We analyzed the whisker to barrel projection in a previously characterized mouse line (Renier et al., 2010) in which the Robo3 gene has been specifically knocked out in rhombomeres 3 (r3) and r5, using the Krox20 promoter (Krox20:Cre;Robo3^lox/lox, hereafter referred to as Robo3^R3-5-cKO, with Krox20:Cre;Robo3^lox/+ referred to as Robo3^R3-5-Het). During development, Robo3 is transiently expressed in r3 neurons, with expression stopping shortly after axon crossing, suggesting that it does not play a role in later stages of development such as axon targeting (Badura et al., 2013; Michalski et al., 2013; Renier et al., 2010). As expected, in situ hybridization confirmed that Robo3 expression is deleted from r3 in 12-day-old (E12) Robo3^R3-5-cKO embryos (Figure 1B, n = 3/3).

To visualize the trigemino-thalamic pathway, we crossed Robo3^R3-5-cKO mice with Tau-lox-Stop-lox-mGFP-IRES-nls-lacZ mice (Tau^GFP) (Hippenmeyer et al., 2005). In E12 controls many GFP+ axons cross the midline at the r3 and r5 level (Figure 1B–D). The ßgal nuclear reporter showed a dense distribution of Krox20+ cell bodies in the ventral region of the PrV nucleus, as expected from previous fate-mapping experiments (Oury et al., 2006). In the Robo3^R3-5-cKO;Tau^GFP mice, the distribution of ßgal+ neurons and organization of barrelettes was normal, indicating that Robo3 deletion did not alter development of the PrV nucleus (Figure 1C and Figure 1—figure supplement 1). By contrast, most of the GFP-labeled axons arising from the PrV failed to cross, although they still projected rostrally towards the forebrain (Figure 1B–D; n = 3/3). At E13, coronal sections at the level of r3 showed that the density of GFP+ commissural axons at the ventral midline was strongly reduced in the Robo3^R3-5-cKO;Tau^GFP mice, but also indicated that a subset of axons still crosses (Figure 1B). Double staining for NeuN and the nuclear cre reporter ßgal in adult sections of Robo3^R3-5-Het;Tau^GFP mice (Figure 1—figure supplement 1; n = 5) showed that 88% of NeuN+ neurons in the ventral part of PrV express ßgal, but that a small subset of the NeuN+ PrV neurons (10.7 ± 1.5%) were not ßgal+ and probably did not express Cre recombinase.

As ephrins and their receptors have been shown to control targeting and orientation of thalamocortical projections for visual or somatosensory axons (Dufour et al., 2003; Pfeiffenberger et al., 2005), we used in situ hybridization to check the pattern of ephrin-A5 mRNA expression at P0 in controls and Robo3^R3-5-cKO mutants. The expression gradients of ephrin-A5 in the cortex and thalamus were not noticeably different in the mutants and the controls (Figure 1E, [n = 2)], suggesting that deletion of Robo3 in the brainstem did not affect expression of patterning cues in the thalamus and cortex.

Three-dimensional imaging of the trajectory of the trigemino-thalamic (TT) tract from the brainstem to the thalamus using iDISCO (Renier et al., 2014) (Figure 2A) revealed that at P4, the GFP+ axons had a similar trajectory in the TT tract in both control and mutant mice (n = 5), although the tract appeared slightly more defasciculated in the mutants. GFP+ axon terminals arborized in the VPM in both control and mutant mice and formed barreloids (Figure 2A,B; Video 1). In all mutants and controls (n = 14 for each genotype), the r3-derived GFP+ (r3-GFP+) axons projected to the dorsolateral VPM containing the barreloids (Figure 2B). Abnormal organization of the whisker barreloids was noted in the mutant VPM. In controls, all the barreloid rows coincided with a dense r3-GFP+ axon territory (Figure 2B; n = 4/4), whereas in mutants two distinct zones were observed (n = 4/4): a lateral VPM domain containing a high density of GFP+ axons and a medial VPM domain with only sparse patches of GFP+ axons (Figure 2B). These two VPM domains contained barreloids, identified by cytochrome oxidase staining, and were of comparable surface area in coronal sections through the middle of the VPM (0.19 ± 0.006 mm² for the lateral dense GFP+ domain and 0.18 ± 0.003 mm² for the medial patches of GFP+ domain, p=0.46). Moreover, they were separated by a cytochrome oxidase-free septum. These observations suggest that in the Robo3^R3-5-cKO mice, the VPM is split into two separate domains, each with different barreloid patterning (although the lateral domain contains the highest density of r3-derived projections, and larger barreloids).

