Reelin is a large secreted glycoprotein that regulates neuronal migration, lamination and establishment of dendritic architecture in the embryonic brain. Reelin expression switches postnatally from Cajal-Retzius cells to interneurons. However, reelin function in interneuron development is still poorly understood. Here, we have investigated the role of reelin in interneuron development in the postnatal neocortex. To preclude early cortical migration defects caused by reelin deficiency, we employed a conditional reelin knockout (RelncKO) mouse to induce postnatal reelin deficiency. Induced reelin deficiency caused dendritic hypertrophy in distal dendritic segments of neuropeptide Y-positive (NPY+) and calretinin-positive (Calr+) interneurons, and in proximal dendritic segments of parvalbumin-positive (Parv+) interneurons. Chronic recombinant Reelin treatment rescued dendritic hypertrophy in Relncko interneurons. Moreover, we provide evidence that RelncKO interneuron hypertrophy is due to presynaptic GABABR dysfunction. Thus, GABABRs in RelncKO interneurons were unable to block N-type (Cav2.2) Ca2+ channels that control neurotransmitter release. Consequently, the excessive Ca2+ influx through AMPA receptors, but not NMDA receptors, caused interneuron dendritic hypertrophy. These findings suggest that reelin acts as a ‘stop-growth-signal’ for postnatal interneuron maturation.
The morphology of dendritic processes is an important parameter that contributes the electrophysiological properties and synaptic connectivity of neurons. In rodents, dendritogenesis begins between embryonic day (E) 15 and E18 (E15-18), but dendrites continue to extend and branch out after neuronal migration is completed until postnatal day 25-30 (P25-30), thereby significantly contributing to postnatal brain growth (Ben-Ari, 2001). In vivo time-lapse imaging has demonstrated that dendritic growth is a highly dynamic process that includes addition and retraction of fine branches (Prigge and Kay, 2018; Wu et al., 1999). Early in development, dendritic growth is regulated by cell-intrinsic programs and molecular signals that regulate various aspects of dendritic development (Jan and Jan, 2003). Neuronal activity plays a key role in dendritic development during early brain maturation. Thus, glutamatergic activity has been found to regulate dendrite morphology. For example, blocking synaptic activity in the optic tectum neurons delayed dendritic development (Rajan and Cline, 1998), and experience-dependent dendritic growth of optic tectum neurons in vivo was shown to be mediated through α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate receptors (AMPARs) (Haas et al., 2006). AMPARs also contribute to the regulation of the dendritic architecture of motoneurons (Inglis et al., 2002; Jeong et al., 2006; Prithviraj et al., 2008; Zhang et al., 2008) and neocortical neurons (Hamad et al., 2011, 2014).
Reelin is an extracellular matrix glycoprotein that regulates brain development and function. The most prominent role of reelin is to control neuronal migration and layer formation in the developing cerebral cortex (Caviness, 1976; Curran and D'Arcangelo, 1998). The canonical reelin signaling cascade involves direct binding of reelin to apolipoprotein E receptor 2 (ApoER2) and very low density lipoprotein receptor (VLDLR), and subsequent activation of the intracellular adapter protein disabled 1 (Dab1) by tyrosine phosphorylation (Cooper and Howell, 1999; D'Arcangelo et al., 1999; Hiesberger et al., 1999; Howell et al., 2000; Trommsdorff et al., 1999). By binding to its receptors, reelin guides the migration of newborn neurons and ensures the proper development of cortical layers (Bock and May, 2016; Cooper and Howell, 1999). Reelin also regulates dendritic growth during embryonic development. For example, dendritic growth of pyramidal cells is disturbed in both homozygous and heterozygous reeler mice (Niu et al., 2004; Pinto Lord and Caviness, 1979). The addition of reelin in vitro has also been shown to increase dendritic growth of hippocampal neurons (Jossin and Goffinet, 2007; Matsuki et al., 2008). In the developing neocortex, reelin promotes dendritic growth of pyramidal cells in early embryonic brain development (Chai et al., 2015; Kohno et al., 2015; Kupferman et al., 2014; Nichols and Olson, 2010; O'Dell et al., 2015). On the other hand, reelin restricts dendritic growth of cortical pyramidal neurons postnatally (Chameau et al., 2009). Moreover, quantitative analysis revealed that forebrain interneurons in reeler mice are hypertrophic with longer dendritic branches when compared with wild type (Yabut et al., 2007). Thus, reelin appears to exert opposing effects on dendritic growth when embryonic and postnatal brain development are compared.
