Thalamic afferents influence cortical progenitors via ephrin A5-EphA4 interactions

The excitatory projection neurons of the different cortical layers are generated in a temporal fashion, with deep layer neurons born first, while upper layer neurons are produced later during corticogenesis, regulated by a complex network of intrinsic and extrinsic mechanisms (Franco and Muller, 2013).

Two main types of cortical progenitors are described. Radial glial cells (RGCs) characterized by Pax6 expression and a defined apicobasal polarity reside in the ventricular zone (VZ) and undergo asymmetric divisions at the onset of neurogenesis, generating neurons directly or predominantly indirectly by producing a second type of precursor termed the intermediate progenitor cells (IPCs) (Chenn and McConnell, 1995; Rao et al., 2000; Kriegstein and Gotz, 2003; Gotz and Huttner, 2005). Tbr2-positive IPCs translocate to the subventricular zone (SVZ) and undergo symmetric divisions, producing either pairs of neurons or self-amplifying prior to neurogenesis, contributing to all cortical layers (Haubensak et al., 2004; Noctor et al., 2004, 2008; Englund et al., 2005; Attardo et al., 2008; Farkas et al., 2008; Sessa et al., 2008; Kowalczyk et al., 2009; Vasistha et al., 2014).

Endogenous regulation of cortical neurogenesis by transcription factors and morphogens, feedback signaling by postmitotic neurons and the regulation of neuronal progenitors by meninges as well as factors of the liquor have all been described (Siegenthaler and Miller, 2008; Seuntjens et al., 2009; Lehtinen and Walsh, 2011; Tiberi et al., 2012; Parthasarathy et al., 2014; Toma et al., 2014). Moreover, paracrine interactions of membrane-bound molecules, such as the Eph/ephrin system, affect the proliferation and differentiation of cortical progenitors (Qiu et al., 2008; North et al., 2009).

Eph-ephrin interactions regulate a variety of neurodevelopmental processes, including axonal guidance, migration and apoptosis, in addition to proliferation (Drescher et al., 1997; Flanagan and Vanderhaeghen, 1998; Depaepe et al., 2005; Zimmer et al., 2008; North et al., 2009; Rudolph et al., 2010), and are crucial for establishing thalamocortical projections (Donoghue and Rakic, 1999; Šestan et al., 2001; Bolz et al., 2004).

Despite intensive discussion as to whether cortical development is mainly regulated by intrinsic and/or extrinsic cues (Rakic, 1988,, 1991; Dehay et al., 2001; Rakic et al., 2009; Zhou et al., 2010; Reillo et al., 2011), the thalamic influence on cortical progenitors and neurogenesis remains debated. Here, we provide further evidence for the extra-cortical regulation of cortical progenitors and hence the final output of neurons by thalamic afferents via the Eph/ephrin system.

Ephrin A5 is expressed in cortical tissue at embryonic as well as postnatal stages (Fig. 1A; supplementary material Fig. S1A). Here, we aimed to address the role of ephrin A5 during cortical neurogenesis and in neocortical organization in adult mice.

We first analyzed radial extension of the cortical layers of the somatosensory cortex of adult ephrin A5-deficient and wild-type mice at postnatal day (P) 37. We performed DAPI staining of coronal brain sections in combination with immunolabeling, using antibodies directed against Cux2 (Nieto et al., 2004; Zimmer et al., 2004) and Otx1/2 (Frantz and McConnell, 1996) to distinguish between the superficial and infragranular layers, respectively. This analysis revealed a significant increase in the radial extension of deep layers V and VI in ephrin A5 knockout animals, compared with wild types, at the expense of radial extension of superficial layers II-IV at P37 (Fig. 1C-E). The overall cortical thickness remained unchanged (supplementary material Fig. S1B). The neuronal density of the different layers, as determined by the number of NeuN (Rbfox3)-labeled neurons among DAPI-stained cells, revealed no significant alterations in the ephrin A5-deficient cortices (supplementary material Fig. S1C-E).

To determine if the phenotype is caused by an accumulation of neurons fated for the supragranular layers in deep layers, BrdU birthdating experiments were performed. BrdU administration at E16.5, when supragranular layer neurons are born, did not reveal ectopically located BrdU-positive neurons in the deep layers of the adult cortex (supplementary material Fig. S1F-H). However, consistent with the increased thickness of the deep layers at the expense of the upper layers, the E16.5 BrdU-labeled cells were shifted towards the pial surface in adults (supplementary material Fig. S1F-H).

Interestingly, the radial extension of the deep layers at the expense of the upper cortical layers observed in ephrin A5 mutants was phenocopied by Epha4 knockout mice (Fig. 1E). This suggests that EphA4 is the mediating receptor and that ephrin A5-EphA4 interactions contribute to the regulation of developmental processes underlying proper cortical layering.

