Spatiotemporal Ablation of Myelinating Glia-Specific Neurofascin (Nfasc^NF155) in Mice Reveals Gradual Loss of Paranodal Axoglial Junctions and Concomitant Disorganization of Axonal Domains

Animals

All animal experiments were carried out according to UNC-IACUC-approved guidelines for ethical treatment of laboratory animals. Transgenic mice used in this study expressing Cre recombinase from the Cnp locus (Cnp-Cre; Lappe-Siefke et al., 2003), the P0 promoter (Feltri et al., 1999), and inducible Plp-CreER (Doerflinger et al., 2003) were generously provided by Drs. Klaus Nave (Max Planck Institute of Experimental Medicine), Laura Feltri (San Rafaele Scientific Institute, Italy), and Brian Popko (University of Chicago), respectively.

Conditional Targeting of Nfasc Gene

The neurofascin (Nfasc) gene spans ~171 kb of the genomic DNA and generates many alternative spliced mRNAs (refer http://www.ensemble.org/mouse). To ablate Nfasc gene expression and function conditionally, a targeting construct was generated in which loxP sites were targeted to flank exons 2 and 3 of the Nfasc gene. A BAC clone (bMQ228D3) from a 129 AB2.2 BAC library containing the targeted Nfasc exons was obtained from the Sanger Center, England. With BAC recombineering technology, a polymerase II promoter directed diphtheria toxin (DT) vector containing short PCR-generated arms homologous to those of the Nfasc DNA sequence was transformed into Escherichia coli DY380 recombinant bacterial strain (heat-induced RED and GAM recombinase complex) containing the Nfasc BAC. By this approach, an ~9.4-kb fragment from Nfasc BAC was successfully retrieved into the DT vector by means of gap repair. The first loxP site was inserted into intron 3 by transforming a PGK-neo cassette flanked with loxP sites. This confers kanamycin resistance to DT-Nfasc plasmid. The DT-Nfasc clone with proper neo cassette integration was transformed into strain EL350, and, upon arabinose-induced Cre recombination in transformed EL350, the neo cassette was removed, leaving the loxP site in intron 3. Similar steps were repeated for the insertion of another neo cassette flanked with FRT and FRT-loxP sites into intron 1. Prior to ES cell targeting, the functionality of the loxP sites as well as the expected restriction enzyme digestion patterns in the targeting vector was confirmed. The targeting construct was linearized with NotI, followed by electroporation into ES cells. The targeted ES cells were screened by PCR amplification and standard molecular biological methods (for details see Bhat et al., 2001).

Antibodies, Immunofluorescence, and Immunoblotting

Three antisera were generated against the two Nfasc protein isoforms as follows. Sequences corresponding to the Nfasc^NF155-specific fibronectin III domain 3 (amino acids KIRV--ASFP), the Nfasc^NF186-specific mucin domain (amino acids TVGT--VYSR), and the common cytoplasmic region (amino acids FIKR--YSLA) were used to immunize guinea pigs and rats as previously described (Bhat et al., 1999, 2001). Other primary antibodies used were guinea pig anti-Caspr antibodies (Bhat et al., 2001), rabbit anticalbindin, mouse pansodium channel, rabbit and mouse antipotassium channel (K_v1.1; Sigma, St. Louis, MO), mouse anti-β-tubulin (Chemicon, Temecula, CA), mouse anti-PLP (ABR Bioreagents), mouse anti-MBP (Sterberger Monoclonals, Lutherville, MD), rabbit anti-NrCAM (Abcam, Cambridge, MA), rat anti-PLP (A. Gow, Wayne State University), and rabbit anti-Cont (J. Salzer, NYU; Rios et al., 2000). The secondary antibodies conjugated to Alexa Fluors-488, -568 and -647 were obtained from Invitrogen Corporation (La Jolla, CA).

Preparation of Teased Sciatic Nerves and Cerebellar and Spinal Cord Sections

Fixed sciatic nerves were teased into individual fibers in ice-cold PBS, mounted on glass slides, and dried overnight at RT, followed by treatment with acetone at −20°C for 20 min. The slides were washed with PBS several times before immunostaining; in some cases, slides were stored at −80°C until processed further. For cerebellar and spinal cord sections, wild-type and mutant mice were deeply anesthetized and transcardially perfused with saline buffer followed by a solution of 4% paraformaldehyde in PBS. For immunofluorescence, cerebelli or spinal cords were sectioned on a Vibratome (Leica) at 30 µm thickness and processed as previously described (Garcia-Fresco et al., 2006).

Transmission Electron Microscopy

TEM was carried out essentially as described by Garcia-Fresco et al. (2006). After intracardiac perfusion, the whole animal was postfixed for 2 weeks at 4°C in the same fixation solution. After an overnight incubation in Millonig’s buffer, nerve fibers, spinal cords, or cerebellar tissue from each animal was postfixed for 1 hr in cacodylate-buffered 1% osmium tetroxide. The tissue was then rinsed, dehydrated in increasing concentrations of cold ethanol, and infiltrated and embedded in PolyBed (Polysciences, Warrington, PA). One-micrometer and ninety-nanometer sections were cut and stained with to-luidine blue or a combination of uranyl acetate and lead citrate, respectively. The 1-µm-thick sections were analyzed with a Nikon Eclipse 800, and the 90-nm-thick sections were analyzed with a JEOL JEM1230 transmission electron microscope equipped with a Gatan Ultrascan digital camera for acquiring images at high resolution.

Image Analysis and Quantitation

Confocal images were captured with a Bio-Rad Radiance 2000 laser scanning system attached to a Zeiss Axioplan2 microscope. The immunofluorescence images shown are Z stacks of five to eight sections with a scan step of 0.25 µm. All scanning parameters were optimized for the wild-type tissues, and the mutant tissues were scanned at identical settings. Quantitative measurements were performed on at least three independent tissue samples for immunofluorescence analysis and two independent samples for TEM analysis. For paranodal numbers and length estimation, only those paranodes were counted that looked morphologically normal and did not display structural distortions or stretching as a result of teasing of the fibers during preparation.

Tamoxifen-Induced Nuclear Translocation of CreER Recombination

The tamoxifen induction experiments were carried out essentially according to Doerflinger et al. (2003). Animals of the genotype Plp-CreER;Nfasc^Flox/+ and Plp-CreER;Nfasc^Flox were intraperitoneally injected with 100 µl of suspension (=1 mg tamoxifen) for 10 consecutive days beginning at P23. The animals were then sacrificed at various time points for phenotypic analysis by light and electron microscopy and for electrophysiological measurements.

