Dynamic integration of subplate neurons into the cortical barrel field circuitry during postnatal development in the Golli-tau-eGFP (GTE) mouse

Abstract

In the Golli-tau-eGFP (GTE) transgenic mouse the reporter gene expression is largely confined to the layer of subplate neurons (SPn), providing an opportunity to study their intracortical and extracortical projections. In this study, we examined the thalamic afferents and layer IV neuron patterning in relation to the SPn neurites in the developing barrel cortex in GTE mouse at ages embryonic day 17 (E17) to postnatal day 14 (P14). Serotonin transporter immunohistochemistry or cytochrome oxydase histochemistry was used to reveal thalamic afferent patterning. Bizbenzimide staining identified the developing cytoarchitecture in coronal and tangential sections of GTE brains. Enhanced green fluorescent protein (GFP)-labelled neurites and thalamic afferents were both initially diffusely present in layer IV but by P4–P6 both assumed the characteristic periphery-related pattern and became restricted to the barrel hollows. This pattern gradually changed and by P10 the GFP-labelled neurites largely accumulated at the layer IV–V boundary within the barrel septa whereas thalamic afferents remained in the hollows. To investigate whether this reorganisation is dependent on sensory input, the whiskers of row ‘a’ or ‘c’ were removed at P0 or P5 and the organisation of GFP-labelled neurites in the barrel cortex was examined at P6 or P10. In the contralateral region corresponding to row ‘a’ or ‘c’ the lack of hollow to septa rearrangement of the GFP-labelled neurites was observed after P0 row removal at P10 but not at P6. Our findings suggest a dynamic, sensory periphery-dependent integration of SPn neurites into the primary somatosensory cortex during the period of barrel formation.

The subplate (SP) is a largely transient layer of the developing cerebral cortex consisting of its earliest born neurons (Kostovic & Rakic, 1990; Allendoerfer & Shatz, 1994; Price et al. 1997). SP is located between the white matter and cortical layer VI and constitutes the first layer below the cortical plate during the peak period of cortical neurogenesis and migration. Thalamic afferents encounter the subplate while they accumulate below the cortical plate prior to invading the cortex and contacting their ultimate target neurons in layer IV (Rakic, 1977; Luskin & Shatz, 1985). It is known that the SP neuron (SPn) population is highly heterogeneous in neurochemical properties, electrophysiology, target projection and morphology (Del Río et al. 2000; Hanganu et al. 2002; Hoerder-Suabedissen et al. 2008b). Subplate-specific ablation studies in the primary visual cortex of cats have allowed us to infer a role for this elusive cell population in carnivores (Ghosh et al. 1990; Ghosh & Shatz, 1992; Kanold et al. 2003). From these studies it has been established that SPn are critical in pioneering corticothalamic connections. They are required for correct thalamocortical targeting to the visual cortex and remodelling of cortical layer IV into ocular dominance and orientation columns (Allendoerfer & Shatz, 1994; Kanold & Shatz, 2006). Subplate projections into the cortical plate are believed to be the anatomical and physiological substrate responsible for the establishment of modular organisation of the visual cortex (Friauf et al. 1990; Kanold et al. 2003; Arber, 2004). A transient cortical circuit has been proposed involving the subplate to mediate input to the cortical plate before and during thalamic innervation of layer IV (Friauf & Shatz, 1991; Allendoerfer & Shatz, 1994; Kanold & Shatz, 2006). Cell-filling studies of subplate neurons in carnivores indeed revealed subplate projections extending to the marginal zone with collaterals to the cortical plate at stages before the full establishment of thalamic innervation (Friauf et al. 1990). Using optical recording with voltage-sensitive dyes, Higashi et al. (2002, 2005) showed that functional thalamocortical connections exist between thalamus and cortex at the time of thalamic fibre arrival from as early as E17 in the rat, earlier than previously described (Hanganu et al. 2002) and suggested in carnivores by Friauf et al. (1990). By P0–2 the activation pattern spreads to the entire depth of the cortex in a diffuse fashion, by P2–4 it is becoming restricted to individual columns spanning the entire depth of the cortex, and subsequently it gets confined to layer 4 and 6 in a pattern resembling the modules of the barrel field (Fig. 1D). Although these results suggest functional integration, the detailed cellular and molecular mechanisms by which SPn are involved in thalamocortical circuit formation is still largely not known (López-Bendito & Molnár, 2003; Kanold, 2004). The possibility of subplate cells acting as ‘amplifiers’ of afferent inputs arising from the thalamus or of cholinergic inputs was raised by Dupont et al. (2006). According to these authors, the relative mature firing properties of subplate cells, their column-like gap-junctional coupling with the developing cortex and the presence of column-like oscillatory networks driven by the subplate demonstrate that these transient cells play an active role in the formation of functional cortical circuits in rodents from P0. Our current knowledge of SPn morphology is limited to Golgi impregnations, lucifer yellow tracing, biocytin filling and retrograde carbocyanine dye-labelling techniques (Marin-Padilla, 1971; Robertson et al. 2000; Hanganu et al. 2002; McConnel et al. 1989) which do not allow visualisation of the entire neurite pattern of SPn into the cerebral cortex during development and are not necessarily specific for the anatomically defined SP. Our inability to selectively and specifically label subplate neurons and their neurites during these phases of development hinders our further understanding. The genetic accessibility and the postnatal establishment of the sensory periphery-related modular pattern makes the barrel field of the mouse primary somatosensory cortex an ideal model system to analyse and manipulate early cortical circuits. Whisker removal during the critical period abolishes the organisation of thalamic afferents into the periphery-related pattern in layer IV and the cytoarchitectonic differentiation of that area (Van der Loos & Woolsey, 1973).

