Reelin differentially shapes dendrite morphology of medial entorhinal cortical ocean and island cells Free

Proper dendrite growth is crucial for appropriate functioning of neurons. The geometry of dendritic morphology determines the manner and quality of neuronal information processing. Most of the studies concerning dendritic growth have shown that the dendritic growth process is extremely dynamic (Prigge and Kay, 2018; Wu et al., 1999). Defects in dendrite development impair the formation of neuronal circuits and information processing between neurons. Dendritic growth mechanisms are regulated by cell intrinsic genetic programs, as well as by extrinsic signalling molecules (Lin et al., 2020). Intrinsic factors play a key role in dendritic development and are mediated through specific transcription factors that determine the neuronal identity, positioning, morphological features and functional electrical properties of the neuronal cells (Ledda and Paratcha, 2017; McAllister, 2000; Schuldiner and Yaron, 2015). On the other hand, extrinsic factors influence local and global mechanisms of dendrite development (Hamad et al., 2023; Valnegri et al., 2015). There are several extrinsic factors that influence dendrite growth, among these is reelin, which has the capacity to regulate dendritic morphology at different developmental stages.

The multimodular signalling protein reelin is a secreted glycoprotein that controls neuronal migration and cortical layering (Caviness, 1976; Curran and D'Arcangelo, 1998). The canonical reelin signalling cascade is triggered through direct binding of reelin to apolipoprotein E receptor 2 (ApoER2) and very low-density lipoprotein receptor (VLDLR). Binding of reelin activates the intracellular adapter protein disabled 1 (Dab1) (Cooper and Howell, 1999; D'Arcangelo et al., 1999; Hiesberger et al., 1999; Trommsdorff et al., 1999). Reelin is best known for guiding the migration of newborn neurons and orchestrating proper cortical layering (Bock and May, 2016; Förster et al., 2010; Frotscher et al., 2009). In addition, it also regulates dendritic growth of cortical neurons during embryonic development. For example, dendritic growth of pyramidal cells is reduced in both homozygous and heterozygous reeler mice (Niu et al., 2004; Pinto Lord and Caviness, 1979). Application of reelin to embryonic hippocampal neurons in vitro has been shown to promote dendritic growth (Jossin and Goffinet, 2007; Matsuki et al., 2008). Moreover, reelin promotes dendritic growth of pyramidal cells during early embryonic neocortical development (Chai et al., 2015; Kohno et al., 2015; Kupferman et al., 2014; Nichols and Olson, 2010; O'Dell et al., 2015). However, studies carried out during early postnatal development have shown that reelin may have an opposite effect; i.e. it may restrict dendritic growth of cortical pyramidal neurons (Chameau et al., 2009). Indeed, reeler mice have been shown to display hypertrophic dendrites in the forebrain (Yabut et al., 2007). Furthermore, inducing reelin deficiency directly after birth caused interneuron dendritic hypertrophy in the somatosensory cortex (Hamad et al., 2021b). Thus, reelin appears to exert opposing effects on dendritic growth depending on the cell type and on the developmental stage.

The MECII has been intensively studied over the past decade, it shows strong connectivity with the hippocampus and with other cortical and subcortical areas (Agster and Burwell, 2013; Kerr et al., 2007; Tukker et al., 2022; Witter et al., 2017). Neocortical prefrontal memory engram cells, which have been shown to be crucial for remote contextual fear memory, were rapidly modulated during initial learning through inputs from both the hippocampal and MEC network (Kitamura et al., 2017). The MECII comprises functionally defined cell types that include grid cells (Doeller et al., 2010), head direction cells (Chadwick et al., 2015; Jacobs et al., 2010), speed cells (Kropff et al., 2015; Pérez-Escobar et al., 2016), border cells (Savelli et al., 2008; Solstad et al., 2008), object-vector cells (Høydal et al., 2019) and conjunctive cells, which represent different combinations of speed, place and head direction (Kropff et al., 2015; Sargolini et al., 2006; Solstad et al., 2008). The MEC also comprises anatomically defined cell types. The two major neuronal population in the MECII are the ocean cells (stellate ocean cells) and island cells (pyramidal island cells). These neuronal subtypes differ in morphology, molecular marker expression, electrophysiological properties and projection targets (Alonso and Klink, 1993; Canto et al., 2008; Fuchs et al., 2016; Kitamura et al., 2014, 2015; Ray et al., 2014; Tang et al., 2014; Varga et al., 2010). Moreover, MECII neuronal spatiotemporal firing properties are remarkably different from those of MECIII neuron (Tang et al., 2015). Ocean cells express reelin and morphologically present ‘stellate-type’ multipolar dendritic trees (Kitamura et al., 2015). These cells project to the dentate granule cells (DGCs) and hippocampal CA3 pyramidal cells (Kitamura et al., 2014; Ray et al., 2014; Varga et al., 2010), which rapidly form a distinct representation of a novel context and drive context-specific activation of downstream hippocampal CA3 cells (Kitamura et al., 2015). On the other hand, island cells express Wolfram syndrome 1 (Wfs1) and calbindin, but they do not express reelin (Kitamura et al., 2014). They also project to CA1 and predominantly to interneurons (Kitamura et al., 2014; Ray et al., 2014; Varga et al., 2010); however, they do not exhibit context-specific Ca²⁺-activity and therefore do not contribute to the discriminatory activation of CA3 cells upon serial exposure to distinct contexts (Kitamura et al., 2015). Based on morphological features and electrophysiological properties, another study has defined ‘intermediate stellate’ (intermediate ocean) and ‘intermediate pyramidal’ (intermediate island) cells (Fuchs et al., 2016). However, by using the same analysis, another study could not unveil these additional clusters of principal neurons in MECII (Winterer et al., 2017). Besides excitatory MECII neurons there are also groups of heterogeneous GABAergic inhibitory interneurons (∼10%) that have local branching patterns (Melzer et al., 2012; Ye et al., 2018).

