Adenomatous polyposis coli regulates radial axonal sorting and myelination in the PNS

During the radial axonal sorting process, which starts around birth in the rodent PNS, immature Schwann cells extend cytoplasmic processes to axon bundles and initiate the sorting of axons based on size. Schwann cells establish a one-to-one ratio with large-caliber axons and thereafter myelinate them. Small-caliber axons remain as bundles (known as Remak bundles) and are surrounded by non-myelinating Schwann cells. Axonal sorting is a highly regulated process that is controlled by concerted signals intrinsic and extrinsic to Schwann cells. Extrinsic signals originate either from the axons and include the signal transducer proteins neuregulin 1 (Nrg1) (Birchmeier and Nave, 2008) and Notch1 (Woodhoo et al., 2009) or from the extracellular matrix, which includes proteins such as laminins, β1 integrin (Itgb1) (Feltri et al., 2002) and Rac1 (Nodari et al., 2007). Schwann cell-intrinsic signals include the transcription factors β-catenin (Ctnnb1) (Grigoryan et al., 2013), Sox10 (Finzsch et al., 2010) and Sox2 (Le et al., 2005).

Adenomatous polyposis coli (APC) is a member of the β-catenin destruction complex that mediates the degradation of β-catenin and thereby is involved in the Wnt signaling pathway (Nathke, 2006). Loss of APC leads to persistent activation of β-catenin, which results in tumor formation in the colon (Nathke, 2006). In addition to its involvement in the Wnt signaling pathway, APC is localized to membrane-protrusive areas and tips, where it binds to the plus end of microtubules and interacts with multiple RNAs and proteins to form APC-containing ribonucleoprotein complexes (Mili et al., 2008; Reilein and Nelson, 2005). APC was also shown to interact with actin and is believed to control actin-based cell protrusion (Okada et al., 2010). The binding between APC and the cytoskeleton stabilizes the cytoskeleton, prevents its collapse (Kroboth et al., 2007) and enables the cell to extend processes (Votin et al., 2005). Interaction of APC with the cytoskeleton is, at least in part, independent of Wnt signaling (Harris and Nelson, 2010). Owing to the role that APC plays in the regulation of β-catenin and because of its interaction with the cytoskeleton, we decided to investigate APC function in radial sorting and myelination of the PNS.

To examine the specific role of APC in Schwann cells we used a P0-Cre (P0 is also known as Mpz) transgenic mouse line (Feltri et al., 1999) in combination with mice containing a conditional Apc allele (Shibata et al., 1997) to generate mice in which APC is specifically eliminated from Schwann cells. Histological analysis revealed that loss of APC in Schwann cells delays radial axonal sorting and myelination in the sciatic nerve of the Apc^lox/lox;P0-Cre mice. Furthermore, APC-deficient Schwann cells displayed aberrant process formation in vitro and a transient proliferative deficiency and delayed differentiation in vivo. Taken together, our data indicate that APC is required by Schwann cells for their timely differentiation to mature, myelinating cells and plays an essential role in the extension of Schwann cell processes, radial axonal sorting and PNS myelination.

Schwann cell-specific conditional Apc knockout mice (Apc^lox/lox;P0-Cre) were generated by mating mice carrying an Apc allele in which exon 14 was flanked by loxP sites (Shibata et al., 1997) with P0-Cre transgenic mice (Feltri et al., 1999). Sciatic nerve-specific excision was verified using PCR primers that exclusively recognize the excised allele (Fig. 1A). In order to study the timing of Apc excision, we quantified the levels of the excised allele in the sciatic nerve at different time points. We found that the excised allele is present at high levels as early as postnatal day 1 (P1) and continues to be detected approximately at the same levels at P60, the latest time point that we examined (Fig. 1B). We also crossed the reporter mouse line ROSA26-stop-EYFP (Srinivas et al., 2001) with the P0-Cre mice and found reporter gene expression in approximately 60% of the sciatic nerve cells at all time points examined (Fig. 1C and Fig. S1).