In the early postnatal and adult brain, the majority of GABAergic interneurons in the neocortex expresses reelin (Pohlkamp et al., 2014). Gamma-aminobutyric acid (GABA) is the main inhibitory neurotransmitter in the mammalian brain and plays a key role in modulating neuronal activity. GABAB receptors (GABABRs) are guanine nucleotide-binding protein (G protein)-coupled metabotropic receptors that modulate Ca2+ and potassium (K+) channels, and elicit both presynaptic and slow postsynaptic inhibition (Benarroch, 2012; Bettler et al., 2004). Presynaptic GABABRs are coupled to Ca2+ channels, regulating the release of neurotransmitters, while postsynaptic GABABRs are coupled to K+ inward rectifying (Kir) channels (Kir3), regulating postsynaptic slow inhibition (Pinard et al., 2010). GABABR1 and GABABR2 subunits were detected in embryonic and postnatal neocortical neurons (López-Bendito et al., 2002). Recently, we have shown that reelin controls early network activity by modulating presynaptic GABABR function. Thus, in the absence of early postnatal reelin expression GABABRs were not functional (Hamad et al., 2021). Conditional reelin deficiency induced in adult mice resulted in enhanced LTP; however, no alterations in dendritic spine density or morphology and no perturbation of cortical layering were observed (Lane-Donovan et al., 2015). Moreover, adult conditionally induced RelncKO specifically in interneurons showed that reelin is neither required to establish nor to maintain cortical layers at that stage (Pahle et al., 2020). The question of the extent to which reelin deficiency affects early postnatal dendritic growth of interneurons remains to be answered. To address this, we investigated the effect of early postnatally induced reelin deficiency on dendritic development of neocortical interneurons of RelncKO mice. We employed the organotypic tissue culture (OTC) system, in which cortical neurons displayed electrophysiological characteristics similar to those seen in vivo (Hamad et al., 2015; Klostermann and Wahle, 1999). We have observed recently that early postnatal RelncKO interneurons exhibited an excessive Ca2+ spike frequency due to a dysfunction of GABABRs (Hamad et al., 2021). As it is evident that neurotransmission-mediated elevation in intracellular Ca2+ levels ([Ca2+]i) plays a major role in dendritic growth, we hypothesized that early postnatal reelin deficiency could lead to dendritic hypertrophy. Our present data show that reelin restricts interneuron dendritic growth to ensure proper development of dendritic architecture.
Interneurons exhibit hypertrophic dendritic growth after conditionally induced reelin deficiency
To assess the role of reelin in interneuronal development, it is necessary to bypass the cortical layer malformations that are seen in the reeler mutant. Therefore, we employed the RelncKO mouse line in this study. This mouse line ubiquitously expresses a fusion protein composed of Cre recombinase and a mutated form of the estrogen receptor (Cre-ERT2), and (Z)-4-hydroxytamoxifen administration induces nuclear Cre activity and knockout of the floxed reelin gene. Therefore, OTCs were prepared from newborn postnatal day 0 (P0) mice and cultured in roller tubes for 10 days. To conditionally induce reelin deficiency, 1 µM (Z)-4-hydroxytamoxifen (4-OHT) was added to OTC from the first day in vitro (1 DIV) for four 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 (Fig. 1A). We have previously shown that postnatal 4-OHT administration induces nuclear Cre activation and knockout of the floxed reelin gene, and that cortical layering is not altered after postnatal loss of reelin in RelncKO mice (Hamad et al., 2021). After birth, reelin is mainly produced by a subset of GABAergic interneurons in the neocortex and hippocampus (Alcántara et al., 1998; Pesold et al., 1998). The specific distribution of reelin-positive interneuron subtypes in the neocortex has been characterized previously (Pohlkamp et al., 2014). That study showed that 63% of NPY+, 50% of Parv+ and 35% of Calr+ interneurons express reelin, while it is expressed by a smaller percentage of interneurons that are characterized by other markers. To explore the role of reelin in dendritic cell growth in these three major interneuron subtypes, we employed genetic constructs with short promotor sequences to drive fluorescent protein expression in specific types of mammalian cortical inhibitory neurons using adeno-associated virus (AAV) vectors (Nathanson et al., 2009). OTCs were transfected at 5 DIV with pAAV-fNPY-GFP, pAAV-fPV-GFP or pAAV-fCR-GFP separately, fixed at 10 DIV and reconstructed for morphometrical quantifications (Fig. 1A, for schematic representation). Our quantitative morphometric analysis of NPY+ interneurons revealed that the length of dendrites and their segments was significantly increased in RelncKO when compared with wild type (Fig. 1B,C), whereas the number of primary dendrites and the soma size were unaltered (Fig. 1D,E). To identify the dendritic segments with differing complexity between RelncKO and wild type, we employed Sholl analysis of dendrites. This analysis revealed a significant increase in dendritic complexity in the distal dendritic segments of RelncKO interneurons (Fig. 1F). Furthermore, the total number of dendritic intersections was significantly increased in RelncKO interneurons (Fig. 1G). Further analyses of Calr+ interneurons revealed the same results as for NPY+ interneurons. Summarized, we found that dendritic length and segments were significantly increased in RelncKO with no change in the number of primary dendrites and soma size (Fig. 2A-D). Similarly, Sholl analysis of Calr+ interneurons revealed a significant increase in dendritic complexity in distal dendritic segments of RelncKO interneurons (Fig. 2E) and the total number of dendritic intersections was significantly increased in RelncKO interneurons (Fig. 2F). Parv+ interneurons had significantly longer dendrites and an increased number of segments in the RelncKO interneurons in comparison with wild-type interneurons (Fig. 3A,B), with no change in the number of primary dendrites and soma size (Fig. 3C,D). However, Sholl analyses showed that, in RelncKO interneurons, the number of proximal rather than distal dendritic intersections was increased (Fig. 3E) and that the total number of dendritic intersections was significantly higher in RelncKO interneurons (Fig. 3F). Taken together, these results suggest that the major reelin expressing interneuron subtypes exhibit dendritic hypertrophy following early postnatal reelin elimination. Because all the interneuron subtypes examined here showed similar hypertrophy in response to reelin deficiency, we decided to transfect OTCs only with EGFP for morphometrical analyses. To distinguish interneurons from pyramidal cells, we used well-established morphological criteria that easily distinguish interneurons based on their dendritic and axonal patterns (see Materials and Methods section).