How does ephrin A5 deficiency affect cortical development resulting in altered cortical layer thickness? The Eph/ephrin system regulates apoptosis of early cortical progenitors at E11 and E12 (Depaepe et al., 2005). However, we did not observe any alterations in the number of caspase 3-positive cells in the embryonic cortex at E13.5 and E16.5 of ephrin A5-deficient mice (supplementary material Fig. S2B-C′). When a TUNEL-based apoptosis assay was performed with E14.5 cortical cells plated on a substrate of recombinant ephrin A5-Fc, or of Fc as control, no difference in the number of apoptotic cells was determined after 5 h in vitro (13.6±3.3%, n=245 for ephrin A5-Fc and 15.9±4.4%, n=145 for control; n.s., P=0.52).

Next, we assessed whether the altered radial layer extensions in adult ephrin A5-deficient mice are elicited by changes in cortical neurogenesis. We focused our analysis predominantly on E13.5 and E16.5, when deep versus upper layer neurons are born (Polleux et al., 1997), and examined the dorsolateral cortex of coronal brain sections of the same anterior posterior level, characterized by the appearance of the medial and lateral ganglionic eminence in the basal telencephalon (supplementary material Fig. S2A). To monitor whether the expansion of deep layers results from an increase in neuronal production at E13.5, we performed immunostaining against Tbr1 as a marker for early postmitotic neurons (Englund et al., 2005; Kolk et al., 2005). The radial extension of Tbr1-positive cells in the dorsolateral cortex representing the cortical plate (CP) at E13.5 was measured in relation to overall cortical thickness. Indeed, compared with wild type, a significant increase in Tbr1-positive cells was observed in ephrin A5-deficient brain sections (Fig. 1F-H). This correlates with augmented numbers of Tbr1-positive cells per unit area, increasing from 24.4±1.1 cells in wild-type (n=46) to 28.9±1.2 in ephrin A5-deficient (n=47) embryos (P≤0.01, three brains per genotype). These results suggest that the number of postmitotic neurons was increased in E13.5 ephrin A5-deficient cortices.

In agreement with the augmented numbers of postmitotic neurons at E13.5, we also revealed an increase in overall cortical thickness in E13.5 ephrin A5 mutant embryos (supplementary material Fig. S2D). However, the cortical thickness returned to wild-type levels with progressing embryonic age (supplementary material Fig. S2E-G). This suggests that the elevated generation of neurons is temporally restricted to the stage of early corticogenesis, when deep layer neurons are born.

To quantify the number of postmitotic cells during the generation of superficial layer neurons, we assessed the rate of cells leaving the cell cycle by BrdU pulsing at E16.5 and subsequent BrdU/Ki67 double labeling after 24 h. The BrdU single-labeled cells represent the fraction that left the cell cycle, which was reduced in ephrin A5-deficient mice (Fig. 1I-K). In wild-type embryos we determined 32.94±2.0% BrdU single-labeled cells, whereas a decrease to 26.1±1.7% was determined in ephrin A5-deficient animals (Fig. 1K). This observation points to reduced neuronal production during the generation of upper layer neurons.

We also performed in situ hybridization experiments with an insulinoma-associated 1 (Insm1)-specific riboprobe at E16.5, to label terminally dividing cells (Farkas et al., 2008; Rosenbaum et al., 2011). Insm1 expression was detected in cortical proliferative zones of wild-type and ephrin A5 knockout cortices (supplementary material Fig. S2H,I). Decreased signal intensities were revealed in ephrin A5 knockout brains predominantly in the SVZ (supplementary material Fig. S2J). These data indicate that the reduced cell cycle exit observed at later stages (E16.5+24 h) results from diminished neurogenic divisions of IPCs residing in the SVZ.

As described above, the altered neuronal output at E13.5 and E16.5 correlates with changes in the thickness of the deep and upper cortical layers in ephrin A5-deficient mice. To analyze whether RGC division is affected in ephrin A5-deficient mice we performed immunostaining against Aspm to label spindle poles in combination with DAPI nuclear staining and analyzed the orientation of the cleavage plane of RGCs in relation to the ventricular surface in ephrin A5-deficient and wild-type embryos (Fig. 2A-A″) (Fish et al., 2006; Asami et al., 2011). In accordance with Asami et al. (2011), the angles of cell division were classified into three groups: with a cleavage plane of 0-30° (horizontal), 30-60° (oblique) or 60-90° (vertical) (Fig. 2B,B′). In wild-type animals, most of the division events analyzed were defined as vertical (59.3%), with 34.5% classified as oblique and 6.2% as horizontal (Fig. 2C). In E13.5 ephrin A5-deficient embryos we detected an 18.2% decrease in vertical divisions to 41.1%, whereas the percentage of obliquely and horizontally dividing apical progenitors increased to 43.3% and 15.6%, respectively (Fig. 2C). These data suggest that ephrin A5 deficiency affects the mode of RGC division, leading to an increase in divisions with unequal inheritance (asymmetric division).