Apc^lox/lox;P0-Cre mice exhibit hindlimb clenching when suspended by the tail, a common indication of neurological dysfunction (Golan et al., 2013; Novak et al., 2011; Porrello et al., 2014) (Fig. 1D), and they also suffer from hindlimb weakness as detected by grip strength analysis (Fig. 1E). In order to examine peripheral nerve function, mutant and wild-type animals were subjected to electrophysiological examination, which showed that conduction velocity and the compound muscle action potential amplitude were both significantly reduced in sciatic nerves of the Apc ^lox/lox;P0-Cre mice (Fig. 1F).

The reduced conduction velocity detected in the sciatic nerves of the Apc^lox/lox;P0-Cre mice suggests the presence of myelination defects in these animals. Therefore, we examined the morphology of sciatic nerves of Apc^lox/lox;P0-Cre and control mice. The number of myelinated fibers was significantly reduced in the sciatic nerve of the Apc^lox/lox;P0-Cre mice at postnatal days (P) 1, 4 and 7 compared with controls (Fig. 2A,E). Nevertheless, the number of myelinated fibers was similar between the two genotypes at P14 – the peak of the PNS myelination – suggesting that the myelination process is delayed in the Apc^lox/lox;P0-Cre mutant animals during the early postnatal period. We also observed occasional abnormal polyaxonal myelination in which individual thin myelin sheaths wrapped multiple small-caliber axons, which affected ∼12% of the axons of these mice at P60 (Fig. 2A; higher magnification in Fig. S2). This finding may be due to abnormal axonal sorting in the Apc^lox/lox;P0-Cre sciatic nerves. In addition, Apc^lox/lox;P0-Cre sciatic nerves contained axons with thinner myelin (higher g-ratios) compared with controls at P7-P60 (Fig. 2B-D), strongly indicating that APC loss results in PNS hypomyelination.

Since we saw profound hypomyelination in Apc^lox/lox;P0-Cre mice compared with control mice at P7, we performed a detailed morphometric analysis of the P7 sciatic nerves. We used 3DEM (serial blockface scanning EM) to further analyze the myelin sheath abnormalities observed in the PNS of the Apc^lox/lox;P0-Cre mice at P7. EM images of serial sections were acquired from sciatic nerves of Apc^lox/lox;P0-Cre and Apc^lox/lox (control) mice and individual fibers were traced and analyzed in both genotypes. We found that the mutant sciatic nerve contained fewer myelinated axons compared with controls, confirming that the mutant sciatic nerve is significantly hypomyelinated. The results are presented in Movies 1-4. Three-dimensional reconstruction of the serial EM images showed that mutant sciatic nerves contained ∼3 times shorter internodes at P7 (Fig. 2F) regardless of the axon diameter (Fig. S3). This phenotype persists to adulthood, since shorter internodes were also detected in teased sciatic nerve fibers stained for the paranodal marker Caspr (Menegoz et al., 1997) at P60 (Fig. 2G,H). The reduced myelin thickness and shorter internodes of the Apc^lox/lox;P0-Cre mice probably explains the electrophysiological abnormalities (Fig. 1F) found in these animals.

APC loss in Schwann cells appears to disrupt radial axonal sorting, since we observed abnormally large bundles of axons in the sciatic nerves of the Apc^lox/lox;P0-Cre mice (Fig. 3B) that were absent in Apc^lox/lox control mice (Fig. 3A). The mutant axonal bundles were approximately three times larger than the bundles found in control mice at P1, P4, P7 and P14 (Fig. 3E), suggesting that the axonal sorting process was disrupted in the mutant animals during the early postnatal period. The axon bundles of the mutant mice were larger because they contained large- and small-caliber axons in contrast to the bundles of the control mice that contained only small-caliber axons (Fig. 3F and G). The simultaneous presence of large- and small-caliber axons in the mutant axonal bundles further indicates that the radial axonal sorting process was disrupted in the Apc^lox/lox;P0-Cre mice. The delayed sorting and hypomyelination observed at P7 was not accompanied by axonal loss because the density of axons was not significantly different between the mutant and control mice at P7 (Fig. 3H).