MECII malfunctioning underlies deficits in spatial navigation and working memory, thereby causing a variety of neurological disorders. The question of to what extent reelin deficiency affects dendrite development of ocean cells and their neighbouring island cells in early development remains to be elucidated. To address this, we investigated the effect of early postnatally induced reelin deficiency on dendritic development of ocean cells in the Reln^cKO mice. Because reelin is a secreted glycoprotein, we hypothesized that reelin deficiency might also influence dendrite development of neighbouring island cells. We used the OTC system because both morphological and electrophysiological characteristics are preserved in vitro and are similar to those seen in vivo (Hamad et al., 2015; Klostermann and Wahle, 1999). We have recently shown that early postnatal Reln^cKO in inhibitory GABAergic interneurons exhibited dendritic hypertrophy in these cells (Hamad et al., 2021b). In this article, we show that reelin is also required for proper dendrite development of ocean cells as well as the neighbouring island cells. Our present data demonstrate that reelin limits ocean cell and island cell dendritic overgrowth to ensure proper development of dendritic architecture. Reelin-expressing ocean cells are the first cell type that shows accumulation of intracellular amyloid during the early stages of Alzheimer's disease (AD) (Kobro-Flatmoen et al., 2016, 2023) and therefore these findings suggest functional relevance of reelin in pathogenesis of AD.

To examine the role of reelin in ocean cell development, it is important to circumvent the cortical layer malformations that are seen in the reeler mutant. Therefore, we used the Reln^cKO mouse line in this study, because it ubiquitously expresses a fusion protein composed of Cre recombinase and a mutated form of the oestrogen receptor (Cre-ERT2). (Z)-4-hydroxytamoxifen (4-OHT) application induces nuclear Cre activity and knockout of the floxed reelin gene. Therefore, OTCs were prepared from MEC of newborn postnatal day 0 (P0) mice and cultured in roller tubes for 10 days. To conditionally induce reelin deficiency, 1 µM 4-OHT was added to the OTCs from the first day in vitro (1 DIV) for 4 consecutive days. Around 5 DIV, OTCs were transfected and 5 days later (10 DIV) the OTCs were either subjected to Ca²⁺ imaging or fixed and stained for morphological analysis. We have recently verified that 4-OHT administration to OTCs induces nuclear Cre activation and knockout of the floxed reelin gene without altering cortical layering in the Reln^cKO mice (Hamad et al., 2021a). During embryonic development, reelin is secreted by Cajal-Retzius (CR) cells, which rapidly decline in number after birth. CR cells selectively undergo apoptosis around the second postnatal week (Anstötz et al., 2014). This means that, after birth, reelin is mainly produced by a subset of GABAergic interneurons in the neocortex and hippocampus (Pesold et al., 1998), with a specific distribution of reelin-positive interneuron subtypes in the neocortex (Pohlkamp et al., 2014). Besides GABAergic interneurons, MECII ocean cells also express reelin (Kitamura et al., 2015). The MEC also comprises anatomically defined cell types. The two major neuronal populations in the MECII are the ocean cells (stellate cells), which express reelin, and island cells (pyramidal cells), which express Wfs1. Initially, the presence of these cells in MECII was verified by staining parasagittal brain slices from wild-type mice around P10 (Fig. 1). In MECII, we observed clusters of island cells labelled with wfs1, which appeared in a curvilinear matrix of bulb-like structures (Fig. 1B,F), as reported previously (Kitamura et al., 2014). The multipolar ocean cells were labelled for reelin (Fig. 1C,G) but show no overlap with the neighbouring island cells (Fig. 1D,I). To investigate the role of reelin in dendritic growth, we first confirmed the knockout of reelin. To this end, wild-type and reelin knockout Reln^cKO mice received a tamoxifen injection on P1 for three consecutive days and were subsequently immunostained using an antibody specific for reelin at 2, 3, 4 and 5 days after injection (Fig. 2). Immunostaining revealed a gradual decrease in reelin expression, with complete disappearance of reelin immunoreactivity observed after 5 days (Fig. 2A-E). Moreover, disabled 1 (Dab1) is an adaptor protein that is essential for neuronal migration and maturation in response to the extracellular protein reelin (D'Arcangelo, 2005). It is noteworthy that Dab1 is upregulated in reeler mice and, conversely, in wild-type mice, Dab1 protein is degraded upon reelin stimulation (Rice et al., 1998). To demonstrate that Dab1 is upregulated in the Reln^cKO mice, we administered tamoxifen at P1 and sacrificed the animals at P10 (Fig. 2F-P). The MECII slices were stained using Dab1- and Wfs1-specific antibodies. The mean fluorescence intensity analyses demonstrated that the immunofluorescence intensity of Dab1 was markedly elevated in the Reln^cKO MECII slices in comparison with the wild-type control, thereby confirming the upregulation of DAB in the Reln^cKO mice.

To explore the role of reelin in MECII ocean cell dendrite growth, we transfected wild-type and Reln^cKO OTCs at 5 DIV with pEGFP-N1 to visualize the dendrites, fixed the tissue at 10 DIV and reconstructed neurons for morphometrical quantifications (Fig. 3A, for schematic representation). Although MECII ocean cells have a multipolar structure similar to inhibitory interneurons, morphologically, they are easily distinguished by the following criteria. First, cortical inhibitory interneurons have a local axon branching pattern (Karube et al., 2004), whereas the excitatory ocean cell axon descends the MECII distally towards the white matter and further to the hippocampus (Kitamura et al., 2015). Second, MECII ocean cells have a higher number of primary dendrites with an average of eight dendrites (Fuchs et al., 2016), in contrast to inhibitory interneurons with an average of four dendrites. Third, the ocean cells dendritic shaft is remarkably rich in spines, whereas the inhibitory interneuron dendrites have only few sparsely distributed spines along the dendritic shaft (Kawaguchi et al., 2006). The quantitative morphometric analysis of identified ocean cells around 10 DIV revealed that dendritic length and segments were significantly increased in Reln^cKO when compared with wild type (Fig. 3B,C and Table S1), whereas the number of primary dendrites remained unaltered (Fig. 3D), suggesting that early postnatal reelin elimination resulted in a hypertrophic dendritic growth in the MECII ocean cells. To corroborate these findings in vivo, wild-type and Reln^cKO mice received a tamoxifen injection on P1 for 3 consecutive days and were euthanized at P10 (see Fig. 3G for schematic representation). The brains were then subjected to Golgi staining for morphological analyses. The three-dimensional reconstruction of dendritic arbours of ocean cells revealed that the mean dendritic length and the number of dendritic segments were significantly increased in the Reln^cKO ocean cells in comparison with wild-type cells (Fig. 3H-L and Table S2), thereby confirming the findings in OTCs.