Figure 1. The Golli-tau-eGFP mouse as a model to study subplate neurite integration into the barrel cortex during development.

A, schematic of the GTE construct (adopted from Jacobs et al. 2007). B, schematic diagram of the sensory pathway from the whiskers to the contralateral barrel cortex. C, schematic diagram of a small sector from a tangential section through a P0 barrel cortex with homogeneous distribution of layer IV neurons (blue), thalamic afferents (red) and GFP neurites (green). At P6, the thalamic projections assume the characteristic periphery-related pattern and target the barrel hallows and the layer IV neurons cluster at the septa (diagram based on Molnár & Molnár, 2006). There are three possible scenarios of the GFP neurite arrangement: (i) non-specific distribution; (ii) intra-barrel patterning; and (iii) inter-barrel patterning. D, schematic representation of the thalamic fibre ingrowth in the mouse primary somatosensory cortex. At embryonic day 11 (E11), the first postmitotic cells migrate to the outer edge of the cerebral wall to form the preplate (PP), which is subsequently split into the marginal zone (MZ, the future layer I) and the subplate (SP) by the arrival of neurons in the true cortical plate (CP). When thalamic axons (red lines) arrive to the cortex at E15, only a densely packed cortical plate is present. The axons start to accumulate in SP, although some axons and side branches penetrate the deep part of the cortical plate (DCP). During the early postnatal period (P0), most thalamic fibres invade the CP and layers V and VI. Thalamic axons assume their characteristic periphery-related pattern and impose a barrel arrangement on cortical layer IV neurons. E, optical recording with voltage-sensitive dyes revealed that thalamocortical projections elicit sustained depolarization in the SP at E18–19. This depolarization spreads to almost the entire thickness of the cortex shortly after birth and concentrate by the end of the first postnatal week (P8) at the periphery-related clusters in layer IV. IZ, intermediate zone; SVZ, subventricular zone; VZ, ventricular zone; WM, white matter. Adapted and modified from Molnár & Hannan (2000) and López-Bendito & Molnár (2003).

In the Golli-tau-eGFP (GTE) mouse, the golli-promoter of the golli-myelin basic protein gene drives reporter gene expression (enhanced green fluorescent protein (eGFP) fused to bovine tau microtubule protein) selectively in subplate neurons (see Fig. 1A for transgenic construct from Jacobs et al. 2007). This model facilitates the study of the integration of SP and their associated neurites through the visualisation of eGFP. It can be effectively used to study SP neurite integration into the neocortical column, as modelled by the individual barrels (Molnár & Molnár, 2006) and so make correlations to the observations obtained in the visual cortical modules of carnivores (Ghosh & Shatz, 1992; Kanold et al. 2003).

The first aim of this study was to document the GFP⁺ neurite arrangement in the barrel cortex during postnatal development of the GTE mouse. We raised three possible scenarios for GFP neurite distribution in relation to the barrels: (i) non-specific distribution, (ii) predominantly inside the barrel hollows (intra-barrel patterning) or (iii) predominantly between the barrels (inter-barrel patterning) (Fig. 1C). Our second aim was then to determine if the arrangement of the GFP⁺ neurites in the barrel cortex is dependent on the sensory whisker periphery.