Figure 2 with 1 supplement see all

Download asset Open asset

Video 1

Download asset

We traced the PrV to VPM projections anterogradely in P4 Robo3^R3-5-cKO mice using carbocyanine dyes. In control mice the PrV-VPM projection was completely crossed, whereas in mutants the VPM received bilateral innervation from the PrV (Figure 2C). Moreover, the position, shape and size of the traced projections in the VPM was reminiscent of the two domains described previously, suggesting that in the mutants, the dense GFP+ lateral region might correspond to abnormal ipsilateral projections from the PrV, while the medial patches might originate from the contralateral side.

These observations indicate that the mutant VPM receives segregated ipsilateral and contralateral trigemino-thalamic inputs. Retrograde injections from the VPM labeled cell bodies on both contralateral and ipsilateral trigeminal PrV nuclei in Robo3^R3-5-cKO mutants (3.3 times (n = 3) more cell bodies were labeled ipsilaterally than contralaterally) (Figure 2D; n = 4/4). Interestingly, the neurons projecting ipsilaterally and contralaterally were mixed in the ventral PrV in the Robo3^R3-5-cKO mutants, in contrast to the segregation of their projections seen in the anterograde tracings.

Overall these data suggest that a large fraction of the rhombomere 3-derived trigemino-thalamic axons project ipsilaterally in Robo3^R3-5-cKO mutants, but that some still project contralaterally either because Cre recombination is incomplete or this occurs after crossing.

It is possible that the time of arrival of the ascending axons in the VPM could be a factor in segregation of the ipsilateral and contralateral domains because of shorter paths in the mutants. Ipsilateral axons in mutants may reach the VPM first and occupy the dorso-lateral quadrant of the VPM normally populated by the larger barreloids in controls. We looked at development of the trigemino-thalamic (TT) tract at E15.5 in Robo3^R3-5-cKO;Tau^GFP mutant embryos, when axons from the PrV have not yet reached their targets in the VPM (Figure 2—figure supplement 1A, n = 3) (Kivrak and Erzurumlu, 2013). In this line, both ipsilateral and contralateral PrV projections to the VPM are GFP+. To selectively label contralateral PrV axons, we injected DiI unilaterally into the PrV. In the TT tract, red axons (DiI+ only), yellow axons (double positive for GFP and DiI), and green axons (GFP+) were organized in a medial to lateral gradient (Figure 2—figure supplement 1A, n = 3), suggesting that ipsilateral and contralateral axons in the TT tract might be pre-sorted before reaching their target. Ipsilateral (green only) axons were always seen next to contralateral axons (red and yellow) in the TT tract up to the rostral-most sections containing the endings/growth cones of the developing axons. Therefore, ipsilateral axons had no measurable lead over contralateral axons before reaching the VPM at this stage. Indeed, in both Robo3^R3-5-Het;Tau^GFP and Robo3^R3-5-cKO;Tau^GFP mutant E18 embryos, the TT have reached the VPM and the terminals fill the whole dorsal region (Kivrak and Erzurumlu, 2013) (Figure 2—figure supplement 1B, n = 3). However, the precise timing of arrival of the contralateral and ipsilateral axons in the mutant VPM is difficult to assess. Although we did not find evidence for delayed arrival of the contralateral TT projections in the VPM, we cannot rule out that ipsilateral projections reach their targets earlier and hence have a competitive advantage to innervate the VPM.