Recombinant reelin rescues hypertrophic dendritic growth of RelncKO interneurons
Next, we investigated whether recombinant reelin might rescue abnormal dendritic growth observed in RelncKO OTCs. To address this question, we treated wild-type and RelncKO OTCs daily from 3 to 9 DIV with recombinant reelin (Fig. 4). The OTCs were transfected at 5 DIV, fixed and stained at 10 DIV for morphometrical analysis. Identified interneurons from each OTC were reconstructed. Our data show that dendritic length and segments were significantly smaller in reelin-treated RelncKO OTCs in comparison with control RelncKO OTCs (Fig. 4A,B). The number of primary dendrites remained unaltered (Fig. 4C), suggesting that recombinant reelin was able to restore dendritic growth of interneurons in RelncKO OTCs to a basic level comparable with the wild-type control group.
No difference in GABA excitation/inhibition shift between wild-type and RelncKO OTCs
In neocortex, the GABA excitation/inhibition shift occurs between the first and second postnatal developmental week (for a review, see Ben-Ari et al., 2007). The timing of the GABA switch can be monitored in developing OTCs by measuring the occurrence and amplitude of GABA-induced Ca2+ responses. To disclose any temporal difference in the occurrence of the GABA switch between wild-type and RelncKO OTCs, we evaluated the percentage of neurons showing GABA-induced Ca2+ transients and the amplitude of such responses after stimulation with 100 µM GABA (Fig. S1A,B) between 2 and 7 DIV. We observed an excitatory GABA action mainly from 2 to 5 DIV and a pronounced inhibitory GABA action starting from 6 DIV onwards (Fig. S1). Thus, the occurrence of GABA excitation/inhibition in our OTCs system supports the existence of depolarizing GABA in early network maturation reported in numerous other studies (Ben-Ari et al., 2012). At all-time points, we found no statistical difference among groups. Moreover, in wild-type and RelncKO OTCs, the excitatory GABA activity disappeared around 6 DIV. We expected a delay in the GABA shift in RelncKO in comparison with wild-type OTCs because the GABABRs were dysfunctional. However, this did not occur, probably because the developmental GABA shift depends on functional expression of other receptors, such as KCC2 and GABAA receptors (Ben-Ari, 2002), which in RelncKO OTCs may still be functional.
Aberrant Ca2+ spike frequency in early postnatally reelin-deficient RelncKO interneurons
To explore possible causes that might underlie the hypertrophic dendritic morphology of RelncKO interneurons, we decided first to measure Ca2+ signal activity because Ca2+ signals are known to exert an important influence on neuronal morphology by regulating the growth and branching of dendrites (Konur and Ghosh, 2005). Therefore, OTCs were transfected at 5 DIV with a genetic construct encoding the Ca2+ indicator GCaMP6s (Chen et al., 2013). GCaMP6s is very sensitive to Ca2+ changes and we found it to be distributed in soma, dendrites and axons of transfected neurons (Fig. 5A,B). At 10 DIV, we recorded Ca2+ signal amplitude, frequency and Ca2+ transient half-width in wild-type and RelncKO interneurons. Our results showed an unchanged amplitude in RelncKO compared with wild-type interneurons (Fig. 5D) and revealed an enhanced frequency and Ca2+ transient half-width in the RelncKO interneurons (Fig. 5E,F), suggesting that RelncKO interneurons exhibit defects at the presynaptic level. GABABRs are well-known to control Ca2+ release at the presynaptic site, and recently we have shown that GABABRs are involved in postnatal reelin signaling (Hamad et al., 2021). To confirm that the excessive Ca2+ activity in the RelncKO is due to the dysfunction of presynaptic GABABRs, we treated wild-type and RelncKO with the GABABRs antagonist CGP. Strikingly, the application of CGP (10 µM) did not influence Ca2+ amplitude either in wild-type or RelncKO interneurons (Fig. 5D). However, blockade of GABABRs in wild-type OTCs dramatically increased the Ca2+ signaling frequency and Ca2+ transient half-width in wild type, but not in RelncKO interneurons (Fig. 5E,F). Taken together, these results indicate a presynaptic dysfunction of GABABRs in glutamatergic, but not GABAergic, terminals in the RelncKO interneurons.
