|Home | About | Journals | Submit | Contact Us | Français|
Mouse models of human disease are helpful for understanding the pathogenesis of the disorder, and ultimately for testing potential therapeutic agents. Here we describe the engineering and characterization of a mouse carrying the I268N mutation in Egr2, observed in patients with recessively inherited Charcot-Marie-Tooth disease type 4E, which is predicted to alter the ability of Egr2 to interact with the Nab transcriptional coregulatory proteins. Mice homozygous for Egr2I268N develop a congenital hypomyelinating neuropathy similar to their human counterparts. Egr2I268N is expressed at normal levels in developing nerve, but is unable to interact with Nab proteins or to properly activate transcription of target genes critical for proper peripheral myelin development. Interestingly Egr2I268N/I268N mutant mice maintain normal weight and have only mild tremor until 2 weeks after birth, at which point they rapidly develop worsening weakness and uniformly die within several days. Nerve electrophysiology revealed conduction block, and neuromuscular junctions showed marked terminal sprouting similar to that seen in animals with pharmacologically induced blockade of action potentials or neuromuscular transmission. These studies describe a unique animal model of CMT whereby weakness is due to conduction block or neuromuscular junction failure rather than secondary axon loss, and demonstrate that the Egr2-Nab complex is critical for proper peripheral nerve myelination.
Charcot-Marie-Tooth disease (CMT) is a common inherited disorder of peripheral nerves characterized by progressive sensory loss and weakness beginning in the feet and legs, and later progressing to the hands (Charcot and Marie, 1886; Tooth, 1886). Remarkable progress has been made in recent years identifying many gene defects that can lead to this phenotype (Suter and Scherer, 2003; Shy, 2004; Nicholson, 2006). Based on nerve electrophysiology, most patients with CMT can be divided into two major forms. CMT type 1 is characterized by median nerve conduction velocity <38 m/s, and genes mutated in this type are largely involved in the development or maintenance of myelinating Schwann cells (Pareyson et al., 2006). CMT type 2 shows preserved nerve conduction velocity with diminished compound motor action potentials suggesting axon loss without myelin dysfunction, and genes mutated in CMT type 2 likely define molecular pathways necessary for axonal stability (Zuchner and Vance, 2006).
The proper formation of myelin by Schwann cells requires a precisely controlled genetic program coordinated by a series of transcription factors including SOX10, SCIP/Oct6, Egr2, and Nab1/Nab2 (Jessen and Mirsky, 2002; Svaren and Meijer, 2008). Loss of these transcription factors disrupts the myelination process, as does persistent overexpression (Ryu et al., 2007). The remarkable precision in transcriptional control necessary for proper myelination is evidenced by the fact that relatively small increases or decreases in the quantity of myelin proteins can lead to myelin dysfunction and disease. For example, loss of one copy of PMP22 leads to hereditary neuropathy with liability to pressure palsy (HNPP), whereas PMP22 duplication leads to CMT1A (Lupski, 1997).
Egr2 is absolutely required for proper peripheral nerve myelination, and serves as a master regulatory transcription factor in both developmental and regenerative myelination (Topilko et al., 1994; Le et al., 2005a; Decker et al., 2006). Furthermore, dominant and recessive mutations in EGR2 have been identified in patients with CMT (denoted CMT1D and CMT4E) (Warner et al., 1998). Egr2 interacts with two transcriptional coregulatory proteins, Nab1 and Nab2, which are able to act as either coactivators or corepressors of Egr2 depending on the promoter context (Russo et al., 1995; Svaren et al., 1996; Sevetson et al., 2000). Importantly, Nab1/Nab2 double knockout mice have severe hypomyelination and early lethality, providing evidence that Nab proteins are also critical for myelin development (Le et al., 2005b). Although null and hypomorphic Egr2 mouse mutants have been instrumental in defining its role in myelin development, they have been difficult to study as models of human CMT in large part due to early lethality.
To further delineate the role of Egr2-Nab interactions in myelin development, and develop an animal model of CMT4E, we generated mice expressing a point mutation found in human CMT patients in Egr2 (I268N) under control of the endogenous locus. In contrast to other Egr2 mouse models, Egr2I268N/I268N mice initially grow normally but develop rapidly progressive weakness two weeks after birth, and electrophysiology and terminal axonal sprouting implicate conduction block as the mechanism of weakness and death in these mice. This confirms that disability in inherited peripheral neuropathy can be due to abnormal action potential propagation or neuromuscular transmission, separable from secondary axonal degeneration.
