.3 MΩ; p<0.001, one-way ANOVA followed by Bonferroni’s multiple comparison test; Figure 5G). This indicates that the small responses of LIIa fan cells cannot be explained by the differences in input resistance among the superficial neurons and may simply be due to the small number of synaptic inputs to LIIa fan cells from LVb neurons. In contrast to MEC, LEC-LVa neurons showed large responses to light stimulation, which matched with the anatomically dense LVb fiber distribution in LEC-LVa (Figure 3L). In line with our previous monosynaptic input tracing study using rabies virus (Ohara et al., 2018), the latency of the EPSP onset, which was similar in all cell types (LIIa neurons, 4.8 ± 1.1; LIIb neurons, 5.0 ± 1.4; LIII neurons, 4.2 ± 0.9; LVa neurons, 4.2 ± 0.6 ms; Figure 5H), points to the LVb-to-LVa connectivity, as well as that to LIII and LII, being largely monosynaptic.

Figure 5 with 1 supplement see all
Lateral entorhinal cortex-layer Vb (LEC-LVb) neurons target layer Va (LVa) pyramidal neurons as well as LII/III pyramidal neurons.

(A) Representative image of semicoronal slice showing expression of oChIEF-citrine in LVb neurons (green) and recorded neurons labeled with biocytin (magenta) in LEC. Inset shows a low-power image of the section indicating the position of the higher-power image. Scale bars represent 500 μm (inset) and 100 μm. (B) Voltage responses to injected current steps recorded from neurons shown in (A): i, pyramidal cell in LVa; ii, pyramidal cell in LIII; iii, pyramidal cell in LII; iv, fan cell in LII. (C) Voltage responses to light stimulation (light blue line) recorded from neurons shown in (A). Average traces (blue) are superimposed on the individual traces (gray). (D–G) The proportion of responding cells (D), excitatory postsynaptic potential (EPSP) amplitude (E), the normalized EPSP based on LIII response (F, one-way ANOVA, F3,75 = 7.675, ***p=0.0002, Bonferroni’s multiple comparison test, **p<0.01, ***p<0.001), and the input resistance (G, one-way ANOVA, F3,101 = 11.75, ***p<0.0001, Bonferroni’s multiple comparison test, *p<0.05, ***p<0.001) was examined for each cell type (error bars: mean ± standard errors). (H) Latency of EPSP onset for LEC neurons to optical activation (one-way ANOVA, F3,47 = 11.65). (I, J) Comparison of the proportion of responding cells (I) and the normalized EPSP based on LIII response (J) between medial entorhinal cortex (MEC) and LEC (error bars: mean ± standard errors; two-tailed unpaired t-test, t21 = 2.239, *p=0.0361). Figure 5—source data 1. See also Figure 5—figure supplement 1.

Figure 5—source data 1

Patch-clamp recording data in lateral entorhinal cortex.

https://cdn.elifesciences.org/articles/67262/elife-67262-fig5-data1-v2.xlsx

The striking difference between MEC and LEC regarding LVb to LVa projections is clear from comparing the proportion of responding neurons (Figure 5I), and the normalized EPSP based on LIII response (Figure 5J) between MEC and LEC. In contrast to the similar responses of LII neurons between the two subregions, the normalized voltage responses of LEC-LVa neurons were significantly larger than those of MEC-LVa neurons (0.93 ± 0.10 vs. 0.47 ± 0.10, p<0.05, two-tailed unpaired t-test; Figure 5J). On a final note, it is apparent that postsynaptic responses in LEC are larger than those in MEC. This difference in response amplitudes is most likely due to the differences in the number of oChIEF-expressing neurons since, as shown in Figure 1E, the proportion of tTA-expressing neurons is higher in LEC than in MEC. We deem it unlikely that these amplitude differences are caused by differences in biophysical properties since no such differences have been reported between matching cell types in LEC and MEC (Canto and Witter, 2012a). We also did not observe striking effects of the maintenance of the apical dendritic tree of LEC-LVa; neurons showed responses irrespective of whether they had dendrites restricted to layer V (n = 14) or additionally showed dendrites extending into layer III (n = 5; Figure 4—figure supplement 2D).

The present data clearly show that neurons in LVb of both LEC and MEC give rise to dense intrinsic projections to more superficial layers and show laminar preferences (Figure 6). We noticed a striking difference between the two entorhinal regions, in that neurons in LEC-LVb seemed to innervate LVa neurons, whereas in dorsal MEC this was rarely the case. In contrast, other intrinsic circuits from LVb to LII/III were very similar in both entorhinal subdivisions, which preferentially targeted pyramidal cells rather than the stellate or fan cells in LII. These data indicate that the weak experimental support for a strong LVb-to-LVa projection in dorsal MEC is not due to technical issues. Rather, the data provide evidence that the intrinsic LV circuitry that supposedly underlies the canonical hippocampal-cortical output circuit in dorsal MEC is differently organized than in other parts of EC.

Schematic diagram of the different local circuits in lateral entorhinal cortex (LEC) and medial entorhinal cortex (MEC) used by layer Vb (LVb) neurons to transfer dorsal hippocampal output.

Local connectivity of LVb neurons in LEC (left, orange) and MEC (right, purple). In both LEC and MEC, LVb neurons are the primary recipients of dorsal hippocampal output, but the transfer to LVa neurons through direct LVb-to-LVa projections is only prominent in LEC. Such projections are sparse and weak in MEC. Neurons in LVa are the output neurons of EC, projecting to the neocortex and other telencephalic subcortical structures. In contrast, in both LEC and MEC we find projections from LVb that target pyramidal cells in LIII, including neurons projecting to CA1 and subiculum, and pyramidal cells in LII. Projections to stellate (MEC) and fan (LEC) cells, which project to the dentate gyrus and CA3, are sparse and weak. The output projections of LII pyramidal neurons are not indicated in the figure, they project to ipsilateral-EC, contralateral-EC, CA1, or other telencephalic structures (Ohara et al., 2019). For clarity reasons, all these projections are indicated schematically as originating from a single LVb neuron, but this is not yet known. f: fan cell; s: stellate cell; p: pyramidal cell.

Discussion

In this study, we experimentally tested the major assumption about the organization of hippocampal-cortical output circuits via entorhinal LVb neurons, considered to be crucial for the normal functioning of the medial temporal lobe memory system, more in particular systems memory consolidation. Our key finding is that LEC and MEC are strikingly different in that excitatory connectivity from LVb to LVa is anatomically denser and electrophysiologically stronger in dorsal LEC than in dorsal MEC. In addition, we present new data that point to three major functionally relevant insights in the organization of the intrinsic translaminar entorhinal network originating from LVb neurons.

First, the present data indicate that LVb pyramidal neurons in LEC and MEC differ with respect to main morphological and electrophysiological characteristics. In contrast, LVa neurons in MEC and LEC are rather similar in these two aspects. Second, we show that projections from principal neurons in LVb in both entorhinal subdivisions preferentially contact pyramidal neurons in LIII and LII. LVb neurons have a sparser connectional relationship with principal neurons in LII that project to the dentate gyrus (DG) and the CA3/CA2 region, i.e., stellate and fan cells. Last, and most important, our data point to a new and challenging circuit difference between the two entorhinal subdivisions with respect to the inputs to LVa neurons, i.e., the output neurons of EC. Whereas in LEC, LVa neurons receive substantial input from LVb neurons, this projection is relatively weak in dorsal MEC. This difference in dorsal MEC, though unexpected in view of previous data including our own rabies tracing data (Ohara et al., 2018), has been recently corroborated in an in vitro study using paired-patch recording (Rozov et al., 2020), and our present corroborating tracing data are in line with previous tracing data in the rat (Köhler, 1986) and monkey (Chrobak and Amaral, 2007).

Layer Vb neurons in LEC and MEC are morphologically and electrophysiologically different

The use of layer-specific TG mice allowed us to differentiate neurons in LVa from those in LVb and to differentiate these layer-specific neuron types between LEC and MEC. This contrasts with previous studies in rats, showing that LV neurons in both LEC and MEC share electrophysiological properties (Canto and Witter, 2012a; Canto and Witter, 2012b; Hamam et al., 2000; Hamam et al., 2002), although these authors did differentiate between LVa and LVb neurons based on morphological criteria and laminar distribution. We corroborate the reported morphological differences and add that the neurons also differ with respect to their electrophysiological properties. The most striking difference between MEC- and LEC-LVb neurons, however, is in the morphology of the apical dendrite. Neurons in MEC-LVb have an apical dendrite that heads straight to the pia, such that distal branches reach all the way up into LI, which is in line with previous studies (Canto and Witter, 2012a; Hamam et al., 2000; Sürmeli et al., 2015). In contrast, the apical dendrites of LEC-LVb neurons have a more complex branching pattern and they do not extend beyond LIII. This indicates that LEC-LVb neurons are unlikely to be targeted by inputs to LEC that selectively distribute to layers I and II, such as those carrying olfactory information from the olfactory bulb and the piriform cortex (Luskin and Price, 1983) as well as commissural projections (Leitner et al., 2016). The LVb neurons in LEC are thus dissimilar to their counterparts in MEC, which are morphologically suited to receive such superficially terminating inputs, as has been shown for inputs from the parasubiculum (Canto et al., 2012) and contralateral MEC (Fuchs et al., 2016). The here reported differences between LVb neurons, with MEC-LVb neurons showing a shorter time constant than LEC-LVb neurons, further indicate that MEC-LVb neurons have a shorter time window to integrate inputs compared to LEC-LVb neurons (Canto and Witter, 2012a; Canto and Witter, 2012b). The differences in AP frequency and half duration of AP may result in differences in the propensity of neurons to show graded persistent firing, which is prominent in MEC LV. Unfortunately, reports of persistent activity in MEC do not differentiate between neurons in LVa and LVb (Egorov et al., 2002a; Fransén et al., 2006). However, up-down state activity originating in LIII particularly entrains neurons in LVb (Beed et al., 2020), indicating that indeed LVb neurons might preferentially show persistent activity. Together, these differences will result in differences in information processing.

Layer Vb neurons preferentially target pyramidal neurons in layers III and II rather than layer II neurons that project to the DG

Both our anatomical and electrophysiological data show that projections from principal neurons in LVb in both entorhinal subdivisions preferentially target pyramidal neurons in LIII and LII. LVb neurons have a weaker relationship with the class of stellate and fan cells in MEC or LEC, respectively. This makes it likely that in both LEC and MEC hippocampal information preferentially interacts with neurons that are part of the LIII-to-CA1/Sub projection system rather than with the LII-to-DG/CA2-3 projecting neurons. Additional target neurons in layer II/III might be the pyramidal neurons that project contralaterally, which in LII belong to the calbindin (CB+) population (Ohara et al., 2019; Steward and Scoville, 1976; Varga et al., 2010), as well as the substantial population of CB+ excitatory intrinsic projection neurons (Ohara et al., 2019). The present findings are in line with a previous study using wild-type mice, reporting that most of the inputs to MEC-LII stellate cell arise from superficial layers, whereas those of MEC-LII pyramidal cells arise from the deep layers (Beed et al., 2010).