To determine the 3D organization of the VPM map in the Robo3^R3-5-cKO mutants, and the origin of the GFP+ patches, we performed whole-mount imaging of brains from Robo3^R3-5-cKO;Tau^GFP mutants and heterozygous controls at P8 using iDISCO (Figure 3)(Belle et al., 2014; Renier et al., 2014). A suitable angle was determined for optimal projection of the barreloids in the thalamic whisker map onto a plane: 3D datasets were oriented at a 45° oblique angle from both coronal and horizontal planes (Figure 3A, Figure 3—figure supplement 1A and Video 2). In the Robo3^R3-5-cKO;Tau^GFPmutants, the GFP dense and patched regions were manually segmented to show their respective 3D domains. In both regions, barreloids were organized in rows reminiscent of the control VPM map (Figure 3A and Figure 3—figure supplement 1B, n = 3).

Figure 3 with 1 supplement see all

Download asset Open asset

Video 2

Download asset

To determine whether each region corresponded to a distinct whisker map, we performed unilateral lesions of the infraorbital nerve (ION) at P0 and the GFP+ projections in the VPM were imaged at P8. In control mice, the unilateral ION lesions caused fusion of the barreloids in the contralateral VPM (Figure 3B and Figure 3—figure supplement 1C; n = 3). In mutants, barreloid fusion was found in both the ipsilateral and the contralateral VPM: ipsilaterally, in the GFP-dense region, and contralaterally in the GFP-patched region (Figure 3B, n = 5). This indicated that, as suggested by the tracing experiments, the GFP-dense region in the VPM receives ipsilateral inputs from the PrV, whereas the GFP-patched region receives contralateral inputs. This also demonstrated that the GFP-patched region carries somatosensory inputs from the infraorbital branch of the trigeminal nerve. Moreover, in addition to fusion of the barreloids, the thalamic map that sustained sensory deprivation was reduced in size, while the adjacent non-deprived map was enlarged (Figure 3C, n = 5). This showed that sensory-activity-based competition defines the final space allocated to each map in the mutant VPM.

To verify whether both maps in the VPM received inputs from the periphery, we performed intact brain c-Fos immunolabeling in control and mutant mice whose whiskers were shaved on the left side, and b, d rows were spared on the right side (n = 3). The intact brain immunolabeling gave us the opportunity to navigate the complex 3D organization of the VPM using arbitrary oblique projection planes (Figure 4A). In controls, two bands of c-Fos+ cells were seen contralateral to the spared whiskers, revealing the barreloids in rows b and d (Figure 4B). In Robo3^R3-5-cKO mice, a dual band pattern in the VPM was seen on both sides of the brain. In the ipsilateral VPM, the bands were visible on the same intersecting plane as controls. On the contralateral VPM, the bands of activity were visible on a more medial plane, at the edge of the VPM annotation.

Figure 4

Download asset Open asset

Overall these experiments provide a model for organization of the VPM in Robo3^R3-5-cKO mutant mice (Figure 3D and Figure 4C). In the mutant VPM, sensory inputs from the ipsilateral PrV establish a dorsolateral map, with dense projections. Adjacent to this map, inputs from the contralateral PrV project to the dorso-medial VPM as discrete patches. These two thalamic maps are organized in rows, reminiscent of the normal thalamic map. Our finding that whisker stimulation triggers activity-related expression in the two VPM maps in the mutant suggests that the barreloid organization is functional (Figure 4C).