Next, we performed the pair-cell assay as a gain-of-function approach to directly show the effect of exogenously applied ephrin A5-Fc on RGC division. E13.5 cortical cells were plated at clonal density and stimulated with recombinant clustered ephrin A5-Fc or with Fc as control. After 24 h in vitro, division of progenitor cells leads to the formation of cell pairs. By staining with antibodies directed against nestin and βIII-tubulin we distinguished between RGCs and postmitotic neurons, respectively. The analysis of cell pairs consisting of either two nestin-positive cells, two βIII-tubulin-positive cells, or one βIII-tubulin-positive and one nestin-positive cell (Fig. 2D-F) revealed a significant increase in the proportion of the mixed βIII-tubulin/nestin-positive combination at the expense of βIII-tubulin-positive postmitotic pairs upon ephrin A5-Fc stimulation (Fig. 2G). The reduction of postmitotic neurons in response to ephrin A5-Fc application is the converse of, and hence consistent with, what we observed in ephrin A5-deficient mice. The pair-cell assay only monitors symmetric proliferative and neurogenic, as well as asymmetric neurogenic, division events. Ephrin A5-dependent changes in asymmetric proliferative divisions leading to the generation of RGC and IPC daughters were not addressed in the pair-cell assay.

Thus, we next examined the appearance of distinct types of cortical progenitors at different embryonic stages in ephrin A5-deficient and wild-type embryos. To distinguish between RGCs and IPCs, we performed immunostainings with antibodies against Pax6 and Tbr2 (Eomes), respectively (Gotz et al., 1998; Englund et al., 2005; Pontious et al., 2008). For quantitative analysis of Pax6-labeled and Tbr2-labeled cells, we divided E13.5 cortices into ten equal horizontal segments, while twenty segments were used for analysis at E16.5 and E18.5, and the cell number per segment was determined. At E13.5, the number of Pax6-expressing RGCs was similar between knockout and wild-type cortices, except for a slight increase restricted to segments 8 and 9 in ephrin A5-deficient mice (supplementary material Fig. S3A-C). As segments 8 and 9 represent regions with high IPC numbers and therefore mainly refer to the SVZ (Fig. 2H-J), the augmentation of RGCs is likely to reflect newly generated IPCs that still express some Pax6 (Englund et al., 2005). Staining for Tbr2-positive IPCs revealed a highly significant increase in segments 3-8 of E13.5 ephrin A5-deficient cortices (Fig. 2H-J), whereby segments 1-7 represent the VZ as illustrated by Pax6 labeling at E13.5 (supplementary material Fig. S3A,B). Hence, the increase in Tbr2-positive cells seems to be caused by a boosted production of IPCs by RGCs, which is consistent with the elevated levels of asymmetric RGC division observed in E13.5 ephrin A5-deficient mice. By contrast, at E16.5 the number of Tbr2-positive progenitors was significantly decreased in segments 6 and 8 (Fig. 2K-M), and this became even more evident at E18.5 (Fig. 2N-P), whereas we observed a slight increase in Pax6-positive RGCs in segments 5-7 of the E16.5 ephrin A5-deficient mice (supplementary material Fig. S3D-F).

The changes in the number of RGCs and IPCs at later stages were also reflected in an analysis of the proportional distribution of Ki67-labeled mitotic cells (E16.5) in the VZ and SVZ, where apical versus basal progenitors reside. In ephrin A5-deficient animals, we observed an increase in the proportion of proliferating cells located in the VZ, while Ki67-labeled cells in the SVZ decreased accordingly (supplementary material Fig. S3G-I).

Taken together, ephrin A5-deficient embryos show increased rates of horizontally dividing RGCs accompanied by slightly elevated numbers of Pax6-positive cells at E13.5 and E16.5. Moreover, we detected changes in the IPC pool, with strongly augmented levels at E13.5 and decreased numbers at E16.5 and E18.5, which correlate with the alterations in neuronal production observed in ephrin A5-deficient mice. This suggests that the defects in the adult cortical layers are mainly provoked by the altered pool of IPCs in ephrin A5 knockout mice, which, at least at E13.5, appears to be a consequence of changes in the RGC division mode.

As described above, adult Epha4 knockout mice mimic the alterations in radial layer extension observed in ephrin A5 mutants (Fig. 1E), suggesting that ephrin A5-induced EphA4 forward signaling in cortical progenitors is involved in the regulation of proliferation and differentiation. Indeed, in situ hybridization experiments revealed that, of all ephrin A5-interacting receptors, only EphA4 is strongly expressed in the cortical proliferative regions (supplementary material Fig. S4). EphA4 expression could be detected from E13.5-E16.5 (Fig. 3A-C). Immunohistochemistry in coronal brain sections revealed EphA4 localization in the membrane of the soma and along the radial processes of nestin-expressing RGCs spanning the whole radial extension of the developing cortex (Fig. 3D-G‴). Consistent with this, binding studies with Alexa 488-labeled ephrin A5-Fc showed binding sites in the intermediate zone (IMZ) and CP in addition to the proliferative zones in living cortical slice cultures (Fig. 3H). EphA4 expression in nestin-positive RGCs was further directly demonstrated by immunostaining of dissociated cortical cells (Fig. 3I-L).