Electrophysiological Studies

Age-matched wild-type, Cnp-Cre;Nfasc^Flox/Flox, P0- Cre;Nfasc^Flox/Flox, and Plp-CreER;Nfasc^Flox/Flox mutant mice were deeply anesthetized with Avertin or sacrificed by cervical dislocation. Left tibial/plantar nerves (mean length 13.5 mm) were carefully dissected free, cleaned of excess connective tissue, and maintained in cooled, oxygenated, modified Bretag’s solution (123 mM NaCl, 3.5 mM KCl, 0.7 mM MgSO₄, 2 mM CaCl₂, 9.5 mM Na gluconate, 1.7 mM NaH₂PO₄, 5.5 mM glucose, 7.5 mM sucrose, 10 mM HEPES; pH 7.40, osmolarity 290) for 2 hr to permit recovery from the acute injury of the dissection. In a dual-compartment ex vivo recording chamber (Stampfli, 1954; Reeh, 1986), the nerves were superfused in oxygenated Bretag’s solution at 30°C. Rectangular wave pulses (0.01 msec) were delivered by an Ag/AgCl suction electrode to the proximal portion of the nerve. One millimeter of the distal portion of the nerve was pulled out of the solution, suspended in a layer of mineral oil, and positioned on a gold recording electrode. A reference electrode was placed in solution close to the gold recording electrode. Evoked compound action potentials (CAP) were amplified and stored as digitized signals with pClamp 10.2 software (Molecular Devices, Union City, CA). Amplitude of the stimulus pulses (10–30 V) was adjusted to ensure a near-maximal stimulation of the A-component of the CAP (Gasser and Grundfest, 1939). In other words, the nerve was stimulated at an intensity sufficient to elicit a stable A-component amplitude and latency. Conduction latency was measured as the time in milliseconds from the start of the stimulus artifact to the start of the first upward deflection of the A-component. Conduction velocity (m/sec) was calculated as the distance between the stimulating and recording electrodes divided by the conduction latency.

Generation and Phenotypic Characterization of Myelinating Glia-Specific Neurofascin (Nfasc^NF155) Null Mutants

The main molecular components of the paranodal axoglial junctional complex include Caspr and its cis interacting partner Cont on the axonal side and the Nfasc^NF155 on the glial side (Fig. 1A; Peles and Salzer, 2000; Pedraza et al., 2001; Bhat, 2003; Salzer, 2003). Recently, Nfasc mutants were generated revealing that Nfasc is required for the formation of paranodal axoglial junctions and organization of the node of Ranvier (Sherman et al., 2005). We first determined the molecular differences between the glial and the neuronal isoforms of Nfasc by analyzing the reported Nfasc cDNA sequences and the mouse genomic DNA databases (for details see http://www.ensembl.org/mouse). This analysis revealed that the major Nfasc isoforms contain 28 exons, with Nfasc^NF155 containing exons 20 and 21, encoding the third fibronectin (FN) III domain (Fig. 1B, exons in gray). Nfasc^NF186 instead contains exons 23 and 24, which are absent in Nfasc^NF155, encoding the mucin (MUC) domain (Fig. 1B, exons in dark gold; Davis et al., 1996; Tait et al., 2000).

To ablate the function of Nfasc^NF155 specifically without interfering with the function of Nfasc^NF186, we first analyzed the structure of the 5′ region of the Nfasc locus. We generated a targeting construct (Nfasc^Flox) such that the coding exons 2 and 3 were flanked by loxP sites. Myelinating glia-specific Cre-mediated recombination resulted in the deletion of exons 2/3, causing translation termination of Nfasc^NF155 in exon 4 (Fig. 1C). A 30-amino-acid polypeptide from the putative initiation codon and 20 unrelated amino acids were produced as a result of frame shift without affecting neuronal Nfasc^NF186 expression (Fig. 1C, red asterisk). A detailed strategy for the generation of the Nfasc^Flox targeting construct is given in Materials and Methods. After germline transmission, heterozygous and homozygous mice were confirmed by PCR amplification across the targeted region. The primer combinations and the products of the PCR amplification from wild-type (+/+), Nfasc^Flox/+, Nfasc^Flox/Flox, and Cnp-Cre-mediated deletion of the floxed exons are shown in Figure 1D. The PCR amplification using primer sets 1 and 4 confirmed that both exons 2 and 3 were deleted in Cnp-cre/+;Nfasc^Flox/Flox (Fig. 1D).

To ablate the expression of Nfasc^NF155 specifically in the myelinating glial cells in the CNS and PNS, we used two well-established myelinating glia-specific Cre lines, Cnp-cre (Lappe-Siefke et al., 2003) and myelin P0-cre (Feltri et al., 1999). The mice of the genotype Cnpcre/+;Nfasc^Flox/Flox, in which only Nfasc^NF155 has been ablated in both the Schwann cells and the oligodendrocytes, are born at the expected Mendelian frequency from heterozygous intercrosses. The Nfasc^NF155 mutant mice are indistinguishable from their wild-type littermates until approximately P8. Starting at about P12, the mice become identifiable by their smaller stature and progressive neurological defects that reach maximal severity toward the beginning of the third postnatal week, and all mutants die at about P16/17. A typical P16 Cnpcre/+;Nfasc^Flox/Flox animal has a posture shown in Figure 1F and is easily differentiated from the wild type shown in Figure 1E. At P16, the defects include hypomotility and severe motor coordination defects. These mice fail to maintain their balance on a stationary beam and are nearly immobile in open field tests (data not shown). We monitored the weight of the Cnp-cre/+;Nfasc^Flox/Flox animals and compared it with that of their wild-type littermates. As shown in Figure 1G, at about P6, the wild-type and Cnp-cre/+;Nfasc^Flox/Flox littermates are of almost identical weight, but, from P8 onward, the Cnp-cre/+;Nfasc^Flox/Flox animals begin to show progressive loss of weight, and, by P16/17, the weight of the Cnp-cre/+;Nfasc^Flox/Flox animals falls below what they had weighed at P6. The Cnp-cre/+;Nfasc^Flox/+ are indistinguishable from their wild-type littermates and live a normal life span without displaying any physical or behavioral phenotypes. In contrast to Nfasc mutants reported by Sherman et al. (2005), which die at P6 when paranodal and nodal organization is just taking place, our Cnpcre/+;Nfasc^Flox/Flox mice die at P16/17, emphasizing the role of individual Nfasc isoforms in paranodal and nodal organization.