Egr2I268N/I268N mice were generated using a strategy identical to that previously described (Le et al., 2005a), except that the I268N point mutation was engineered into the targeting construct (Supp. Fig 1). Mouse embryonic stem (ES) cell colonies were placed under selection for neomycin resistance, and the resulting colonies screened by Southern blotting. Genomic DNA from positive ES clones was isolated, and the I268N mutation was confirmed by direct sequencing of a PCR product spanning the mutation site. ES cells were injected into C57Bl6 blastocysts to derive chimeric mice, and mice carrying the Egr2I268N locus were back-crossed to C57Bl6 mice for 6 generations prior to analysis. Egr2I268N/I268N mice became severely weak after P14, and were given food and water at the floor of the cage and monitored daily for progression. Mice were euthanized if they were unable to right themselves, or had signs of labored breathing, which uniformly took place between P18 and P21.
Sciatic nerve and HEK293T lysates were analyzed by immunoblotting using standard techniques. Nerves were collected in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton-X100, complete protease inhibitor cocktail (Roche) and sonicated. Primary antibodies used were rabbit anti-Flag (1:1000, Covance), mouse anti-Myc (1:1000, Abcam), mouse anti-HA (1:1000, Covance), and mouse anti-Nab2 (1:1000) (Abdulkadir et al., 2001). For co-immunoprecipitation experiments, HEK293T cells were transfected with the indicated expression plasmids using the Fugene transfection reagent (Roche). Two days after transfection cells were collected in low-IP buffer from the nuclear complex Co-IP kit (Active Motif) prepared according to the manufacturer’s instructions, and incubated with Protein G-agarose beads bound to either anti-Myc antibodies, or M2-anti-Flag agarose conjugated antibodies (Sigma). Immunoprecipitates were washed in Co-IP buffer, and then analyzed by SDS-PAGE and immunoblotting. Generation of expression plasmids for Flag and HA-tagged mouse Egr2, Flag-Egr2I268N, myc-tagged mouse Nab2 was described previously (Svaren et al., 1996; Warner et al., 1999).
Mice were euthanized and perfused with 1% paraformaldehyde/2.4% glutaraldehyde in 100 mM cacodylate buffer (pH 7.4), and sciatic nerves were dissected and postfixed for 16 h in the same buffer. Nerves were transferred to 100 mM cacodylate buffer (pH 7.4) and processed for plastic and electron microscopic sections. Nerve segments were dehydrated and embedded in epon-araldite. One-micron sections of nerve were stained with toluidine blue for light microscopy evaluation. For electron microscopy, thin sections were prepared and imaging was performed using a JEOL 1200 electron microscope. We calculated g-ratios by mathematical division of axonal diameter by fiber diameter, that is, axonal diameter/(axonal + myelinating sheath diameter). Axons counts were performed from electron micrographs of sciatic nerve preparations. Axons > 1µm were counted in 2 fields, and total axon counts were calculated as (#axons/area × total area of the nerve). Skin was collected following perfusion of the mice with 4% paraformaldehyde, paraffin embedded, cut into 6-µM sections and stained with hematoxylin and eosin.
Fresh frozen sciatic nerves were embedded in OCT (Tissue-TEK), cut into 10-µM sections and fixed for 15−20 min in 4% paraformaldehyde. Sections were cut at 6 µM and antigen retrieval was performed in 0.01 M sodium citrate for 30 min. Blocking was performed in PBS-BB (PBS with 1% NFD-Milk, 1% bovine albumin and 0.2% Triton-X100) for 30 min at 25°C. Nerve were incubated with mouse monoclonal BrdU (1:200, Roche), rabbit polyclonal Pou3f1 (1:1,000, generous gift from J. Bermingham) or rabbit polyclonal Ki67 (1:1000, Vector Laboratories) in blocking buffer at 4 °C for 14−18 h. Prior to incubation with BrdU antibody nerve sections were treated with 4N HCl for 10 min. Fluorescently labeled secondary antibodies (goat anti-mouse Cy3 for BrdU; goat anti-rabbit Cy3 for SCIP and Ki67) were applied for 1.5 h at 25 °C in PBS-BB. Neurofilament staining of embryonic day 10.5 mice was performed as previously described (Le et al., 2005b). Briefly, embryos were dissected, fixed in 4% paraformaldehyde at 4°C for 18 h, quenched of endogenous peroxidase activity and blocked in PBS with 4% BSA and 1% Triton at 4°C for roughly 18 h. Neurofilament antibody (Developmental Studies Hybridoma Bank clone 2H3, 1:50) was incubated in blocking solution for 48 h at 4 °C. Goat anti-mouse HRP secondary was then incubated in at 4°C for 18 h, with signal development by diaminobenzidine tetrahydrochloride (DAB) (Sigma).