EphA4 is a high-affinity binding partner for ephrin A5 (Davis et al., 1994). By applying recombinant ephrin A5-Fc clustered with Alexa 488-labeled anti-human IgG to dissociated E14.5 cortical single cells, we showed that ephrin A5-Fc interacts with EphA4 receptors of cortical precursors as identified by Ki67 and EphA4 staining (Fig. 3M-T). As illustrated by anti-phosphotyrosine (PY99) labeling, EphA4 receptors of cortical cells were activated upon ephrin A5-Fc stimulation (Fig. 3U-W), which we already reported in a previous study (Zimmer et al., 2007). In summary, these data show EphA4 receptor expression by cortical progenitors, and that these are activated upon ephrin A5-Fc binding.

The results presented so far suggest that ephrin A5-induced EphA4 forward signaling in RGCs is involved in the regulation of the pool of IPCs and thus the neuronal output fated for the infra- and supragranular layers. Robust EphA4 expression was observed in cortical progenitors at the mRNA and protein levels. As Eph-ephrin signaling is dependent on cell-cell contact owing to the localization of the proteins in the cell membrane, we next examined which cells express ephrin A5 ligands. Paracrine Eph-ephrin signaling by cortical progenitors has already been described (Qiu et al., 2008; North et al., 2009). Furthermore, a feedback signaling by postmitotic neurons (Seuntjens et al., 2009; Parthasarathy et al., 2014; Toma et al., 2014) also represents a potential mechanism, as EphA4-expressing radial processes of RGCs could serve as a structural scaffold.

Ephrin A5 mRNA was already shown in E14.5 preparations of cortical tissue by RT-PCR (Fig. 1A). Although we observed ephrin A5 expression by non-radioactive in situ hybridization in other regions of the developing brain, including the thalamus, it was not detected in cortical structures at E13.5-E16.5 (Fig. 4A-C). To circumvent sensitivity limitations of non-radioactive in situ hybridization, we performed single-cell RT-PCR, which provides high sensitivity and cellular resolution. Individual embryonic cortical cells were manually isolated and limited 3′-end cDNA libraries of the transcriptome were generated. With sequence-specific primers against Pax6, Tbr2 and HuD (Elavl4), we discriminated between RGCs, IPCs and postmitotic neurons, respectively. Whereas EphA4 expression was detected in most of the Pax6-positive RGCs (n=13 of 15 tested cells) as well as in some Tbr2-expressing IPCs (n=4 of 7 tested cells) confirming the data presented above, ephrin A5 was not detected in cortical progenitors (Fig. 4D,E). Furthermore, ephrin A5 transcripts were also absent from single-cell cDNA libraries, which were classified as postmitotic cortical transcriptomes according to HuD expression (Fig. 4F) (Fujiwara et al., 2008; Gauthier-Fisher et al., 2009). However, ephrin A5 was detected in single-cell libraries isolated from the preoptic area known to express this ligand (Fig. 4D,E; n=3) (Zimmer et al., 2008), as well as in cDNAs generated from single-cell-equivalent dilutions (10 pg quantities) of embryonic brain tissue RNA that served as positive controls. Thus, consistent with the results of the in situ hybridization experiments shown in Fig. 4A-C, ephrin A5 was not detected in cortical progenitors and postmitotic neurons. Hence, a paracrine EphA4 activation induced by ephrin A5-expressing neighboring cortical precursors or a feedback signaling of postmitotic neurons seems unlikely.

Thalamic ephrin A5 mRNA expression at E13.5 was observed in the mantle zone (Fig. 4A, arrow), from which the first axons grow to the developing cortex and which is positive for L1 (L1cam) (Fig. 5I, arrow). Placing a DiI crystal into the E16.5 dorsolateral cortex tags the thalamic ventrobasal complex, a region that corresponds to the ephrin A5-expressing thalamic nuclei at E16.5 and that is known to innervate the somatosensory cortex (supplementary material Fig. S5A,B) (Uziel et al., 2006).

At later stages of cortical development ephrin A5 affects the topography and branching of thalamocortical afferents (Bolz et al., 2004). Thus, the effects observed in the ephrin A5 mutants could be indirect due to defects in thalamocortical development evoked by ephrin A5 deficiency. However, we observed no gross alterations in the growth of L1-positive thalamic fibers during the timecourse of cortical neurogenesis in ephrin A5 mutants as compared with wild types (supplementary material Fig. S5C,D) (Fukuda et al., 1997). Moreover, thalamic explant cultures of wild-type and ephrin A5 knockout thalami (E14.5+2 days in vitro) did not reveal any differences in the axonal outgrowth index, which is defined as the ratio between the area of the whole explant and the area of the explant without axons (supplementary material Fig. S5E-G). This suggests a direct action of ephrin A5-expressing thalamic axons on EphA4-expressing cortical progenitors, rather than any indirect effect evoked by alterations of thalamic axonal projections due to ephrin A5 deficiency.