The relatively sparse projection from MEC LVb neurons to LII stellate cells and the more massive projection to LII pyramidal cell was unexpected for two reasons. First, both the stellate and the CB+ population of layer II pyramidal neurons contain grid cells (Hafting et al., 2005; Tang et al., 2014) and hippocampal excitatory inputs are required for the formation and translocation of grid patterns (Bonnevie et al., 2013). Though our data do not exclude that LVb inputs can reach LII stellate cells indirectly through LIII- and LII-pyramidal cells (Ohara et al., 2019; Winterer et al., 2017), they do indicate that the two populations of grid cells, stellate vs. CB+ cells, might differ with respect to the strength of a main excitatory drive from the hippocampus.

Second, re-entry of hippocampal activity, i.e., the presence of recurrent circuits, has been proposed as one of the mechanisms for temporal storage of information in a neuronal network (Edelman, 1989; Iijima et al., 1996). Re-entry through LII-to-DG has been observed in in vivo recordings under anesthesia in rats, although this was examined with current source density analysis, which is not optimal to exclude multisynaptic responses (Kloosterman et al., 2004). Such multisynaptic inputs could be mediated by pyramidal neurons in LIII and LII, both of which do contact layer II-to-DG projecting neurons (Ohara et al., 2019; Winterer et al., 2017). Our current data strongly favor the circuit via LIII-CA1/subiculum in both entorhinal subdivisions to mediate a recurrent hippocampal-entorhinal-hippocampal circuit. The importance of this layer III recurrent network is corroborated by the observation that entorhinal LIII input to the hippocampus field CA1 plays a crucial role in associating temporally discontinuous events and retrieving remote memories (Lux et al., 2016; Suh et al., 2011).

The density and strength of the excitatory projection from LVb to LVa appear more prominent in dorsal LEC compared to dorsal MEC

Ever since the seminal observation in monkeys and rats of a hippocampal-cortical projection mediated by layer V of the EC (Kosel et al., 1982; Rosene and Van Hoesen, 1977), the canonical circuit underlying the hippocampal-cortical interplay, necessary for memory consolidation (Buzsáki, 1996; Eichenbaum et al., 2012), is believed to use EC LV neurons that receive hippocampal output and send projections to the neocortex. More recent studies in rats and mice indicated that neurons in LVb likely are the main recipients of this hippocampal output stream (Sürmeli et al., 2015) and that principal neurons in LVa form the main source of outputs to neocortical areas (Ohara et al., 2018; Sürmeli et al., 2015). We showed that in LEC as well as ventrally in MEC the LVb-to-LVa connectivity is relatively well developed, in line with our previous study (Ohara et al., 2018). In contrast, the relatively sparse connection from LVb-to-LVa in dorsal MEC reported here indicates that at least in dorsal MEC the canonical role of EC LV neurons to mediate hippocampal information transfer to downstream neocortical areas might require a revision. Note that in the present study we focused on the presumed direct excitatory connectivity from LVb-to-LVa neurons using a newly derived TG mouse line. Although we above provided data to argue that we find it unlikely that the reported difference in connectivity between LEC and MEC might be caused by a different preference for a specific cell type in MEC vs. LEC, we cannot completely exclude that option. We further cannot exclude that the apparent differences between dorsal MEC, on the one hand, and ventral MEC and LEC, on the other hand, might be modulated through differences in local circuits, resulting in state changes in the LV network, such as reported in MEC depending on activity in LIII (Beed et al., 2020). Likewise, differences in local inhibitory circuits or in the in vivo membrane potential and spike threshold of LVa neurons might be relevant, but in vivo these are unknown. However, our in vitro data do not support the latter possibility. The effectiveness of functional connections may further be influenced by incoming inputs to LVa and/or LVb, such as those from the claustrum (Kitamura et al., 2017), medial septum, medial prefrontal cortex, and retrosplenial cortex (Ohara et al., 2018). Adding to this complexity is a recent report that there is a direct projection from the intermediate/ventral hippocampus to neurons in MEC LVa (Rozov et al., 2020). To assess these complex interactions, further in vivo studies are clearly required.

Finally, it is of interest that LVa, the entorhinal-output layer, is thicker in LEC than in MEC, which might be taken to strengthen our proposal that LEC might be the more relevant player in mediating the hippocampal-cortical interplay relevant for systems memory consolidation (Buzsáki, 1996; Eichenbaum et al., 2012; Frankland and Bontempi, 2005). However, studies that have functionally linked the LVa-output projection with memory consolidation are based on data obtained in MEC (Kitamura et al., 2017). In our view, this more likely reflects the strong focus on functions of MEC circuits rather than LEC circuits ever since the discovery of the grid cell (Hafting et al., 2005; Moser et al., 2017). With the discovery of LEC networks coding for event sequences (Bellmund et al., 2019; Montchal et al., 2019; Tsao et al., 2018), this is likely to change. It is clear that our current suggestion that LEC might be more relevant than MEC in mediating the export of information from the dorsal hippocampus to the neocortex needs to be substantiated in vivo. The functional relevance of the similarities between networks in LEC and MEC mediating re-entry into the hippocampal formation with an apparent preference to target CA1 and subiculum is likewise in need of in vivo studies in order to understand the functional consequences of the present data.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or referenceIdentifiersAdditional
information
Strain, strain background
(Mus musculus)
MEC13-53DBlankvoort et al., 2018Not commercially available, but upon request we can send the mouse line. Send email to Dr. Stefan Blankvoort; stefan.blankvoort@ntnu.no.
Strain, strain background (Mus musculus)tetOGCaMP6-mCherryBlankvoort et al., 2018Not commercially available, but upon request we can send the mouse line. Contact see above.
Strain, strain background (Mus musculus)GAD1GFPTamamaki et al., 2003The animals are bred in house after obtaining breeding pairs from Dr Yuchio Yanagawa;yanagawa@med.gunma-u.ac.jp.
Strain, strain background (Mus musculus)Gt(ROSA)26Sortm9(CAG-tdTomato)HzeThe Jackson Laboratory007909
RRID:IMSR_JAX:007909
Genetic reagent (virus)AAV-TRE-tight-GFP (serotype 2/1)Nilssen et al., 2018Viral Vector Core at Kavli Institute for Systems Neuroscience; contact Dr Rajeevkumar Nair Raveendranrajeevkumar.r.nair@ntnu.no.
Genetic reagent (virus)AAV-TRE-tight-oChIEF-citrine (serotype 2/1)Nilssen et al., 2018Viral Vector Core at Kavli Institute for Systems Neuroscience; contact see above.
Genetic reagent (virus)AAV-CMV-FLEX-mCherry (serotype 2/1)This paperViral Vector Core at Kavli Institute for Systems Neuroscience; contact see above.
Genetic reagent (virus)AAVrg-pmSyn1-EBFP-creAddgene51507
Genetic reagent (virus)AAV1.CAG.tdTomato.WPRE.SV40Upenn viral coreAV-1-PV3365
AntibodyAnti-GFP
(chicken polyclonal)
Abcamab13970
RRID:AB_300798
(1:500)
AntibodyAnti-GFP
(rabbit polyclonal)
Thermo Fisher ScientificA11122
RRID:AB_221569
(1:2000)
AntibodyAnti-PCP4
(rabbit polyclonal)
Sigma AldrichHPA005792
RRID:AB_1855086
(1:300)
AntibodyAnti-Ctip2
(rat monoclonal)
Abcamab18465
RRID:AB_2064130
(1:3000)
AntibodyAnti-NeuN
(guinea pig polyclonal)
MilliporeABN90P
RRID:AB_2341095
(1:1000)
AntibodyAnti-NeuN
(mouse monoclonal)
MilliporeMAB377
RRID:AB_2298772
(1:1000)
AntibodyAnti-PHA-L
(rabbit)
Vector LaboratoriesAS-2300
RRID:AB_2313686
(1:1000)
AntibodyGoat anti-chicken IgG (AF 488)Thermo Fisher ScientificA11039
RRID:AB_2534096
(1:400)
AntibodyGoat anti-rabbit IgG (AF 546)Thermo Fisher ScientificA11010
RRID:AB_2534077
(1:400)
AntibodyGoat anti-rabbit IgG (AF 635)Thermo Fisher ScientificA31576
RRID:AB_10374303
(1:400)
AntibodyGoat anti-rat IgG (AF 633)Thermo Fisher ScientificA21094
RRID:AB_2535749
(1:400)
AntibodyGoat anti-guinea pig IgG (AF 647)Thermo Fisher ScientificA21450
RRID:AB_2735091
(1:400)
AntibodyGoat anti-guinea pig IgG (AF 488)Thermo Fisher ScientificA11073
RRID:AB_2534117
(1:400)
AntibodyGoat anti-mouse IgG (AF 488)Thermo Fisher ScientificA11001
RRID:AB_2534069
(1:400)
AntibodyStreptavidin, Alexa Fluor 546 conjugateThermo Fisher ScientificS11225
RRID:AB_2532130
(1:600)
AntibodyCy3 streptavidinJackson Immuno Research016-160-084
RRID:AB_2337244
(1:400)
AntibodyNeurotrace 640/660 deep-red fluorescent Nissl stainThermo Fisher ScientificN21483
RRID:AB_2572212
(1:200)
Chemical compound, drugBiotinylated dextran amineInvitrogenD1956
Chemical compound, drugPhaseolus vulgaris leucoagglutininVector LaboratoriesL-1110
Software, algorithmPatchmasterHeka Eletronik
Software, algorithmClampfitMolecular Devices
Software, algorithmMATLAB, 2018aMathWorks
Software, algorithmImage Jhttp://rsb.info.nih.gov/ij
Software, algorithmGraphPad Prism, version 5GraphPad software

Animals

All animals were group housed at a 12:12 hr reversed day/night cycle and had ad libitum access to food and water. Mice of the TG MEC13-53D enhancer strain expressing tTA in PCP4-positive entorhinal LVb neurons (Blankvoort et al., 2018) were used for whole-cell recordings (n = 38) and histological assessment of specific transgene expression (n = 7). To characterize the tTA expression patterns in this mouse line, MEC13-53D was crossed with a tetOGCaMP6-mCherry line (Blankvoort et al., 2018; n = 2). Other TG mouse lines, GAD1GFP (Tamamaki et al., 2003; n = 4) and Gt(ROSA)26Sortm9(CAG-tdTomato)Hze (Madisen et al., 2010; n = 2), were used to characterize entorhinal neurons in layers Va and Vb. We further used C57BL/6N mice to characterize the morphology of entorhinal LVa neurons (n = 2) and examine the projection of entorhinal LVb neurons and PER neurons in wild-type mice (n = 2). The projection of PER/POR neurons was also examined in MEC13-53D (n = 2). Information on the availability of animals is summarized in the Key Resources Table. All experiments were approved by the local ethics committee and were in accordance with the European Communities Council Directive and the Norwegian Experiments on Animals Act (#17898, #22312).