To demonstrate that we had specifically generated an Nfasc^NF155 null allele in myelinating glia, we carried out Western blot analysis using antisera generated against the C-terminal domain of Nfasc (NFCT), which recognizes both Nfasc^NF155 and Nfasc^NF186. As shown in Figure 1H, the wild-type (+/+) spinal cord lysate shows both Nfasc^NF155 and Nfasc^NF186 protein isoforms (arrows), whereas Cnp-cre/+;Nfasc^Flox/Flox animals show only the Nfasc^NF186 isoform and lacks the Nfasc^NF155 isoform (red asterisk). These results indicate specific loss of Nfasc^NF155 in the myelinating glial cells. To establish that the Nfasc^NF155 isoform is absent from the paranodes of the myelinated axons, we immunostained teased sciatic nerve fibers with isoform-specific antibodies against Nfasc^NF155 and Nfasc^NF186 (Fig. 1B). As shown in Figure 1I,J and in the merged image in Figure 1I″, wildtype (+/+) fibers showed specific paranodal localization of Nfasc^NF155 (Fig. 1I, green), and nodal localization of Nfasc^NF186 (Fig. 1I′, red). The Cnp-cre/+;Nfasc^Flox/Flox mutant sciatic nerve fibers lacked immunoreactivity against Nfasc^NF155 (Fig. 1J, green) but showed normal nodal localization of Nfasc^NF186 (Fig. 1J′, red) in the merged image (Fig. 1J″). Similarly P0-cre/+;Nfasc^Flox/Flox mice revealed loss of Nfasc^NF155 specifically in the peripheral nerve fibers both by immunoblot and by immunohistochemical analyses and not in the central myelinated fibers (data not shown). Taken together, our data demonstrate that we have generated myelinating gliaspecific Nfasc^NF155 mutants without altering the expression and localization of Nfasc^NF186 at the node of Ranvier.

Axonal Domain Organization is Disrupted in Nfasc^NF155 Mutant Myelinated Axons

To determine the relationship between expression of the Nfasc isoforms with paranodal biogenesis, we first analyzed the developmental expression profile of Nfasc^NF155 in myelinated sciatic nerve fibers by using immunohistochemistry. As shown in Figure 2A, at P0, Nfasc^NF155 is expressed at the sites where paranodes are beginning to form. At this stage, Nfasc^NF186 localizes to the node, which shows a broad morphology (Fig. 2Aa, red). Similarly, we analyzed P1, P3, P5, P10, and P15 sciatic nerves, which revealed that Nfasc^NF155 is expressed at the paranodes during their formation. At P10 and P15, as paranode formation is completed, Nfasc^NF186 localization becomes more concentrated at the node (compare Fig. 2Aa–f). We next analyzed the distribution of Nfasc^NF155 along with the axonal paranodal protein Caspr and nodal Nfasc^NF186 to determine their precise colocalization during paranode formation. As shown in Figure 2B, Nfasc^NF155 (green) colocalizes with Caspr at the paranodes (blue) at all postnatal developmental stages beginning with P0 through P15. The immunoreactivity for Caspr at P0 is qualitatively lower compared with later stages. From P3 onward, Nfasc^NF155 and Caspr showed complete colocalization (Fig. 2Bc–f). To establish the comparative expression levels of Nfasc^NF155 and Nfasc^NF186 during postnatal development, we carried out Western blot analysis of P1, P3, P6, P8, P12, and P15 spinal cords using NFCT antibodies that recognize both isoforms. As shown in Figure 2C, Nfasc^NF186 protein levels are comparatively higher than Nfasc^NF155 at early stages (Fig. 2C, compare P1 and P3). At later stages, Nfasc^NF155 expression levels increase compared with earlier stages and with Nfasc^NF186 (Schafer et al., 2006). These results are consistent with the increase in Nfasc^NF155 fluorescence observed at later developmental stages in the immunohistochemical analysis shown in Figure 2A,B. Thus Nfasc^NF155 expression pattern coincides with paranode formation during postnatal development.

To establish the consequence of loss of Nfasc^NF155 in the Schwann cells in Cnp-cre/+;Nfasc^Flox/Flox animals, we analyzed mutant sciatic nerve fibers by coimmunostaining against Caspr and Cont paranodal proteins along with nodal Nfasc^NF186. Wild-type nerve fibers from P17 animals show the expected localization of Nfasc^NF155 at the paranodes (Fig. 2D, red), and Nfasc^NF186 localization at the node (Fig. 2D′, green, see also merged image in Fig. 2D″). Similarly, immunostaining with NFCT antibodies reveals proper paranodal localization of Nfasc^NF155 (Fig. 2E–E″). In Cnp-cre/+;Nfasc^Flox/Flox animals, NFCT immunoreactivity indicates specific loss of Nfasc^NF155 at the paranodes without affecting Nfasc^NF186 (Fig. 2F, red), as is seen with Nfasc^NF186 specific immunoreactivity (Fig. 2F′, green; also shown in the merged image in Fig. 2F″,F″′). Immunostaining against Caspr in the wild-type (Fig. 2G, red) and Cnp-cre/+;Nfasc^Flox/Flox animals (Fig. 2H, red) and against Cont in the wild-type (Fig. 2I, red) and Cnp-cre/+;Nfasc^Flox/Flox animals (Fig. 2J, red) shows that both Caspr and Cont fail to localize at the paranodes in the Cnp-cre/+;Nfasc^Flox/Flox animals. These results indicate that glial Nfasc^NF155 is essential for paranodal Caspr and Cont localization and that the localization of nodal Nfasc^NF186 is not affected in Cnp-cre/+;Nfasc^Flox/Flox mutant fibers.

A hallmark of myelinated axons is the localization of voltage-gated Na⁺ channels at the nodes and the delayed rectifier Shaker-type K⁺ channels at the juxtaparanodes and to regions of the internode that oppose the noncompacted myelin membranes, e.g., the Schmidt-Lanterman incisures and the internal mesaxon (Arroyo et al., 1999). To determine whether loss of Nfasc^NF155 at the paranodal junctions in Cnp-cre/+;Nfasc^Flox/Flox mice affects the distribution of these channels, we carried out immunofluorescence staining of sciatic nerves. In both wild-type (Fig. 2K) and Cnp-cre/+;Nfasc^Flox/Flox (Fig. 2L) mice, Na⁺ channels (Fig. 2K,L, green) were restricted to the nodes of Ranvier. Unlike the localization of Na⁺ channels, the distribution of the K⁺ channels (Fig. 2K,L, K_v1.1, red) was markedly different between the wild-type and the Cnp-cre/+;Nfasc^Flox/Flox mice. In wild-type nerves (Fig. 2K), K⁺ channels were completely separated from the Na⁺ channels by the paranodal region, as expected. In contrast, Cnp-cre/+;Nfasc^Flox/Flox mice showed mislocalized K⁺ channels in the paranodal region, immediately adjacent to and, in some cases, slightly overlapping with Na⁺ channels (note the areas of overlap appear as yellow in the merged images; Fig. 2L). These data indicate that glial Nfasc^NF155 at the paranodal axoglial junctions is required for the segregation of Na⁺ and K⁺ channels into distinct domains.