Methods

Animals

All animal experiments were approved by a local ethical review committee and conducted in accordance with personal and project license under the UK Animals (Scientific Procedures) 1986 Act. Golli-tau-eGFP (GTE) mice were obtained from a local breeding colony at UCLA. To study the developmental pattern of GFP-labelled neurite integration into the cerebral cortex we analysed various developmental stages with a time resolution of 2–4 days from E17–P14 mice (see Table 1 for details). Both hemispheres of a total of 47 GTE brains were used for this study. The pups were deeply anaesthetized with pentobarbitone (Euthatal 150 mg kg⁻¹i.p.; Merial Animal Health Ltd, Harlow, UK), and perfused through the heart with 4% paraformaldehyde (PFA; TAAB, Reading, UK) in phosphate-buffered saline (PBS). The brains were post-fixed in 4% PFA overnight. To obtain the E17 brains, timed pregnant mice were killed by cervical dislocation and the embryos were collected with caesarean section and kept on ice. The brains were dissected and immersed into 4% PFA overnight for fixation.

Table 1.

Number of hemispheres used for the various ages and experimental protocols obtained from 47 animals

Treatment	E17	P0	P2	P4	P6	P8	P10	P14
Bizbenzimide	2C	1C	3C	2C	4C, 1T	1T	3C, 1T	2C, 1T
Cytochrome oxidase	—	1T	1T	1T	2T	2T	2T	—
DiI + Bizbenzimide	—	1C	1C	1C	1C	1C, 1T	1C, 1T	1C
Whisker removal	—	—	—	—	4C	—	4T, 8C	—
5HTT					2C		2C	2C
Total (n)	1	2	4	4	11	4	16	5

Abbreviations: T, tangential sections; C, coronal sections.

Sectioning

Serial coronal sections of 40 μm thickness at the level of the somatosensory cortex were obtained by cutting brains embedded in 5% agarose (Sigma-Aldrich) on a vibraslicer (Leica, VT1000S). For the serial tangential sections (40–60 μm thickness) a sliding freezing microtome (Microm, HM 400) was used. In these hemispheres the concavity of the cortex was reduced by dissecting out most of the subcortical brain tissue and gently pressing the cortex on the flat surface of the sectioning stage during freezing with crushed dry ice.

Histochemistry (nuclear and cytochrome oxidase staining)

To reveal the cortical layers and cytoarchitecture, sections were counterstained with the fluorescent nuclear stain bisbenzimide, (1 μg ml⁻¹ (Invitrogen) in 0.1 m phosphate-buffered saline (PBS), pH 7.4). After imaging the GFP and bizbenzimide staining pattern we decoverslipped and dismounted the relevant tangential sections (at least one brain for each stage) and processed them for cytochrome oxidase histochemistry according to Wong-Riley & Welt (1980). Sections were washed and then placed in a staining solution (cytochrome C (0.3 mg ml⁻¹, Sigma), 3,3′diaminobenzidine (0.55 mg ml⁻¹, Sigma) and sucrose (45 mg ml⁻¹) in 0.1 m PBS). Sections were incubated in this solution at 37°C in the dark for approximately 12 h. Sections were then mounted onto slides, dehydrated and coverslipped.

Immunohistochemistry

To be able to assess the formation of periphery-related thalamic afferent pattern in relation to the GFP-labelled subplate neurites, serotonin transporter (5HTT or serotonin 5-hydroxy-tryptamine) immunohistochemistry was performed on coronal sections on the level of the somatosensory cortex at P6, P10 and P14 GTE mouse. Sections were washed in 0.1 m PB, blocked in 0.3% PB and 2% Triton goat serum and incubated overnight at 4°C in rabbit anti-serotonin antibody (1 : 2000, Calbiochem). The antibody staining was visualised using a fluorescent secondary antibody (Cy3-conjugated goat anti-rabbit, 1 : 500 (Jackson ImmunoResearch, West Grove, PA, USA)). To boost and to verify the GFP signal. we used fluorescent immunohistochemistry anti-GFP. After blocking as above, the sections were incubated overnight with an anti-GFP primary antibody (Molecular Probes, 1 : 1000). The antibody staining was visualised using a fluorescent secondary antibody (Alexa488-conjugated goat anti-rabbit, 1 : 500 (Invitrogen)).

Thalamocortical axon tracing using carbocyanine dye DiI

In order to visualise the thalamocortical afferents, DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) crystals (Molecular Probes) were embedded into the thalamic ventrobasal nuclei of the GTE mouse brains using a fine application pin. These brains were incubated for anterograde labelling in PBS azide (0.1 μg ml⁻¹) for 6 weeks in the dark at room temperature and sectioned (Molnár et al. 2006).