Surgical procedures and virus/tracer injections

Request a detailed protocol

Animals were anesthetized with isoflurane in an induction chamber (4%, Nycomed, airflow 1 l/min), after which they were moved to a surgical mask on a stereotactic frame (Kopf Instruments). The animals were placed on a heating pad (37°C) to maintain stable body temperature throughout the surgery, and eye ointment was applied to the eyes of the animal to protect the corneas from drying out. The animals were injected subcutaneously with buprenorphine hydrochloride (0.1 mg/kg, Temgesic, Indivior), meloxicam (1 mg/kg, Metacam Boehringer Ingelheim Vetmedica), and bupivacaine hydrochloride (Marcain 1 mg/kg, Astra Zeneca), the latter at the incision site. The head was fixed to the stereotaxic frame with ear bars, and the skin overlying the skull at the incision site was disinfected with ethanol (70%) and iodide before a rostrocaudal incision was made. A craniotomy was made around the approximate coordinate for the injection, and precise measurements were made with the glass capillary used for the virus injection. The coordinates of the injection sites are as follows (anterior to either bregma [APb] or transverse sinus [APt], lateral to sagittal sinus [ML], ventral to dura [DV] in mm): LEC (APt +2.0, ML 3.9, DV 3.0), MEC (APt +1.0, ML 3.3, DV 2.0), nucleus accumbens (NAc) (APb +1.2, ML 1.0, DV 3.8), retrosplenial cortex (RSC) (APb −3.0, ML 0.3, DV 0.8), PER (APb −4.5, ML 4.5, DV 1.5), and POR (APt +1.1, ML 3.3, DV 0.9). Viruses were injected with a nanoliter injector (Nanoliter 2010, World Precision Instruments) controlled by a microsyringe pump controller (Micro4 pump, World Precision Instruments); 100–300 nl of virus was injected with a speed of 25 nl/min. The capillary was left in place for an additional 10 min after the injection, before it was slowly withdrawn from the brain. Finally, the wound was rinsed, and the skin was sutured. The animals were left to recover in a heating chamber, before being returned to their home cage, where their health was checked daily.

For electrophysiological studies, young MEC13-53D mice (5–7 weeks old) were injected with a tTA-dependent AAV (serotype 2/1) carrying either GFP or a fused protein of oChIEF, a variant of the light-activating protein channelrhodopsin2 (Lin et al., 2009), and citrine, a yellow fluorescent protein (Griesbeck et al., 2001). The construction of these viruses, AAV-TRE-tight-GFP and AAV-TRE-tight-oChIEF-citrine respectively, has been described in Nilssen et al., 2018. Data on availability of viral constructs are summarized in the Key Resources Table. These samples were also used to characterize the transgenic mouse line and also the projection patterns of entorhinal LVb neurons. To label LVa neurons, retrograde AAV expressing enhanced blue fluorescent protein (EBFP) and Cre recombinase (AAVrg-pmSyn1-EBFP-cre, Addgene #51507) was injected into either NAc or RSC of Gt(ROSA)26Sortm9(CAG-tdTomato)Hze. LVa neurons were also labeled in C57BL/6N mice by injecting AAVrg-pmSyn1-EBFP-cre in NAc while injecting AAV-CMV-FLEX-mCherry in LEC/MEC. The pAAV-FLEX-mCherry-WPRE construct was created by first cloning a FLEX cassette with MCS into Cla1 and HindIII sites in pAAV-CMV-MCS-WPRE (Agilent) to create pAAV-CMV-FLEX-MCS-WPRE. The sequence of the FLEX cassette was obtained from Atasoy et al., 2008. Subsequently, the mCherry sequence was synthesized and cloned in an inverted orientation into EcoR1 and BamH1 sites in pAAV-CMV-FLEX-MCS-WPRE to make pAAV CMV-FLEX-mCherry-WPRE. AAV-CMV-FLEX-mCherry was recovered from pAAV CMV-FLEX-mCherry-WPRE as described elsewhere (Nair et al., 2020; Nilssen et al., 2018).

For anterograde tracing experiments in wild-type animals, either 2.5% phaseolus vulgaris leucoagglutinin (PHA-L; Vector Laboratories, #L-1110) or 3.5% 10 kDa biotinylated dextran amine (BDA; Invitrogen, #D1956) was injected iontophoretically with positive 6 µA current pulses (6 s on; 6 s off) for 15 min. To label projections from PER in C57BL/6N mice, AAV1.CAG.tdTomato.WPRE.SV40 (Upenn viral core, cat. AV-1-PV3365) was injected iontophoretically with positive 5 µA current pulses (5 s on; 5 s off) for 5 min.

Acute slice preparation

Request a detailed protocol

Two to three weeks after AAV injection, acute slice preparations were prepared as described in detail (Nilssen et al., 2018). Briefly, mice were deeply anesthetized with isoflurane and decapitated. The brain was quickly removed and immersed in cold (0°C) oxygenated (95% O2/5% CO2) artificial cerebrospinal fluid (ACSF) containing 110 mM choline chloride, 2.5 mM KCl, 25 mM D-glucose, 25 mM NaHCO3, 11.5 mM sodium ascorbate, 3 mM sodium pyruvate, 1.25 mM NaH2PO4, 100 mM D-mannitol, 7 mM MgCl2, and 0.5 mM CaCl2, pH 7.4, 430 mOsm. The brain hemispheres were subsequently separated and 350-μm-thick entorhinal slices were cut with a vibrating slicer (Leica VT1000S, Leica Biosystems). We used semicoronal slices for LEC recording, which were cut with an angle of 20° with respect to the coronal plane to optimally preserve neurons and local connections of LEC (Canto and Witter, 2012b; Nilssen et al., 2018). In case of MEC recordings, horizontal slices were prepared (Canto and Witter, 2012a; Couey et al., 2013). Slices were incubated in a holding chamber at 35°C in oxygenated ASCF containing 126 mM NaCl, 3 mM KCl, 1.2 mM Na2HPO4, 10 mM D-glucose, 26 mM NaHCO3, 3 mM MgCl2, and 0.5 mM CaCl2 for 30 min and then kept at room temperature for at least 30 min before use.

Electrophysiological recording

Request a detailed protocol

Patch-clamp recording pipettes (resistance: 4–9 MΩ) were made from borosilicate glass capillaries (1.5 outer diameter × 0.86 inner diameter; Harvard Apparatus) and back-filled with internal solution of the following composition: 120 mM K-gluconate, 10 mM KCl, 10 mM Na2-phosphocreatine, 10 mM HEPES, 4 mM Mg-ATP, 0.3 mM Na-GTP, with pH adjusted to 7.3 and osmolality to 300–305 mOsm. Biocytin (5 mg/ml; Iris Biotech) was added to the internal solution in order to recover cell morphology. Acute slices were moved to the recording setup and visualized using infrared differential interference contrast optics aided by a ×20/1.0 NA water immersion objective (Zeiss Axio Examiner D1, Carl Zeiss). Electrophysiological recordings were performed at 35°C and slices superfused with oxygenated recording ACSF containing 126 mM NaCl, 3 mM KCl, 1.2 mM Na2HPO4, 10 mM D-glucose, 26 mM NaHCO3, 1.5 mM MgCl2, and 1.6 mM CaCl2. LVb in both LEC and MEC was identified through the presence of the densely packed small cells, and LVb neurons labeled with GFP were selected for recording. LVa neurons were selected for recording on the basis of their large soma size and the fact that they are sparsely distributed directly superficial to the smaller neurons of LVb. Gigaohm resistance seals were acquired for all cells before rupturing the membrane to enter whole-cell mode. Pipette capacitance compensation was performed prior to entering whole-cell configuration, and bridge balance adjustments were carried out at the start of current-clamp recordings. Data acquisition was performed by Patchmaster (Heka Elektronik) controlling an EPC 10 Quadro USB amplifier (Heka Elektronik). Acquired data were low-pass Bessel filtered at 15.34 kHz (for whole-cell current-clamp recording) or 4 kHz (for whole-cell voltage-clamp recording) and digitized at 10 kHz. No correction was made for the liquid junction potential (13 mV as measured experimentally). Data were discarded if the resting membrane potential was ≥ −57 mV or/and the series resistance was ≥40 MΩ.

Intrinsic membrane properties were measured from membrane voltage responses to step injections of hyperpolarizing and depolarizing current (1 s duration, −200 pA to 200 pA, 20 pA increments). Acquired data were exported to text file with MATLAB (MathWorks) and were analyzed with Clampfit (Molecular Devices). The following electrophysiological parameters analyzed were defined as follows:

  • Resting membrane potential (Vrest; mV): membrane potential measured with no current applied (I = 0 mode).

  • Input resistance (Mohm): resistance measured from Ohm’s law from the peak of voltage responses to hyperpolarizing current injections (−40 pA injection).

  • Time constant (ms): the time it took the voltage deflection to reach 63% of peak of voltage response at hyperpolarizing current injections (−40 pA injection).

  • Sag ratio (steady-state/peak): measured from voltage responses to hyperpolarizing current injections with peaks at −90 ± 5 mV, as the ratio between the voltage at steady state and the voltage at the peak.

  • AP threshold (mV): the membrane potential where the rise of the AP was 20 mV/ms.

  • AP peak (mV): voltage difference from AP threshold to peak.

  • AP half-width (ms): duration of the AP at half-amplitude from AP threshold.

  • AP maximum rate of rise (mV/ms): maximal voltage slope during the upstroke of the AP.

  • Fast AHP (in mV): the peak of AHP in a time window of 6 ms after the membrane potential reached 0 mV during the repolarization phase of AP.

  • Medium AHP (mV): the peak of AHP in a time window of 200 ms after fast AHP.

  • AP frequency after 200 pA inj. (Hz): frequency of APs evoked with +200 pA of 1-s-long current injection.

  • Adaptation: measured from trains of 10 ± 1 APs as [1 -(Last/First)], where Last and First are, respectively, the frequencies of the last and first interspike interval.

  • DAP: depolarizing voltage deflection after AP. DAP was defined based on previous studies (Canto and Witter, 2012a; Canto and Witter, 2012b; Hamam et al., 2000; Hamam et al., 2002).

Optogenetic stimulation and patch-clamp data analysis oChIEF+ fibers were photostimulated with a 473 nm laser controlled by a UGA-42 GEO point scanning system (Rapp OptoElectronic) and delivered through a ×20/1.0 NA WI objective (Carl Zeiss Axio Examiner.D1). Laser pulses had a beam diameter of 36 μm and a duration of 1 ms. The tissue was illuminated with individual pulses at a rate of 1 Hz in a 4 × 5 grid pattern. The grid was positioned such as to allow for the light stimulation to cover across layer I to Va. Laser power was fixed to an intensity that evokes inward currents (EPSCs) but not action potentials. The voltage- or current traces from individual stimulation spots were averaged over 4–6 individual sweeps to create an average response for each point in the 4 × 5 grid. The stimulation point that showed the largest voltage response was used for further analysis. Deflections of the average voltage trace exceeding 10 SDs (±10 SDs) of the baseline were classified as synaptic responses. Postsynaptic potentials were calculated as the difference between the peak of the evoked synaptic potential and the baseline potential measured before stimulus onset. The latency of optical activation was defined as the time interval between light onset and the point where the voltage trace exceeded 10% amplitude. Data analysis was performed using MATLAB. Only slices with at least one successful synaptic response to photostimulation were included in the analysis. There were no outliers that were removed from the data.

All data presented in the figures are shown as mean ± standard errors. Prism software was used for data analysis (GraphPad software), and one-way ANOVA with Bonferroni’s multiple comparison test was used to compare the electrophysiological properties and voltage responses between each cell types. To analyze mediolateral gradient in LVb-to-LVa connectivity of MEC, we used Pearson correlation coefficient. A principal component analysis based on the 12 electrophysiological properties (Figure 2—source data 1) was conducted in MATLAB. For this purpose, all variables were normalized to a standard deviation of 1.