Reelin is a secreted protein that can act in an autocrine or paracrine manner. Our observation that the absence of autocrine reelin release from ocean cells resulted in dendritic hypertrophy of ocean cells prompted us to investigate whether paracrine reelin release into their closely neighbouring island cells also might influence island cell dendritic morphology. To test this hypothesis, we reconstructed island cell dendrites from the Reln^cKO and control wild-type OTCs from MECII. Island cells can easily be identified by their single large prominent apical dendrite emerging from the cell soma located in MECII and multiple small basal dendrites (Fuchs et al., 2016; Naumann et al., 2016), thereby resembling the typical neocortical pyramidal cell morphology found in other areas of the neocortex. The quantitative morphological analyses of the 3D reconstructed island cells around 10 DIV showed that the apical dendritic length and segments were significantly increased in Reln^cKO when compared with wild type (Fig. 4A,B and Table S3). Similarly, the length of basal dendrites and dendritic segments of island cells was significantly longer in Reln^cKO than in the wild-type OTCs (Fig. 4C,D). To corroborate these findings in vivo, wild-type and Reln^cKO mice received a tamoxifen injection at P1 (for 3 consecutive days) and were euthanized at P10. Their brains were then subjected to Golgi staining. The three-dimensional reconstruction of dendritic arbours of island cells revealed that the mean apical and basal dendritic length, as well as the number of apical and basal dendritic segments, were significantly increased in the Reln^cKO ocean cells in comparison with wild-type cells (Fig. 4G-L and Table S4), thereby confirming the findings in OTCs. These data suggest that paracrine release of reelin from ocean cells into island cells plays a regulatory role in dendrite growth of the island cells.

To confirm the molecular identity of the reconstructed island cells, we transfected wild-type and Reln^cKO OTCs at 5 DIV with EGFP plasmid and allowed expression for 5 days. At 10 DIV, the OTCs were fixed and stained for Wfs1. Only cells expressing the Wfs1 marker were reconstructed in both groups (Fig. S1A-F). The quantitative morphological analyses of the 3D reconstructed island cells at 10 DIV revealed that the apical dendritic length and segments were significantly increased in Reln^cKO when compared with wild type (Fig. S1G,H). Similarly, the length of basal dendrites and dendritic segments of island cells was significantly higher in Reln^cKO than in the wild-type OTCs (Fig. S1I,J), thereby corroborating the analyses performed based solely on morphological criteria.

To investigate which dendritic compartments from ocean cells and island cells might differ morphologically between Reln^cKO and wild type, we employed Sholl analysis of dendrites. The analysis revealed a significant increase in dendritic complexity in proximal dendritic intersections the apical and basal dendrites of Reln^cKO island cells (Fig. 5A,C). Furthermore, the total number of dendritic intersections of apical and basal dendrites was significantly increased in Reln^cKO island cells (Fig. 5B,D). These data suggest that absence of reelin released from ocean cells caused proximal dendrite hypertrophy of the neighbouring island cells. On the other hand, Sholl analyses of ocean cell dendritic complexity showed that deficiency of autocrine reelin action affected the entire dendritic compartment of these cells (Fig. 5E). The total number of dendritic intersections was also remarkably increased in the Reln^cKO ocean cell dendrites (Fig. 5F). Together, these results imply that paracrine release of reelin from ocean cells specifically affects the proximal dendritic compartment of the neighbouring island cells.

The next step was to determine whether secreted reelin from wild-type OTCs might reverse the aberrant dendritic hypertrophy of the island cells in the Reln^cKO OTCs. To address this, we conducted a co-culture experiment in which OTCs were grouped as follows: wild type+wild type, Reln^cKO+Reln^cKO or wild type+Reln^cKO (Fig. 6A). In the Reln^cKO+Reln^cKO co-culture, the apical and basal dendritic length ,and segments of island cells reconstructed from Reln^cKO+Reln^cKO OTCs were significantly increased in comparison with the reconstructed wild type+wild type OTCs (Fig. 6B-E, Tables S5 and S6). However, in the wild type+Reln^cKO co-culture, the apical and basal dendritic length, and segments of island cells reconstructed from Reln^cKO OTCs were unaltered from those island cells from the wild type+wild type co-culture (Fig. 6B-E). These findings indicate that reelin secreted by wild-type OTCs into the incubation medium restored the dendritic morphology of the co-cultured Reln^cKO OTCs to a basal level comparable with that of the wild type+wild type co-culture control group.

To explore the potential mechanisms that might cause the hypertrophic dendritic morphology of Reln^cKO ocean cells and island cells, we performed single-cell Ca²⁺-imaging to quantify amplitude and frequency of Ca²⁺ signals. Calcium signalling influences dendrite dynamics through modulation of the cytoskeleton near the calcium entry site and also activates specific transcription factors that can induce dendritic growth (Konur and Ghosh, 2005). For that purpose, OTCs were transfected at 5 DIV with a genetic construct encoding the Ca²⁺ indicator GCaMP6s (Chen et al., 2013). The GCaMP6s construct very sensitively detects Ca²⁺ concentration changes and is distributed to soma, dendrites and axon when overexpressed in neurons (Fig. 7). The Ca²⁺ signal amplitude and frequency were recorded at 10 DIV in wild-type and Reln^cKO ocean cells and island cells. The analyses showed that the amplitude of Ca²⁺ signals of ocean cells and island cells did not differ from the control wild-type group (Fig. 7E,K and Table S7). However, recorded ocean and island cells in Reln^cKO displayed enhanced frequency of Ca²⁺ signals (Fig. 7F,L and Table S7). As the frequency of Ca²⁺ signals is controlled via presynaptic mechanisms, this implies that Reln^cKO ocean and island cells signalling is defective at the presynaptic level. Taken together, these results suggest that Reln^cKO hypergrowth is due to an enhanced Ca²⁺ signal frequency in the recorded neurons. Because in both ocean and island cells the calcium signals and dendrite growth are similarly affected in the Reln^cKO mice, we decided to focus on latter to explore in more detail the mechanism that might underlie dendritic overgrowth.