To characterize further the role of Nfasc^NF155 in myelinated axons, we coimmunostained sciatic nerve fibers from wild-type (Fig. 2M,N) and Cnp-cre/+;Nfasc^Flox/Flox (Fig. 2O) mice and analyzed them at low magnification. As shown in Figure 2M, an entire internodal segment flanked with two nodes was immunostained against Na⁺ channels (green arrows), and paranodes were immunostained against Caspr (blue) and K⁺ channels (K_v1.1, red) with distinct localization at the juxtaparanodal region and Schmidt-Lanterman incisures (red arrows) and inner mesaxon spanning the internodal region. Immunostaining of the wild-type sciatic nerves with antibodies against NFCT (Fig. 2N, green), Na⁺ channels (Fig. 2N, blue), and K⁺ channels (Fig. 2N, K_v1.1, red) revealed that Nfasc^NF155 is strongly expressed in the internodal region at Schmidt-Lanterman incisures (green arrows). The Nfasc^NF155 immunoreactivity either overlaps or is present adjacent to K⁺ channel immunoreactivity (Fig. 2N, red and green arrows). Immunostaining of sciatic nerve fibers from Cnp-cre/+;Nfasc^Flox/Flox mice revealed that Nfasc^NF155 immunoreactivity along the internodes at the Schmidt-Lanterman incisures is absent (Fig. 2O, green) and that the localization of K⁺ channels at or near the incisures is compromised (Fig. 2O, red). These data suggest that Nfasc^NF155 expressed by myelinating Schwann cells may be required for the formation of internodal specializations.

To establish that loss of Nfasc^NF155 also affected central myelinated axons, we carried out immunofluorescence analysis of the spinal cords from wild-type and Cnp-cre/+;Nfasc^Flox/Flox mice. As shown in Figure 2P, wild-type spinal cord white matter immunostained against NFCT (blue) and NF186 (red) showed paranodal Nfasc^NF155 (blue) and nodal Nfasc^NF186 localization (pink, because both antibodies recognize Nfasc^NF186). The spinal cord white matter from Cnp-cre/+;Nfasc^Flox/Flox mice, immunostained with the same antibodies as in Figure 2P, showed no immunoreactivity against Nfasc^NF155 but showed normal localization of Nfasc^NF186 (Fig. 2Q, pink), indicating specific loss of Nfasc^NF155 in the oligodendrocytes. We also analyzed the domain organization in the spinal cord white matter by using coimmunostaining against Caspr (Fig. 2R,S, blue), NF186 (green) and K_v1.1 (red). As observed in the PNS, the spinal cord also revealed loss of the paranodal domain and absence of segregation between the juxtaparanodal and the nodal components (Fig. 2S).

To determine further whether loss of glial Nfasc^NF155 phenocopied the presence of axonal swellings in the cerebellar Purkinje neurons, as was observed in Caspr (Garcia-Fresco et al., 2006; Pillai et al., 2007), Cont (Boyle et al., 2001), and CGT (Garcia-Fresco et al., 2006) mutants, we carried out immunostaining of the cerebella from wild-type and Cnp-cre/+;Nfasc^Flox/Flox mice with anticalbindin antibodies. As shown in Figure 2T,U, wild-type Purkinje neuron axons displayed normal axonal morphology. In contrast, Purkinje axons from Cnp-cre/+;Nfasc^Flox/Flox mice displayed extensive presence of axonal swellings (Fig. 2V,W). These swellings are reminiscent of those observed in Caspr and CGT mutants, which we had previously shown to be caused by cytoskeletal abnormalities and accumulation of the cellular organelles at the disrupted paranodal regions (Garcia-Fresco et al., 2006). To establish further that loss of glial Nfasc^NF155 did not alter the protein expression levels of myelin or paranodal proteins, we carried out Western blot analysis of brain lysates from wild-type and Cnp-cre/+;Nfasc^Flox/Flox mice. As shown in Figure 2X, the protein levels of the paranodal proteins Caspr and Cont and the myelin proteins proteolipid protein (PLP) and myelin basic protein (MBP) between the wild-type and Cnp-cre/+;Nfasc^Flox/Flox mice revealed no significant differences. These results indicate that loss of glial Nfasc^NF155 does not alter the steady-state levels of paranodal or myelin proteins. Taken together, our data demonstrate that loss of glial Nfasc^NF155 results in the disruption of the paranodal region and failure to segregate the axonal domains in the myelinated axons.

Nfasc^NF155 Mutants Fail To Establish Paranodal Axoglial Junctions

As established above, myelinating glia-specific loss of Nfasc^NF155 did not alter myelin protein levels; therefore, we wanted to determine whether histological organization of the nervous system, including myelination, was affected. Light microscopy did not reveal any obvious abnormalities in the histological organization or the extent of myelination (data not shown). As shown in Figure 3, the myelinated sciatic nerve fibers from the wild-type (Fig. 3A) and Cnp-cre/+;Nfasc^Flox/Flox (Fig. 3B) mice were indistinguishable, and morphometric analysis of the myelinated fibers in the PNS showed no significant differences in axon and nerve fiber diameters, myelin thickness, or the g value (ratio of axon to fiber diameter; data not shown). These results, together with the normal myelin protein levels (Fig. 2X), suggest that lack of Nfasc^NF155 does not result in any significant alterations in the myelination process.

Because immunofluorescence analysis established that Nfasc^NF155 was localized to the myelin side of the paranodal region, and loss of Nfasc^NF155 specifically in myelinating glia resulted in mislocalization of Caspr and Cont, we examined the ultrastructure of the paranodal region in the PNS and CNS. As shown in Figure 3, low-magnification images of the sciatic nerve nodal and paranodal regions from wild-type (Fig. 3C) and Cnp-cre/+;Nfasc^Flox/Flox (Fig. 3D) mice reveal no major anatomical or organizational abnormalities. At higher magnification, the paranodal region in the wild-type showed myelin loops arrayed sequentially and in close apposition to the axonal membrane (Fig. 3E). Between the loops and the axon, periodic densities corresponding to the septate-like transverse bands were readily apparent (Fig. 3E, black arrowheads). In the Cnp-cre/+;Nfasc^Flox/Flox sciatic nerve fibers, the organization of the paranodal loops was preserved; however, the characteristic transverse bands were consistently absent (Fig. 3F,G, black arrowheads). In addition, the spacing between the paranodal loops and the axolemma was often abnormally wide, and the normal indentation of the axolemma by the paranodal loops was often absent (compare Fig. 3F–H with E). A consistent abnormality observed was the absence of the regular array of transverse bands between the paranodal loops and the axon. This was further confirmed by analyzing the Schwann cell-specific P0-cre/+; Nfasc^Flox/Flox sciatic nerve fibers (Fig. 3I). Interestingly, the P0-cre/+; Nfasc^Flox/Flox mice do not show any significant peripheral neurological deficits until P90. However, it remains to be seen whether older P0-cre/+;Nfasc^Flox/Flox mice will develop any severe peripheral nerve functional abnormalities. Furthermore, the Cnp-cre/+;Nfasc^Flox/Flox sciatic nerve fibers showed accumulation of organelles, such as mitochondria, and nodal membrane folds resulting from cytoskeletal abnormalities at the nodal/paranodal region (Fig. 3G). Together these data reveal that loss of Nfasc^NF155 in Schwann cells results in the loss of paranodal axoglial septate junctions.