To show that ephrin A5 is expressed in thalamic axons and to circumvent the lack of specific antibodies for ephrin A5, we first applied Alexa 488-labeled recombinant EphA3-Fc protein to thalamic explant cultures prior to fixation in order to visualize ephrin A ligands, according to previous studies (Zimmer et al., 2011). As illustrated in Fig. 5A-D, we detected EphA3-Fc binding sites along βIII-tubulin-labeled thalamic axons indicating ephrin A ligand expression. Consistent with region-specific thalamic ephrin A5 expression (Fig. 4A-C), not all thalamic explants revealed EphA3-Fc binding sites (Fig. 5N,O). This excludes nonspecific labeling by the Alexa 488-conjugated anti-human IgG antibody used for detection of EphA3-Fc binding sites. Axonal mRNA localization and protein synthesis have already been described (Jung et al., 2012). To investigate whether ephrin A5 transcripts, in particular, can be identified in thalamic axons, we manually isolated axonal parts of thalamic explants (Fig. 5E-G) and performed RT-PCR. As a housekeeping gene we used β-actin (Actb), which was detected in axonal compartments but not in the negative controls. Indeed, we identified ephrin A5 transcripts in samples of thalamic axons (Fig. 5H). These data suggest that exogenous ephrin A5 ligands are imported into the cortex by thalamic axons.

Auladell et al. (2000) provided evidence that thalamic afferents reach the cortex as early as E13. As L1 was used in other studies to label thalamocortical axons (Fukuda et al., 1997; Ohyama et al., 2004; Demyanenko et al., 2011), we performed immunohistochemical staining of L1 at E13.5, E14.5 and E16.5. Fibers positive for L1 were detected in the lateral cortex at E13.5, becoming increasingly evident with progressing developmental stages (Fig. 5I-K; supplementary material Fig. S5C). However, L1 is also expressed by early cortical projection neurons (Espinosa and Stryker, 2012). For this, we performed anterograde labeling of thalamic axons with the lipophilic dialkylaminostyryl dye (DiA) as an alternative approach. Labeling in living tissue is 6 mm/day faster than in fixed tissue (0.2-0.6 mm/day) due to active dye transport (Lukas et al., 1998). Thus, we injected DiA into the thalamus of freshly prepared brains at E13.5, which were then kept in culture for 10 h. After fixation, the tissue was stored for 7 days at 37°C to ensure complete axonal labeling. In agreement with Auladell et al. (2000), we observed DiA-positive thalamic fibers reaching the cortex at E13.5 (Fig. 5L,M). These data indicate that the arrival of the first thalamocortical fibers is coincident with the alterations that we observe in the dorsolateral cortex of E13.5 ephrin A5-deficient mice (Figs 1 and 2; supplementary material Figs S1-S3).

The data described so far showed that ephrin A5-EphA4 interactions are involved in the regulation of cortical progenitors. As ephrin A5 was not detected in cortical cells but in the thalamus and thalamic axons, we hypothesize an exogenous regulation of cortical precursors by thalamic afferents. To address whether ephrin A-expressing thalamic axons affect the proliferation of cortical precursors, we performed a modified pair-cell assay in the presence of ephrin A ligand-positive or -negative thalamic axons. We co-cultured E13.5 thalamic explants and dissociated cortical cells (E13.5) labeled with Cell Tracker Green at clonal density. The ephrin A ligand-expressing thalamic axons were identified by binding of recombinant EphA3-Fc clustered with an Alexa 488-labeled anti-human IgG antibody, as EphA3 represents a high-affinity receptor for ephrin A ligands but does not interact with ephrin B ligands, in contrast to EphA4 (Fig. 5C,P). We determined the number of cell pairs localized within the area covered by thalamic axons around the explants that consisted of either two nestin-positive progenitors, two βIII-tubulin-positive postmitotic neurons, or mixed pairs composed of a nestin-positive cell and a βIII-tubulin-expressing cell (Fig. 5N-P). We detected a significant increase in the proportion of the mixed nestin/βIII-tubulin-positive combination accompanied by a decrease in βIII-tubulin-positive cell pairs and nestin-positive cell pairs in the presence of EphA3-Fc-positive axonal fibers as compared with EphA3-Fc-negative explants (Fig. 5Q). This is reminiscent of what was found for E13.5 cell pairs stimulated with recombinant ephrin A5-Fc (Fig. 2G). These in vitro data suggest that ephrin A-expressing thalamic fibers influence the proliferation and differentiation of cortical progenitors.