To determine if the delay in PNS myelination in Apc^lox/lox;P0-Cre mice was correlated with an alteration in Schwann cell development, we examined the expression of Schwann cell lineage markers in P7 sciatic nerves. Scip (also known as Pou3f1) and Krox20 (also known as Egr2) are key transcriptional factors in the Schwann cell differentiation program. Scip is a marker of premyelinating Schwann cells (Bermingham et al., 1996) and Krox20 is a marker of myelinating Schwann cells (Topilko et al., 1994). We found increased numbers of premyelinating Schwann cells labeled for Scip (Fig. 4A,B) and reduced numbers of myelinating Schwann cells labeled for Krox20 in P7 sciatic nerves of the Apc^lox/lox;P0-Cre mice compared with controls (Fig. 4C and D), suggesting that the differentiation of Schwann cells into myelin-forming cells is delayed. We also examined cell division in the mutant sciatic nerves using the proliferation marker Ki67. Although there was no difference in the number of Ki67⁺ cells (Fig. 4E,F) or the total number of Schwann cell nuclei (Fig. S4) present in the mutant and control nerves at P7, when examined at P1, a reduction in Ki67⁺ cells was detected in the Apc^lox/lox;P0-Cre sciatic nerve, suggesting aberrant Schwann cell proliferation in the absence of APC (Fig. S4).

In order to further characterize the delayed differentiation of the mutant Schwann cells, we analyzed the expression of several genes that are known to be differentially expressed along the Schwann cell lineage in sciatic nerves (Eccleston et al., 1987; Jessen and Mirsky, 2005; Woodhoo et al., 2009). We found that APC loss results in the upregulation of genes specifically expressed in immature, premyelinating Schwann cells such as Scip, Sox2 and Galc (Fig. 5A) and downregulation of myelin-specific genes expressed in mature myelinating Schwann cells (Fig. 5B).

APC loss might dysregulate the Wnt/β-catenin signaling pathway in Schwann cells that has been shown to be crucial for PNS myelination (Grigoryan et al., 2013). APC is a member of the β-catenin destruction complex that mediates the degradation of β-catenin. In the absence of APC, β-catenin does not undergo degradation resulting in constant activation of the Wnt signaling pathway (Nathke, 2006). Consistent with this, we found that expression of the Wnt signaling target genes Axin2, Myc and Lef were all upregulated in the sciatic nerves of the Apc^lox/lox;P0-Cre mice at P7, indicating that loss of APC activates the Wnt signaling pathway in APC-deficient Schwann cells (Fig. 5C). This data indicates that the role of APC in controlling the Wnt signaling pathway potentially contributes an essential function in the differentiation program of Schwann cells.

It was previously shown that APC forms a complex with actin (Okada et al., 2010) and is also localized to the plus end of microtubules, where it interacts with the cytoskeleton (Nathke, 2006). In agreement with these observations, primary Schwann cells in culture displayed APC staining in processes, where it colocalized with F-actin and α-tubulin, and in lamellipodia, where it colocalized with F-actin (Fig. S5). Prior reports indicated that the interaction of APC with the cytoskeleton may regulate cell process formation and extension in radial glia, astrocytes and neurons (Imura et al., 2010; Wang et al., 2011; Yokota et al., 2009). In order to test the possibility that APC also controls process extension in Schwann cells, we traced and measured the length of these cells in 3DEM serial sciatic nerve sections of Apc^lox/lox;P0-Cre and Apc^lox/lox control mice. We found that mutant Schwann cells in vivo were on average much shorter compared with control cells (Fig. 6A). Since the Schwann cells in the control mice were very long and many of them expanded over the length of the 3DEM sections, the number of Schwann cells available for this analysis was limited. Therefore, we cultured primary Schwann cells from sciatic nerves of Apc^lox/lox;P0-Cre and Apc^lox/lox control mice using an immunopanning technique (Brosius Lutz, 2014). Using primary Schwann cell cultures stained for the Schwann cell marker S100, we detected shorter Schwann cell processes in the Apc^lox/lox;P0-Cre mice (Fig. 6B,C).

It has been shown previously that activation of Wnt signaling results in increased lamellipodia formation and an increased number of processes in Schwann cells (Grigoryan et al., 2013). Similarly, we found that APC loss leads to increased numbers of lamellipodia in Apc^lox/lox;P0-Cre mutant Schwann cells (Fig. 6D) when plated on laminin. In order to assess whether APC ablation affects lamellipodia formation through a laminin/β1 integrin-dependent mechanism, we plated the cells on the β1 integrin-independent substrate vitronectin. This assay showed, however, that the change in lamellipodia formation observed in the APC-ablated cells was not β1 integrin dependent since a similar increase was observed in cells plated on vitronectin (Fig. S6). The APC mutant Schwann cells showed normal numbers of cell processes (Fig. 6E).