Histology, immunohistochemistry, and imaging of electrophysiological slices

Request a detailed protocol

After electrophysiological recordings, the brain slices were put in 4% paraformaldehyde (PFA, Merck Chemicals) in 0.1 M phosphate buffer (PB) for 48 hr at 4°C. Slices were permeabilized 5 × 15 min in phosphate buffered saline containing 0.3% Triton X-100 (PBS-Tx) and were immersed in a blocking solution containing PBS-TX and 10% Normal Goat Serum (NGS, Abcam: AB7481) for 3 hr at room temperature. To visualize targeted entorhinal LVb neurons expressing either GFP or oChIEF-citrine, slices were incubated with a primary antibody, chicken anti-GFP (1:500, Abcam, #ab13970), diluted in the blocking solution for 4 days at 4°C. Note that citrine is derived from Aequorea victoria GFP (Griesbeck et al., 2001), and thus, the signal can be amplified with GFP antibodies. After this, the sections were washed 5 × 15 min in PBS-Tx at room temperature and incubated in a secondary antibody, goat anti-chicken (1:400, Alexa Fluor 488, Thermo Fisher Scientific, #A11039), overnight at room temperature. This secondary antibody incubation was accompanied with the fluorescent conjugated streptavidin (1:600, Alexa Fluor 546, Thermo Fisher Scientific, #S11225) and Neurotrace 640/660 deep-red fluorescent Nissl stain (1:200, Thermo Fisher Scientific, #N21483) in order to stain cells filled with biocytin and to identify the cytoarchitecture. Slices were rinsed in PBS-TX (3 × 10 min) and dehydrated by increasing ethanol concentrations (30, 50, 70, 90, 100, and 100%, 10 min each). They were treated to a 1:1 mixture of 100% ethanol and methyl salicylate for 10 min before clearing and storage in methyl salicylate (VWR Chemicals).

To image the recorded neurons with a laser scanning confocal microscope (Zeiss LSM 880 AxioImager Z2), the slices were mounted in custom-made metal well slides with methyl salicylate and coverslipped. Overview images of the tissue were taken at low magnification (Plan Apochromat ×10, NA 0.45) to confirm the location of the recorded neurons and at higher magnification (Plan Apochromat ×20, NA 0.8) to determine the morphology of the recorded neurons. Both overview images and high-magnification images were obtained as z-stacks that included the whole extent of each recorded cell to recover the full cell morphology. The morphology of LVb neurons of MEC/LEC was classified based on previous studies (Canto and Witter, 2012a; Canto and Witter, 2012b; Hamam et al., 2000; Hamam et al., 2002).

Histology, immunohistochemistry, and imaging of neuroanatomical tracing samples

Request a detailed protocol

After 2–3 weeks of survival, virus- or tracer-injected mice were anesthetized with isoflurane before being euthanized with a lethal intraperitoneal injection of pentobarbital (100 mg/kg, Apotekerforeningen). They were subsequently transcardially perfused using a peristaltic pump (World Precision Instruments), first with Ringer's solution (0.85% NaCl, 0.025% KCl, 0.02% NaHCO3) and subsequently with freshly prepared 4% PFA in 0.1 M PB (pH 7.4). The brains were removed from the skull, postfixed in PFA overnight, and put in a cryo-protective solution containing 20% glycerol, 2% DMSO diluted in 0.125 m PB. A freezing microtome was used to cut the brains into 40-μm-thick sections, which were collected in six equally spaced series for processing.

To enhance the GFP/citrine signal in AAV-TRE-tight-GFP/oChIEF-citrine-infected entorhinal LVb neurons and in GAD67-positive neurons, sections were stained with primary (1:400, chicken anti-GFP, Abcam; 1:2000, rabbit anti-GFP, Thermo Fisher Scientific #A11122) and secondary antibodies (1:400, Alexa Fluor-488 goat anti-chicken IgG, Thermo Fisher Scientific; 1:400, Alexa Fluor-546 goat anti-rabbit IgG, Thermo Fisher Scientific #A11010). This amplification of GFP/citrine signal enabled us to examine the axonal distribution of the infected LVb neurons. Due to the strong neurites labeling, however, it became difficult to visualize cell bodies of the targeted neurons. For such reasons, we occasionally used the native fluorophore (GFP/citrine) to visualize the cell bodies, while using immunolabeling (anti-GFP/citrine [anti-GFP]) to identify the neurites.

To identify the LVb border and characterize the TG mouse line, LVb neurons were visualized with primary (1:300, rabbit anti-PCP4, Sigma Aldrich #HPA005792; 1:3000, rat anti-Ctip, Abcam #ab18465) and secondary antibodies (1:400, Alexa Fluor 633 goat anti-rat IgG, Thermo Fisher Scientific # A21094; 1:400, Alexa Fluor-546 goat anti-rabbit IgG; 1:400, Alexa Fluor 635 goat anti-rabbit IgG, Thermo Fisher Scientific # A31576). For delineation purpose, sections were stained with primary (1:1000, guinea pig anti-NeuN, Millipore #ABN90P; 1:1000, mouse anti-NeuN, Millipore #MAB377) and secondary antibodies (1:400, Alexa Fluor 647 goat anti-guinea pig IgG, Thermo Fisher Scientific #A21450; 1:400, Alexa Fluor 488 goat anti-guinea pig IgG, Thermo Fisher Scientific #A11073; 1:400, Alexa Fluor 488 goat anti-mouse IgG, Thermo Fisher Scientific #A11001). PHA-L was visualized with primary (1:1000, rabbit anti-PHA-L, Vector Laboratories AS-2300) and secondary antibodies (1:400, Alexa Fluor 635 goat anti-rabbit IgG), while BDA was visualized with Cy3-streptavidin (1:400, Jackson ImmunoResearch 016-160-084).

For all immunohistochemical staining except for Ctip2 staining, we used the same procedure. Sections were rinsed 3 × 10 min in PBS-Tx followed by 60 min incubation in a blocking solution containing PBS-Tx with either 5% NGS or 3% bovine serum albumin (BSA). Sections were incubated with the primary antibodies diluted in the blocking solution for 48 hr at 4°C, rinsed 3 × 10 min in PBS-Tx, and incubated with secondary antibodies diluted in PBS-Tx overnight at room temperature. Finally, sections were rinsed 3 × 10 min in PBS. For Ctip2 staining, sections were heated to 80°C for 15 min in 10 mM sodium citrate (SC, pH 8.5). After cooling to room temperature, the sections were permeabilized by washing them three times in SC buffer containing 0.3% Triton X-100 (SC-Tx) and subsequently pre-incubated in a blocking solution containing SC-Tx and 3% BSA for 1 hr at room temperature. Next, sections were incubated with the primary antibodies diluted in the blocking solution for 48 hr at 4°C, rinsed 2 × 15 min in SC-Tx and 1% BSA, and incubated with secondary antibodies diluted in SC-Tx and 1% BSA for overnight at room temperature. Finally, sections were rinsed 3 × 10 min in PBS. After staining, sections were mounted on SuperfrostPlus microscope slides (Thermo Fisher Scientific) in Tris-gelatin solution (0.2% gelatin in Tris-buffer, pH 7.6), dried, and coverslipped with entellan in a toluene solution (Merck Chemicals, Darmstadt, Germany).

Coverslipped samples were imaged using an Axio ScanZ.1 fluorescent scanner, equipped with a ×10 objective, Colibri.2 LED light source, and a quadruple emission filter (Plan Apochromat ×10, NA 0.45, excitation 488/546, emission 405/488/546/633, Carl Zeiss). To quantify the colocalization of GFP, PCP4, and Ctip2 immunolabeling, confocal images were acquired in sections taken at every 240 μm throughout the EC using a confocal microscope (Zeiss LSM 880 AxioImager Z2) with a ×40 oil objective (Plan Apochromat ×40 oil, NA 1.3, Carl Zeiss). The number of immunohistochemically labeled neurons was quantified in a fixed Z-level of the confocal images using Image J software (http://rsb.info.nih.gov/ij).

Definition of EC and delineations of its layers

Request a detailed protocol

The EC of rodents is typically defined as a ventral area of the cerebral cortex, delineated laterally by the rhinal sulcus and medially by the hippocampal formation. It shows a well-established typical laminar structure comprising two superficial and two deep cell layers, separated by a cell-free layer, or lamina dissecans (Witter, 2011). Here we divide EC into LEC and MEC, mainly defined based on cytoarchitectural differences in LV, LIII, and LII. MEC has more columnar arrangement of LVb neurons compared to LEC. Layer III in MEC shows a sharp border with the lamina dissecans and has a very homogeneous appearance, whereas in LEC the border is less clear and the cellular organization is much more irregular. Finally, LEC can be identified by the presence of a superficial LII (LIIa) with densely packed neurons that tend to cluster into islands.

The borders of the superficial layers (I–III) and the thin acellular layer IV (lamina dissecans) were delineated as previously described (Witter, 2011). Layers Va and Vb were differentiated based on cell size, cell density, cell marker, and the projection patterns. LVb neurons are densely packed small cells that are PCP4-positive, whereas LVa is made up of sparsely distributed large cells, which project to various cortical and subcortical regions (Figure 1—figure supplement 1; Kitamura et al., 2017; Ohara et al., 2018; Sürmeli et al., 2015). The border between layer Vb and VI is more difficult to identify since PCP4 staining also labels LVI neurons. We determined this border based on the cell density that decreases in LVI and the overall prominent orientation of neurons parallel to the pial surface (Witter, 2011). In case of MEC, the border can also be identified since the typical columnar organization of LVb stops upon entering LVI.

Statistics

Statistical analyses were performed using GraphPad Prism (GraphPad Software) or MATLAB (MathWorks). The details of tests used are described with the results. Differences between the groups were tested using paired and unpaired t-tests. Group comparisons were made using one-way ANOVA followed by Bonferroni post-hoc tests to control for multiple comparisons. All statistical tests were two-tailed, and thresholds for significance were placed at *p<0.05, **p<0.01, and ***p<0.001. All data are shown as mean ± standard errors. No statistical methods were used to pre-determine sample size, but the number of mice and cells for each experiment is similar with previous studies in the field (Doan et al., 2019; Nilssen et al., 2018). Mice were randomly selected from both sexes, and all experiments were successfully replicated in several samples. No blinding was used during data acquisition, but electrophysiological data analyses were performed blind to groups.

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files. We provide source data for Figure 1, 2, 3, 4, 5, Figure 1-figure supplement 2, and Figure 4-figure supplement 2.

References

  1. Book
    1. Edelman GM
    (1989)
    The Remembered Present: A Biological Theory of Consciousness
    Basic Books.
  2. Thesis
    1. Nilssen ES
    (2019)
    Synaptic connectivity of principal cells in layer II of the lateral entorhinal cortex
    PhD Thesis.
  3. Book
    1. Witter MP
    (2011)
    The hippocampus
    In: Paxinos G, Puilles L, Watson C, editors. The Mouse Nervous System (First Edition). Academic. pp. 112–139.

Decision letter

  1. Katalin Toth
    Reviewing Editor; University of Ottawa, Canada
  2. Laura L Colgin
    Senior Editor; University of Texas at Austin, United States
  3. Matthew F Nolan
    Reviewer; University of Edinburgh, United Kingdom
  4. Desdemona Fricker
    Reviewer; CNRS UMR8119, France

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The study addresses a fundamentally important question regarding the connectivity of LVb and LVa in the medial and lateral entorhinal cortex. The authors suggest that the strength of the LVb to LVa connection is stronger in the LEC than in the MEC. This finding will have important implications on studies investigating circuit functions between the hippocampus and the EC. The differential operational principles in the lateral and medial entorhinal cortex is a novel element to the circuit underlying the hippocampal-cortical interplay.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Local circuits of layer Vb-to-Va mediate hippocampal-cortical outputs in lateral but not in medial entorhinal cortex" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Matthew F Nolan (Reviewer #2); Desdemona Fricker (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The study addresses a fundamentally important question regarding the connectivity of LVb and LVa in the medial and lateral entorhinal cortex. The authors suggest that LVb to LVa connection exists in the LEC but not in the MEC. This finding would have important implications on studies investigating circuits functions between the hippocampus and the EC. All three reviewers found the central question important and the data novel. However, there are several technical issues that limit the robustness of the authors claim.