Hypertrophy of RelncKO interneurons is due to presynaptic GABABR dysfunction
GABABRs are present in GABAergic neuronal terminals (as autoreceptors), and in glutamatergic and other terminals (heteroreceptors) (Benarroch, 2012). Previously, we have shown that presynaptic GABABRs at the glutamatergic but not GABAergic neuronal terminals were defective in RelncKO mice and the postsynaptic site was not affected as no significant change in the Ca2+ amplitude was observed (Hamad et al., 2021). The main function of presynaptic GABABRs is to block the N-type (Cav2.2) Ca2+ channels, which in turn control neurotransmitter release (Benarroch, 2012). Thus, the increased Ca2+ frequency in RelncKO interneurons should be responsible for the observed hypertrophic growth. To test this hypothesis, we reconstructed interneurons from EGFP-transfected wild-type and RelncKO OTCs in presence of the GABABRs antagonist CGP. 10 µM CGP was applied daily to the OTCs. Our quantitative morphometric analysis of selected interneurons revealed that CGP increased dendritic length and segments in wild-type but not in RelncKO interneurons (Fig. 6A,B), whereas the number of primary dendrites was unaltered (Fig. 6C). Moreover, using Sholl analysis, we found that CGP increased both distal and proximal dendritic intersections of wild-type but not of RelncKO interneurons (Fig. 6D). Indeed, CGP increased the total number of dendritic intersections of wild type but did not alter that of RelncKO interneurons (Fig. 6E). Together, these results suggests that dendritic hypertrophy seen in the RelncKO is caused by malfunctioning of presynaptic GABABRs. Therefore, we assumed the hypertrophic growth of RelncKO interneurons might be due to the lack of GABABRs-mediated inhibition of the N-type (Cav2.2) Ca2+ channels at the presynaptic site. To test this, OTCs were transfected at 5 DIV with EGFP and the N-type (Cav2.2) Ca2+ inhibitor conotoxin (2 µM) was applied to OTCs from 5 to 10 DIV. At 10 DIV, the OTCs were fixed, stained and subjected to morphometric analysis. Our quantitative morphometric analysis of selected interneurons revealed that conotoxin did not affect either dendritic length or dendritic segments of wild-type interneurons; however, it reduced dendritic length and segments in RelncKO interneurons (Fig. 7A,B). The numbers of primary dendrites were the same in all treatment groups (Fig. 7C). In addition, Sholl analysis showed that conotoxin reduced the number of dendritic intersections of RelncKO but not wild-type interneurons (Fig. 7D,E). These findings suggest that reelin restricts dendritic growth by limiting neurotransmitter release, a mechanism that is controlled through GABABR-mediated inhibition of N-type (Cav2.2) Ca2+ channels.
AMPAR, but not NMDAR, blockade rescues dendritic hypertrophy of RelncKO interneurons
We have shown that the enhanced Ca2+ frequency in recorded interneurons resulted in hypertrophic dendritic growth in RelncKO interneurons. Glutamate-mediated transmission converging on Ca2+ signaling plays a key role in enhancing dendritic growth. Thus, the excessive Ca2+ influx through the main glutamate receptors (AMPARs or NMDARs) might be the reason for dendritic hypertrophy in the RelncKO interneurons. To test this, OTCs were transfected at 5 DIV with EGFP and the NMDARs antagonist APV (50 µM) was applied to OTCs from 5 to 10 DIV. At 10 DIV, the OTCs were fixed, stained and subjected for morphometrical analyses. Our quantitative morphometric analysis of selected interneurons revealed that the APV did not alter dendritic growth of either wild-type or RelncKO interneurons (Fig. S2), suggesting that NMDARs are not involved in dendritic hypertrophy seen in RelncKO interneurons. However, when we tested the AMPAR antagonist GYKI (100 µM), we observed a reduction in dendritic length and in the number of segments, but not in the number of primary dendrites of wild-type and RelncKO interneurons (Fig. 8A-C). Expectedly, the number of dendritic intersections was reduced in proximal and distal dendrites of wild-type and RelncKO interneurons (Fig. 8D,E). These findings suggest that the hypertrophic growth effect seen in RelncKO interneurons might be mediated by excessive Ca2+ influx through AMPARs.