The proteins APC, GSK3β and Axin2 compose the β-catenin destruction complex that mediates the degradation of β-catenin (Nathke, 2006). In the absence of APC, this complex disassembles, β-catenin does not undergo degradation and the Wnt signaling pathway remains activated (Nathke, 2006). In order to assess whether the lamellipodia-related phenotype observed in the APC-ablated cells is dependent on the β-catenin destruction complex, we modulated Wnt signaling with small molecules that target either GSK3β or Axin2. To activate Wnt signaling we used CHIR99021, which is a potent inhibitor of GSK3β, and to inhibit Wnt signaling, we used XAV939, which is an inhibitor of tankyrase 1 and 2 (TNKS), the enzymes responsible for the degradation of the axin-GSK3β complex (Huang et al., 2009). By inhibiting TNKS activity, XAV939 thereby promotes the degradation of β-catenin, resulting in inhibition of the Wnt signaling pathway (Fig. 7). In wild-type Schwann cells, activation of Wnt signaling with CHIR99021 resulted in increased lamellipodia formation and inhibition of the pathway with XAV939 resulted in reduced lamellipodia formation (Fig. 7C). In the absence of β-catenin inhibition, we speculated that the APC-ablated cells would be insensitive to the modulation of Wnt signaling. Consistent with this hypothesis, lamellipodia formation in the APC mutant Schwann cells was resistant to both Wnt pathway activation and inhibition (Fig. 7C). Moreover, neither drug had an effect on the extension of processes in wild-type or APC mutant Schwann cells, suggesting that this phenotype is not dependent on the Wnt signaling pathway (Fig. 7B,D).

In the present study, we used the well characterized P0-Cre transgenic mouse line (Feltri et al., 1999) to conditionally ablate APC expression from Schwann cells. The resultant Apc^lox/lox;P0-Cre animals exhibit hindlimb weakness and impaired axonal conduction in the sciatic nerve. We showed that radial axonal sorting and PNS myelination are delayed in these animals. Our data suggest that the delayed sorting and myelination observed in the Apc^lox/lox;P0-Cre animals are likely to be the result of delayed Schwann cell differentiation. Our data also suggest that APC loss affects lamellipodia formation and process extension in Schwann cells during PNS myelination.

The role of APC in CNS myelination has been investigated; APC was shown to be important for CNS myelination and remyelination (Fancy et al., 2009). APC is expressed transiently in oligodendrocyte lineage cells during development, where it regulates formation of processes (Lang et al., 2013). Ablation of APC from oligodendrocytes altered the expression of genes involved in actin and microtubule polymerization, as well as genes involved in the generation of the cytoskeleton (Lang et al., 2013). The mechanism by which APC controls CNS myelination is complex and may be attributed to aberrant regulation of Wnt signaling and/or to abnormal expression of cytoskeletal proteins in the oligodendrocytes that lack APC (Lang et al., 2013).

Since APC is expressed in Schwann cells, we speculated that it might also be important for PNS myelination. Previous studies have conditionally inactivated APC expression in the developing neural crest, which resulted in apoptosis of cephalic and cardiac neural crest cells at about 11.5 days post coitum, resulting in craniofacial and cardiac anomalies at birth (Hasegawa et al., 2002). Nevertheless, the Cre driver line that was used in this study activated loxP recombination in Schwann cells as well as neuronal cells and ventral craniofacial mesenchyme (Yamauchi et al., 1999). Therefore, the specific role of APC in Schwann cells has not been addressed. We showed here that loss of APC in Schwann cells delays differentiation, resulting in disrupted radial axonal sorting and diminished PNS myelination during the early postnatal period. This delay in radial axonal sorting and PNS myelination results in a persistent PNS neuropathy as displayed by hindlimb weakness and electrophysiological abnormalities in adult animals. This neuropathy is probably the result of the shorter internodes and reduced myelin thickness observed in these animals.