Next, we determined how organization of the VPM in the mutant influences formation of the somatosensory map in the cerebral cortex. Tangential sections through layer 4were stained for cytochrome oxidase and Vglut2 immunoreactivity to label thalamocortical afferents (Nahmani and Erisir, 2005) (Figure 5A and Figure 5—figure supplement 1A). A striking abnormality in the layout of thalamic afferents was noted in the posteromedial barrel subfield (PMBSF, which corresponds to the representation of the large whiskers) of S1 in Robo3^R3-5-cKO mice (n = 5/5; Figure 5A). Large barrels were extra-numerous (52 ± 2 barrels in mutant PMBSF vs. 33 ± 0 in controls), and were reduced in size (0.04 mm² ±0.01 per barrel in mutants, compared with 0.09 mm² ±0.02 in controls, p<0.0001). Moreover, they were arranged into eight rather than the usual five whisker rows, with clear delineation of two separate cortical zones, a central zone and a peripheral zone, each containing distinct barrel rows (Figure 5A and Figure 5—figure supplements 1–3). These abnormalities were similar in both hemispheres and at all ages analyzed (with only slight individual variations; n = 25/25; Figure 5—figure supplements 1–3). To map the functional whisker representation in this unusual map, we monitored activation of the immediate early gene c-Fos following a 1-hour exposure to an enriched sensory environment (Staiger et al., 2000). In mice with unilateral trimming of the whiskers (Figure 5B), strong c-Fos labeling is normally observed only in the S1 contralateral to the intact whiskers. In Robo3^R3-5-cKO mice with unilateral whisker trimming, c-Fos was activated in the PMBSF of both hemispheres (Figure 5B). Contralateral to the intact whiskers, c-Fos activation was visible in the peripheral barrel rows (contra domain; Figure 5B). Ipsilateral to the intact whiskers, a mirror image was noted with c-Fos activation in the central barrel rows (Ipsi domain; Figure 5B). In both contralateral and ipsilateral patterns, c-Fos+ cells were detected in all layers from the columns of the stimulated barrels (Figure 5B-Figure 5—figure supplement 3B). These results show that in Robo3^R3-5-cKO mice, the crossed and uncrossed trigemino-thalamic inputs are mapped as two segregated domains, with the ipsilateral map nested within the contralateral map (Figure 5B).

Figure 5 with 3 supplements see all

Download asset Open asset

The segregation of the two maps led to the question of their topographic organization. Two scenarios are possible: (i) the thalamic afferents follow topographic molecular guidance cues expressed in the cortex, with complementary receptor expression in the thalamus; in this case one would expect that neighboring whiskers of the ipsi- and contralateral maps lie in register with one another; (ii) the thalamic afferents are clustered following sensory activity-based rules leading functionally coordinated afferents to cluster together; in this scenario, the topographical rule of near neighbors would prevail over molecular gradients. To analyze the topographic alignment of the crossed and uncrossed somatosensory maps, we monitored c-Fos expression in S1 after clipping all whiskers except one row or one arc of whiskers on one side (Figure 5C and data not shown). When the five posterior-most whiskers of the whisker pad (A1-E1; Figure 5C) were left intact in control mice, this resulted in activation of c-Fos in a caudal arc of five barrels exclusively in the contralateral S1 (Figure 5C). Likewise, when the second whisker row (B1-B4; Figure 5C) was left intact, the corresponding row of barrels was activated in the contralateral S1 (Figure 5C). In mutants, c-Fos activation was bilateral, with labeling in both the central (ipsi) and peripheral (contra) PMBSF domains (Figure 5C). The general orientation of the two nested maps was similar and resembled that of control mice, likely because of the patterning activity of morphogens that determine the polarity of the map (Fukuchi-Shimogori and Grove, 2001, 2003). However, unlike the visual or auditory bilateral maps, we observed that there were discontinuities in organization of the bilateral somatotopic map, such as a lack of topographic proximity of the ipsi/contra representation for a given barrel or barrel row. Rather, there appeared to be clear separation and independence of the ipsi- and contralateral inputs. Taken together, these results favor the hypothesis that nearest-neighbor’s interactions prevail to some extent over the molecular pre-patterning to organize continuous representation of the periphery for each map. However, the molecular gradients still contribute to maintain the general orientation of the maps.