In the current study, we revealed a previously unrecognized role of reelin in early postnatal interneuron dendritic development. We found that postnatally induced reelin deficiency caused abnormal dendritic growth of interneurons. Chronic recombinant reelin treatment rescued dendritic hypertrophy in Relncko interneurons. By testing the major neocortical interneuron subtypes that express reelin, we found that NPY+ and Calr+ interneurons displayed dendritic hypertrophy at the distal dendritic segments, whereas the Parv+ interneurons displayed hypertrophic dendritic growth of proximal segments of RelncKO interneurons, suggesting that reelin regulates dendritic growth differentially in various classes of inhibitory interneurons. In contrast to early embryonic stages, when reelin was found to promote dendritic development, our results show that postnatal reelin secretion acts as a ‘stop-growth-signal’ for developing interneurons. It has been suggested that hypertrophy of interneurons observed in the postnatal reeler mutant cortex might be a secondary effect caused by the lamination defects in the reeler neocortex (Chameau et al., 2009), which implies that interneurons differentiate in a different microenvironment when compared with wild type. We have previously demonstrated by immunohistochemical staining with antibodies against layer-specific markers, that cortical neurons in postnatal RelncKO did not alter their characteristic layer-specific marker expression in the absence of reelin (Hamad et al., 2021). Combined with our present results, we conclude that interneuron hypertrophy, which had been seen previously in the reeler mutant, is not a secondary effect of malpositioning of neurons during early embryonic development in reeler, but rather due to the lack of a function for the reelin that is postnatally expressed by wild-type interneurons.
Next, we investigated possible mechanisms affected by postnatal reelin deficiency that might result in hypertrophic dendritic growth of interneurons. We have recently demonstrated that reelin controls early network activity by modulating presynaptic GABABR function (Hamad et al., 2021). As Ca2+ is a key molecule in the regulation of dendritic growth, we examined in this study Ca2+ amplitude and frequency in RelncKO interneurons in the presence of the GABABR antagonist CGP. Our data show that RelncKO interneurons displayed an excessive Ca2+ spike frequency. Presynaptic GABABRs are coupled to N-type (Cav2.2) Ca2+ channels (Pinard et al., 2010). GABABRs inhibit N-type (Cav2.2) Ca2+ channels, which in turn decrease neurotransmitter release (Benarroch, 2012). GABABR inhibition of N-type (Cav2.2) Ca2+ channels was absent in RelncKO interneurons because the N-type (Cav2.2) Ca2+ channels blocker conotoxin was able to restore length of dendrites and dendritic segments of RelncKO interneurons to levels comparable with wild type. Thus, it is likely that increased Ca2+ frequency in RelncKO interneurons is responsible for the hypertrophic dendritic growth of RelncKO interneurons. Finally, we could demonstrate that hypertrophic growth of RelncKO interneurons is mediated by excessive Ca2+ influx through AMPARs, but not NMDARs. Accordingly, the AMPAR GluR1 subunit has been shown to promote dendritic growth of neocortical interneurons (Hamad et al., 2011). Furthermore, surface expression of AMPARs was increased in a knock-in mouse model in which the wild-type GABABR was replaced with a S783A-mutated version that cannot be phosphorylated (Terunuma et al., 2014). That NMDARs are not involved in reelin-mediated interneuron dendritic growth is expected, as recent data have shown that NMDARs do not affect the neocortical dendritic growth of interneurons (Gonda et al., 2020). During embryonic development, GABA is the main neurotransmitter, and it is not hyperpolarizing at this stage, as GABABR lacks coupling between G proteins and potassium channels until the end of the first postnatal week (Fukuda et al., 1993). The non-hyperpolarizing GABABR activation in early embryonic development has been shown to promote dendritic growth (Bony et al., 2013). However, when GABABR shifts to hyperpolarization after birth, GABABRs are able to couple to G proteins and potassium channels. We have previously shown that GABABR intracellular signaling via Gαi/o proteins is impaired in RelncKO mice (Hamad et al., 2021). Thus, in the absence of GABABR-mediated inhibition in RelncKO, the resulting excessive Ca2+ might cause abnormal dendritic growth.
In general, the most prominent function of reelin in early development is guiding migration of newborn neurons and control of proper cortical layer development (for a review, see Förster et al., 2006; Jossin, 2020; Lee and D'Arcangelo, 2016). The role of reelin in dendritic development is a dualistic one, depending on the developmental stage. Thus, during embryonic development, (1) reelin has been shown to promote growth of embryonic pyramidal cells in dissociated hippocampal cell cultures of reeler mutant mice (Niu et al., 2004), and (2) to increase dendritic growth of hippocampal neurons (Jossin and Goffinet, 2007; Matsuki et al., 2008). (3) Neocortical pyramidal cell apical dendrites failed to extend and to contact the marginal zone when reelin signaling was suppressed (Olson et al., 2006), and (4) reorganized their dendrites in a tangential orientation to the marginal zone (O'Dell et al., 2015). (5) Reelin induced branching of the leading processes of migrating neurons and that of the apical processes of radial glial cells (Chai et al., 2015). (6) Deletion of the reelin C-terminal region around E14.5 resulted in a reduction of pyramidal cells apical dendritic length around P7 (Kohno et al., 2015). (7) Reelin was found to specify the molecular identity of the pyramidal neuron distal dendritic compartment (Kupferman et al., 2014). In turn, opposite effects were observed during postnatal development: (1) the N-terminal region of reelin restricted dendritic growth of neocortical apical pyramidal neurons (Chameau et al., 2009); and (2) morphometric analysis of interneurons in the adult reeler neocortex and hippocampus revealed hypertrophic growth with longer dendritic branches in reeler when compared with wild-type interneurons (Yabut et al., 2007). (3) Newly generated neurons in the adult hippocampus respond to reelin overexpression with faster development of the dendritic tree within the first 2 weeks, whereas dendritic complexity was reduced after 8 weeks (Teixeira et al., 2012). Taken together, these data suggest that reelin has opposing effects on morphological neuronal maturation during embryonic when compared with postnatal dendritic development.