For teased nerve immunohistochemistry, P14 mice were sacrificed and sciatic nerves removed and postfixed in 4% para for 1h at 4°C. Teased nerves were dried and stored at −20°C, prior to staining were postfixed with acetone for 10m at −20°C, then blocked in 5% fish skin gelatin/0.1% triton in PBS for 1h. Rabbit primary antibodies, were incubated in blocking buffer overnight, slides were washed with PBS + 0.1% triton and incubated in secondary goat anti-mouse Cy3 antibody for 1h at RT. For mouse antibodies the Vector MOM immunodetection kit was used (Vector Laboratories), with visualization via alexa-488 streptavidin. Rabbit polyclonal antibodies and dilutions used were: Kv1.1 – 1:100 (Abcam); Nav1.6 – 1:100 (Chemicon); Kv1.2 – 1:100 (Chemicon); Pan Nav – 1:250 (Chemicon). Mouse monoclonal to Caspr was used at 1:1000 (gift from Dr. Elior Peles).
Sciatic nerves were collected from P14 mice, placed in Trizol (Invitrogen), homogenized, and total RNA was generated according to the manufacturer’s instructions. First strand RT libraries were produced from total RNA using MMLV reverse transcriptase (Invitrogen), and analyzed using intron-spanning gene specific primers with 2X Sybr-Green qPCR mix (Clontech) on an ABI Taqman machine. Levels of GAPDH transcript were used for normalization.
To visualize the presynaptic component of the neuromuscular junction we used Thy1-YFP16 mice, which express YFP in all spinal motor neurons (Feng et al., 2000). Egr2I268N/+ mice were bred with Thy1-YFP/+ mice maintained on a C57Bl6 background, and then Egr2I268N/+: Thy1-YFP/+ mice were bred to Egr2I268N/+ mice. Mice were euthanized at P14 or P18, perfused with 4% paraformaldehyde, and the diaphragm or extensor digitorum longus muscles were removed. Muscles were post-fixed in 4% paraformaldehyde, stained with Alexa-594 α-bungarotoxin (Invitrogen) for 30 min at room temperature, and whole mounted onto slides for imaging.
Electrophysiology was performed on mice at postnatal day 18–20. Mice were anesthetized with avertin, and placed on a heating pad. Subcutaneous platinum subdermal EEG electrodes (0.4mm diameter, 12mm length; Viasys) were used. Stimulating electrodes were placed just above the left ankle and the left sciatic notch for nerve stimulation. Recording electrodes were placed in the footpad. Evoked CMAPs were obtained using a Viking Quest electromyography machine (Nicolet, Madison, WI, USA) using supramaximal stimulation, and distance between the two sites of stimulating electrodes was used to calculate conduction velocity.
Myelination in Schwann cells is critically dependent on proper gene dosage, and overexpression of wild-type forms of genes such as MPZ and PMP-22 are themselves capable of producing a demyelination neuropathy in rodents (Sereda et al., 1996; Wrabetz et al., 2000). To maintain this precise stoichiometry we created a mouse model of inherited neuropathy with a point mutation in Egr2 using a knock-in approach to incorporate the I268N mutation into the endogenous mouse Egr2 genomic locus (Supp. Fig 1). This leaves the entire locus intact, with only a small loxP scar present in the single intron, thereby maintaining all necessary control elements for proper Egr2 expression. Indeed protein lysates from wild-type and Egr2I268N/I268N mice showed identical protein levels of Egr2 in postnatal day 14 (P14) sciatic nerve (Fig 1A).