In the mouse models studied here, the space occupied by the barrel map was not increased in mutants (2.32 ± 0.05 mm² in controls vs 1.90 ± 0.12 mm² in mutants, p=0.04), unlike results from other mouse models with duplication of the S1 map where a second S1 map is formed at the expense of other cortical areas (Fukuchi-Shimogori and Grove, 2001, 2003). This suggests that the ipsi- and contralateral thalamic inputs compete to occupy a defined cortical space in S1. Accordingly, individual barrels in mutants were roughly half the size (44 ± 2%) of those in controls. As with the VPM, we looked again at the consequences of unilateral deprivation of whisker inputs induced by a neonatal (P1) lesion of the ION (Waite and Cragg, 1982). In control mice, barrel fusion was observed in the S1 contralateral to the lesion (Figure 6A; n = 3/3). In mutants, barrels fused in the maps corresponding to the ipsi- and contralateral representations of the lesioned whisker pad (Figure 6A; n = 5/5). Furthermore, the representation of the unlesioned side expanded at the expense of the fused-map. This suggests that there is sensory activity-dependent competition for cortical space between the ipsilateral and contralateral sensory inputs in the mutant S1.

Figure 6

Download asset Open asset

Finally, we checked whether the ipsilateral and contralateral maps were functionally isolated. We took advantage of the ClearMap pipeline (Renier et al., 2016) to compare c-Fos activity patterns in the whole brain in an unbiased way, in the unilateral whisker stimulation protocol (n = 3 per group) (Figure 7A and Figure 7—figure supplement 1). We looked for brain regions that exhibited left-right differences that were opposite in controls and mutants. As expected, the barrel cortex and VPM exhibited statistically significant differences between the shaved and stimulated sides of the brain, that were opposite between controls and mutants (Figure 7A). Of note, the column of activity detected in d row registered precisely in the same position in control and mutants, showing that the absolute position of the d row in the brain is the same in the aberrant ipsilateral map of the mutant mice a in controls. We then isolated c-Fos+ cells from the upper cortical layers (n = 4). In upper layers, contrary to layer 4, activity patterns are not restricted to the stimulated barrels (Figure 7B), but expand over adjacent barrels, due to downstream cortical integration (Kaliszewska et al., 2012; Peron et al., 2015). We looked at the effect of a patterned sensory deprivation created by trimming rows b and d on one side and shaving all whiskers on the other side. Expansion of c-Fos+ cells was observed within both maps between activated rows in the upper cortical layers 2–3 (Figure 7B), but not across the boundaries of each map. This suggests that the ipsilateral and contralateral whisker maps have little to no direct horizontal integration in the upper cortical layers.

Figure 7 with 1 supplement see all

Download asset Open asset

Contrary to previous observations of experimental map duplication (Fukuchi-Shimogori and Grove, 2001), the present orientation of the two whisker maps was similar in the general rostrocaudal and mediolateral axes, indicating that matching gradients of guidance molecules and their receptors was most likely unchanged, which was confirmed for ephrin-A5 (Figure 1E). This contrasts with the mirror image organization of the sensory maps obtained when inducing novel sources of molecular gradients in the somatosensory cortex (Fukuchi-Shimogori and Grove, 2001). This also contrasts with observations of map duplication in the visual system, caused by a change in the retinal axon crossing at the midline (Petros et al., 2008; Rebsam et al., 2009), by lack of one eye (Trevelyan et al., 2007) or by lack of one hemisphere (Muckli et al., 2009).

The normal orientation of the ipsilateral map in the mutants is surprising. One might have expected that switching laterality of the normally crossed projections would flip the axis of the ipsilateral map from the contralateral map to account for the chiral organization of the left and right sides of the face. As both maps in the mutants respect the normal orientation, the ascending tract from the ipsilateral side may either undergo torsion en route to the VPM to correct the orientation based on molecular gradients present in the lemniscal pathway. Alternatively, the correction of the orientation may occur only at the target site in the VPM based on the gradients of expression of Eph receptors/ephrin ligands in a process akin to the visual tectum (Feldheim et al., 1998; Tessier-Lavigne, 1995). If true, this hypothesis implicates the presence of additional mechanisms of axonal pruning and refinement, as seen during the post-targeting development of visual projections (Nakamura and O'Leary, 1989; Simon et al., 2012) to correct the final orientation of the ipsilateral map.