1. While the transgenic animal used in these experiments in elegant and novel, it only labels a subpopulation of the neurons. There is a possibility of selective labelling of neurons with distinct connectivity patterns. The authors would need to show that their approach is not leading to false negative results due to the selective visualization of those neurons that project more modestly to the LVa.

2. The specificity of the injection to the LEC/MEC should be better documented and potential spread to the perirhinal or postrhinal cortex carefully excluded.

3. The findings are presented as LVb to LVa connection did not exist at all in the MEC, however the data shows that the connection is there but it is significantly less dense than in the LEC. Given the graded finding, if the authors aim to show their central claim regarding the lack of mediation of hippocampo-cortical outputs by this connection in the MEC, this would require the addition of functional studies.

Reviewer #1:

The current study by Ohara et al. describes differences in the connectivity patterns between LVb to LVa. The study builds on the authors previous study (Ohara et al. 2018) where they showed the intrinsic connectivity of LVb neurons in the MEC and LEC. The focus of the current study is the difference the authors observed in the strengths of connectivity between LVb and LVa in the MEC and LEC. The authors suggest that the in MEC Vb neurons do not provide substantial direct unput to LVa neurons. The manuscript emphasizes the functional importance of difference as the authors suggest that "……hippocampal -cortex output circuit is present only in LEC, suggesting that episodic systems consolidation predominantly uses LEC-derived information and not allocentric spatial information from MEC."

The study uses a newly developed mouse line to investigate connectivity differences, this is a nice technical approach and the experimental data is of high quality. While the data is solid, the authors tend to overinterpret their findings from the functional point of view. While the observed difference is quite interesting, it is unclear what the impact is on information flow in the MEC and LEC and to which degree they differ functionally. The authors assume major differences and their work is framed based on these expected differences, but the manuscript does not provide data that would demonstrate functionally distinct features.

1. Throughout the text the authors treat their findings as it was 'all-or-none' i.e the LEC has a direct connection between LVb and LVa while the MEC does not. This does not seem to be the case based on their data, the data shows that connectivity in the MEC is less robust but it is definitely there. The difference seems to be quantitative and not qualitative.

2. Due to this problem, the authors seem to be overinterpreting their data by suggesting that the information flow must be significantly different conceptually in the LEC and the MEC and this would have important implications for memory consolidation. It is impossible to draw these conclusions based on the data presented, as there are no experiments investigating the functional, network level consequences of these connectivity differences.

3. The electrophysiology experiments provide information about the basic parameters of the investigated cells, but these lack a physiological context that would allow the authors to evaluate the consequences of these differences on information flow and/or processing in the MEC and the LEC.

4. The study is using a novel transgenic mouse line to differentiate between LVb and LVa neurons, while this is definitely a strength of the study, this strategy allows the authors to visualize ~50% of LEC and ~30% MEC neurons. Since the authors aim to prove a negative (MEC does not have direct connection) the fact that ~70% of the neurons are not labelled could be problematic.

Reviewer #2:

The study investigates key components of the entorhinal circuits through which signals from hippocampus are relayed to the neocortex. The question addressed is important but the stated claim that layer 5b (L5b) to layer 5a (L5a) connections mediate hippocampal-cortical outputs in LEC but not MEC appears to be an over-interpretation of the data. First, the experiments do not test hippocampal to L5a connections, but instead look at L5b to L5a connections. Second, the data provide evidence that there are L5b to L5a projections in LEC and MEC, which contradictions the claim made in the title. These projections do appear denser in LEC under the experimental conditions used, but possible technical explanations for the difference are not carefully addressed. If these technical concerns were addressed, and the conclusions modified appropriately, then I think this study could be very important for the field and would complement well recent work from several labs that collectively suggests that information processing in deep layers of MEC is more complex than has been appreciated (e.g. Sürmeli et al. 2015, Ohara et al. 2018, Wozny et al. 2018, Rozov et al. 2020).

1. An impressive component of the study is the introduction of a new mouse line that labels neurons in layer 5b of MEC and LEC. However, in each area the line appears to label only a subset (30-50%) of the principle cell population. It's unclear whether the unlabelled neurons have similar connectivity to the labelled neurons. If the unlabelled neurons are a distinct sub-population then it's difficult to see how the experiments presented could support the conclusion that L5b does not project to L5a; perhaps there is a projection mediated by the unlabelled neurons? I don't think the authors need include experiments to investigate the unlabelled population, but given that the labelling is incomplete they should be more cautious about generalising from data obtained with the line.

2. For experiments using the AAV conditionally expressing oChIEF-citrine, the extent to which the injections are specific to LEC/MEC is unclear. This is a particular concern for injections into LEC where the possibility that perirhinal or postrhinal cortex are also labelled needs to be carefully considered. For example, in Figure 3D it appears the virus has spread to perirhinal cortex. If this is the case then axonal projections/responses could originate there rather than from L5b of LEC. I suggest to exclude any experiments were there is any suggestion of expression outside LEC/MEC or where this cannot be ruled out through verification of the labelling. Alternatively, one might include control experiments in which the AAV is targeted to perirhinal and postrhinal cortex. Similar concerns should be addressed for injections that target the MEC to rule out spread to the pre/parasubiculum.

3. It appears likely from the biocytin fills shown that the apical dendrites of some of the recorded L5a neurons have been cut (e.g. Figure 4A, Figure 4-Supplement 1D, neuron v). Where the apical dendrite is clearly intact and undamaged synaptic responses to activation of L5b neurons are quite clear (e.g. Figure 4-Supplement 1D, neuron x). Given that axons of L5b cells branch extensively in L3, it is possible that any synapses they make with L5a neurons would be on their apical dendrites within L3. It therefore seems important to restrict the analysis only to L5a neurons with intact apical dendrites; a reasonable criteria would be that the dendrite extends through L3 at a reasonable distance (> 30 μm?) below the surface of the slice.

4. Throughout the manuscript the data is over-interpreted. Here are some examples:

– The title over-extrapolates from the results and should be changed. A more accurate title would be along the lines of "Evidence that L5b to L5a connections are more effective in lateral compared to medial entorhinal cortex".

– "the conclusion that the dorsal parts of MEC lacks the canonical hippocampal-cortical output system" seems over-stated given the evidence (see comments above).

– Discussion, para 1, "Our key finding is that LEC and MEC are strikingly different with respect to the hippocampal-cortical pathway mediated by LV neurons, in that we obtained electrophysiological evidence for the presence of this postulated crucial circuit in LEC, but not in MEC". This is misleading as there is also evidence for L5b to L5a connections in MEC, although this projection may be relatively weak. Recent work by Rozov et al. demonstrating a projection from intermediate hippocampus to L5a provides good evidence for an alternative model in which MEC does relay hippocampal outputs. This needs to be considered.

5. What proportion of responses are mono-synaptic? How was this tested?

Reviewer #3:

The present manuscript focuses on a subpopulations of layer 5 neurons in medial and lateral entorhinal cortex and its functional connections to target neurons in layers 2, 3 and 5. The authors show a difference in LVb-to-LVa connectivity between MEC and LEC. The results suggest that the entorhinal output circuit via LVb-to-LVa is present primarily in LEC.

The work relies on and is made possible by a newly described transgenic mouse (TG) where LVb neurons can be labeled and stimulated with light. The authors showed that these neurons are largely co-labeled with PCP4, a marker for LVb. They compared the apical dendritic extent from TG labeled cells (LVb) and Nac retrogradely labeled cells (LVa) in medial and lateral EC. The intrinsic electrophysiological properties of LVa and LVb neurons were measured and used for PCA showing segregation according to sublayer and region. The axonal distribution and translaminar local connections of LVb neurons form the TG mice were then examined. Cells were recorded in vitroin vitro and filled with biocytin, both from MEC and LEC, with multiple cells in a same slice, documented with exquisite high quality images. The study of the LVb translaminar connectivity via a direct comparison of postsynaptic responses in neurons in different layers in a same slice is gold standard for this type of functional connectivity analysis. There is also an investigation of mixed excitatory-inhibitory postsynaptic response sequences, and evidence for a dorso-ventral gradient in LVb-to-LVa connectivity in MEC is given.

The study combines TG mice, immunolabeling, retrograde labeling, morphological analysis and in vitro electrophysiology with optogenetic photo-stimulation. While it builds on already published work by the same group and others, by comparing the local target neurons of LVb in MEC and in LEC, the manuscript provides a unique contribution to the literature on the laminar circuit organization in the Entorhinal Cortex. In view of the central position of this area in the hippocampal memory systems of the rodent brain, these results are of interest to a broader neuroscience audience. It is also a nice example of a bottom-up approach, where data on the entorhinal translaminar connectivity may influence and constrain theories of hippocampal-cortical processing.

1. Almost all TG labelled neurons are positive for PCP4 but not so vice versa, only 45.9 and 30.P% of PCP4 + neurons in LEC and MEC are labeled in the TG mouse (page 5) leaving open the possibility that the TG mous labels a (specific?) subset of LVb neurons. Did you test whether TG labeled LVb cells co-localize with Ctip2 ?

2. The direct comparison of translaminar connectivity of LVb neurons is very convincing. But if your main conclusion (title) concerns the difference of LVb-to-LVa connectivity between MEC and LEC, it would have been more appropriate to test that in a same slice. I think this point should not stand in the way of publication, but while the data strongly support conclusions on the laminar differences of LVb connectivity, the evidence for differences in LVb-to-LVa connectivity between MEC and LEC is a bit weaker and more indirect.

3. Postsynaptic responses (in mV) in LEC are about twice as high in amplitude as in MEC (Figure 4E vs Figure 5E), across all layers. Please discuss possible reasons, and possible impact on the circuit function. Is the probability to initiate action potentials higher in LEC ?

4. Give the onset latencies of postsynaptic excitatory potentials induced by LVb photostimulation. Are latencies monosynaptic? Or also polysynaptic? Ideally this could be tested by applying a cocktail of TTX-4-AP.

5. Figure 4 S3, Figure 5 S2. Analysis of inhibition. What is the cut-off criterium to say inhibition is present or not? It might be more appropriate to give the I/E ratio.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Local projections of layer Vb-to-Va are more effective in lateral than in medial entorhinal cortex" for further consideration by eLife. Your revised article has been reviewed by three reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Laura Colgin as the Senior Editor.

The study addresses a fundamentally important question regarding the connectivity of LVb and LVa in the medial and lateral entorhinal cortex. The authors suggest that the strength of the LVb to LVa connection is stronger in the LEC than in the MEC. This finding would have important implications on studies investigating circuits functions between the hippocampus and the EC. All three reviewers found the central question important and the data novel. However, the text overstates the functional implications of the findings and an important alternative explanation regarding potential differences in cut dendrites haven't been fully excluded.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

There are two major issues that need to be addressed in the revised in the manuscript.

1. The current version still overstates the differences in the effectiveness of the connections; the functional data is not robust enough to draw these conclusions and the claims should be restated.