Many neuropathy associated Egr2 mutants have alterations in the DNA binding domain, and act in a dominant negative manner to inhibit myelin gene expression, consistent with their dominant inheritance pattern (Nagarajan et al., 2001). However, Egr2-I268N leads to recessive disease, and is able to bind normally to an Egr2 consensus DNA sequence, indicating that it has a different mechanism than loss of DNA binding (Warner et al., 1999). The I268N mutation is located in the R1 domain, a region in Egr family members that interacts with Nab proteins (Russo et al., 1993), though this has not been demonstrated for Egr2. Therefore we expressed tagged Egr2 or Egr2-I268N together with Nab2 in 293T cells, and examined their ability to directly interact via coimmunoprecipitation. Wild-type Egr2 and Nab2 were coimmunoprecipitated regardless of whether the complex was pulled down with epitope-tagged Egr2, or Nab2 (Fig 1B,C). In contrast, Egr2 carrying the I268N mutation did not coimmunoprecipitate with Nab2, indicating that Egr2-I268N is physically unable to bind Nab2. This provides a molecular basis for the prior observation that Nab2 does not modulate transcription of an artificial promoter system with Egr2-I268N (Warner et al., 1999), and supports the idea that abnormalities present in Egr2I268N/I268N mice are due the inability of Egr2-I268N to interact with Nab proteins.
Normal peripheral myelination in mice begins at around postnatal day 1 (P1), and is completed from P14-P30 (Hahn et al., 1987; Bermingham et al., 1996; Mirsky and Jessen, 1996). Egr2I268N/I268N mice weigh the same as their wild-type and heterozygous (Egr2I268N/+) littermates through the first two weeks of life, in contrast to Egr2 null, Egr2 hypomorphic, and Nab1/Nab2 null mice which are runted from birth (Topilko et al., 1994; Le et al., 2005a; Le et al., 2005b). However, Egr2I268N/I268N mice rapidly lose weight after P14 and inevitably die by P21 (Fig 2A, B). The decline in function is striking, with only mild slowness and gait abnormality present at P14, but rapid ascending weakness leading to complete hindlimb paralysis and death within 4–7 days (see Supplemental movie 1–Supplemental movie 3). On gross inspection the sciatic nerve appears translucent in Egr2I268N/I268N mice as compared with wild-type, indicating abnormal myelination (Fig 2C, D).
To examine in detail the myelin abnormality in Egr2I268N/I268N mice, we performed histological and electron microscopic analysis of the sciatic nerve at P14. Toluidine-blue stained plastic sections showed decreased numbers of myelinating profiles in Egr2I268N/I268N nerves as compared to wild-type (Fig 3A, B). Electron microscopy revealed that there were frequent large caliber axons without any myelin, and that when present myelin sheaths were abnormally thin (Fig 3C–F). Additionally, Schwann cell nuclei frequently appeared irregularly shaped. No onion-bulb formations nor myelin debris were present, indicating that no demyelination or remyelination had taken place by P14. Analysis of dorsal and ventral nerve roots showed that the abnormal myelination was present equally in sensory and motor nerves (Supp. Fig. 2).
To further characterize the abnormality in myelin development in Egr2I268N/I268N mice, we performed immunohistochemistry on P14 sciatic nerves for markers of Schwann cell differentiation. Immunostaining for Pou3f1/Oct6/SCIP, a transcription factor important in the promyelinating phase of Schwann cell development, showed increased numbers of Pou3f1 positive cells in Egr2I268N/I268N nerves as compared to wild-type (Fig 4A, B). Furthermore, more dividing nuclei were present in P14 sciatic nerves from Egr2I268N/I268N mice compared to wild-type, as visualized by Ki67 immunostaining (Fig 4B,C). These findings indicate that many Schwann cells in Egr2I268N/I268N mice are unable to progress beyond the promyelinating phase and continue to proliferate, similar to that seen in Egr2 hypomorphic and Nab1/2 null mice (Le et al., 2005a; Le et al., 2005b).
Lastly we directly examined mRNA levels of Egr2 target genes involved in myelination by quantitative RT-PCR on P14 sciatic nerve. We examined several target genes of Egr2 transcriptional regulation, as evidenced by the presence of Egr2 binding sites (Maier et al., 2003; Leblanc et al., 2005; LeBlanc et al., 2006), increased expression in Schwann cells with forced Egr2 expression (Nagarajan et al., 2001), and decreased mRNA levels present in Egr2 hypomorphic mice (Le et al., 2005a). As expected, Egr2 mRNA levels were identical in wild-type and Egr2I268N/I268N nerves, in agreement with Egr2 protein levels (Fig 1A) and confirming that transcriptional regulation of the Egr2I268N knockin locus remains intact. In contrast, mRNA levels of the Egr2 target genes Pmp22, Periaxin, Mpz, and Gjb1 (Connexin 32) are all diminished 2–5 fold (Fig 4E). These results indicate that Egr2I268N/I268N mice have congenital hypomyelination with most Schwann cells arrested in the promyelinating phase, and lack of transcriptional activation of multiple Egr2 target genes associated with myelination. This argues that Egr2-Nab1/2 complexes are critical transcriptional regulators of Schwann cell development, that loss of the interaction between Egr2 and Nab proteins is sufficient to recapitulate loss of either Egr2 or Nab1/2, and that the lack of functional Egr2-Nab complexes is responsible for the severe peripheral neuropathy in patients with the Egr2I268N mutation.