Aside from the Nfasc^NF155 localization to paranodes in the PNS, it also localizes to CNS paranodal junctions; therefore, we examined the ultrastructure of this region in the spinal cord and cerebella of Cnp-cre/+;Nfasc^Flox/Flox mice. As shown in Figure 4A, in the wild-type spinal cord white matter, the paranodal loops are arrayed in close apposition to the axonal membrane, which establishes the septate-like transverse bands (black arrowheads). In contrast, the paranodal loops in Cnp-cre/+;Nfasc^Flox/Flox spinal cord white matter come in close apposition with the axolemma, but the regular array of transverse bands between paranodal loops and the axon is absent (Fig. 4B,C, black arrowheads). The most striking paranodal abnormalities often observed in Cnp-cre/+;Nfasc^Flox/Flox are the overlapping myelin loops of two adjacent myelinating glia that obscure the nodal regions (Fig. 4D, OL). In the example shown in Figure 4D, paranodal loops from the myelin glial segment on the left (labeled OL) overlap the loops of the myelin segment on the right. Additional defects observed in Cnp-cre/+;Nfasc^Flox/Flox included astrocytic processes that were interposed between the paranodal loops and the axon (Fig. 4D, black arrows) and everted and disorganized paranodal loops that are rarely observed at the paranodes in the wild-type spinal cords (data not shown). These data demonstrate that loss of Nfasc^NF155 in oligodendrocytes results in paranodal abnormalities and the loss of paranodal axoglial junctions.

Myelinating Glia-Specific Nfasc^NF155 Mutants Display Organelle Accumulation and Degeneration of Cerebellar Purkinje Axons

Our immunofluorescence analysis of the cerebella from Cnp-cre/+;Nfasc^Flox/Flox mice revealed Purkinje neuron axonal swellings as previously observed in Cont (Boyle et al., 2001), Caspr, and CGT mutants (Garcia-Fresco et al., 2006). To establish whether Purkinje axons from the Cnp-cre/+;Nfasc^Flox/Flox mice also undergo axonal degeneration and organelle accumulation, similarly to Caspr, Cont, and CGT mutants, we carried out cross-sections of the wild-type cerebellar white matter and found normal morphology of the myelinated axons with no signs of any vacuolation or degeneration (Fig. 4E, black arrows). In contrast, in the Cnp-cre/+;Nfasc^Flox/Flox cerebellar white matter, axonal degeneration was very apparent (Fig. 4F, black arrows). Higher magnifications revealed severe structural abnormalities and vacuolation in the degenerating axons (Fig. 4G, black arrows). These characteristics are displayed by axons that are in the terminal stages of degeneration (Palay and Palay, 1974). In an effort to determine the effect of Nfasc^NF155 deletion on the axonal structure, we performed TEM analysis of the Purkinje axon swellings in Cnp-cre/+;Nfasc^Flox/Flox mice. We observed large axonal accumulations of organelles, that included mitochondria (m) and smooth endoplasmic reticulum (SER; Fig. 4H). These axonal accumulations dramatically altered the axonal structure and may ultimately lead to axonal degeneration as observed in Figure 4F,G. Consistently with our immunofluorescence data, we found that the axonal swellings were frequently formed within or in close proximity to the paranodal region, as previously observed in Caspr mutants (Einheber et al., 2006; Garcia-Fresco et al., 2006). Taken together, the immunofluorescence and ultrastructural analyses show that loss of Nfasc^NF155 at the paranodal axoglial junctions leads to paranodal cytoskeletal disorganization and the eventual degeneration of Purkinje axons.

Ablation of Nfasc^NF155 in Adult Myelinating Glia (Nfasc^NF155AD) Reveals Gradual Disorganization of Axonal Domains

The ablation of Nfasc^NF155 in myelinating glia during early development showed that Nfasc^NF155 is required for the formation of the paranodal axoglial junctions but not myelination per se. However, these studies did not allow us to address how the axoglial junctions and axonal domains are maintained throughout adult life in myelinated axons. To address this question, we took advantage of a tamoxifen inducible-Cre line, developed by Doerflinger et al. (2003), which is specifically expressed in myelinating glia (Plp-CreER) to ablate Nfasc^NF155 during postnatal development, when axoglial junctions and axonal domains have been well formed and established. P23 Plp-CreER;Nfasc^Flox/Flox mice and age-matched or littermate wild-types were injected with tamoxifen for the next 10 days, until P33 (Fig. 5A, for details see Materials and Methods). The myelinated sciatic nerve fibers and spinal cords of the injected mice were analyzed by immunofluorescence over a period of 10–90 days postinjection for alterations in Nfasc^NF155 paranodal localization and the domain structure at the nodal/paranodal regions. As shown in Figure 5B, at P33, Plp–CreER;Nfasc^Flox/Flox mice showed normal paranodal localization of Nfasc^NF155 (green) and Nfasc^NF186 (red). After 20 days postinjection, Nfasc^NF155 levels at the paranodes began to decline (Fig. 5C, green). By 40 days postinjection, Nfasc^NF155 levels were drastically reduced (Fig. 5D, green), and, by postinjection day 60, Nfasc^NF155 immunoreactivity at the paranodes was essentially absent (Fig. 5E, green). In control mice, which also received tamoxifen, Nfasc^NF155 localized normally to the paranodes and was indistinguishable from that in the untreated wild types (data not shown). Next, we followed the localization and stability of the axonal paranodal protein Caspr in control and Plp–CreER;-Nfasc^Flox/Flox treated mice. As shown in Figure 5F, at P33, wild-type, control treated mice showed normal localization of Caspr (Fig. 5F, green) and nodal Nfasc^NF186 (Fig. 5F′, red; also seen in the merged image, Fig. 5F″). In Plp–CreER;Nfasc^Flox/Flox fibers, Caspr immunoreactivity began to decline gradually, to the point at which only traces of paranodal Caspr were seen by 90 days after tamoxifen injection (compare Caspr localization in Fig. 5G–K with F). In contrast, the localization of Nfasc^NF186 was not affected in Plp-CreER;Nfasc^Flox/Flox fibers (compare Fig. 5G′–K′ with F′). The Caspr immunoreactivity present after 60 and 90 days postinjection was also observed in the nodal region overlapping with Nfasc^NF186 (Fig. 5J″,K″), suggesting that Nfasc^NF155 is required to maintain Caspr localization at the paranodes.