Thalamic fibers invade the developing cortex along the IMZ and upper SVZ (Auladell et al., 2000), whereas cortical progenitors reside in the VZ and SVZ. As EphA4 expression was observed along the radial processes of cortical progenitors, this scaffold might enable contact-dependent ephrin A5/EphA4 signaling between growing thalamic axons and neuronal precursors. To address this, we placed ephrin A5-Fc-coated or control Fc-coated protein-A-agarose beads in the CP/IMZ of organotypic coronal slice cultures of E14.5 ephrin A5 knockout brains. After 2 days in vitro we performed immunostaining against Tbr2, to investigate whether ectopic ephrin A5-Fc influences IPC numbers in ephrin A5-deficient cortices. This revealed an increase in Tbr2-positive cells in response to the ephrin A5-Fc-coated beads compared with the control after 2 days in vitro (supplementary material Fig. S5H-J). Thus, ectopically applied ephrin A5-Fc outside of the proliferative zones affects IPC production by regulating the division mode of RGCs. As we observed elevated levels of Tbr2-expressing cells when placing the ephrin A5-Fc-coated beads at E14.5+2 days in vitro into slices of ephrin A5-deficient brains, this can be regarded as a partial rescue of the IPC reduction observed in E16.5 knockout mice (Fig. 2).

Taken together, these data suggest that extra-cortical ephrin A5 ligands imported by invading thalamic axons regulate the proliferation and differentiation of cortical progenitors by activating EphA4 forward signaling and thereby the laminar organization of the cerebral cortex.

In this study, we provide evidence for the extra-cortical regulation of RGC division, regulating proper IPC levels at early and late stages of cortical neurogenesis by thalamic axons via ephrinA5/EphA4 signaling, thereby controlling the neuronal output fated for the deep and upper cortical layers.

Control of cortical proliferation by the paracrine activation of Eph receptors was reported previously (Qiu et al., 2008; North et al., 2009). Here, we describe a contribution of ephrin A5 to the EphA4-dependent control of cortical progenitors. Interestingly, EphA4 has already been shown to regulate cortical progenitor proliferation; EphA4 overexpression enlarges the progenitor pool, whereas reduced EphA4 receptor forward signaling results in fewer mitotically active cells mediated by ephrin B1-EphA4 interactions (North et al., 2009).

IPCs represent an intermediate stage from RGCs to neurons (Gotz and Huttner, 2005; Franco and Muller, 2013). Upon generation from RGCs in the VZ, IPCs translocate to the SVZ, where they undergo symmetric proliferative and neurogenic divisions (Attardo et al., 2008; Noctor et al., 2008). Even if IPCs are restricted in their proliferative potential, they expand the pool of neurogenic progenitors (Haubensak et al., 2004; Noctor et al., 2004) and contribute to all cortical layers (Smart, 1973; Haubensak et al., 2004; Attardo et al., 2008; Farkas et al., 2008; Sessa et al., 2008; Kowalczyk et al., 2009; Vasistha et al., 2014). It has even been suggested that the majority of cortical projection neurons derive from IPCs, whereas only ∼10% seem to be generated directly from RGCs (Attardo et al., 2008; Sessa et al., 2008; Kowalczyk et al., 2009). In support of this, we observed distinct changes in IPC numbers that concur with the neuronal output at early and late stages of neurogenesis in ephrin A5-deficient embryos and the altered thickness of infra- and supragranular adult cortical layers.

Our observations, including alterations in the RGC cleavage plane (Fig. 2A-C) and augmented IPC numbers in regions that represent the VZ (Fig. 2H-J), point to an ephrin A5 deficiency-induced shift in RGC division towards an elevated IPC production at E13.5. Furthermore, the increase in the number of both IPCs and postmitotic neurons in E13.5 ephrin A5-deficient cortices is in contradiction with the elevation of self-amplifying basally dividing cells observed after forced Insm1 expression, which occurs at the expense of neuronal production (Farkas et al., 2008). For this, an increase in proliferative basal cell division seems unlikely.

In contrast to IPCs, which showed significantly higher cell numbers in most of the regions of the E13.5 cortex, the slight increase in Pax6-positive cells was restricted to the upper segments, in which the majority of the Tbr2-positive IPCs are located. Thus, these cells could represent differentiating IPCs, which still express Pax6 consistent with the observation of Englund et al. (2005), thus revealing cells co-expressing both markers.

As the genetic ablation of ephrin B1 leads to premature neurogenesis and loss of cortical progenitors (Qiu et al., 2008), the switch towards decreased levels of IPCs at E16.5 and E18.5 might result from precocious augmentation at the subsequent expense of IPCs. Another possible scenario to explain the reduced IPC numbers is a reciprocal effect on the RGC division mode at E16.5 as compared with E13.5 mediated by distinct intracellular signaling molecules that evoke differential effects induced by the same ligand-receptor interaction at certain developmental stages. For ephrin A5-induced EphA4 receptor activation, a switch in the cellular response that is dependent on Src family kinase activity has already been shown (Zimmer et al., 2007).