Presynaptic GABA B receptors (GABA_BRs) have the ability to control Ca²⁺-release at the presynaptic site (Benarroch, 2012). Recently, we have shown that GABA_BRs are involved in postnatal reelin signalling in the cortex (Hamad et al., 2021a). In brief, the GABA_BR agonist baclofen failed to activate and the antagonist CGP failed to block GABA_BRs in inhibitory interneurons of Reln^cKO somatosensory cortex (Hamad et al., 2021a). Hypothetically, the increased Ca²⁺-frequency in Reln^cKO island cells could be the main cause of dendritic hypertrophy in these cells. To test this hypothesis, we reconstructed EGFP-transfected wild-type and Reln^cKO island cells in presence of the GABA_BRs antagonist CGP. The quantitative morphometric analysis of reconstructed island cells revealed that CGP significantly increased apical and basal dendritic length (Fig. 8A,C, Tables S8 and S9) and dendritic segments (Fig. 8B,D) of wild-type island cells but not Reln^cKO, suggesting that dendritic hypertrophy seen in the Reln^cKO is caused by malfunctioning of presynaptic GABA_BRs. In absence of reelin, the GABA_BRs were not able to block N-type (Ca_v2.2) Ca²⁺-channels at the presynaptic site, thereby causing excessive frequency of calcium signals and overgrowth of cortical interneurons in the somatosensory cortex (Hamad et al., 2021b). We therefore assumed that, similarly, the hypertrophic growth of Reln^cKO island cells might be due to the absence of GABA_BR-mediated inhibition of the N-type (Ca_v2.2) Ca²⁺ channels at the presynaptic site. To validate this hypothesis, we transfected OTCs at 5 DIV with EGFP. The N-type (Ca_v2.2) Ca²⁺-inhibitor conotoxin (2 µM) was applied daily to OTCs from 5 to 10 DIV (Fig. S2). The quantitative morphometric analysis of island cells showed that conotoxin did not affect either apical or basal dendritic length and dendritic segments of wild-type island cells (Fig. S2A-D and Tables S10 and S11). However, it reduced apical and basal dendritic length and segments of Reln^cKO island cells (Fig. S2A-D). In parallel, to confirm that GABA_BR-mediated inhibition of N-type (Ca_v2.2) Ca²⁺ channels was defective in Reln^cKO island cells, we performed Ca²⁺ imaging of wild-type and Reln^cKO island cells at 10 DIV. Both OTC groups were treated with conotoxin (2 µM) daily from 5 to 10 DIV (Fig. S2I-K). Ca²⁺ imaging analyses revealed that conotoxin did not affect the Ca²⁺ amplitude of either wild-type or Reln^cKO island cells (Fig. S2I). However, it reduced the Ca²⁺ frequency of Reln^cKO island cells (Fig. S2J). Taken together, these results suggest that reelin signalling through GABA_BR-mediated inhibition of N-type (Ca_v2.2) Ca²⁺ channels is important to maintain normal calcium signalling, which prevents excessive outgrowth of island cell dendrites.

The observation that enhanced Ca²⁺-frequency in recorded island cells resulted in hypertrophic dendritic growth in Reln^cKO could also be due to excessive glutamatergic excitation. Glutamatergic transmission converging on Ca²⁺-signalling plays a key role in enhancing dendritic growth (Cline, 2001). Reelin has been shown to modulate NMDA receptor activity in neocortical neurons (Chen et al., 2005). Therefore, it is possible that excessive Ca²⁺-influx through NMDARs might be an additional cause of dendritic hypertrophy of Reln^cKO island cells. To test this hypothesis, OTCs were transfected at 5 DIV with EGFP, and the NMDAR antagonist APV (50 µM) was applied to OTCs from 5 to 10 DIV. At 10 DIV, the OTCs were fixed, stained and reconstructed (Fig. S3). The quantitative morphometric analysis of selected island cells showed that the APV treatment can reduce apical and basal dendritic length and segments in both wild-type and Reln^cKO island cells (Fig. S3, Tables S12 and S13), suggesting that NMDARs are involved in dendritic hypertrophy seen in Reln^cKO island cells. These findings suggest that the hypertrophic growth of Reln^cKO island cells is also mediated through an excessive excitation of the NMDARs. The AMPARs play a crucial role in interneuron development (Hamad et al., 2011, 2014). To test whether the growth effects observed in the island cells might be secondary to the effects in the interneurons, OTCs were transfected at 5 DIV with EGFP, and the AMPAR antagonist CNQX (10 µM) was applied to OTCs from 5 to 10 DIV. At 10 DIV, the OTCs were fixed, stained and reconstructed (Fig. S4). Quantitative morphometric analysis of selected island cells revealed that the CNQX treatment can reduce apical and basal dendritic length (Fig. S4A,C, Tables S14 and S15) and segments (Fig. S4B,D) in both wild-type and Reln^cKO island cells. These results suggest that AMPARs are involved in the dendritic hypertrophy observed in Reln^cKO island cells. The findings indicate that the hypertrophic growth of Reln^cKO island cells is a consequence of the effect observed in the interneurons.

As mentioned above, presynaptic GABA_BRs do not function properly in the absence of reelin. Consequently, we sought to ascertain whether the function of GABA_BRs might be influenced by canonical reelin signalling in order to prevent dendritic overgrowth in MECII island cells. To address this question, we transfected OTC with EGFP-N1 at 5 DIV. The OTCs were treated for 5 consecutive days with RAP, an inhibitor of low-density lipoprotein receptor-related proteins (including VLDL and APOER2; reelin binds to VLDLR and ApoER2 to regulate Dab1 tyrosine phosphorylation). Chronic RAP treatment (50 μg/ml) in wild-type OTCs was found to increase apical and basal dendrite length (Fig. S5A,C, Tables S16 and S17) and mean number of apical and basal segments of island cells (Fig. S5B,D). As anticipated, Reln^cKO OTCs treated with RAP demonstrated no alterations in dendritic growth parameters (Fig. S5A-D). These observations in MECII island cells underscore the necessity for GABA_BRs to engage in crosstalk with reelin signalling for optimal functionality.