2. Previously raised issue regarding the potential impact of cut dendrites is still not resolved. The authors would need to provide more in depth analysis to address/exclude this possibility fully.

Other, smaller issues listed in the reviewers' reports that also would need to take into consideration during the revision.

Reviewer #1:

The authors have improved the manuscript based on previous comments. However, I still think that the functional aspect and implications are overstated in the revised manuscript. The physiological data demonstrate the connectivity's strengths (and hence supports the anatomy data), but this alone won't be sufficient to comment on the functional effectiveness of the connections, which would also depend on other factors that are not investigated here.

Reviewer #2:

This study presents data about the connectivity from subpopulations of neurons in layer 5b of the medial and lateral entorhinal cortices to neurons in more superficial layers. This data is likely to make an important contribution to understanding the circuit basis for memory and spatial cognition.

The major weakness of the study is that the claim that L5b -> L5a connections are more effective in the LEC than MEC continues to be over-stated given the evidence. There is a risk here that this will have an impact, but beyond what the data actually establish and possibly in a way that will hinder rather than advance the field. In my opinion, the more specific claim, that differences in L5b -> L5a connectivity suggest different processing mechanisms with as yet unknown functional impact would be sufficiently important to justify publication in eLife.

A remaining technical weakness is that the possibility of cut dendrites explaining differences in synaptic responses is not addressed sufficiently convincingly.

It should be possible to address these issues without additional experimentation.

The strengths of the manuscript are the combined anatomical and electrophysiological approach that leverages a novel mouse line, and the likely importance of the observations to investigation of circuit mechanism for spatial cognition and memory.

While the manuscript is greatly improved there are remaining weaknesses.

1. While the over-interpretation of the data is reduced compared to the previous version of the manuscript, this still substantially compromises the study. The manuscript continues to make claims about functional effectiveness of connections that are not sufficiently supported by the data. A more specific claim that glutamateric connections from L5b to L5a appear denser in MEC than LEC, and that this suggests different operating principles for computation in these two regions, would be well justified by the data, would be of wide interest, would do much to stimulate future work and would not lead to erroneous interpretations down the line. Specific statements that over-interpret the data should be amended.

2. The Discussion could do a much better job of considering caveats in the interpretation of the data. In particular, relatively greater excitatory connectivity in a brain slice does not necessarily imply more effective functional connectivity. Alternative scenarios to consider could include the following. 1. It's unclear from the present data whether the relative difference in excitatory input from L5b to L5a is matched by a difference in feedforward or feedback inhibition. If it is then the pathways may turn out to be similarly effective. Testing this convincingly will require in vivo experiments. 2. Either pathway could be subject to neuromodulation in vivo. It's conceivable this could substantially modify or even reverse their apparent relative effectiveness. 3. The membrane potential and spike threshold of L5a neurons in vivo is unknown. If the membrane potential of L5a neurons in MEC is more depolarized in vivo than in LEC, then the L5b to L5a pathway could turn out to be more effective. The discussion should also consider the fact that the mouse line used labels only a sub-set of the neurons in L5b of LEC and MEC. The possibility that connectivity from other L5b neurons may differ should be clearly noted.

3. The possibility that differences in cut dendrites explain the differences in synaptic responses between LEC and MEC, and within the MEC, is still not ruled out. The revised manuscript now gives an indication of the number of dendritic branches in MEC that are > 150 μm. This is insufficient as layer 3 can be on the order of 300 μm or more in width between its superficial and deep borders, while the axons from L5b extend well into L2. To make the claims convincing it's important that adequate quantification of remaining apical dendritic length (and if possible diameter/surface areas) in LEC and at each dorsoventral level of MEC is included (for both cut and not-cut neurons).

4. I couldn't find anywhere whether any connections were tested with glutamate receptor antagonists to confirm they are synaptic and glutamatergic. It's unlikely but not impossible that some short latency responses reflect low level expression of ChR2 in non-5b neurons (in my experience this might not be detectable from fluorescence). It would be good to know this can be ruled out.

Reviewer #3:

I have read the authors' responses and I feel that they are adequate. I have no further comments.

https://doi.org/10.7554/eLife.67262.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

[…] All three reviewers found the central question important and the data novel. However, there are several technical issues that limit the robustness of the authors claim.

1. While the transgenic animal used in these experiments in elegant and novel, it only labels a subpopulation of the neurons. There is a possibility of selective labelling of neurons with distinct connectivity patterns. The authors would need to show that their approach is not leading to false negative results due to the selective visualization of those neurons that project more modestly to the LVa.

This is certainly a correct concern, since it is well known that layer Vb in rodents comprises multiple neuron types, both in LEC and MEC. In the initial version, we tried to address this concern in the discussion, by mentioning that traditional anterograde tracing methods result in similar patterns. We indeed did not provide additional experimental data to substantiate that this is not a concern. We have now added experimental data on the sparsity of projections from the general population of neurons in LVb of MEC to LVa, in both rats and mice (Figure 3—figure supplement 1. D, E). We further maintained referencing existing literature in the rat (Köhler, 1986) and in the present version added a reference to a study in the monkey (Chrobak and Amaral, 2007), both substantiating the apparent differences between LEC and MEC regarding Vb to Va projections in wildtype animals using anterograde tracing.

2. The specificity of the injection to the LEC/MEC should be better documented and potential spread to the perirhinal or postrhinal cortex carefully excluded.

This is a fair criticism, which we originally assumed was covered sufficiently in the discussion since it is known that deep layers of PER and POR in the rat do not substantially contribute to projections to LEC and MEC (Burwell and Amaral, 1998). However, since we do have experimental data in both wildtype and the TG mice, we have added this as a supplementary figure to the manuscript (Figure 3—figure supplement 1A—C). In none of these experiments did we notice substantial projections to neither LEC or MEC, so the very sparse spill of the AAV infection into PER/POR in our experimental cases with injections in LVb of MEC or LEC is unlikely to have affected our results.

3. The findings are presented as LVb to LVa connection did not exist at all in the MEC, however the data shows that the connection is there but it is significantly less dense than in the LEC. Given the graded finding, if the authors aim to show their central claim regarding the lack of mediation of hippocampo-cortical outputs by this connection in the MEC, this would require the addition of functional studies.

We do agree that we have emphasized and simplified our findings too much, using inappropriate wording. We will respond to this in more detail below, but want to add here that our data now only show that the projection is less dense (and indeed significantly less as our new, added analyses show), but also that in MEC the projection shows less synaptic strength.

Reviewer #1:

[…] The study uses a newly developed mouse line to investigate connectivity differences, this is a nice technical approach and the experimental data is of high quality. While the data is solid, the authors tend to overinterpret their findings from the functional point of view. While the observed difference is quite interesting, it is unclear what the impact is on information flow in the MEC and LEC and to which degree they differ functionally. The authors assume major differences and their work is framed based on these expected differences, but the manuscript does not provide data that would demonstrate functionally distinct features.

We appreciate that this reviewer, as do the other reviewers, indicate that we have used incorrect wordings, leaving the reader with the impression that there is no projection from LVb to LVa in dorsal MEC at all, and that we provide data supporting functional consequences. We have carefully rephrased this throughout the manuscript.

1. Throughout the text the authors treat their findings as it was 'all-or-none' i.e the LEC has a direct connection between LVb and LVa while the MEC does not. This does not seem to be the case based on their data, the data shows that connectivity in the MEC is less robust but it is definitely there. The difference seems to be quantitative and not qualitative.

We have carefully corrected our wording and now also emphasize that the lesser connectivity is much more prominent in dorsal MEC and less so in ventral MEC. Although the data were already in the original manuscript (Figure 4—figure supplement 2.), we ignored this in how we phrased our final conclusions (see also our responses below).

2. Due to this problem, the authors seem to be overinterpreting their data by suggesting that the information flow must be significantly different conceptually in the LEC and the MEC and this would have important implications for memory consolidation. It is impossible to draw these conclusions based on the data presented, as there are no experiments investigating the functional, network level consequences of these connectivity differences.

We appreciate this comment and as correctly mentioned by the reviewer, we indeed suggest, but not conclude, that functional differences of LEC and MEC regarding memory consolidation may exist. We do however realize that we seem to conclude that such is the case which is incorrect indeed. We carefully corrected our wordings throughout so that it would not lead to any misunderstanding.

3. The electrophysiology experiments provide information about the basic parameters of the investigated cells, but these lack a physiological context that would allow the authors to evaluate the consequences of these differences on information flow and/or processing in the MEC and the LEC.

We most certainly agree with the comment that our data do not allow us to evaluate the consequences of the reported differences on information flow and we refrain from aiming to do so. We merely suggest that our data point to potentially interesting differences that need to be explored. It would take quite some time to obtain the results requested by the reviewer, which is beyond what we feel feasible at this point in time. In the original manuscript, we already included some functional comments related to main differences in physiological properties of neuron types in layers II and III. In the revised manuscript, we added some comments concerning potential functional consequences of the reported differences in neurons in LVa and LVb and between LVb neurons in LEC versus MEC. However, since firing patterns of neurons result from the complex interactions of morphological structure, the types and somatodendritic distribution of voltage-gated channels and many other factors, we refrain from detailed functional speculations since these are not related to the core message we aim to convey with this paper.

4. The study is using a novel transgenic mouse line to differentiate between LVb and LVa neurons, while this is definitely a strength of the study, this strategy allows the authors to visualize ~50% of LEC and ~30% MEC neurons. Since the authors aim to prove a negative (MEC does not have direct connection) the fact that ~70% of the neurons are not labelled could be problematic.

As indicated above (response to general comment 1), we added a supplementary figure which shows the projection of layer Vb neurons in MEC of wild-type mice, which was also confirmed in wild-type rats (Figure 3—figure supplement 1). These additional data show that MEC LVb neurons project more heavily to LIII than to LVa, thus indicating that our results are not due to the labeling of only a subpopulation in MEC-LVb. The sparse MEC LVb-to-LVa connectivity has also been reported in a

recent paper (Rozov et al., 2020), which we now reference in our discussion. Note that also the PCP4 data, representing a much larger population of LVb neurons, show a similar lack of axons in LVa in MEC (Figure 1—figure supplement 1). We have emphasized this is the revised version.

Reviewer #2:

The study investigates key components of the entorhinal circuits through which signals from hippocampus are relayed to the neocortex. The question addressed is important but the stated claim that layer 5b (L5b) to layer 5a (L5a) connections mediate hippocampal-cortical outputs in LEC but not MEC appears to be an over-interpretation of the data. First, the experiments do not test hippocampal to L5a connections, but instead look at L5b to L5a connections. Second, the data provide evidence that there are L5b to L5a projections in LEC and MEC, which contradictions the claim made in the title. These projections do appear denser in LEC under the experimental conditions used, but possible technical explanations for the difference are not carefully addressed. If these technical concerns were addressed, and the conclusions modified appropriately, then I think this study could be very important for the field and would complement well recent work from several labs that collectively suggests that information processing in deep layers of MEC is more complex than has been appreciated (e.g. Sürmeli et al. 2015, Ohara et al. 2018, Wozny et al. 2018, Rozov et al. 2020).

1. An impressive component of the study is the introduction of a new mouse line that labels neurons in layer 5b of MEC and LEC. However, in each area the line appears to label only a subset (30-50%) of the principle cell population. It's unclear whether the unlabelled neurons have similar connectivity to the labelled neurons. If the unlabelled neurons are a distinct sub-population then it's difficult to see how the experiments presented could support the conclusion that L5b does not project to L5a; perhaps there is a projection mediated by the unlabelled neurons? I don't think the authors need include experiments to investigate the unlabelled population, but given that the labelling is incomplete they should be more cautious about generalising from data obtained with the line.