Whisker removal

Newborn pups were anaesthetised on ice and the top row (row ‘a’) of whiskers on the right side of the snout was removed at P0 in four GTE mice by plucking followed by cauterisation of the row to prevent regrowth of the removed whiskers according to the methods described by Van der Loos & Woolsey (1973). The mice were warmed, then returned to their litter and raised in the same environment as control mice and perfused at P10. A similar procedure was performed on additional experiments where row ‘c’ of whiskers on the right side of the snout were removed at P0 or P5 and mice perfused at P6 (n= 4) or P10 (n= 8). We refer to the P5 whisker removal experiments as ‘delayed’ ablation experiments.

Microscopy and imaging processing

We analysed sections at the level of the barrel cortex for the presence of barrels (bisbenzimide), subplate neurite patterning (GFP labelling) and thalamic periphery-related pattern into barrels (cytochrome oxidase or 5HTT immunohistochemistry) using fluorescent and bright field microscopy. Serial tangential and coronal sections were examined. An upright microscope, (Leica DMR) was used and images were acquired at ×10 and ×20 objectives using the attached digital camera (Leica DC 500) and Leica IM50 4.0 software. Fluorescent field images (GFP and bisbenzimide) were captured prior to the cytochrome oxidase staining of the same sections. Images were assembled into figures using Adobe Photoshop 7.0 software.

Results

Thalamocortical projections establish periphery-related pattern by P3–4 in the GTE mouse

Thalamocortical axons reach their target in layer IV of the neocortex and arborise there, guided by various factors including attractive and repulsive molecules and neural activity (Yamamoto et al. 2002; Molnár et al. 2007). The developmental patterning of the thalamic axons in layer IV has been elucidated by previous studies using several methodologies, including thalamic fibre tracing or cytochrome oxidase staining (Rice et al. 1985; Agmon et al. 1995). Serotonin transporter (5HTT) is specifically expressed in thalamocortical axon terminals (Lebrand et al. 1998; Rebsam et al. 2002) and therefore provides an additional sensitive method for the detection of the periphery-related clustering in early postnatal mice (Cases et al. 1996). We found 5HTT immunohistochemistry a more reliable method to reveal large populations of thalamic projections, whereas carbocyanine dyes give better individual fibre resolution (data not shown). It has been demonstrated previously that thalamic afferents and the layer IV neurons are both homogenously distributed in the putative primary somatosensory cortex shortly after birth (P0) (Fig. 1C). Subsequently, at P3/4 thalamocortical projections assume a periphery-related pattern and impose the cytoarchitectonic pattern observed on layer IV neurons (Schlaggar & O’Leary, 1991). Here, we confirmed this developmental timeframe for the establishment of the barrel patterning in the GTE mouse. The thalamocortical arrangement was primarily visualized by 5HTT immunohistochemistry that we found to be the most compatible method with the GFP labelling (example in Fig. 3I-Q) but was also confirmed by cytochrome oxidase staining (example in Fig. 2G and J). The cytoarchitectonic arrangement of layer IV neurons, as revealed by bisbenzimide counterstain into barrels was also first apparent at P4 (n= 4) (Fig. 3D) as expected (Rice et al. 1985; Hoerder-Suabedissen et al. 2008b).

Figure 3. Fluorescence micrographs of coronal sections through the developing barrel cortex of the GTE mouse at ages E17–P14 (A–H) and thalamic axon segregation pattern in relation to the GFP-labelled neurites in the GTE mouse from P6 to P14 (I–Q).

GFP neurites are shown in green, thalamic afferents (labelled with 5HTT immunohistochemistry) in red and the bisbenzimide nuclear counterstain shown in blue. A–C, GFP-labelled neurites extend radially towards the pial surface from cell bodies situated in SP and to lesser extent in cortical plate (layer VI). By P4 and P6 GFP-positive fibres cluster in layer IV, but subsequently this pattern is changed by the retraction of the GFP-labelled neurites from the hollows of layer IV. Filled arrowheads in D and E indicate intra-barrel GFP densities; arrows in F indicate barrel septa; open arrowheads in G and H indicate inter-barrel GFP concentration. I, L and O, thalamic axons were visualized by serotonin immunohistochemistry for serotonin transporter (5HTT) which show a clear segregation in a periphery-related pattern at P6, P10 and P14. To relate the subplate neurites in relation to the thalamic afferents, coronal sections through the barrel cortex of a GTE mouse were immunostained for 5HTT. At P6, the GFP neurite innervation from the subplate is confined to the barrels (filled arrowheads in J) similarly to the serotonin transporter immunoreactivity pattern (I) but by P10 and P14 the GFP neurite innervation and the serotonin immunohistochemistry pattern do not coincide with each other (L, M, O and P). GFP-labelled fibres are confined to the septa at P10 and P14 (open arrowheads in M and P). Abbreviations: mz marginal zone; cp, cortical plate; sp, subplate; wm, white matter; IV, layer IV; hp, hippocampus. Scale bar, 100 μm (A–C, I–Q) and 200 μm (D–H).