In a previous study, we demonstrated that reelin restricts dendritic growth of interneurons in the somatosensory neocortex (Hamad et al., 2021b). To investigate the impact of reelin on interneuron dendritic growth in the MECII in vivo, we generated wild-type and Reln^cKO mice that express enhanced green fluorescent protein (EGFP) under the direction of the mouse GAD67 (GAD67-wt and GAD67-Reln^cKO transgenic mice, respectively; for details, refer to the Material and Methods section). The reelin knockout was induced in vivo by tamoxifen injection at P1. At P10, the animals were sacrificed and dendritic arbours of GAD67-wt and GAD67-Reln^cKO interneurons of the MECII were reconstructed (Fig. 9 and Table S18). The quantitative morphological analyses demonstrate that the elimination of reelin after birth results in an increased dendritic growth of GAD67-Reln^cKO interneurons in comparison with the GAD67-wt interneurons (Fig. 9A and Table S18). In a similar fashion, mean dendritic segments of GAD67-Reln^cKO interneurons were found to be increased in comparison with the GAD67-wt interneurons (Fig. 9B), whereas the primary dendrites were unaltered (Fig. 9C). These findings indicate that reelin plays a role in regulating dendritic growth of interneurons in the MECII, in a manner analogous to that observed in the somatosensory cortex.

Here, we have investigated a new role of reelin in regulating early postnatal dendritic development of MECII island cells and ocean cells and the inhibitory interneurons. The induction of reelin deficiency at an early postnatal developmental stage caused hypertrophic growth of MECII ocean cells, which normally express reelin, possibly acting in an autocrine manner. On the other hand, proximal dendrites of the MECII island cells that do not express reelin also underwent dendritic hypertrophy, possibly through the lack of paracrine action of reelin released from ocean cells onto island cells. The blockade of reelin binding to its receptors ApoER2 and VLDLR by RAP mimicked the observed effect in Reln^cKO OTCs. There are no specific selective antagonists for ApoER2 and VLDLR; therefore, we used RAP to broadly block LRP family members. Co-culture of Reln^cKO OTCs with reelin-secreting wild-type OTCs rescued the deficient dendritic hypergrowth observed in Reln^cKO island cells. It is conceivable that the secreted reelin from CR cells, which is known to regulate cortical projection neuron morphology (Enck and Olson, 2023), may also contribute to the regulation of dendritic growth of MECII neurons. It is important to note that CR cells undergo apoptosis around the second postnatal week of development (Anstötz et al., 2014; Chin et al., 2007; Pesold et al., 1998). This may result in the presence of these cells until that point, ensuring proper dendritic maturation. However, the apical dendrites of cortical neurons are inhibited by chondroitin sulphate proteoglycans (CSPGs), which are highly expressed in the marginal zone. A recent study has demonstrated that reelin signalling in the marginal zone counteracts the inhibitory action of CSPGs on apical dendrites in the cortex (Zluhan et al., 2020). Consequently, it can be excluded that reelin produced by CR cells is responsible for the growth inhibition of MECII neurons, as it is observed that reelin is important for inhibiting dendritic overgrowth of stellate and island cells in MECII. Another source of reelin in the MECII is the GABAergic inhibitory interneurons. These cells express reelin, and we have recently demonstrated that reelin is essential for preventing dendritic overgrowth in the somatosensory cortex (Hamad et al., 2021b). Similarly, our findings in the current study indicate that early postnatal reelin elimination in vivo also results in dendritic overgrowth of the MECII in vivo. Collectively, these results suggest that ocean cells or/and interneurons are responsible for ensuring proper dendritic growth of the MECII neurons.

In Reln^cKO island cells, only the proximal dendritic compartment was affected. It has previously been shown that reelin signalling is essential for establishing and maintaining the molecular identity of the distal dendritic compartment of cortical pyramidal neurons, where HCN1 and GIRK1 channels are enriched, i.e. in the distal tuft dendrites of both hippocampal CA1 and neocortical pyramidal neurons (Kupferman et al., 2014). However, conditional reelin knockout mice did not show an altered HCN1 distribution in CA1 neurons, in contrast to cortical pyramidal neurons (Meseke et al., 2018). The current study shows that reelin acts specifically on the proximal dendritic compartment of MECII island cells. Moreover, this compartment-specific effect of reelin on dendrite growth was not observed in ocean cells. The different action of reelin on these cell types might suggest a different distribution of reelin receptors across the dendrites, or it might involve other cell intrinsic mechanisms. Further studies are required answer this question.

Neuronal dendritic hypertrophy can lead to pathological hyperexcitability of neuronal circuits, which in turn can contribute to disorders such as epilepsy, schizophrenia, bipolar disorder or autism spectrum disorders (Courchesne and Pierce, 2005; Hu et al., 2017; Ishii et al., 2016). Therefore, it is indispensable to have ‘stop growth signals’ to transform dynamic, immature dendrites into more stable mature dendrites and establish a balanced neuronal network connectivity. Interestingly, rodent MECII and the human caudal entorhinal cortex are structurally similar (Naumann et al., 2016). The proper growth of MECII neurons is of great importance because MECII dysfunction induces deficits in spatial navigation and semantic, episodic and working memory, potentially causing a wide range of neurological diseases. For example, it has been reported that in mild Alzheimer's disease (AD) a profound loss of MECII neurons in human brain occurs (Gómez-Isla et al., 1996). Interestingly, before the onset of clinically overt AD, it has been demonstrated that entorhinal cortex-based virtual reality navigation tasks can differentiate between individuals with mild cognitive impairment that are at low or high risk of developing dementia (Howett et al., 2019). Moreover, hippocampal temporal lobe epilepsy is associated with selective lesions in MECII causing significant cognitive impairments in rats (Gloveli et al., 1997). Furthermore, in temporal lobe epilepsy, epileptiform activity can originate in the superficial MEC, thereby leading to neuronal degeneration in a rat model of epilepsy (Du et al., 1995). Interestingly, the temporal lobe epilepsy has been reported to be associated with reelin deficiency (Leifeld et al., 2022). Furthermore, MECII neurons were found to be involved in tonic but not phasic pain modulation in rats (Zhang et al., 2014). Indeed, in humans, deep brain stimulation of the entorhinal region, but not the hippocampus, enhanced spatial memory when applied during learning (Suthana et al., 2012). Taken together, these studies suggest that correct MECII neurons development is essential for efficient integration into the network so as to ensure their proper cognitive functions. The current study shows that reelin is required for this proper maturation of MECII ocean and island cells.