Here, we provide evidence for an extra-cortical regulation of neuronal precursors by thalamic afferents expressing ephrin A5. As we did not detect ephrin A5 in cortical progenitors or postmitotic neurons, consistent with previous reports (North et al., 2009; Deschamps et al., 2010), a paracrine signaling of cortical precursors (Qiu et al., 2008; North et al., 2009) or feedback signaling of CP neurons back to progenitors (Seuntjens et al., 2009; Parthasarathy et al., 2014; Toma et al., 2014) seems unlikely. However, we identified ephrin A5 transcripts in the developing thalamus and thalamic axons. Axonal mRNA transport, localization and translation, as has also been described for ephrins, serves to provide a rapid response to guidance cues during axonal growth and synapse formation (Jung et al., 2012).

Thalamic afferents have been shown to contribute to cortical arealization (Rakic, 1988; Dehay et al., 1996a,b), to influence the survival and identity of postmitotic cortical neurons (Zhou et al., 2010; Sato et al., 2012; Li et al., 2013) and to affect cortical progenitor proliferation in vitro by a diffusible factor (Dehay et al., 2001). Here, we follow the idea of thalamic regulation of cortical progenitors by suggesting that thalamic ephrin A5 interacts with EphA4 receptors of RGCs to regulate the proper levels of IPCs. The radial processes of RGCs provide the structural scaffold for local interactions of membrane-bound EphA4 receptors with ephrin A5 ligands expressed by thalamic afferents, growing along the upper SVZ and IMZ (Auladell et al., 2000). In support of this, we observed increased IPC numbers in cortical slice cultures of ephrin A5-deficient mice in response to ephrin A5-Fc-coated protein-A-agarose beads placed in the CP/IMZ.

The idea of afferents affecting the proliferation of telencephalic structures was proposed by Gong and Shipley (1995), who showed that pioneer olfactory axons modulate the cell cycle kinetics of olfactory bulb precursors. A thalamic influence on cortical proliferation was further shown for cortical progenitors in ferrets, affecting the tangential expansion of the cortex (Reillo et al., 2011). In Celsr3/Dlx mutant mice lacking thalamocortical and corticothalamic connections, the thalamic impact on cortical neurogenesis is suggested to be rather subtle, based on laminar architecture and BrdU birthdating analysis in adults. Interestingly, BrdU labeling of the upper cortical layers in the Celsr3/Dlx mutant mice revealed a shift towards the pial surface and a reduction in the radial extension of the BrdU-labeled upper layer neurons (Zhou et al., 2010), similar to our observation in ephrin A5 knockout mice (supplementary material Fig. S1F-H). However, the subtle defects observed in Celsr3/Dlx mutant and ephrin A5-deficient mice indicate an involvement of thalamic axons in the fine-tuning of cortical neurogenesis.

Thalamocortical projections have been reported to arrive in the cerebral cortex prior to the generation of the majority of cortical neurons (Shatz and Luskin, 1986; Auladell et al., 2000). Deep layer neurons are generated between E12.5 and E14.5, whereas upper layer neurons are born from E15.5-E18.5 (Polleux et al., 1997). Auladell et al. (2000) describes the first thalamocortical axons as arriving at E13 in the murine dorsolateral cortex. Consistently, by DiA labeling in living brain cultures we traced fibers in the dorsolateral cortex labeling the E13.5 thalamus, suggesting that thalamic ephrin A5 instructs cortical progenitors from E13.5 on. However, in the context of the effects on neuronal numbers that we observed at E13.5, we cannot discount the possible influence of other sources of ephrin A5, the detection of which was beyond the scope of this study.

Ephrin A5-EphA receptor interactions regulate the topography of different thalamocortical projections at later stages (Donoghue and Rakic, 1999; Šestan et al., 2001; Bolz et al., 2004). As we did not observe gross alterations in thalamic axonal growth between wild type and ephrin A5 knockouts we suggest a direct influence of ephrin A5 on cortical progenitors.

In conclusion, we propose an influence of invading thalamic axons on the balance of basal cortical progenitors, thereby regulating cortical neurogenesis via ephrin A5-EphA4 interactions.

Time-staged embryos (E1 being the day of insemination) and adult animals (P37) of the ephrin A5 knockout (Knoll et al., 2001), EphA4 PLAP mice (Leighton et al., 2001) and wild-type (C57/BL6) mice, bred and maintained under standard conditions (food and water ad libitum, 12-h light/dark cycle), were used. All animal procedures were performed in accordance with guidelines for the care and use of laboratory animals of the Friedrich Schiller University, Jena, and the University Hospital Jena, Germany.