We then wanted to determine consequences of loss of Nfasc^NF155 in adult myelinated fibers on the maintenance of axonal domains, being the juxtaparanodes, the paranodes, and the nodes. As shown in Figure 5L–L″′, wild-type control mice showed distinct localization of K_v1.1 at the juxtaparanodes (Fig. 5L, red), Caspr at the paranodes (Fig. 5L′, blue), Nfasc^NF186 at the nodes (green), and together in the merged image (Fig. 5L″′). In Plp-CreER;Nfasc^Flox/Flox fibers, within 10 days after injection, the juxtaparanodal K_v1.1 channels (Fig. 5M, red) began to migrate toward the nodal region. By 30 days postinjection, K_v1.1 channels (Fig. 5O, red) had reached closer to the nodal region and began to overlap with the remaining Caspr (Fig. 5O′, blue) at the paranodes. By 60 days after injection, no discrete paranodal region was observed; Caspr immunoreactivity was low, and K_v1.1 channels had almost reached the node (merged image in Fig. 5P″′). By 90 days after injections, K_v1.1 channels essentially abutted nodal Nfasc^NF186, and the paranodal domain was essentially nonexistent (compare Fig. 5M″′–Q″′ with L″′). A quantitative analysis of the paranodal and other domain changes after tamoxifen injection is presented in Table I. Together these data indicate that glial Nfasc^NF155 is required for the maintenance of paranodal axoglial junctions and that axoglial junctions are critical for the maintenance of the paranodal region and axonal domains in myelinated axons.

As observed in Figure 5CH″,N″′, 20 days after injection, Nfasc^NF155 and Caspr already show reduced protein levels; therefore, we wanted to determine whether we could identify fibers that were in the process of losing axonal domains, by showing partial overlap of domain markers. As shown in Figure 5R–R″′,S–S″′, as soon as Caspr localization at the paranodes becomes aberrant (Fig. 5R′, blue), K_v1.1 begins to migrate toward the paranodes (Fig. 5R′, red). When Caspr localization is intact, for example, in the right paranode (Fig. 5S′, blue), K_v1.1 remains at the juxtaparanodes (Fig. 5S, red, and S″′, merged image). The localization of Nfasc^NF186 (Fig. 5R″,S′, green) is not affected by loss of Nfasc^NF155. These data suggest that, as soon as axoglial junctions begin to disassemble, axonal domain disorganization ensues, and juxtaparanodal components begin their migration toward the nodal domain.

To determine whether central myelinated axons also follow the same pattern as the peripheral fibers, we analyzed the spinal cord white matter in Plp-CreER;N-fasc^Flox/Flox mice. As shown in Figure 5T–T″′, wild-type controls show normal domain organization, whereas the adult animals that lose Nfasc^NF155 show gradual disorganization of the axonal domains (Fig. 5U–W″′). One noteworthy feature of the central fibers is that, even in the presence of robust Caspr immunoreactivity at the paranodes (Fig. 5V′,W′, blue), the K_v1.1 channels from the juxtaparanodes (Fig. 5V,W, red) persisted in shifting closer to the node (merged images in Fig. 5V″′,W″′). These results suggest that it is the axoglial junctions, not the mere presence of Caspr, that maintains the axonal domains in the myelinated axons. To establish further the time line of loss of Nfasc^NF155 in adults after injection and the consequences of this on other paranodal proteins, we carried out immunoblot analysis of the spinal cords from age-matched P33 wild-type and Plp-CreER;Nfasc^Flox/Flox tamoxifen-injected animals. As shown in Figure 5X, the levels of Nfasc^NF155 began to decline steadily by 10 days after injection. By 20 days postinjection, nearly half of the Nfasc^NF155 protein remained, and nearly none of Nfasc^NF155 protein could be detected on the immunoblots by postinjection day 60, which is consistent with our immunofluorescence observations (Fig. 5C–E). However, the protein levels of Caspr (Fig. 5Y) and the paranode-enriched protein 4.1B (Fig. 5Z) were not significantly affected. Taken together, these data demonstrate that paranodal axoglial junctions are critical for the maintenance of axonal domains in myelinated axons.

Nfasc^NF155AD Mutants Display Gradual Loss of Paranodal Axoglial Junctions

Because the light microscopic immunofluorescence analyses in tamoxifen-injected Plp-CreER;Nfasc^Flox/Flox mice revealed disorganization of axonal domains, we wanted to determine the structural changes that occur in the electron-dense transverse bands, or septa, at the para-nodal axoglial junctions. To address this, we carried out electron microscopic examination of spinal cord white matter and sciatic nerves of tamoxifen-injected wild-type and Plp-CreER;Nfasc^Flox/Flox mice. As shown in Figure 6A, P33 wild-type nerves showed normal apposition of paranodal myelin loops with the axolemma, with distinct septa at the interface (Fig. 6, black arrows). The region between the black arrowheads is shown at a higher magnification in Figure 6A, inset. In contrast, the myelinated fibers from Plp-CreER;Nfasc^Flox/Flox 20 days after injection (Fig. 6B), showed atypical transverse bands (septa) from the juxtaparanodal side (left in Fig. 6B), with a loosening of the junctions and diffuse electron-dense structures in the cleft (Fig. 6B, see inset), whereas the paranodal loops toward the nodal side (right in Fig. 6B) maintained distinct transverse bands as seen in the wild type (Fig. 6A). After 20 days of injection, the mutant fibers showed progressive disorganization of the septal morphology (see Fig. 6, inset) and the axolemma detaching from the myelin loops (Fig. 6C, asterisks). Further analysis of Plp-CreER;Nfasc^Flox/Flox mice after 40 (Fig. 6D,E) and 90 (Fig. 6F) days of injection, revealed that, by 40 days, some electron density is still present in the myelin/axon cleft at the paranodes, but no presence of intact transverse bands is observed (insets in Fig. 6D,E). By 90 days postinjection, no transverse bands were visible between the myelin loops and the axolemma, and the electron densities noticed at earlier stages had diminished (Fig. 6F, inset). Similarly, we analyzed the sciatic nerves from wild-type and Plp-CreER;Nfasc^Flox/Flox injected mice. As shown in Figure 6G, wild-type fibers displayed typical morphology of the axoglial transverse bands (Fig. 6G, inset), whereas mutant fibers after 20 days of injection revealed disorganization of axoglial junctions (Fig. 6H). Accordingly, the junctions appeared to disassemble from the juxtaparanodal side (left in Fig. 6H) and progressed gradually toward the node. As shown in Figure 6I at higher magnification, the paranodal loops toward the nodal region side still maintain intact axoglial junctions with their typical morphology. By 90 days after injection, all the axoglial junctions were lost (Fig. 6J, see inset); however, electron densities on the paranodal axonal side were still observed. The presence of these densities indicates that the remnants of the paranodal axonal scaffold that links axoglial junctions to axonal cytoskeleton may remain intact well beyond the disruption of axoglial junctions (Garcia-Fresco et al., 2006). Taken together, the ultrastructural analysis reveals that the paranodal axoglial junctions disorganize gradually, and this disorganization begins from the juxtaparanodal side and progresses gradually toward the nodal side. This conclusion is strengthened by the observations that the juxtaparanodal components, like K_v1.1, begin to shift toward the nodal area immediately after axoglial junctions begin to disappear.