See our response to comment 4 of reviewer 1.

2. For experiments using the AAV conditionally expressing oChIEF-citrine, the extent to which the injections are specific to LEC/MEC is unclear. This is a particular concern for injections into LEC where the possibility that perirhinal or postrhinal cortex are also labelled needs to be carefully considered. For example, in Figure 3D it appears the virus has spread to perirhinal cortex. If this is the case then axonal projections/responses could originate there rather than from L5b of LEC. I suggest to exclude any experiments were there is any suggestion of expression outside LEC/MEC or where this cannot be ruled out through verification of the labelling. Alternatively, one might include control experiments in which the AAV is targeted to perirhinal and postrhinal cortex. Similar concerns should be addressed for injections that target the MEC to rule out spread to the pre/parasubiculum.

See our response above to the general comment 2. We have added supplementary data with AAV injection in deep PER (Figure 3—figure supplement 1A, B) in TG-animals as well as anterograde tracer injections in wildtype mice. We hardly observed labeled fibers in LEC originating from deep PER, which indicate that our data collected in LEC were not affected by the sparse neurons in deep PER labeled by our AAV-injections. We included similar experimental data for POR (Figure 3—figure supplement 1C). Regarding spread of AAV from MEC into pre/parasubiculum, we feel that this is not an issue, since we report a diminished projection to LVa, not an added component. Moreover, it is very well established that pre/parasubiculum preferentially if not exclusively project to superficial layers. However, the reviewer is correct that these inputs might target neurons in LVa, as we have shown in the rat (Canto et al., 2012). So, this might have resulted in an overestimation of the inputs to LVa neurons, but even so, the input synaptic strength and proportion of responsive neurons never reaches the level as seen in LVa of LEC. Note that we did stimulate the slices throughout all layers, and we have emphasized this in the method section of the paper (see also our reply to comment 3).

3. It appears likely from the biocytin fills shown that the apical dendrites of some of the recorded L5a neurons have been cut (e.g. Figure 4A, Figure 4-Supplement 1D, neuron v). Where the apical dendrite is clearly intact and undamaged synaptic responses to activation of L5b neurons are quite clear (e.g. Figure 4-Supplement 1D, neuron x). Given that axons of L5b cells branch extensively in L3, it is possible that any synapses they make with L5a neurons would be on their apical dendrites within L3. It therefore seems important to restrict the analysis only to L5a neurons with intact apical dendrites; a reasonable criteria would be that the dendrite extends through L3 at a reasonable distance (> 30 μm?) below the surface of the slice.

As correctly pointed out by the reviewer, all L5a neurons which respond to the light stimulation indeed have intact apical dendrites in L3. However, we also did observe many neurons with apical dendrites (> 150 μm in L3) which do not respond in dorsal- (N=3) and middle-level slices (N=6). These data indicate that the differences of LVb-to-LVa connectivity along the dorso-ventral axis were not due to the severing of apical dendrites. We added the information of severed apical dendrites of L5b neurons in “Figure 4—figure supplement 2”. Note that in all experiments we systematically optically stimulate in all layers of the slice allowing us to assess the possible impact of axons in layers III and II.

4. Throughout the manuscript the data is over-interpreted. Here are some examples:

– The title over-extrapolates from the results and should be changed. A more accurate title would be along the lines of "Evidence that L5b to L5a connections are more effective in lateral compared to medial entorhinal cortex".

We thank the reviewer for this suggestion and have changed our title accordingly.

– "the conclusion that the dorsal parts of MEC lacks the canonical hippocampal-cortical output system" seems over-stated given the evidence (see comments above).

We have rephrased the sentence and carefully checked the manuscript to correct any such overstatements.

– Discussion, para 1, "Our key finding is that LEC and MEC are strikingly different with respect to the hippocampal-cortical pathway mediated by LV neurons, in that we obtained electrophysiological evidence for the presence of this postulated crucial circuit in LEC, but not in MEC". This is misleading as there is also evidence for L5b to L5a connections in MEC, although this projection may be relatively weak. Recent work by Rozov et al. demonstrating a projection from intermediate hippocampus to L5a provides good evidence for an alternative model in which MEC does relay hippocampal outputs. This needs to be considered.

Thank you for suggesting this paper, but please note that this paper was published after we submitted the paper to eLife. We now included this paper in our discussion, and carefully stated that our suggestion concerning a weaker output hippocampal-cortical circuit is the case in the dorsal hippocampal-medial entorhinal circuit in the text and also in the conclusion figure (Figure 6).

5. What proportion of responses are mono-synaptic? How was this tested?

This is an interesting question, but we did not test this directly, using standard pharmacological approaches. We have added information on latency of the measured responses in Figures 4 and 5 (see also the request of Reviewer 3). In a series of experiments carried out in parallel in mouse LEC LII, using similar viral and laser stimulation protocols, we measured latencies ranging from 4.5 – 6.8 msec, which consistently were shown to be monosynaptic when tested using pharmacological measures (applying TTX and 4-AP to the bath; E.S. Nilssen, 2019 Synaptic connectivity of principal cells in layer II of the lateral entorhinal cortex. PhD Thesis NTNU, ISSN 1503-8181). Based on the present latency data, we find it likely that most recorded responses are monosynaptic, though we cannot exclude that some of the connections, particularly the ones from MEC LVb to LVa neurons and to LII stellate cells include or represent disynaptic responses. We have inserted a short note on this in the manuscript. When we planned these experiments, we did not deem pharmacological checks on monosynaptic contacts all that relevant; even if there are disynaptic or polysynaptic responses, the fact that the overall number of responding cells and the synaptic strengths is lower in LVa in dorsal MEC compared to ventral MEC and LEC stands.

Reviewer #3:

[…] 1. Almost all TG labelled neurons are positive for PCP4 but not so vice versa, only 45.9 and 30.P% of PCP4 + neurons in LEC and MEC are labeled in the TG mouse (page 5) leaving open the possibility that the TG mous labels a (specific?) subset of LVb neurons. Did you test whether TG labeled LVb cells co-localize with Ctip2 ?

We haven’t checked whether TG labeled LVb neurons co-localize with Ctip2. However, as shown in “Figure 1—figure supplement 2”, we confirmed that Ctip2-positive neurons incorporate PCP4positive neurons, and in addition, labels also PCP4-negative neurons (yellow and white arrowheads). Therefore, it is likely that the TG labeled LVb neurons co-localize with Ctip2.

2. The direct comparison of translaminar connectivity of LVb neurons is very convincing. But if your main conclusion (title) concerns the difference of LVb-to-LVa connectivity between MEC and LEC, it would have been more appropriate to test that in a same slice. I think this point should not stand in the way of publication, but while the data strongly support conclusions on the laminar differences of LVb connectivity, the evidence for differences in LVb-to-LVa connectivity between MEC and LEC is a bit weaker and more indirect.

We are not entirely sure what the reviewer is getting at. If the point is that the difference between LEC and MEC is not tested in the same slice, that is a correct statement, but to do that is technically unrealistic. In line to our replies to the previous reviewers, the main point is that in dorsal MEC the projections from LVb to LVa are very different from what we observe in LEC and ventral MEC.

However, to get dorsal MEC and LEC in the same slice is impossible and although dorsal and ventral MEC might be captured in one sagittal section, one need to inject AAV at multiple dorsoventral levels to do such recordings. The risk of having too much spread of virus in such cases would be too high and we do not find it ethically acceptable to use many animals in the hope of getting one (at best) successful animal. The number of animals used and the number of neurons we recorded from provide a reliable statistical sample to conclude that there are differences in the local circuits.

3. Postsynaptic responses (in mV) in LEC are about twice as high in amplitude as in MEC (Figure 4E vs Figure 5E), across all layers. Please discuss possible reasons, and possible impact on the circuit function. Is the probability to initiate action potentials higher in LEC ?

We thank the reviewer for this suggestion, and we agree that we should have discussed this. We don’t have any data that indicate that the probability to initiate action potentials higher in LEC than in MEC and this is in line with data in two published studies in the rat where elementary biophysical properties were systematically compared between neurons in all layers of LEC and MEC in postnatal rats (Canto et al. 2012a and b) As discussed at some length in Canto and Witter (2012a) most studies carried out in adult rodents are in line with these observations. We deem it most likely that the differences in the generated postsynpatic amplitudes are due to the differences in the number of oChiEF-expressing neurons and thus the resulting differenced in labeled axonal densities. As shown in Figure 1E, the proportion of tTA-expressing neurons is higher in LEC than in MEC. We have inserted a comment along these lines in the revised manuscript.

4. Give the onset latencies of postsynaptic excitatory potentials induced by LVb photostimulation. Are latencies monosynaptic? Or also polysynaptic? Ideally this could be tested by applying a cocktail of TTX-4-AP.

Please see our reply to major comment 5 of reviewer 2 above.

5. Figure 4 S3, Figure 5 S2. Analysis of inhibition. What is the cut-off criterium to say inhibition is present or not? It might be more appropriate to give the I/E ratio.

We agree that our analysis on the inhibition may not be appropriate. In view of the very low number of observations we do not have sufficient data to provide proper I/E ratios, and since the presence or absence of inhibition does not affect our conclusion, we removed the data related to the inhibition.

[Editors’ note: what follows is the authors’ response to the second round of review.]

[…] All three reviewers found the central question important and the data novel. However, the text overstates the functional implications of the findings and an important alternative explanation regarding potential differences in cut dendrites haven't been fully excluded.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

There are two major issues that need to be addressed in the revised in the manuscript.

1. The current version still overstates the differences in the effectiveness of the connections; the functional data is not robust enough to draw these conclusions and the claims should be restated.

2. Previously raised issue regarding the potential impact of cut dendrites is still not resolved. The authors would need to provide more in depth analysis to address/exclude this possibility fully.

Other, smaller issues listed in the reviewers' reports that also would need to take into consideration during the revision.

We are very happy with the overall positive evaluation of the revised version. Both reviewer 1 and 2 indicate clearly that although the data as such are very relevant for the field and merit publication, we are overstating the functional consequences of our findings. Though we do appreciate this evaluation and have revised the paper accordingly, we want to emphasize that during the preparation of the first version as well as the revision we have consulted some experts/theoreticians in the field of memory consolidation who convinced us about the potential importance of our findings related to memory consolidation and our choice for wording has been strongly influenced by these detailed and stimulative discussions.

We have thoroughly addressed all issues as indicated in detail below.

Reviewer #1:

The authors have improved the manuscript based on previous comments. However, I still think that the functional aspect and implications are overstated in the revised manuscript. The physiological data demonstrate the connectivity's strengths (and hence supports the anatomy data), but this alone won't be sufficient to comment on the functional effectiveness of the connections, which would also depend on other factors that are not investigated here.

We thank the reviewer for the overall positive evaluation of the revised version. We fully agree with the reviewer that we do not provide experimental data allowing us to conclude on the functional effectiveness of the connections. However, we respectfully disagree that our data would not allow us to comment on this aspect or point to potential relevant functional consequences. We are aware of the fact that many other factors will impact these functional consequences in vivo and have elaborated on that a bit more in the discussion (pages 18/19, lines 460-472 of the revised manuscript). We have scrutinized all comments on functional effectiveness and reworded them to the best of our skills, particularly guided by the detailed suggestions of reviewer 2 (see below).