APC is a member of the β-catenin destruction complex that mediates the degradation of β-catenin. In the absence of APC, β-catenin does not undergo degradation, resulting in persistent activation of the Wnt signaling pathway (Nathke, 2006). Accordingly, we showed that loss of APC activates the Wnt signaling pathway in Apc^lox/lox;P0-Cre sciatic nerves, resulting in upregulation of genes specifically expressed in immature premyelinating Schwann cells and downregulation of myelin-specific gene expression.

Activating the Wnt signaling pathway in Schwann cells by distinct approaches results in similar phenotypes; the β-catenin gain-of-function (GOF) mutation (Grigoryan et al., 2013) and APC depletion in our study both result in upregulation of genes specifically expressed in premyelinating Schwann cells and delayed myelination. Both the β-catenin GOF mutants and the Apc^lox/lox;P0-Cre mutants showed upregulation of Sox2 gene expression, which is an inhibitor of Schwann cell differentiation and myelination (Le et al., 2005). Sox2 has been shown to downregulate Krox20 expression (Le et al., 2005), which could explain the reduction in Krox20⁺ cells present in the Apc^lox/lox;P0-Cre mice. Therefore, we suggest that APC controls Schwann cell differentiation and PNS myelination soon after birth by inhibiting the Wnt signaling pathway in Schwann cells.

Nevertheless, despite very similar effects on myelination and on Schwann cell differentiation, the β-catenin GOF and the APC depletion mutations appear to have opposite effects on axonal sorting: the β-catenin GOF mutation accelerates the radial axonal sorting process, whereas the APC ablation described here delays axonal sorting despite the activation of the Wnt signaling pathway. Our observations indicate that APC might affect radial axonal sorting through a Wnt-independent mechanism: by using a pharmacological approach, we demonstrated that the abnormal extension of processes in APC-ablated Schwann cells is likely to be the result of a Wnt signaling-independent mechanism whereas the aberrant lamellipodia formation observed in these cells appears to depend on Wnt signaling. Proper lamellipodia formation is important for radial axonal sorting and for PNS myelination (Nodari et al., 2007), therefore its aberrant formation may explain, at least in part, the sorting and myelination defects in the Apc^lox/lox;P0-Cre mice. In addition, during PNS development, radial axonal sorting is contingent on normal extension of processes by Schwann cells (Feltri et al., 2016). Therefore, the inability of the APC-ablated Schwann cells to extend normal processes, which was observed in vivo and in vitro, may explain the sorting failure that occurs in the Apc^lox/lox;P0-Cre mice.

In the PNS, proteins involved in the Wnt signaling pathway are expressed by both neurons and Schwann cells. In Schwann cells, Wnt signaling activity is highest at E15.5-E17.5 and declines subsequently (Grigoryan et al., 2013). Our data suggest that the Wnt signaling pathway plays a fundamental role in Schwann cell maturation and PNS myelination, as suggested previously (Grigoryan et al., 2013; Jacob et al., 2011).

In addition, we observed a reduction in Schwann cell proliferation in P1 animals in the absence of APC expression. This result was surprising in that APC was identified as a tumor suppressor and in most situations, its inactivation results in uncontrolled cellular proliferation. Nevertheless, the inactivation of APC in CNS radial glial cells also results in a reduced proliferative capacity of these cells (Yokota et al., 2009). Despite the reduced Schwann cell proliferation observed in the newborn mice in the absence of APC, Schwann cell proliferation and the total number of cells present in the mutant sciatic nerve recovered to normal levels by P7. The extent to which this transient Schwann cell proliferative defect contributes to the overall mutant phenotype remains to be determined.

The work presented here reveals a novel role of APC in PNS myelination. Our genetic and pharmacological results indicate that APC controls Schwann cell process extension, lamellipodia formation and differentiation through Wnt signaling-dependent as well as Wnt signaling-independent mechanisms. Therefore, our study expands knowledge of the molecular mechanisms controlling radial axonal sorting and PNS myelination by Schwann cells, which could support our efforts to better understand PNS disorders.