The discrepancy between the roles of reelin during embryonic versus postnatal development might be partially explained by differential effects of the different proteolytic fragments of reelin. At early embryonic stages, reelin is secreted by Cajal-Retzius (CR) cells, which disappear around P14 by undergoing selective cell death through apoptosis (Anstötz et al., 2014). After its secretion from CR cells, reelin cleavage produces at least five fragments that could be detected using antibodies against N-terminal, central and C-terminal reelin epitopes (Jossin et al., 2004, 2007). The secreted reelin fragments containing the C-terminal fragment, including full-length reelin, appear to remain localized in the marginal zone (Jossin et al., 2007). Because the C-terminal of reelin is important for promoting dendritic growth (Kohno et al., 2015), pyramidal cell apical dendrites, localized in the vicinity of the marginal zone, benefit from locally deposited C-terminal fragments to promote their dendritic growth. With the decline in the number of CR cells during early postnatal development, interneurons (mostly in layers II/III) become the dominant cell type that expresses reelin in the neocortex (Pohlkamp et al., 2014). It has been shown that the cleaved N-terminal fragment diffuses, is transported to distant regions in layers II/III (Koie et al., 2014) and acts to oppose the C-terminal fragment, i.e. as a stop-growth-signal for dendritic growth (Chameau et al., 2009). Thus, the observed discrepancies of the reelin effects might be due to the differential action of the various reelin proteolytic fragments.
Proper growth and arborization of dendrites is crucial for correct functioning of the nervous system. Although research in the past two decades has markedly increased our general understanding of the cellular and molecular mechanisms of dendritic growth, so far, relatively little is known about the signals that limit dendritic growth. While many signaling molecules have been implicated in promoting dendritogenesis, only few molecules have been identified as dendritic growth-limiting factors. For example, in mouse cerebellar organotypic slice cultures, the activation of class I metabotropic glutamate receptors (mGluR1) led to a very strong inhibition of dendritic growth, resulting in Purkinje cells with very small stubby dendrites (Sirzen-Zelenskaya et al., 2006). Furthermore, in optic tectum neurons, the expression of a specific peptide inhibiting Ca2+/calmodulin-dependent protein kinase II activity blocked the growth restriction that normally occurs during maturation of tectal cell dendritic trees (Wu and Cline, 1998; Zou and Cline, 1999). A previous study on the reeler mutant neocortex (Yabut et al., 2007) and our present data obtained from the RelncKO neocortex have shown that neocortical interneurons are hypertrophic in absence of reelin. Moreover, reelin is involved in the development of vertical columnar structures in the mouse presubicular cortex, where reelin may act as a stop signal for growth and branching of postnatal pyramidal cell apical dendrites (Nishikawa et al., 2002). Neocortical interneuron growth defects can lead to pathological hyperexcitability of neuronal circuits, which in turn can contribute to disorders such as epilepsy, schizophrenia or bipolar disorder (Hu et al., 2017). Moreover, the increased interneuron growth caused by reelin deficiency might affect the function of neuronal circuits in autism (Courchesne and Pierce, 2005). Moreover, genetic studies have shown that the reelin is associated with a number of psychiatric diseases, including schizophrenia, bipolar disorder and autistic spectrum disorder (Ishii et al., 2016). Thus, dendritic overgrowth caused by the absence of reelin appears to be causally linked to neurological disorders. Despite our detailed knowledge of the molecular events in dendritic development, the specific molecular factors that act as a ‘stop-growth-signal’ are just beginning to be investigated. ‘Stop-growth-signals’ might be required to transform dynamic, immature dendrites into more stable, mature dendrites and to establish a balanced neuronal network connectivity.
MATERIALS AND METHODS
All mouse experiments were reviewed and approved by the local ethic commission. A license for animal experiments has been obtained from the State Agency for Nature, Environment and Consumer Protection in North Rhine-Westphalia with permission number 84-02.04.2016.A383. The guidelines of the German Animal Welfare Act were respected according to the Federal German law.
Reelin conditional knockout mice (RelncKO)
Animals were housed in a standard 12 h light cycle and fed ad libitum with standard mouse chow. The generation of the conditional RelncKO line has been previously described (Lane-Donovan et al., 2015, 2016). To obtain conditional reelin knockout mice (Relnflox/flox CAG-CreERT2 mice), we crossed Relnflox/flox mice with hemizygous tamoxifen-inducible Cre recombinase-expressing mice (CAG-CreERT2) (Hayashi and McMahon, 2002). For the experiments, only Relnflox/flox CAG-CreERT2 male mice were selected and then crossed with Relnflox/flox female mice to generate Relnflox/flox wildtype (wild type) and Relnflox/flox CAG-CreERT2 (RelncKO) siblings as verified by PCR. The cKO mouse line ubiquitously expresses a fusion protein comprising Cre recombinase and a mutated form of the estrogen receptor (Cre-ERT2). The RelncKO mice require a material transfer agreement from the University of Texas Southwestern Medical Center. No exclusion criteria were pre-determined and the study was exploratory. No blinding or randomization was performed to allocate subjects in the study. No sample size calculation was performed in this study.