Figure 2. Age specific rearrangement of the GFP positive neurite and the effect of whisker removal at birth.

A–F, fluorescence micrographs of the barrel cortex in the tangential plane of the Golli-tau-eGFP (GTE) mouse revealed a changing pattern of GFP-labelled neurite distribution in layer IV between P2–14. The initially homogeneous distribution (A–C) changes to hollow (P6, C), then to septa (P10–14, D–F). G–L, the effect of neonatal whisker removal on the GFP neurite patterning in the P10 barrel cortex. Bright field (G and J) and fluorescence micrographs (H, I, K and L) comparing the two barrel cortices in the tangential sections of a Golli-tau-eGFP mouse at P10. G, H and I are from the control side (ipsilateral hemisphere to the whisker removal) and J, K and L are from the same mouse contralateral to the whisker ablation from row ‘a’ at P0. Dashed rings encircle the row ‘a’ region of the barrel cortex. Scale bar for A–F, 200 μm; for G–L, 300 μm.

The pattern of GFP-labelled neurites and thalamocortical afferents are in register in the barrel cortex until P6

In all the E17, P0 and P2 brains studied, GFP⁺ neurites were clearly seen to project through the cortical plate and into the marginal zone in a homogenous fashion (Figs 2A–B and 3A–C). We believe this GFP signal can be attributed to termination and arborisation of neurites originating from the subplate, because no GFP⁺ cells were seen in the marginal zone in any of the brains studied (n= 47). At P4, GFP-labelled fibres started to assume regular patterning in layer IV, which was observed in all brains studied on coronal sections (n= 3). GFP fibres occupied the barrel hollows, as delineated by the bisbenzimide counterstain, suggesting intra-barrel patterning (Fig. 3D). At P6 (n= 5), the concentration of GFP neurites within the barrel hollows was greater than at P4 (Figs 2C and 3E) and all individuals showed the same intrabarrel projection pattern. However, by P8 the characteristic intrabarrel pattern started to diminish in both coronal and tangential sections (Figs 2D and 3F). To correlate the patterning established by thalamic afferents with the subplate neurite clustering in layer IV of the somatosensory cortex, we aligned the images obtained by immunohistochemistry to 5HTT with the GFP-labelling pattern in the same coronal sections. At P6 the thalamic axons grouped in layer IV inside the barrel hollows (Fig. 3I) together with the GFP neurites (Fig. 3J and K).

By P8–10 the GFP-labelled neurites accumulate in the septa, while thalamocortical afferents remain in the barrel hollow

By P10 and P14, the cortical projecting GFP neurites accumulated increasingly along layer V/IV boundary, directly below the barrels of layer IV. Interestingly, neurites reaching layer IV started to concentrate between the barrels (Fig. 3G–H) which could be even more clearly observed in the tangential sections through layer IV (Fig. 2E–F). These inter-barrel projections spanned the thickness of layer IV at P10 but not through the entire height of layer IV at P14 forming a characteristic undulating band along the lower border of layer IV (Fig. 3G–H). Correlating the patterning of thalamic efferents and subplate neurites, we observed that at P10, the thalamic axons were still concentrated inside the barrel hollows (Fig. 3L) while the GFP-labelled neurites concentrated onto the septa between barrels (Fig. 3M and N). This pattern remained until P14, the latest stage studied here (Figs 2E and F, and 3O–Q).

Lack of GFP neurite rearrangement in areas corresponding to neonatal whisker removal