Reviewer #2:

This study presents data about the connectivity from subpopulations of neurons in layer 5b of the medial and lateral entorhinal cortices to neurons in more superficial layers. This data is likely to make an important contribution to understanding the circuit basis for memory and spatial cognition.

The major weakness of the study is that the claim that L5b -> L5a connections are more effective in the LEC than MEC continues to be over-stated given the evidence. There is a risk here that this will have an impact, but beyond what the data actually establish and possibly in a way that will hinder rather than advance the field. In my opinion, the more specific claim, that differences in L5b -> L5a connectivity suggest different processing mechanisms with as yet unknown functional impact would be sufficiently important to justify publication in eLife.

A remaining technical weakness is that the possibility of cut dendrites explaining differences in synaptic responses is not addressed sufficiently convincingly.

It should be possible to address these issues without additional experimentation.

We thank the reviewer for supporting the publication of our data and manuscript in eLife and we are grateful for the detailed level of comments and suggestions to reword our manuscript to improve how to accurately convey the data and the conclusions, as well as how we evaluate and interpret our data.

The strengths of the manuscript are the combined anatomical and electrophysiological approach that leverages a novel mouse line, and the likely importance of the observations to investigation of circuit mechanism for spatial cognition and memory.

While the manuscript is greatly improved there are remaining weaknesses.

1. While the over-interpretation of the data is reduced compared to the previous version of the manuscript, this still substantially compromises the study. The manuscript continues to make claims about functional effectiveness of connections that are not sufficiently supported by the data. A more specific claim that glutamateric connections from L5b to L5a appear denser in MEC than LEC, and that this suggests different operating principles for computation in these two regions, would be well justified by the data, would be of wide interest, would do much to stimulate future work and would not lead to erroneous interpretations down the line. Specific statements that over-interpret the data should be amended.

We thank the reviewer for this constructive remark, and we have reformulated our main conclusions and suggestions throughout accordingly and tried to clearly differentiate between conclusions and suggestions. The relevant changes include sentences in the abstract, the introduction and the discussion mainly. They can be easily spotted in the version in which we indicated the main changes we made in the revised version (see also our comment to point two).

2. The Discussion could do a much better job of considering caveats in the interpretation of the data. In particular, relatively greater excitatory connectivity in a brain slice does not necessarily imply more effective functional connectivity. Alternative scenarios to consider could include the following. 1. It's unclear from the present data whether the relative difference in excitatory input from L5b to L5a is matched by a difference in feedforward or feedback inhibition. If it is then the pathways may turn out to be similarly effective. Testing this convincingly will require in vivo experiments. 2. Either pathway could be subject to neuromodulation in vivo. It's conceivable this could substantially modify or even reverse their apparent relative effectiveness. 3. The membrane potential and spike threshold of L5a neurons in vivo is unknown. If the membrane potential of L5a neurons in MEC is more depolarized in vivo than in LEC, then the L5b to L5a pathway could turn out to be more effective. The discussion should also consider the fact that the mouse line used labels only a sub-set of the neurons in L5b of LEC and MEC. The possibility that connectivity from other L5b neurons may differ should be clearly noted.

We appreciate the suggestions of the reviewer and we fully agree that all the suggested, and even more scenarios might be relevant. To cover them all would go far beyond the scope of the paper or would simply result in a listing of possibilities with references. Of course, we cannot exclude any of the ones suggested, though some are more plausible than others and that is true for the many additional alternatives one could think of. We do however feel that the observed differences in connectivity, as nicely formulated by the reviewer under point 1 allow for some speculative comments of why this might be relevant. We have rephrased the relevant parts of the discussion accordingly.

The reviewer is apparently not satisfied with the additional supplementary data and the arguments we provided to address the potential risk of selectivity in the population of neurons in the particular TG animal. Of course, we cannot exclude that option but in the previous revision we provided arguments to address this. This was embedded in the result section and we feel that it would interrupt the flow of the manuscript to move that into the discussion. Our arguments are on page 8, lines 202-211. We have reiterated this now in the discussion in two sentences on page 18, lines 457460 which reads: ‘Note that in the present study we focussed on the presumed direct excitatory connectivity from LVb-to-LVa neurons using a newly derived TG mouse line. Although we above provided data to argue that we find it unlikely that the reported difference in connectivity between LEC and MEC might be caused by a different preference for a certain cell type in MEC versus LEC, we cannot completely exclude that option.

3. The possibility that differences in cut dendrites explain the differences in synaptic responses between LEC and MEC, and within the MEC, is still not ruled out. The revised manuscript now gives an indication of the number of dendritic branches in MEC that are > 150 μm. This is insufficient as layer 3 can be on the order of 300 μm or more in width between its superficial and deep borders, while the axons from L5b extend well into L2. To make the claims convincing it's important that adequate quantification of remaining apical dendritic length (and if possible diameter/surface areas) in LEC and at each dorsoventral level of MEC is included (for both cut and not-cut neurons).

We have now provided supplementary data (Figure 4—figure supplement 2D) providing detailed information on how far dendrites extend towards superficial layers for each neuron and correlated that parameter with the evoked postsynaptic responses. The data indicate that it is unlikely that the severing of the apical dendrites caused the differences in the response between MEC- and LEC-LVa neurons. Our material does not allow to provide information of dendritic dimensions other than distribution across layers, so we have therefor not included diameter and surface area information.

We further did not include potential differences in receptor/channel distributions, all of which might result in differences not only between neurons with or without cut dendrites, but also between MEC and LEC neurons with intact dendrites. We thus suggest, and agree with the reviewer, that a parsimonious explanation is the striking difference in apparent anatomical density of the projections from LVb to LVa.

4. I couldn't find anywhere whether any connections were tested with glutamate receptor antagonists to confirm they are synaptic and glutamatergic. It's unlikely but not impossible that some short latency responses reflect low level expression of ChR2 in non-5b neurons (in my experience this might not be detectable from fluorescence). It would be good to know this can be ruled out.

We did not use any pharmacological approaches, as suggested by the reviewer, to test whether contacts are synaptic and glutamatergic. Note that we analyzed the transgenic line extensively and we report that the line almost exclusively labels non-GABAergic neurons: ‘using a GAD67 transgenic line expressing green fluorescent protein (GFP), we showed that the percentage of double-labelled (PCP4+, GAD67+) neurons among total GAD67-positive neurons is very low in both LEC and MEC (4.3% and 2.3% respectively, Figure 1—figure supplement 2)’.

Therefore, we assume that most of the TG-targeted neurons are excitatory, likely glutamatergic. This is certainly something we will explore in the future as we plan to do an ‘in vivo’ study using the same mouse line.

We cannot exclude that a number of tetracycline-controlled transactivator (tTA, Tet-Off) neurons expressed too weakly to be reported in the tTA-dependent reporter mouse which expresses mCherry and that such non-reported neurons might express ChR2 at a level that would impact our results, particularly through representing an unknown class of non-glutamatergic neurons, but we deem that rather unlikely. Since we were looking for overall connectivity differences or similarities between LEC and MEC, and in view of the striking differences between responses in LVa in dorsal LEC versus dorsal MEC we did not explore either of these possibilities further.

https://doi.org/10.7554/eLife.67262.sa2

Article and author information

Author details

  1. Shinya Ohara

    1. Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    2. Laboratory of Systems Neuroscience, Tohoku University Graduate School of Life Sciences, Tohoku, Japan
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    shinyaohara@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0681-5086
  2. Stefan Blankvoort

    Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0993-3790
  3. Rajeevkumar Raveendran Nair

    Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9755-0874
  4. Maximiliano J Nigro

    Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Conceptualization, Formal analysis, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6611-9125
  5. Eirik S Nilssen

    Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Conceptualization, Formal analysis, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6997-3343
  6. Clifford Kentros

    Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3850-3829
  7. Menno P Witter

    Kavli institute for Systems Neuroscience, Center for Computational Neuroscience, Egil and Pauline Braathen and Fred Kavli Center for Cortical Microcircuits, NTNU Norwegian University of Science and Technology, Trondheim, Norway
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    menno.witter@ntnu.no
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0285-1637

Funding

Kavli Foundation (endowment)

  • Menno P Witter

Ministry of Education, Culture, Sports, Science and Technology (KAKENHI (#19K06917))

  • Shinya Ohara

Norwegian Research Council (#197467)

  • Menno P Witter

Norwegian Research Council (#223262)

  • Menno P Witter

Norwegian Research Council (#227769)

  • Menno P Witter

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Grethe M Olsen and Paulo Girao for help with histological preparations and microscopic imaging, and Paulo Girao and Yasutaka Honda for MATLAB programming. Bente Jacobsen and Thanh Doan contributed the tracing material for Figure 3—figure supplement 1, and associated results described.

Ethics

Animal experimentation: All experiments were approved by the local ethics committee and were in accordance with the European Communities Council Directive and the Norwegian Experiments on Animals Act (#17898, #22312).

Senior Editor

  1. Laura L Colgin, University of Texas at Austin, United States

Reviewing Editor

  1. Katalin Toth, University of Ottawa, Canada

Reviewers

  1. Matthew F Nolan, University of Edinburgh, United Kingdom
  2. Desdemona Fricker, CNRS UMR8119, France

Publication history

  1. Received: February 5, 2021
  2. Accepted: March 25, 2021
  3. Accepted Manuscript published: March 26, 2021 (version 1)
  4. Version of Record published: April 16, 2021 (version 2)

Copyright

© 2021, Ohara et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 860
    Page views
  • 131
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    2. Neuroscience
    Lloyd Davis et al.
    Tools and Resources

    Synthetic strategies for optically controlling gene expression may enable the precise spatiotemporal control of genes in any combination of cells that cannot be targeted with specific promoters. We develop an improved genetic code expansion system in C. elegans and use it to create a photo-activatable Cre recombinase. We laser-activate Cre in single neurons within a bilaterally symmetric pair to selectively switch on expression of a loxP controlled optogenetic channel in the targeted neuron. We use the system to dissect, in freely moving animals, the individual contributions of the mechanosensory neurons PLML/PLMR to the C. elegans touch response circuit, revealing distinct and synergistic roles for these neurons. We thus demonstrate how genetic code expansion and optical targeting can be combined to break the symmetry of neuron pairs and dissect behavioural outputs of individual neurons that cannot be genetically targeted.

    1. Neuroscience
    Charlotte Vanacker et al.
    Research Article Updated

    GnRH neurons are the final central neural output regulating fertility. Kisspeptin neurons in the hypothalamic arcuate nucleus (KNDy neurons) are considered the main regulator of GnRH output. GnRH and KNDy neurons are surrounded by astrocytes, which can modulate neuronal activity and communicate over distances. Prostaglandin E2 (PGE2), synthesized primarily by astrocytes, increases GnRH neuron activity and downstream pituitary release of luteinizing hormone (LH). We hypothesized that glial fibrillary acidic protein (GFAP)-expressing astrocytes play a role in regulating GnRH and/or KNDy neuron activity and LH release. We used adeno-associated viruses to target designer receptors exclusively activated by designer drugs (DREADDs) to GFAP-expressing cells to activate Gq- or Gi-mediated signaling. Activating Gq signaling in the preoptic area, near GnRH neurons, but not in the arcuate, increases LH release in vivo and GnRH firing in vitro via a mechanism in part dependent upon PGE2. These data suggest that astrocytes can activate GnRH/LH release in a manner independent of KNDy neurons.