In conclusion, although the initial wiring of the brain relies largely on genetically encoded processes, our results illustrate the remarkable plasticity of the mammalian brain and its ability to accommodate changes in afferent wiring in evolution to create new maps and bilateral representations. These results illustrate also the brain's ability, in the context of developmental brain disorders, to compensate for major axon guidance defects that otherwise would lead to severe brain dysfunction (Jen et al., 2004; Muckli et al., 2009; Williams et al., 1994).

All animal procedures were carried out in accordance with institutional guidelines (UPMC, Charles Darwin ethic committe and INSERM). Mice were anesthetized with ketamine (Virbac) and xylazine (Rompun). The day of vaginal plug is embryonic day 0 (E0) and the day of birth corresponds to postnatal day 0 (P0).

The Robo3 conditional knockout, Krox20:Cre knock-in and Tau^GFP lines were as previously described (Hippenmeyer et al., 2005; Renier et al., 2010; Voiculescu et al., 2000). Unless otherwise mentioned, controls were Robo3^lox/lox or Krox20:Cre;Robo3^lox/+ animals. Double heterozygotes were always similar to wild-type mice. Mice were genotyped by PCR.

The following primers were used for genotyping: the conditional Robo3 allele, 5'-CCA AGG AAA AAC TTG AGG TTG CAG CTA G-3' and 5'-GAT TAG GGG AGG TGA GAC ATA GGG-3', the Krox20:Cre allele, 5'-AGT CCA TAT ATG GGC AGC GAC-3' and 5'-ATC AGT GCG TTC GAA CGC TA-3', and the Tau^GFP allele, 5'-GAG GGC GAT GCC ACC TAC GGC AAG-3' and 5'-CTC AGG GCG GAC TGG GTG CTC AGG-3'. All PCR had 34 cycles with an annealing temperature of 58°C.

In the Tau^GFP line, upon Cre recombination in neurons, the Stop cassette is excised leading to permanent expression of a myristoylated GFP in axons and of ß-galactosidase in nuclei (Hippenmeyer et al., 2005).

Mice were perfused transcardially with 4% PFA in 0.12 mM phosphate buffer. Cortices were flattened between microscope slides and post-fixed in 4% PFA and vibratome (Leica Microsystems, Germany) sectioned at 50 µm. Hindbrains and thalamus were post-fixed in 4% PFA, cryoprotected in 30% sucrose and sectioned at 35 µm with a freezing microtome (Microm, France).

For cytochrome oxidase staining (Melzer et al., 1994), sections were incubated at room temperature for 24 hours in 10% sucrose, 0.3 g/L cytochrome C from equine heart (Sigma), 0.02 g/L catalase from bovine liver (Sigma-Aldrich, Saint Louis, MO) and 0.25 g/L DAB (Sigma-Aldrich). The endogenous fluorescence of the GFP in Krox20:Cre;Tau^GFP was not affected after treatment and could be imaged on the same sections; however, the GFP signal was further enhanced by immunostaining.

For immunohistochemistry, neonatal and adult brains were processed as described previously (Marillat et al., 2002). The following primary antibodies were used: guinea pig anti-Vglut2 (1:1000, Millipore AB-2251), rabbit anti-βGal (1:1000, 55976 Cappel, Cochranville, PA), rabbit anti-c-Fos (1:1000, sc-52 Santa-Cruz biotechnology, Santa Cruz, CA, on sections, or 1:2000, 226–003 Synaptic Systems, Germany, for iDISCO+ studies), rabbit anti-GFP (1:300, A11122 Thermo Fisher, Waltham, MA, on sections), and chicken anti-GFP (1:800, ab13970 Abcam UK on sections or 1:2000, GFP-1020 Aves, Tigard, OR, for iDISCO studies). The following secondary antibodies were used on sections: donkey anti-mouse, anti-rabbit and anti-guinea pig coupled to CY3 or CY5 (1:600, Jackson Laboratories, West Grove, PA), donkey anti-mouse, anti-rabbit and anti-chicken coupled to Alexa Fluor 488, 568 or 657 (1:600, Thermo Fisher) for sections and iDISCO studies. Sections were counterstained with Hoechst 33258 (10 µg/mL, Sigma-Aldrich).