We investigated whether the age-specific GFP neurite distribution was dependent on the sensory periphery. At P0 all whiskers of row ‘a’ were removed on the right hand side of the snout in the GTE mouse. From P0–P5, the sensory environment is believed to exert its influence on the patterning of the somatosensory cortex (Van der Loos & Woolsey, 1973). The mice were raised as their control littermates and perfused at P10. At this age the barrel cortex is well developed and a consistent inter-barrel GFP neurite pattern was established (Fig. 2E and H). Also, to control for any variation in GFP neurite patterning within the barrel cortex, a row of whiskers was removed unilaterally to allow comparison with the contralateral row ‘a’. Analysis of the tangential sections revealed a clear periphery-related change in the contralateral barrel cortex resulting from whisker deprivation, while the ipsilateral barrel cortex was unaffected. Cytochrome oxidase staining of thalamic afferents revealed the altered periphery-related pattern of the whiskers in row ‘a’. Plasticity-related mild changes extending to rows ‘b’ and ‘c’ were also observed (compare Fig. 2G with J). Bisbenzimide staining of the cytoarchitectonic organisation confirmed the absence of barrels in the region of row ‘a’ in comparison to the unaffected rows and contralateral hemisphere (Fig. 2I and L). The GFP neurite organisation was homogenous in the region of row ‘a’ and indistinguishable from the area surrounding the barrel cortex (Fig. 2K). The inter-barrel GFP pattern in the remaining regions of the barrel field, corresponding to the territory of rows B–E (Fig. 2K), was unaffected. This arrangement was confirmed in adjacent sections at the level of layer IV in all GTE mice studied for the effects of row ‘a’ whisker removal (n= 4).

We extended our investigation on the effect of sensory input on the rearrangement of the GFP⁺ neurites to additional stages. The whiskers from row ‘c’ were removed on the right hand side of the snout in the GTE mouse at the age of P0 and the contralateral and ipsilateral barrel field were examined at the age of P5 or P10 (Fig. 4A and C). In order to investigate the effect of the whisker removal at a later stage, we performed ‘delayed’ whisker ablation experiments at P5 and the distribution of the GFP neurite innervation on the barrel field was studied at P10 (Fig. 4B).

Figure 4. The effect of the timing of row ’c’ whisker removal on the GFP neurite patterning in the somatosensory cortex of the Golli-tau-eGFP mouse.

Three different protocols were investigated: row ‘c’ was removed at P0 and studied at P6 (A), row ‘c’ was removed at P5 and studied at P10 (B) and row ‘c’ was removed at P0 and studied at P10 (C). Fluorescence micrographs illustrate the GFP (left), 5HTT (middle) and GFP and 5HTT merged patterning (right column) on coronal sections at P6 (A) and P10 (B and C) from the contralateral (labelled with 1) and ipsilateral (labelled with 2) side to the whisker removal. Arrowheads in B1 and C1 point to the area with remaining GFP neurite innervation towards the barrel hollow, corresponding to the region innervated by the removed whiskers row ‘c’. There is no clear change in the GFP neurite patterning at P6 after P0 whisker ablation. This further supports the observations (illustrated in Figs 1 and 2), that the GFP innervation patterning on the barrel cortex changes from an intra-barrel to an inter-barrel pattern after P6. Scale bar, 200 μm.

Analysis of coronal sections throughout the somatosensory cortex at P6 after P0 row ‘c’ removal showed no noticeable difference on the GFP neurite patterning in the contralateral side when compared to the ipsilateral barrel field, suggesting the effect of sensory input in the reorganisation of the GFP neurite rearrangement might take place at a later stage than P5 (Fig. 4A1–A1″, A2–A2″). In contrast, GFP neurite accumulation to the barrel hollow was observed on the contralateral corresponding region innervated by row ‘c’ of the barrel cortex at P10 brains after row ‘c’ removal at P0 (Fig. 4C1–C1″) or at P5 (Fig. 4B1–B1″). The cortical territory innervated by the ablated whiskers from row ‘c’ showed residual neurite innervation within the corresponding barrel hollows (arrowheads in B1 and C1), suggesting that the reduced sensory experience between P5 and P10 had reduced influence on the GFP neurite patterning (Fig. 4B and C). The 5HTT staining (middle column in Fig. 4) revealed the periphery-related arrangement of the thalamic efferents at P6 (A1′ and A2′) and P10 (B1, B2, C1′ and C2′).

Discussion

The present study of the GTE mouse demonstrated that the organisation of GFP neurites in the barrel cortex is age specific and occurs during the early postnatal period (within the timeframe studied, E17–P14). GFP-labelled neurites extend into the cortical plate from E17 and by P4 they establish characteristic intra-barrel patterning. This occurs at the same age as the characteristic periphery-related thalamic fibre clustering can be detected by cytochrome oxidase staining or with serotonin transporter immunohistochemistry. Both thalamic afferent and subplate neurite patterns become more defined to the intra-barrel region by P6. However, between P10 and P14 the GFP-labelled neurite pattern gets inversed to inter-barrel patterning with GFP neurites concentrated to the septa of the barrel cortex. Pari passu thalamic afferents remain clustered within the barrels. We propose that the GFP neurite rearrangement is modulated by the sensory periphery because the neonatal ablation of a whisker row results in the failure of GFP neurites to change from intra- to extra-barrel pattern in the region corresponding to whisker deprivation in comparison to the controls. Our comparison of the barrel field at P6 and P10 after P0 row ‘c’ whisker removal (data not shown) suggests that most changes in the GFP neurite patterning occur after P6. There are several questions remaining to be addressed before the functional significance of the age specific GFP neurite arrangements and their dependence on the sensory periphery can be considered.