All samples were stored at −20°C until PCR analysis. DNA from samples of ear, tail and brain tissue were isolated with ReliaPrep gDNA kit (A205, Promega). All procedures were performed according to protocols provided by the manufacturer. 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); Relnflox/flox mice, forward primer 5′-GCTCTGGCCAAGCTTTATC-3′ and reverse primer 5′-CGCGATCGATAACTTCGTATAGCATAC-3′ (1200 bp); for detection of CAG-CreERT2, forward primer 5′-ATTGCTGTCACTTGGTCGTGG-3′, reverse primer 5′-GGAAAATGCTTCTGTCCGTTTGC-3′ (200 bp). The amplification products were verified on a 2% agarose gel in TBE buffer.
Organotypic cultures and pharmacological treatment
OTCs were prepared from newborn postnatal day 0 (P0) mice. All solutions used for OTC preparation were sterile and all preparations were performed in a laminar air flow bench with horizontal counter flow (Horizontal Flow, ICN, Biomedicals). The mice were briefly anesthetized and decapitated. The skull was removed gently, and the cortex was placed on the chopper plate (McIllwain). Somatosensory cortex was cut into 350 μm slices, and the slices were gently transferred into ice-cold buffered salt solution (GBSS, 24020117, Gibco) containing 25 mM D-glucose. After 30 min of recovery, the slices were transferred onto coverslips (12×24 mm, Kindler). A mixture of chicken plasma (Sigma) and GBBS/thrombin (Merck) was mixed in a 2:1 ratio and then allowed to coagulate for 45 min. The coverslip was transferred into a roller tube (Nunc) filled with 750 μl semiartificial medium and placed in a roller incubator at 37°C. To induce the knockout, 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. A graphical flow chart for OTCs experimental procedures is shown in Fig. 1A. For the quantitative morphological analysis experiments, the following drugs were used: CGP35348 (CGP, 10 µM, 1245/10, Tocris), APV (50 µM, 0190/10, Tocris), GYKI 52466 (GYKI, 100 µM, 1454, Tocris) and ω-Conotoxin GVIA (conotoxin, 2 µM, 1085, Tocris).
Preparation of recombinant reelin
For preparation of reelin-containing supernatants and control supernatants, HEK 293-cells transfected with plasmid pcDNA3 containing full-length reelin cDNA and stably expressing reelin (D'Arcangelo et al., 1997; Förster et al., 2002), and control HEK 293-cells transfected with a plasmid encoding green fluorescent protein (GFP) were grown in DMEM (Dulbecco's modified eagle medium, low glucose, Invitrogen) with 10% fetal calf serum (FCS, Invitrogen), 1% penicillin-streptomycin (P/S, 10,000 U/ml, Invitrogen) and 0.9 g/l G418 (Geneticin, Invitrogen) for 2 days to reach confluence. The serum-containing medium was then replaced by serum-free medium, and the cells were incubated for 2 days at 37°C, 5% CO2. Conditioned medium of reelin-synthesizing 293-cells and the control medium of GFP-expressing 293-cells were collected and centrifuged at a speed of 4000 g for 5 min to remove dead cells (Chai et al., 2009). Supernatants were concentrated 20× with Amicon 100 centrifugal filter devices. For experimental application, the concentrated supernatant was diluted in culture medium to yield a 1× final concentration and reelin content (or absence of reelin in control-cell supernatants) was confirmed by western blotting using the reelin-specific monoclonal antibody G10. To estimate reelin concentration, cumulative density of the known reelin bands in the purified sample (∼70% purity) was determined and then compared with protein standards on the same Coomassie-stained gel prior to western blotting. A standard concentration of ∼5 nM was used, because it corresponds to ∼10× the Kd of reelin binding to its receptors (∼500 pM), thus ensuring maximal activation of the reelin pathway (see Hiesberger et al., 1999; Weeber et al., 2002).
Biolistic transfection and expression plasmids
Transfection was performed using a Helios Gene Gun (Bio-Rad) as described previously (Hamad et al., 2020; Wirth and Wahle, 2003). In brief, cartridges were prepared by coating 10 mg gold particles (Ø=1 µm; Bio-Rad) with genetically encoded Ca2+ indicator pGP-CMV-GCaMP6 s (GCaMP6 s) for Ca2+ imaging experiment. The pGP-CMV-GCaMP6s was a gift from Douglas Kim & GENIE Project (Addgene plasmid #40753) (Chen et al., 2013). For morphological quantification of interneurons, we prepared gold particles containing plasmid encoding enhanced green fluorescent protein (pEGFP-N1; Clontech). For morphological quantification of selective expression inhibitory neuron subtypes, we used the following plasmids separately: pAAV-fCR-GFP (Addgene plasmid #22911), pAAV-fNPY-GFP (Addgene plasmid #22912) and pAAV-fPV-GFP (Addgene plasmid #22914). All three pAAV constructs were a gift from Edward Callaway (Nathanson et al., 2009). 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. The blockers were washed out 3 h after transfection.