We investigated possible mechanisms that explain the cause of dendritic hypertrophy in island cells. Recently, we have shown that reelin controls early network activity by modulating presynaptic GABA_BR function (Hamad et al., 2021a), and that in the absence of reelin, dendrites of inhibitory interneurons in the somatosensory cortex were hypertrophic (Hamad et al., 2021b). As Ca²⁺ is a key molecule in regulating dendritic growth, we examined Ca²⁺ amplitude and frequency in Reln^cKO ocean cells and island cells in the presence of the GABA_BR antagonist CGP. The data showed that both Reln^cKO ocean cells and island cells display an abnormally excessive Ca²⁺-spike frequency. Presynaptic GABA_BRs are usually coupled to N-type (Ca_v2.2) Ca²⁺ channels (Pinard et al., 2010). Presynaptic GABA_BRs are known to inhibit presynaptic N-type (Ca_v2.2) Ca²⁺ channels, which results in a decrease in neurotransmitter release (Benarroch, 2012). The fact that the N-type (Ca_v2.2) Ca²⁺ channel blocker conotoxin was able to restore the length of dendrites and dendritic segments, and to restore Ca²⁺ frequency of Reln^cKO island cells suggests that the increase of the Ca²⁺ frequency seen in Reln^cKO island cells due to the dysfunction of GABA_BRs is the likely cause of the dendritic hypertrophy in these cells. Similarly, we have previously described a related mechanism that causes dendrite hypertrophy in inhibitory interneurons of the somatosensory cortex (Hamad et al., 2021b), which has shown that glutamatergic AMPARs but not NMDARs are responsible for dendritic hypertrophy found in Reln^cKO interneurons (Hamad et al., 2021b), as recent data have shown that NMDARs do not affect neocortical dendritic growth of interneurons (Gonda et al., 2020). By contrast, the current study shows that reelin action in the MECII ocean cells and island cells does involve glutamatergic transmission through NMDARs to regulate dendritic growth. The present study demonstrates that AMPAR blockade can reduce dendritic growth in both wild-type and Reln^cKO island cells, suggesting that the AMPARs are involved in the dendritic hypertrophy observed in Reln^cKO island cells. A recent study has shown that reelin regulates the phosphatase STEP, which plays an important role in neurodegeneration, as well as the expression of calcium-permeable AMPARs, which play a role in memory formation (Durakoglugil et al., 2021). Because AMPARs play a crucial role in interneuron development (Hamad et al., 2011, 2014), our findings indicate that the hypertrophic growth of Reln^cKO island cells may be secondary to the effects observed in the interneurons. Proper growth and arborization of dendrites of MECII neurons is crucial for correct neuronal physiology and function (Tukker et al., 2022). Disruption of dendritic development can contribute to neurological disorders associated with cognitive deficits (Kweon et al., 2017). Exploring the mechanisms that regulate dendritic growth will pave the way for future studies to improve our understanding of neurological diseases. This will lead to a deeper understanding of the mechanisms that underlie the regulation of MECII neuronal development, which, in the adult stage, is essential for control spatial navigation and memory.

The guidelines of the German Animal Welfare Act were respected according to law. All mouse experiments were reviewed and approved by the local ethic commission. Licences for animal experiments have been obtained from the State Agency for Nature, Environment and Consumer Protection in North Rhine-Westphalia in Germany (permission number 84-02.04.2016.A383) and from the United Arab Emirates University Animal Ethics Committee of the United Arab Emirates University (permission number ERA_2023_2970).

Animals were housed in a standard 12 h light cycle and fed ad libitum with standard mouse chow. Reln^cKO mice have been generated previously (see Lane-Donovan et al., 2015, 2016). In brief, to obtain conditional reelin knockout mice (Reln^flox/flox CAG-Cre^ERT2 mice), we crossed Reln^flox/flox mice with hemizygous tamoxifen-inducible Cre recombinase expressing mice (CAG-Cre^ERT2) (Hayashi and McMahon, 2002). For the experiments, only Reln^flox/flox CAG-Cre^ERT2 male mice were selected and then crossed with Reln^flox/flox female mice to generate Reln^{flox/flox wildtype} (wt) and Reln^flox/flox CAG-Cre^ERT2 (Reln^cKO) siblings, as verified by PCR. The cKO mouse line ubiquitously expresses a fusion protein comprising Cre recombinase and a mutated form of the oestrogen receptor (Cre-ERT2).

To study the effect of reelin on interneuron dendritic growth in medial entorhinal cortex, we generated wild-type and Reln^cKO mice that express enhanced green fluorescent protein (EGFP) under the direction of the mouse GAD67. To do this, we have crossed the Reln^flox/flox wild-type mice with GAD67-EGFP transgenic mice (Oliva et al., 2000) to generate Reln^flox/flox-GAD67-EGFP mice (hereafter, GAD67-wt). Moreover, we crossed the Reln^flox/flox CAG-^CreERT2 mice with GAD67-EGFP transgenic mice to generate Reln^flox/flox CAG-^CreERT2-GAD67-EGFP mice (hereafter, GAD67-Reln^cKO). Thereafter, only GAD67-wt female mice were selected and then crossed with GAD67-Reln^cKO male mice to generate either GAD67-wt mice or GAD67-Reln^cKO siblings, as verified by PCR. The Reln^flox mice (B6.129-Reln^tm1Her/J; Stock 037348) are available from the Jackson Laboratories.