Mice were housed and studied according to the University of Chicago's Animal Care and Use Committee (IACUC) guidelines. PNS-specific conditional knockout Apc mice (Apc^lox/lox;P0-Cre) were generated by mating mice carrying an Apc allele in which exon 14 is flanked by loxP sites (Shibata et al., 1997) with the P0-Cre transgenic mouse line (Feltri et al., 1999). Littermate Apc^lox/lox mice served as controls. Mice were genotyped by PCR analysis of tail genomic DNA using the primers P3 (5′-GTTCTGTATCATGGAAAGATAGGTGGTC-3′) and P4 (5′-CACTCAAAACGCTTTTGAGGGTTGATTC-3′). The excised allele was detected using primers P5 (5′-GAGTACGGGGTCTCTGTCTCAGTGAA-3′) and P3 (Shibata et al., 1997). The P0-Cre reporter line was generated by mating ROSA26-stop-EYFP mice (Srinivas et al., 2001) with the P0-Cre transgenic mouse line (Feltri et al., 1999).

We measured muscle strength in mouse hindlimbs using a computerized grip strength meter (model # 0167-005L) from Columbus Instruments (Columbus, OH). Two trials of 10 measurements per trial were performed for each animal with a 2 min resting period between each measurement. The average force (newton, N) was calculated for each group.

Electrophysiology was performed with a Nicolet Viking Quest Laptop System (VikQuest Port 4ch-7) from Nicolet Biomedical (Madison, WI). Recording needle electrodes were placed subcutaneously in the footpad. Supramaximal stimulation of sciatic nerves was performed with a 0.1-0.2 ms pulse, stimulating distally at the ankles and proximally at the sciatic notch with needle electrodes. Latencies, conduction velocities and amplitudes of compound muscle action potentials were measured. Results from stimulation of bilateral sciatic nerves were averaged for each animal, with n representing the number of animals in each group.

Schwann cells were isolated from Apc^lox/lox;P0-Cre and Apc^lox/lox mouse sciatic nerves using a sequential immunopanning protocol (Brosius Lutz, 2014). Briefly, 30 sciatic nerves from P7 pups were dissected and enzymatically dissociated with a collagenase and dispase mixture at 37°C. After enzymatic dissociation and gentle trituration, the single-cell suspension was sequentially incubated on dishes coated with antibodies against CD45 (macrophage depletion), Thy1.2 (fibroblast depletion) and O4 (Schwann cell selection). O4⁺ Schwann cells were trypsinized from the O4-coated panning plate and seeded onto poly-D-lysine and laminin-coated 12 mm glass coverslips. In some experiments, the cells were seeded onto poly-D-lysine, vitronectin or bovine serum albumin (BSA)-coated plates.

Animals were anesthetized by intraperitoneal injection with avertin (0.5 mg/g) and the sciatic nerve tissues were harvested and snap-frozen in isopentane precooled in liquid nitrogen. Cross-sections were cut from the fresh frozen tissue, fixed for 10 min in 4% paraformaldehyde, washed in PBS and subjected to immunostaining. The antibodies used were against Krox20 (Covance, PRB-236P, 1:250), Scip (Santa Cruz Biotechnology, sc11661, 1:250), Ki67 (eBiosciences, A15, 1:250), GFP (Invitrogen, A21311, 1:500), Cspr (Abcam, ab34151, 1:500), S100b (Dako, z0311, 1:500), APC (Santa Cruz Biotechnology, sc896, 1:150) and α-tubulin (Invitrogen, A11126, 1:250). Phalloidin was used for F-actin labeling (Thermo Fisher, A12381, 1:50). Confocal images of the sciatic nerves were generated on a Leica SP2-AOBS laser scanning microscope (Germany) equipped with an argon laser that excites at 488 nm and a helium/neon laser that excites at 543 nm, using a 20× or 40×/1.4 objective and a pinhole size of 1.0 airy disk units.

Apc^lox/lox;P0-Cre or littermate Apc^lox/lox mice (served as controls) were perfused intracardially through the left ventricle with saline for 2 min, followed by perfusion with a solution of 2.5% glutaraldehyde and 4% paraformaldehyde in a 0.1 M sodium cacodylate buffer. Sciatic nerves were harvested and post-fixed in the same fixative at 4°C for 2 weeks and sciatic nerve samples were embedded in epoxy resin. Ultrathin cross-sections of the sciatic nerve were cut, double stained with uranyl acetate and lead citrate and observed on a Philips CM 10 electron microscope. G-ratios were calculated from the sciatic nerves of three or four mice for each genotype. The g-ratios were calculated as axon diameter/fiber diameter as described previously (Auer, 1994).