At 10 DIV, OTCs were fixed with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4) and warmed to 36°C for 1 h. After washing twice in Tris-buffered saline [TBS; 50 mM Tris and 150 mM NaCl (pH 7.6)] and permeabilization in TBST (TBS, 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 goat anti-mouse biotinylated secondary antibody was added (1:300; E043201-8, Dako). After several washes in TBS buffer, the slices were subjected to the ABC-horseradish peroxidase method using diaminobenzidine as a chromogen.
3D neuron reconstruction
EGFP-immunostained cells were reconstructed with the Neurolucida system (MicroBrightField) at 1000× magnification. Because we used the short promoter sequences to drive fluorescent protein expression in specific types of cortical inhibitory interneurons (transfected with pAAV constructs), there was no need for distinction criteria. For all interneurons subtypes (transfected with EGFP), we used well-established criteria that easily distinguish interneurons from pyramidal cells based on their dendritic and axonal pattern (Hamad et al., 2014; Karube et al., 2004; Kawaguchi et al., 2006). In brief, interneurons were multipolar with sparsely spiny dendrites and axons branching within the dendritic tree into horizontal or columnar arbors. For interneuron dendrites, 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), number of primary dendrites and soma surface area. 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).
Ca2+ imaging using spinning disc laser microscopy
The OTCs were transfected with the Ca2+ sensor GCaMP6s and expression allowed to take place for 5 days. The slices were then transferred to a recording chamber mounted on a fixed stage of an inverted microscope and perfused with artificial cerebrospinal fluid [ACSF; 125 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgSO4, 25 mM NaHCO3, 1.25 mM NaH2PO4, 25 mM glucose (pH 7.4)] (3-5 ml/min) at 32±2°C. Fluorometric Ca2+ recordings were made using a Visiscope spinning-disc confocal system CSU-W1 (Visitron) featuring a spinning disk unit CSU-W1-T2 and an 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 three 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 (background-corrected increase in fluorescence divided by the resting fluorescence). Raw data delivered in the form of a linear 16-bit intensity scale were plotted as fluorescence intensity versus time. Pyramidal cell or interneuron 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. Spike half-width was calculated as the width of the spike at half-maximal amplitude (Weir et al., 2014). For the GABA shift experiment, Ca2+ imaging was performed with the Ca2+ indicator Oregon Green BAPTA-1 Acetoxymethyl (OGB-1 AM) (O6807, Molecular Probes). OTCs were loaded at different developmental time window (2-7 DIV) according to our previously published protocol (Hamad et al., 2015). After loading, the slices were washed several times and fluorometric recording were performed. The fluorescence ΔF/F0 was used to express Ca2+ concentrations. This parameter was recorded in areas of interest (AOIs) corresponding to neuronal cell bodies, and analyzed along sequential images to follow temporal changes. Changes over baseline higher than 0.05 ΔF/F0 in response to 100 µM GABA (0344, Tocris) were considered depolarizing events.
Statistical analysis was performed with Sigma Stat 12 (SPSS Incorporated). Comparisons between two groups were performed with Students’ unpaired t-test when the normality test (Shapiro-Wilk) was passed, otherwise with a 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 they passed the normality test. 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 Edward Callaway for providing all the pAAV constructs. We thank also Dr Douglas Kim for providing the pGP-CMV-GCaMP6s construct.
Conceptualization: M.I.K.H., J.H., E.F.; Methodology: M.I.K.H., P.P., S.D., O.R., A.J., N.M., J.L., I.J.; Validation: M.I.K.H., N.M., E.F.; Formal analysis: M.I.K.H., P.P., S.D., O.R., A.J., J.L., I.J.; Investigation: M.I.K.H., P.P., S.D., O.R., A.J., N.M., J.H., E.F.; Resources: J.H.; Data curation: M.I.H., E.F.; Writing - original draft: M.I.K.H.; Writing - review & editing: M.I.K.H., I.J., G.R., J.H., E.F.; Supervision: G.R., J.H., E.F.; Project administration: E.F.; Funding acquisition: J.H., E.F.
This work was supported by the National Institutes of Health (1RF1 AG053391-01, NHLBI 5R37 HL063762-20, NINDS/NIA 5R01 NS093382-05 and NS108115-02), the BrightFocus Foundation (A20135245 and A2016396S), the Harrington Discovery Institute, a Circle of Friends Pilot Synergy grant and the Bluefield Project to Cure FTD to J.H. E.F. was supported by FoRUM of the Ruhr-Universität Bochum. Deposited in PMC for release after 12 months.
Peer review history
The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199718
The authors declare no competing or financial interests.