All newly born pubs were subjected to DNA genotyping. DNA from samples of ear, tail and brain tissue were isolated with ReliaPrep gDNA kit (A205, Promega). The amounts of DNA isolated from the various samples were determined by spectrophotometry with the Genova Nano system (Jenway). DNA was amplified by PCR. PCR reactions were performed in a total volume of 50 μl reaction mixture containing 200 ng of template DNA, Soriano buffer (0.67 M Tris, 0.16 M ammonium sulphate, 67 mM MgCl₂, 67 µM EDTA and 50 mM β-mercaptoethanol), Taq polymerase, 2 µl DMSO and 10 mM dNTPs. For genotyping, we used the following primers: wild-type mice, forward primer 5′-ATAAACTGGTGCTTATGTGACAGG-3′ and reverse primer 5′-AGACAATGCTAACAACAGCAAGC-3′ (450 bp); Reln^flox/flox mice, forward primer 5′-GCTCTGGCCAAGCTTTATC-3′ and reverse primer 5′-CGCGATCGATAACTTCGTATAGCATAC-3′ (1200 bp); CAG-Cre^ERT2, forward primer 5′-ATTGCTGTCACTTGGTCGTGG-3′ and reverse primer 5′-GGAAAATGCTTCTGTCCGTTTGC-3′ (200 bp); GAD67 transgene, forward primer 5′-ATCCAGTTTGTTTTGCCCCTAAAGG-3′ and reverse primer 5′-CTCTACTGAGCCAGTATGGCTGTACAGG-3′; Gad1 (internal control), forward 5′-CCCCACGCGTGATCACTGAGCGACGAGAAAAGCTAC-3′ and reverse 5′-CCCCACGCGTGATCAGAGCTTTGATCTTGGGAGC-3′ (Oliva et al., 2000). The amplification products were verified on a 2% agarose gel in TBE buffer.

The preparation of MEC OTCs was performed from postnatal day 0 (P0) mice. The identification and cutting of MEC OTCs described was performed according to a previously published protocol for the preparation of MEC parasagittal acute slices (Pastoll et al., 2012). All reagents used for OTCs preparation were sterile and all preparations were performed in a laminar air flow bench with horizontal counter flow (Horizontal Flow, ICN, Biomedicals). For each outcome measure, we used 6 wild-type and 6 Reln^cKO mice (each experiment was repeated two or three times). In total, 320 animals were used for this study. From each animal, we obtained three or four MEC OTCs. This study was not pre-registered. No exclusion criteria were pre-determined and the study was exploratory. No randomization was performed to allocate subjects in the study. No sample size calculation was performed in this study. Before decapitation, mice were briefly anesthetized on ice. The skull was removed gently (after 30 min of recovery), the brain was placed on a chopper plate (McIllwain) and the MEC was cut cortex from both hemispheres in the sagittal plane until the identification of the lateral most extent of the MEC (typically ∼600 μm from the lateral surface). Afterwards, 350 μm parasagittal sections from both hemispheres were cut until the medial extent of the MEC was reached, identified by the absence of the thick white band around the external capsule. The slices were transferred onto coverslips (12×24 mm, Kindler). Chicken plasma (Sigma) and GBBS/thrombin (Merck) were mixed in a proportion of 2:1 and then allowed to coagulate for 45 min. The coverslip was transferred into a roller tube (Nunc) filled with 750 μl semi-artificial medium and placed in a roller incubator at 37°C. For the knockout induction, the OTCs were directly stimulated after preparation with 1 µM (Z)-4-hydroxytamoxifen (4-OHT) (3412, Tocris) for 5 consecutive days and kept for experiments until 10 DIV. For the quantitative morphological analysis experiments, the following drugs were used: CGP35348 (CGP, 10 µM, 1245/10, Tocris), APV (50 µM, 0190/10, Tocris) and ω-conotoxin GVIA (conotoxin, 2 µM, 1085, Tocris). As an inhibitor of low-density lipoprotein receptor-related proteins, we used the recombinant Mouse LRPAP Protein [LDL receptor-related protein-associated protein 1; also named receptor-associated protein (RAP), 50 ng/ml, 4480-LR; R&D Systems]. RAP is used to block the canonical reelin pathway by blocking reelin from binding to VLDL and APOER2 (Herz et al., 1991).

Biolistic transfection was performed using the Helios Gene Gun (Bio-Rad) as described previously (Hamad et al., 2020). In brief, cartridges were prepared by coating 10 mg gold particles (Ø=1 µm; Bio-Rad) with the genetically encoded Ca²⁺-indicator pGP-CMV-GCaMP6 s (GCaMP6 s) for Ca²⁺-imaging experiments. The pGP-CMV-GCaMP6 s was a gift from Douglas Kim & GENIE Project (Addgene plasmid 40753) (Chen et al., 2013). For morphological quantification of MECII ocean cells and island cells, we prepared gold particles containing plasmid encoding enhanced green fluorescent protein (pEGFP-N1; Clontech). To prevent excitotoxicity during transfection, glutamate receptors were temporarily blocked with 3 mM kynuric acid (K3375, Sigma) and 50 µM APV (0190/10, Tocris) before blasting. Three hours after transfection the blockers were washed out.