Origin and nature of the eGFP neurites

It is not known if the GFP neurites projecting into the cortex are axons or dendrites. Friauf and colleagues reported axon collaterals of SPn entering the cortical plate and arborising in layer IV (Friauf et al. 1990). It is likely that some of the GTE neurites projecting to layer IV are axons because tau has been shown to be a neuron-specific marker particularly for axons (Binder et al. 1985). However, the possibility that eGFP is expressed in large dendrites cannot be excluded at present. Indeed, Jacobs et al. (2007) demonstrated co-localization of MAP2 and GFP in neurons expressing the tau-GFP transgene in the embryonic Golli mouse. Co-localization of MAP2 with some of the GFP-positive processes from subplate cells towards upper cortical layers was described, suggesting the possibility that eGFP is expressed in larger subplate dendrites. There are examples for tau expressed in dendrites and MAP2 expressed in axons in the literature (Hirokawa et al. 1996; Kosik & Finch, 1987); therefore we need to study the nature of the GFP⁺ neurites in layer IV in GTE mouse by additional methods.

There are several cortical cell types where dendritic remodelling is reported around this period of development. Koester & O’Leary (1992) and Kasper et al. (1994) described apical dendrite remodelling in layer V callosal projection neurons and Peinado & Katz (1990) in cortical spiny stellate cells. Recently we observed such dendritic remodelling in subplate neurons with long-range projections to subcortical targets or to the contralateral hemisphere between P2 and P7 (Hoerder et al. 2006).

Whilst the GTE mouse is a very convenient model to study subplate cell integration into the intra- and extra-cortical circuitry, it does have limitations. We cannot neglect the possibility that the GFP neurite pattern in layer IV may also be derived from GFP projections of scattered GFP⁺ cells located in layer VIa or V, even though the density of cells in these layers is considerably less then that of the SP (Aye et al. 2006; Jacobs et al. 2007). Furthermore, little is known about the characteristics of the eGFP-labelled cells of the GTE mouse. Previous work by Antonini & Shatz (1990) suggested that the subplate contains GABAergic interneurons and glutamatergic pioneer neurons. Hevner & Zecevic (2006) described a large variety of neurons including pioneer neurons, glutamatergic projection neurons, mature interneurons and migrating interneurons in the subplate. Our own study revealed that not all subplate neurons express GFP in the GTE mouse. At P8, approximately 60% of all NeuN-positive cells in the subplate (VIb) are GFP positive (Aye et al. 2006). According to our study, 20% of the GFP⁺ subplate neurons are GABA immunoreactive. Although we found a considerable population of calretinin, calbindin and parvalbumin-positive neurons in the subplate and layer VIb but very few were GFP positive (below 3%). More work is needed on the characterisation of the GFP⁺ neurons in GTE mouse, in particular the ones developing projections to the barrel cortex. We cannot exclude the possibility of this neurite remodelling occurring due to the cell death of a certain population(s) of subplate/layer VIa cells during development. Although there is evidence that a considerable population of subplate cells undergo apoptosis during early development (Ferrer et al. 1990; Rakic & Zecevic, 2000) we have observed numerous GFP⁺ cells in the subplate and layer VIa until postnatal age P14 (see also Jacobs et al. 2007). We have also observed that a considerable population of subplate cells survive to adulthood and they continue to express markers which were expressed during embryonic and early postnatal development (Hoerder-Suabedissen et al. 2008b). Recently a microarray-based search for SP-specific markers has yielded numerous markers (Hoerder-Suabedissen et al. 2008b) including DOPA decarboxylase (DDC), Complexin3 and MoxD1. Quantitative co-localisation studies of such subplate markers in the GTE mouse will be very useful in the characterization of the eGFP⁺ cells. Interestingly, DDC immunohistochemistry also revealed a periphery-related pattern in the P8 mouse cortex (Hoerder-Suabedissen et al. 2008b) similar to the pattern observed in the GTE mouse. However, DDC does not co-localise with GFP in the P8 GTE mouse barrel cortex (M.C. Piñon, unpublished observations). This may suggest that there are additional subplate neurons with intracortical projections others than the ones expressing the Golli promoter. We should also consider reports in the literature where subplate subpopulations are transformed into layer VIb/VII or white matter cells during development (Valverde & Facal-Valverde, 1988; Woo et al. 1991; Reep, 2000; Clancy et al. 2001).