Sections were examined using a fluorescent microscope (DM6000, Leica Microsystems) equipped with a CoolSnapHQ camera (Princeton Instruments, Trenton, NJ), a confocal microscope (FV1000, Olympus, Japan), or a slide scanner (Nanozoomer, Hamamatsu, Japan).

Adult mice or P4, P8 pups were euthanized with a rising gradient of CO₂ and fixed with intracardiac perfusion of 4% PFA in PBS. All harvested samples were post-fixed overnight at 4°C in 4% PFA in PBS.

Fixed samples were washed in PBS for 1 hour twice, then in 20% methanol (in ddH₂O) for 1 hour, 40% methanol for 1 hour, 60% methanol for 1 hour, 80% methanol for 1 hour, and 100% methanol for 1 hour twice. Samples were then bleached with 5% H₂O₂ (1 vol of 30% H₂O₂ for 5 volumes of methanol, ice cold) at 4°C overnight. After bleaching, samples were re-equilibrated at room temperature slowly and re-hydrated in 80% methanol in H₂O for 1 hour, 60% methanol/H₂O for 1 hour, 40% methanol/H₂O for 1 hour, 20% methanol/H₂O for 1 hour, and finally in PBS/0.2% TritonX-100 for 1 hr twice.

Pre-treated samples were then incubated in PBS/0.2% TritonX-100/20% DMSO/0.3M glycine at 37°C for 36 hours, then blocked in PBS/0.2% TritonX-100/10% DMSO/6% donkey serum at 37°C for 2 days. Samples were then incubated in primary antibodies: chicken anti-GFP (1:2000, Aves GFP-1020), rabbit anti-c-Fos (1:2000, Synaptic Systems, 226–003) in PBS-Tween 0.2% with heparin 10 µg/mL (PTwH)/5% DMSO/3% donkey serum at 37°C for 4 to 7 days. Samples were then washed in PTwH for 24 hours (five changes of the PTwH solution over that time), then incubated in secondary antibody donkey anti-rabbit-Alexa647 from Thermo Fisher or donkey anti-chicken from Jackson Immunoresearch at 1:500 in PTwH/3% donkey serum) at 37°C for 4 to 7 days. Samples were finally washed in PTwH for 1 day before clearing and imaging. Immunolabeled brains were dehydrated in 20% methanol (in ddH₂O) for 1 hour, 40% methanol/H₂O for 1 hour, 60% methanol/H₂O for 1 hour, 80% methanol/H₂O for 1 hour, and 100% methanol for 1 hour twice. Samples were incubated overnight in 1 vol of methanol/2 vol of dichloromethane (DCM, Sigma-Aldrich 270997−12 × 100 ML) until they sank to the bottom of the vial (plastic Eppendorf tubes were used throughout the process). The methanol was then washed for 20 minutes twice in 100% DCM. Finally, samples were incubated (without shaking) in dibenzyl ether (DBE, Sigma-Aldrich 108014–1 KG) until clear (about 30 minutes) and then stored in DBE at room temperature.

Cleared samples were imaged in sagittal orientation (right lateral side up) on a light-sheet microscope (Ultramicroscope II, LaVision Biotec, Germany) equipped with a sCMOS camera (Andor Neo, UK) and a 2X/0.5 objective lens (MVPLAPO 2x) equipped with a 6 mm working distance dipping cap. Version v144 of the Imspector Microscope controller software was used. The microscope is equipped with LED lasers (488 nm and 640 nm) with three fixed light-sheet generating lenses. Scans were made at the 0.8X zoom magnification (1.6X effective magnification), with a light-sheet numerical aperture of 0.1. Emission filters used are 525/50 and 680/30. The samples were scanned with a step-size of 3 µm using the continuous light-sheet scanning method with the included contrast blending algorithm for the 640 nm channel (20 acquisitions per plane with a 50 ms exposure), and without horizontal scanning for the 480 nm channel (50 ms exposure). To speed up the acquisitions, both channels were acquired in two separate scans.