For the morphological assessment, OTCs were fixed around 10 DIV with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4) and warmed to 36°C for 1 h. After several washes with TBS (Tris-buffered saline: 50 mM Tris and 150 mM NaCl at pH 7.6) and permeabilization in TBST (TBS and 0.1% Triton X), OTCs were blocked for 1 h with 1% normal goat serum in TBST. The OTCs were incubated for 24 h at room temperature with the primary antibody mouse anti-EGFP (1:1000, G6795, Sigma). After washing twice in TBS, the secondary antibodies were added accordingly: goat anti mouse biotinylated (1:300, E043201-8, Dako). After several washes in TBS buffer, the slices were subjected to ABC-horseradish peroxidase method using diaminobenzidine as a chromogen. For the visualization of island cells and ocean, the following primary antibodies were used: rabbit anti-Wfs1 (1:3000; 11558-1-AP, Proteintech) for the detection of the island cells and mouse anti-reelin G10 (1:1000, MAB5364, Merck) for ocean cells detection. For the detection of Dab1, we used mouse anti Dab1 (1:500, sc-271136, Santa Cruz Biotechnology). Afterwards, the OTCs were washed 3×15 min with TBST and incubated with secondary antibodies (anti-mouse IgG coupled to Alexa594 and anti-rabbit IgG coupled to Alexa488 in TBS; A-11005 and A-11008, ThermoFisher, 1:500) for 30 min at 23°C. After three repetitive washings with TBS-Tween, the OTCs were mounted with sRIMS mounting medium [70% sorbitol w/v in 0.02 M phosphate buffer with 0.01% sodium azide (pH 7.5)] containing TO-PRO-3 Stain (ThermoFisher, 1:1000) for nuclear counterstaining. DAB1 fluorescence intensity was analyzed with the integrated density tool of ImageJ software.

Tissue preparation and staining were performed according to the manufacturer's instructions using the FD Rapid GolgiStain kit (FD NeuroTechnologies; PK401 Cell Systems Biology). After decapitation at P10, mice brains were removed and immediately washed with water, and then immersed in the impregnation solution (equal volumes of solutions A and B). Two weeks later, the brains were transferred to solution C for 1 week. A vibratome was then used to cut 90 μm parasagittal sections from both hemispheres to the medial extent of the MEC, identified by the absence of the thick white band around the external capsule. A maximum of 16 sections containing the MEC from both hemispheres were prepared per animal. The brain slices were then rehydrated in milli-Q water, immersed in solutions D and E for 10 min for staining, rinsed again in milli-Q water, dehydrated and coverslipped with a water-based mounting medium.

EGFP-immunostained cells were reconstructed with the Neurolucida system (MicroBrightField) at 1000× magnification. The morphology of the ocean cells and island cells has been previously described. As mentioned in the Results section, we adopted the following criteria to distinguish ocean cells from the multipolar inhibitory interneurons in MECII: stellate ocean cells have a larger number of primary dendrites than interneurons, are very abundant in spines and their axons extend towards the white matter instead of having local branching pattern remaining. For quantification of ocean cell morphological parameters, we calculated mean dendritic length (total number of dendritic length divided by the number of primary dendrites), mean number of dendritic segments (total number of dendritic segments divided by the number of primary dendrites) and number of primary dendrites. MECII island cells have one apical dendrite emerging from the cell soma and multiple small basal dendrites. Spines are abundant on island cell dendrites; they are morphologically very similar to pyramidal cells that are found in somatosensory or motor cortex. For quantification of island cell morphological parameters, we calculated apical dendritic length and mean basal dendritic length (total number of dendritic length divided by the number of primary dendrites), total apical dendritic segments and mean number of basal dendritic segments (total number of basal dendritic segments divided by the number of primary dendrites). Sholl analysis of the number of dendrite intersections at 10 µm interval distance points starting from the soma was performed to identify the area where dendritic complexity changed (Sholl, 1953; Zagrebelsky et al., 2010).

To monitor calcium signalling in ocean cells and island cells, OTCs were transfected with the Ca²⁺ sensor GCaMP6s and its expression allowed for 5 days. At 10 DIV, the slices were transferred to a recording chamber mounted on a fixed stage of an inverted microscope and perfused with ACSF (3–5 ml/min) at 32±2°C. Composition of ACSF is 125 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgSO₄, 25 mM NaHCO₃, 1.25 mM NaH₂PO₄ and 25 mM glucose at pH 7.4. Fluorometric Ca²⁺ recordings were made using a Visiscope spinning-disc confocal system CSU-W1 (Visitron) featuring a spinning disc unit CSU-W1-T2 and a sCMOS digital scientific grade camera (4.2 Mpixel rolling shutter version) on an inverted Nikon Ti-E motorized microscope using a CFI P-Fluor 20× objective (NA 0.5, WD=2.10 mm). Images were acquired at 3 frames per second with exposure times of 330 ms with VisiView image acquisition software (Visitron). The biosensor GCaMP6s was excited at 488 nm. Emitted fluorescence was collected through an ET 525/50 filter for GCaMP6s. Fluorometric data are expressed as ΔF/F⁰, also defined as calcium transients, which is the background-corrected increase in fluorescence intensity divided by the resting fluorescence. Only ΔF/F⁰ values that are at least 10% of the resting fluorescence were considered as calcium transients and therefore counted in the analyses. Frequency was calculated as the number of calcium transients occurring in a given time window (3 min). Raw data delivered in the form of a linear 16-bit intensity scale were plotted as fluorescence intensity versus time. Ocean cell or island cell somata were chosen as the region of interest (ROI). The background fluorescence measured near a ROI was then subtracted from these raw data. The baseline fluorescence (F⁰) was calculated as an average of 20 frames in a time window without neuronal activity (as judged by visual inspection). Subsequently, data were normalized to the mean fluorescence intensities [ΔF/F⁰=(F–F⁰)/F⁰], allowing the comparison of data across experiments.

We used Sigma Stat 12 (SPSS Incorporated) for the statistical analyses. Comparisons between two groups were performed with Students’ unpaired t-test when normality test (Shapiro-Wilk) passed, otherwise with Mann–Whitney test. More than two groups were compared with one-way-ANOVA and a Holm-Sidak multiple comparison test for post-hoc analysis if a normality test was passed. If normality failed, we ran one-way ANOVA on Ranks followed by Tukey's multiple comparison test for post-hoc analysis to isolate the significant groups. Results were considered statistically significant at P<0.05.

We thank Dr Douglas Kim for providing the pGP-CMV-GCaMP6s construct. We also thank Marwa Ibrahim and Jeannette Willms for technical support, and Katja Rumpf and Corinna Wojczak from the sectioning core facility at the Ruhr University of Bochum for their support.