The high-density population of GFP⁺ neurites in the GTE mouse makes the tracing of individual GFP neurites from the SP into the cortex not possible. Thus, studies on the neurochemical properties of the GFP-labelled cells of the SP may not be directly correlated with the dynamic GTE patterning in the barrel cortex. Finding specific markers for the SP population with cortical projections will be especially important to further dissect cellular and molecular mechanisms and to be able to design mouse transgenic models to monitor and manipulate this particular subplate population.

Timing of the intra-barrel to inter-barrel pattern shift correlates to various anatomical and physiological changes in the barrel field

Our estimate for the intra-to-inter-barrel shift to occur is between P6 and P10. However, we observed some variation between individual GTE brains for the timing of the establishment of the periphery-related patterning (P4–6) and for the intra-barrel to inter-barrel switch (P8–10). The explanation of this variability might be related to methods (coronal versus tangential sectioning), the possible fading of the GFP signal and the natural fluctuation of the developmental stage between individuals of the same litter. Recently, a strong correlation has been demonstrated between the time the barrels appear and the bodyweight of the newborn pup by Hoerder-Suabedissen et al. (2008a).

The GTE neurites may project within the barrel hollows at P4–6 (P8) to impose the cytoarchitectonic pattern in layer IV, reflecting the sensory periphery-related pattern of the whiskers. However, it is less obvious why GFP neurites concentrate within the barrel septa and become established here at P10 when the thalamic afferents have formed functional connections within layer IV. The intra-barrel/inter-barrel GFP neurite patterning transition period correlates to the critical period during which the barrels can be modified by sensory perturbations (Van der Loos & Woolsey, 1973), to the substantial changes of cortical activation pattern elicited from thalamus (Molnár et al. 2000; Higashi et al. 2002, 2005), and to the ability to elicit LTP in thalamocortical slices (Crair & Malenka, 1995). Assuming that the GFP⁺ neurites are SPn derived, we can speculate on the functional significance of this dynamic integration pattern. Analogous to the visual cortex, barrels in the rat contain functional minicolumns, subbarrels, which have an angular preference (Bruno et al. 2003). Subbarrel patterning has been shown to occur between P8 and P10 in the rat (Land & Erickson 2005; Louderback et al. 2006), which coincides with the intra-barrel/inter-barrel GFP neurite patterning transition in this study at P6–10. Kanold and colleagues have shown in a SP-specific ablation study that SPn are required for strengthening and selective remodelling of thalamocortical synaptic connections which underlie development of visual cortical columns in the cat (Kanold et al. 2003). Therefore, by analogy, it may be plausible for the SPn in the rodent to be involved in the refinement of layer IV circuitry i.e. a role of the SPn in the functional maturation of the barrel cortical column.

The SP has major clinical importance and it is postulated to be involved in several neurodevelopmental cognitive disorders (Polleux & Lauder, 2004; Harrison & Weinberger, 2005). Furthermore, the selective susceptibility of the SP to neonatal hypoxia results in developmental conditions such as periventricular leukomalacia and cerebral palsy (Volpe, 2001; McQuillen & Ferriero, 2004). Therefore, to understand the mechanisms underlying these neurodevelopmental disorders better, knowledge of how the SP integrates into the developing cerebral cortex and the molecular mechanisms responsible for them will be necessary. The information provided in this study is one of the first steps to establish a genetically accessible mouse model system at the level of the developing cortical column.

Glossary

Abbreviations

DDC

DOPA decarboxylase

DiI,1

1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate

E

embryonic day

eGFP

enhanced green fluorescent protein

GTE

Golli-tau-eGFP

5HTT

serotonin 5-hydroxy-tryptamine

LTP

long tem potentiation

MAP2

microtuble associate protein 2

MoxD1

monooxygenase D1 gene

P

postnatal day

PBS

phosphate-buffered saline

PFA

paraformaldehyde

SP

subplate

SPn

subplate neurons

Author contributions

All authors contributed to the planning and design of the experiments as well as to the analysis and critical interpretation of the experimental data. They all approved the intellectual content and final version of the manuscript to be published. E.J. performed the whisker removal experiments and collected the material from the local breeding colony of GTE-eGFP mice at UCLA (Geffen School of Medicine, Semel Institute for Neuroscience, Neuroscience Research Building, Los Angeles, CA 90095-7332, USA). All subsequent analysis, documentation and drafting of the article were performed by M.C.P., A.J. and Z.M. (Department of Physiology, Anatomy and Genetics, The University of Oxford, Oxford, UK).