Peripheral Nerve Conduction Is Severely Affected in Nfasc^NF155 and Modestly Affected in Nfasc^NF155AD Mutants

To determine whether the loss of the paranodal axoglial junctions and/or the altered distribution of ion channels observed in the Cnp-Cre;Nfasc^Flox/Flox, P0-Cre;Nfasc^Flox/Flox, and Plp-CreER;Nfasc^Flox/Flox affected the electrophysiological properties of peripheral nerves, we performed extracellular recordings from tibial/plantar nerves of litter- or age-matched wild-type (+/+) and mutant mice. Nerves were stimulated at an intensity and duration sufficient to elicit a stable A-component amplitude and latency (for details see Materials and Methods). The electrophysiological measurements from various genotypes at different ages are given in Table II, and representative plots are presented in Figure 7. In Cnp-Cre;Nfasc^Flox/Flox nerves, conduction velocity (CV) was reduced by 50% at P13 (8.47 m/sec compared with wild-type at 16.83 m/sec; Fig. 7A). At P16.5, Cnp-Cre;Nfasc^Flox/Flox fibers revealed a CV of 9.5 m/sec compared with a wild-type CV of 17 m/sec (Fig. 7B). At P18, P0-Cre/+;Nfasc^Flox/Flox nerves had a 41% reduction in CV compared with wild-type (+/+, 17.9; P0-Cre, 12.6; Fig. 7C). In addition to the substantial delay in conduction, Cnp-cre;Nfasc^Flox/Flox nerves exhibited altered action potential amplitude and duration. Cnp-Cre;Nfasc^Flox/Flox demonstrated a greater than 50% decrease in amplitude and 50% increase in duration (Fig. 7A,B). This was also observed in P0-Cre/+;Nfasc^Flox/Flox nerves (Fig. 7C, Table II). In Plp-CreER;Nfasc^Flox/Flox fibers, the CV was modestly reduced, with a 24% decrease compared with wild-type (Fig. 7D). A further reduction in CV was observed in older Plp-CreER;Nfasc^Flox/Flox nerves (Fig. 7E, Table II). Taken together, the electrophysiological measurements indicate an important role for paranodal axoglial junctions in ensuring normal action potential propagation in myelinated axons.

Nfasc^NF155 Mutants Display Severe Motor Coordination Defects

Previous work from our laboratory has shown the importance of Caspr/paranodin to paranodal axoglial junction formation (Bhat et al., 2001). Circumstantial evidence suggests that glial Nfasc^NF155 can associate with Caspr and its axonal partner Cont, but no phenotypic analysis of a glial-specific knockout of Nfasc^NF155 has been performed, and there is much debate over the exact mechanism(s) of how these proteins associate (Charles et al., 2002; Gollan et al., 2003; Bonnon et al., 2007). Here we find, with glial-specific Cnp-Cre, that Nfasc^NF155 mutant mice exhibit severe tremors, motor paresis, ataxia, and immobility and die at postnatal day 16 or 17, similarly to both Caspr and Cont mutant mice (Berglund et al., 1999; Bhat et al., 2001; Boyle et al., 2001). We believe that these neurological defects reflect the function of Nfasc^NF155 in the myelinating glial cells and not the neuronal Nfasc^NF186, because Cnp-Cre expression has been shown to be restricted to glia within the CNS and PNS (Lappe-Siefke et al., 2003). Additionally, we have shown that Nfasc^NF186 proteins levels remain unchanged and that the nodal complex remains intact in these animals. These findings indicate for the first time the importance of glial-specific Nfasc^NF155 in the formation of paranodal axoglial junctions and in coordination of motor function.

Axon–Glial Interactions at the Nodes and Paranodes of Nfasc^NF155 Mutants

The proposed role of glial Nfasc^NF155 in the organization of the axoglial junctions was suggested by the mutant phenotypes observed in Nfasc mutants that lack both Nfasc^NF155 and Nfasc^NF186 (Sherman et al., 2005). These mutant mice have severe amyelination and dismyelination as well as a reduction in the levels of myelin-specific proteins, such as MBP and MAG (Zonta et al., 2008). In the current study, we found that glia-specific loss of Nfasc^NF155 did not alter the myelin ultra-structure but instead resulted in the absence of transverse bands, the hallmark of axoglial junctions. These findings provide formal evidence that Nfasc^NF155 is a key glial component of axoglial junctions and is consistent with its localization to the paranode (Tait et al., 2000; Charles et al., 2002; Sherman et al., 2005). Interestingly, other paranodal/nodal abnormalities observed in Nfasc^NF155 mutants are also seen in Caspr mutants, suggesting further that loss of either of the paranodal components generates identical phenotypes. Furthermore, the similarities in phenotype between these mutant may be attributed to cytoskeletal abnormalities created by the disruption of the axoglial junctions as seen previously (Einheber et al., 2006; Garcia-Fresco et al., 2006; Sousa and Bhat, 2007).

In the CNS of Nfasc^NF155 mutants, the morphological organization of the paranodal region is severely affected, as evidenced by the appearance of everting and overriding glial paranodal loops. These organizational defects can be attributed to a failure of the paranodal loops to make physical adhesions with the axon in the Nfasc^NF155 mutant mice. Although the paranodal loops of these mutant mice fan outward, the structure of myelin remains unchanged and suggests that the formation and stabilization of the paranode is independent of myelin stabilization. Upon further examination, it was seen that the overriding loops often occluded the node (Fig. 4D), and, although nodal proteins remain localized, such as Nfasc^NF186, Na channels, and Ank_G (data not shown), the effect of such an obstruction on nodal function remains unknown. The change in the paranodal structure in Nfasc^NF155 mutants may be a result of a change in the trapezoidal geometry of the myelin sheath, as is observed in Caspr mutants (Bhat et al., 2001). Other striking abnormalities observed in Nfasc^NF155 mutants are the axonal swellings and degeneration of cerebellar Purkinje axons (Fig. 2V, W, Fig. 4F–H). In earlier studies, we found that Caspr mutant mice exhibited similar abnormalities. We proposed then that such phenotypes were a result of the disorganization of the paranodal axonal cytoskeleton that results in organelle transport defects at the paranodes and eventual axonal degeneration (Einheber et al., 2006; Garcia-Fresco et al., 2006; Sousa and Bhat, 2007). Our findings indicate further that axoglial junction disruption leads to axonal transport defects and degeneration and indicate a role for the glial cytoskeleton in the maintenance and stabilization of the septate-like junctions as an anchor for the association of glia with axons.