For the preparation of adult brains, mice were transcardially perfused with PBS (pH 7.4) and 4% paraformaldehyde (PFA) solution (pH7.4) and post-fixed overnight in 4% PFA followed by treatment with 10% and 30% sucrose in PBS overnight. For the preparation of time-staged embryonic brains, mice were deeply anesthetized with 10% chloralhydrate in PBS (pH 7.4) and prepared embryonic brains were used fresh or fixed for 5 h in 4% PFA followed by treatment with 10% and 30% sucrose overnight.

For organotypic slice preparation the protocol described by Zimmer et al. (2011) was used. The preparation of dissociated cells was performed according to Zimmer et al. (2007). Cells were seeded at a density of 300 cells/mm² for immunohistochemistry and stimulation experiments and at 150 cells/mm² for the DAB TUNEL-based apoptosis detection assay (see supplementary methods). Thalamic explants were prepared according to Ruediger et al. (2013).

E13.5 cortical cells plated at clonal density in DMEM/F12 supplemented with 10% FBS, 0.4% methyl cellulose, 72 mM glucose, 100 U/ml penicillin, 100 µg/ml streptomycin and 0.4 mM L-glutamine were incubated for 24 h in vitro, supplementing the medium with either 5 µg/ml ephrin A5-Fc (R&D Systems) or Fc protein as control (Rockland Immunochemicals) clustered with 50 µg/ml anti-human IgG antibody (Alexis Biochemicals). Cell pair identity was determined after 24 h by immunostaining for nestin and βIII-tubulin. We included only cell pairs in our analysis that showed staining of both cells.

Dissociated cortical cells were treated with Cell Tracker Green (Life Technologies) and washed before adding to explants at clonal density on coverslips coated with 19 µg/ml laminine and 5 µg/ml poly-L-lysine (Gibco) in DMEM/F12 supplemented with 10% FBS, 0.4% methyl cellulose, 72 mM glucose, 100 U/ml penicillin, 100 µg/ml streptomycin and 0.4 mM L-glutamine. After 24 h at 37°C and 5% CO₂, 5 µg/ml EphA3-Fc (R&D Systems) pre-clustered with 50 µg/ml Alexa 488-conjugated anti-human IgG (Alexis Biochemicals) was added for 30 min at 37°C. After fixation in 4% PFA in PBS for 30 min, cell pair identity was determined by immunostaining against nestin and βIII-tubulin.

Stimulation of dissociated cells and organotypic brain slices with 5 µg/ml recombinant ephrin A5-Fc or EphA3-Fc (R&D Systems) clustered with 50 µg/ml anti-human IgG Alexa 488 (Chemicon) was performed for 3 h at 37°C and 5% CO₂. For the application of either ephrin A5-Fc-coated or Fc-protein-coated protein-A-agarose (Santa Cruz Biotechnology), beads were treated overnight with 5 µg/ml recombinant ephrin A5-Fc or Fc at 4°C on a shaker.

DiA (in 25% DMSO, 25% ethanol, 50% DMEM/F12) was pressure injected with an extended glass capillary into the embryonic thalamus of E13.5 brains. After washing in Krebs buffer, brains were incubated in 3 ml culture medium (DMEM/F12 supplemented with 20 mM glucose, 100 U/ml penicillin, 100 µg/ml streptomycin, 55 mM mercaptoethanol) on an orbital shaker (50 rpm). After 10 h incubation at 37°C and 5% CO₂, brains were fixed in 4% PFA in PBS (pH 7.4) and stored at 37°C for 7 days. Brains were coronally dissected into 300 µm slices.

For immunohistochemistry, fixed embryonic brains (5 h in 4% PFA in PBS) or single cells (15 min in 4% PFA in PBS) were used. Histochemistry is performed as described by Zimmer et al. (2011). Primary antibodies were applied overnight, secondary antibodies for 2 h. For further details, see the supplementary methods.

In situ hybridizations were performed as described by Zimmer et al. (2011) using digoxigenin-labeled riboprobes spanning the sequence 75-761 for Epha4 (accession number 007936) and 106-637 for Efna5 (accession number 010109). Riboprobes were obtained by in vitro transcription using DIG-11-UTP (Roche) from cDNA fragments cloned in pBluescript I KS (Stratagene).

The protocol for global cDNA synthesis was from Iscove et al. (2002) with the modifications presented in the supplementary methods. Preparation of cortical tissue for PCR and details of single-cell RT-PCR and RT-PCR of thalamic axons are described in the supplementary methods.

A detailed description of the image analyses performed, including quantification of cortical progenitors and characterization of cell pairs, measurements of radial extension of cortical structures/layers, analysis of the Insm1 signal and determination of division angles, together with statistical considerations, is presented in the supplementary methods.

We thank Christine Raue for excellent technical assistance and Sandra Clemens for animal breeding (Institut für Allgemeine Zoologie, Friedrich-Schiller Universität Jena, Germany). We thank Hans Zempel for correcting the manuscript. The L1 (rat) antibody developed by P. H. Patterson was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at the University of Iowa, Department of Biology, Iowa City, IA, USA.