For 3DEM analysis, P7 Apc^lox/lox;P0-Cre or littermate Apc^lox/lox mice (served as control) were perfused with 2.5% glutaraldehyde and 4% paraformaldehyde in a 0.1 M sodium cacodylate buffer and the sciatic nerves were dissected, postfixed in the same fixative and submitted to Renovo Neural Inc. for 3DEM analysis. Briefly, 500 μm tissue was stained with heavy metals, embedded in Epon resin and mounted onto pins (detailed protocol available from Renovo Neural). Serial blockface micrographs (analogous to serial sectioning) were obtained using a Zeiss Sigma VP scanning electron microscope equipped with a Gatan 3View in-chamber ultramicrotome. Series of approximately 250 images were acquired at 2.25 kV with a resolution of 10 nm per pixel and 100 nm per section, resulting in a field analysis of 204.8 μm (x)×61.44 μm (y)×25 μm (z). Images were registered and resized as necessary using ImageJ/FIJI software (http://fiji.sc/). For Schwann cell length analysis, we traced and measured the length of Schwann cells containing a classic Remak bundle, Schwann cells containing large diameter axons (one or more) and Schwann cells containing very small diameter axon bundles.

Total RNA was extracted from 26-32 sciatic nerves of P7 Apc^lox/lox;P0-Cre or littermate Apc^lox/lox mice (served as controls) using the TRIzol reagent (Invitrogen). RNA integrity was verified using Agilent chip (Agilent Technologies). cDNA was synthesized using iScript cDNA synthesis kit (Bio-Rad) according to the manufacturer's instructions. Real-time qRT-PCR was performed using iQ SYBER Green Supermix (Bio-Rad) according to the manufacturer's instructions. Gapdh was used as internal control gene for normalization. The following set of primers were used: Myc forward, 5′-TTGGAAACCCCGCAGACAG-3′ and Myc reverse, 5′-GCTGTACGGAGTCGTAGTCG-3′; Axin2 forward, 5′-AAGCCCCATAGTGCCCAAAG-3′ and Axin2 reverse, 5′-GGGTCCTGGGTAAATGGGTG-3′; Lef1 forward, 5′-ATGCACGTGAAGCCTCAACA-3′ and Lef1 reverse, 5′-AGCTGCACTCTCCTTTAGCG-3′; Mbp forward, 5′-ACACGAGAACTACCCATTATGGC-3′ and Mbp reverse, 5′-CCAGCTAAATCTGCTGAGGGA; Galc forward, 5′-GCCTACGTGCTAGACGACTC-3′ and Galc reverse, 5′-AGAACGATAGGGCTCTGGGT-3′; P0 forward, 5′-ACCTCTCAGGTCACGCTCTA-3′ and P0 reverse, 5′-CATGGCACTGAGCCTTCTCTG-3′; Mag forward, 5′-TTTGGACGTCAAGTACCCCC-3′ and Mag reverse, 5′-GGTACAGGCTCTTGGCAACT-3′.

For quantification of the Apc-excised allele we used genomic DNA extracted from sciatic nerves using Red Extract Kit (Sigma). The excised allele was amplified using SYBR Green Extract-N-Amp qPCR kit (Sigma) according to the manufacturer's instructions and the Apc gene P4 and P3 primers. The Il2 gene was also amplified as an internal control gene for normalization using the primers 5′-CTAGGCCACAGAATTGAAAGATCT-3′ (forward) and 5′-GTAGGTGGAAATTCTAGCATCATCC-3′ (reverse).

Statistical significance was determined using one-way ANOVA for electrophysiological measurements, Wilcoxon signed-rank test for intermodal length, and Student's t-test for all the rest. Differences were considered to be statistically significant if P<0.05. All results are expressed as the mean±s.e.m.

We thank Dr Laura Feltri and Dr Lawrence Wrabetz for providing the P0-Cre mice. We thank Erdong Liu, Gloria Wright, Ani Solanki and Yimei Chen for skilful technical assistance and Andrew Roholt (Renovo Neural) for assistance with serial blockface scanning electron microscopy analysis.