Keeping Neighbors Apart: Paranodal Junctions and Maintenance of Axonal Domains

A major observation of this study is that the organization of axonal domains in myelinated axons requires myelinating glia-specific expression of Nfasc^NF155 and that, along with Caspr and Cont, it is able to establish the transverse bands/septa required at the paranodal axon–glial interface. Recent work has suggested that intact paranodal axoglial domains alone, formed by reintroducing Nfasc^NF155 into a complete Nfasc null background, may act to cluster and form the node independent of the neuronal-specific Nfasc^NF186 (Zonta et al., 2008). Interestingly, these mutant mice were not able to survive past P6, the same expiry as the complete null mice lacking both Nfasc^NF155 and Nfasc^NF186. Here we find that Na⁺ and K⁺ channels are able to cluster in the absence of the glial Nfasc^NF155 and intact paranodal junctions. This suggests that these junctions are not required for the clustering of ion channel domains and nodal proteins but instead are likely necessary for their continued segregation (see below). This is consistent with previous observations demonstrating that clustering of nodal components occurs prior to mature paranodal junction formation in the developing PNS (Melendez-Vasquez et al., 2001) and in CGT mutant mice that fail to form paranodal axoglial junctions (Coetzee et al., 1996; Dupree et al., 1999). Accordingly, in the absence of the paranodal axoglial junctions in Nfasc^NF155, Caspr, Cont, and CGT mutants, the nodal Na⁺ and juxtapara-nodal K⁺ channel domains remain largely segregated, with only occasional overlaps into the nodal region. These phenotypes indicate that distinct molecular interactions, possibly involving specific axonal and glial scaffolding proteins, maintain nodal protein complexes to their respective domains (Peles and Salzer, 2000; Pedraza et al., 2001; Bhat, 2003; Sherman and Brophy, 2005).

To address the importance of axoglial domain stabilization to the segregation of axonal domains, i.e., the node, paranode, and juxtaparanode, in adults we utilized a tamoxifen-inducible Plp-CreER to knock out Nfasc^NF155 after stablilization of these domains was complete. Upon observation, we found that, 20 days posttamoxifen injection, Nfasc^NF155 levels were reduced by 50% and that only after 90 days of injection is near-complete loss of Nfasc^NF155 seen. These findings indicate that Nfasc^NF155 is quite stable once it is incorporated into axoglial junctions. In addition, we found that, with increasing time after injection, Caspr staining is gradually reduced and that the juxtaparanodal K⁺ channels migrated toward the nodal region. The concomitant shift of the clustered juxtaparanodal K⁺ channels toward the node suggests that these complexes are mobile and are not anchored by, or within, the axonal/glial cytoskeleton. However, it is unclear why the K⁺ channels move toward the nodal area instead of diffusing throughout the internode; if it was a simple diffusion in the membrane, then one would expect that they should diffuse in either direction. Interestingly, these new findings are in direct contrast to findings from the Cnp-Cre;Nfasc^Flox mutant mice shown here, which fail to form paranodal axoglial junctions but still maintain segregated nodal and juxtaparanodal domains. These results indicate that two independent functions exist to segregate and maintain axonal domains during development and in mature animals. During development, each axonal domain is assembled and stabilized independently of the other and thus retains its anchorage at the respective domain, whereas, in the adult, maintenance of the domains requires the paranode to act as a fence to segregate the domains. This suggests that, with time, as these animals progressively lose the paranodal axoglial junctions, that both the juxtaparanodal complex and the nodal complex may diffuse throughout the internodes, which may lead to severe depolarization and abnormal saltatory conduction.

Conduction Velocity and the Role of Paranodal Axoglial Junctions

Electrophysiological data on Nfasc^NF155 mutant mice generated during embryonic development reveal that population nerve CVs are severely reduced in peripheral nerve fibers. These defects may reflect changes in the properties of fast-conducting myelinated fibers and may also be caused by a conduction block in a subset of fibers and/or slower propagation in the remaining fibers. Another possibility for the weakened CVs observed in Nfasc^NF155 mutant mice is that the disorganized paranodal region may cause current leakage into the periaxonal space. Leakage of the current may in turn increase the capacitance of the axonal membrane, leading to slow action potential propagation (Chiu and Ritchie, 1980; Boyle et al., 2001). Interestingly, it might be expected that in the adult Nfasc^NF155 mutant mice CVs would be slowed concomitantly with the degradation of axoglial junctions; however, we found that both the CV and the CAP peak-to-peak amplitudes were not dramatically altered in the mutant mice compared with control mice of the same age. These results were seen even in mice examined 6 months postinjection that had near-complete loss of Nfasc^NF155 protein levels and disassembled paranodal junctions (Fig. 7). It is not clear why the loss of axoglial junctions during adult stages has a less dramatic effect on the electrophysiological properties of the peripheral nerve fibers than loss in early developmental stages. One possibility is that many structural features are maintained in adult mutant nerves that allow the conduction of action potentials at near-normal or at a slightly reduced velocity. Alternatively, it could be that different classes of fibers in these adult nerves conduct differently and what we observed is a summation of all the fibers. It is unlikely that the reduced change in CVs is due to the injection and variability of its penetration to all glial cells, insofar as our quantitative measurements revealed that greater than 90–95% of the Schwann cells lost Nfasc^NF155 after 10 days of tamoxifen treatment (Table I), so the modest changes in CV and amplitude in adult fibers may be due to the structural integrity of the nerve fibers and slow disorganization of the axonal domains over time. It would be interesting to see how the Nfasc^NF155 mutant mice CVs faired through extended time, in that it has been shown that human demyelinating neuropathies, where axonal domains are compromised, such as multiple sclerosis and Charcot-Marie-Tooth disease, CVs are severely reduced (Nicholson, 1999; Wolswijk and Balesar, 2003; Coman et al., 2006). Further studies on Nfasc^NF155 mutant mice may reveal more similarities to human neurological disorders and provide a system for studying human disease. In summary, we have demonstrated that glial Nfasc^NF155 is critical for the formation of the paranodal axoglial junctions and that these junctions are central to the organization and maintenance of the axonal domains in myelinated axons.