Maximum 3D projections in Figure 2A and all panels of Figures 3, 4 and 7B were performed using Imaris (Bitplane, http://www.bitplane.com/imaris/imaris), and generated from manual 3D segmentation of the raw data using the surface tool. ClearMap (Renier et al., 2016) (https://www.idisco.info) was used to quantify and register c-Fos+ cells in Figure 7A. Parameters were set as previously described, and automated isolation of the cortex was done using scripts available online.

Antisense riboprobes were labeled with digoxigenin-11-D-UTP (Roche Diagnostics) as described previously (Marillat et al., 2002) by in vitro transcription of mouse cDNAs encoding robo3 or an exon specific probe of robo3 targeting the floxed region (Renier et al., 2010).

4% PFA fixed P4 pups were injected with small crystals of 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (DiI, Thermo Fisher) and 4-(4-(dihexadecylamino)styryl)-N-methylpyridinium iodide (DiA, Invitrogen) using glass micropipettes. For anterograde tracing, the dye crystals were injected unilaterally in the PrV. For retrograde tracing of the PrV nuclei, the cortices were removed to expose the thalamus and DiI or DiA crystals were at the level of the VPM.

Brains were kept at 37°C for 4 weeks. Brains were cut in 80 µm sections with a vibratome (Leica) and counterstained with Hoechst.

P0-P1 pups were cold anesthetized, and an incision was made between the whisker pad and the eye. The nerve was cut with scissors under a dissecting scope. The pups were allowed to recover for 10 days and then perfused.

P20-P30 mice were anesthetized with ketamine, and all whiskers were trimmed on the left side. In different experiments, either all whiskers were spared on the right side, or only selected whisker rows or arcs were spared. Mice were allowed to recover from anesthesia for 6 to 12 hours, and then left alone in a large (1 m x 60 cm) ‘enriched’ cage in the dark for 1 hour before being perfused and processed for c-Fos immunostaining.

Areas were calculated with NDPview (Hamamatsu) from cytochrome oxidase staining at P10 or Vglut2 staining when done in adults. For quantification of map areas from P10 flattened cortices, the surface in controls was limited to the first four barrels in row a, four in row b, six in row c, seven in row d, and eight in row e. In Krox20:Cre;Robo3^lox/lox mutants, the central (ipsi) map area comprises the domain bordered by a thick Vglut2-negative boundary. The peripheral (contra) map area was limited to the barrels located immediately above and below the border of the central ipsi map. To determine individual barrel areas in adults, only the largest unambiguous barrels were measured (first three barrels for rows e, d, c, and first barrels for rows a and b). Areas were assessed on the tangential section showing the most complete map of the PMBSF.

The areas of the VPM nucleus were calculated with NDPview from frontal sections of P4 Krox20:Cre;Tau^GFP mice stained with cytochrome oxidase and immunostained for GFP, at a mid-level of the VPM, where the barreloid organization was the most obvious.

Results are presented as means ± SEM. Differences of the means between two sample sets were assessed by two-tailed non-parametric Mann-Whitney test. Statistics were carried out with Prism (Graphpad Software, La Jolla, CA).

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

We thank P Charnay for providing the Krox20:Cre line, and S Arber for the Tau^GFP mice. We also thank N Narboux-Nême for technical help. FMR laboratory was supported by the Swiss National Science Foundation (31003A_149573) and the Novartis Research Foundation. The teams of AC and PG are part of the Ecole des Neurosciences de Paris training network.

Animal experimentation: All animal procedures were carried out in accordance to institutional guidelines and approved by the UPMC University ethic committee (Comité Charles Darwin, authorization # 03787.02). All surgery was performed under ketamine/xylazine anesthesia, and every effort was made to minimize suffering.

© 2017, Renier et al.

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