Aside from its function in peripheral nerve development, Egr2 is also important in hindbrain development, and Egr2 null mice have loss of rhombomeres 3 and 5 and typically die at birth (Schneider-Maunoury et al., 1993). In contrast, hindbrain abnormalities are variable in Egr2 hypomorphic mice, and do not occur in Nab1/2 null mice (Le et al., 2005a; Le et al., 2005b). We performed neurofilament immunostaining on E10.5 embryos and examined hindbrain segmentation in Egr2I268N/I268N mice. Unlike Egr2 null and hypomorphic mice, Egr2I268N/I268N mice did not show absence of rhombomeres 3 or 5, or changes in the projection pattern of cranial nerves V or VII. However, 33% of Egr2I268N/I268N mice had partial or complete fusion of cranial nerves IX and X (Fig 5A–C). Surprisingly, 15% of Egr2I268N/+ mice showed a similar defect in cranial nerve IX/X development, indicating that this abnormality can be inherited as a dominant trait.
Nab1/2 null mice have several abnormalities in addition to hypomyelination, including epidermal hyperplasia and hyperkeratosis (Le et al., 2005b). In contrast, the skin was histologically normal in Egr2I268N/I268N mice, indicating that this aspect of Nab1/2 function is likely mediated through other Egr family members (Supplemental Fig 2).
These data indicate that though hindbrain development is abnormal in a minority of Egr2I268N/I268N mice, it cannot be responsible for the uniformly rapid decline and lethality in these mice.
Patients with the Egr2I268N/I268N mutation have congenital hypomyelination with severe slowing of nerve conduction velocity of ~3 m/s (Szigeti et al., 2007). We performed electrophysiology on Egr2I268N/I268N mice between P18-P21, in animals that showed severe lower extremity weakness or paralysis (Fig 6A, B). Nerve conduction velocity was markedly slowed in Egr2I268N/I268N mice, with an average nerve conduction velocity of ~3 m/s, significantly slower than wild-type or Egr2I268N/+ mice at this age (~20 m/s; Fig 6C). Interestingly, Egr2I268N/I268N mice showed marked temporal dispersion and diminished CMAP amplitude on proximal stimulation, with relative preservation of distal CMAP amplitude and duration (distal/proximal CMAP amplitude ratio: 1.22 ± 0.3 in wild-type vs. 0.46 ± 0.3 in Egr2I268N/I268N; p<0.001). This suggests that congenital hypomyelination in Egr2I268N/I268N mice leads to variable slowing and loss of action potential conduction along the course of the nerve, and that conduction block likely contributes to the rapidly progressive weakness in these animals.
Disruption of node of Ranvier organization is a common finding in both acute demyelination (Rasband et al., 1998) and chronic dysmyelination (see Poliak and Peles, 2003 for review). In rodents, peripheral nodes typically express the voltage-gated sodium channel Nav1.2 at birth, but switch during development (prior to P7) to express Nav1.6 (Rasband and Trimmer, 2001). We found that Nav1.6 is present in nodes from P14 sciatic nerve of Egr2I268N/I268N mice (Fig 7A,B), whereas Nav1.2 is absent (not shown), indicating that the normal developmental switch between these channels is intact. However, staining with Caspr (a marker of the paranode) appeared dispersed and was occasionally diminished or absent. Furthermore Kv1.1, which is normally localized to the juxtaparanode (Fig 7C), was found in all cases to either (i) overlap with Caspr indicating spread into the paranode (Fig 7D); ii) be dispersed into the internode (Fig 7E); or (iii) be completely absent (Fig 7F,G). These abnormalities are similar to that seen in other models of inherited dysmyelination, (Poliak and Peles, 2003; Ulzheimer et al., 2004; Devaux and Scherer, 2005) and likely contribute to the alterations in nerve electrophysiology seen in the Egr2I268N/I268N mice.
Weakness and sensory loss in patients with demyelinating CMT is typically associated with secondary axon loss (Sancho et al., 1999; Krajewski et al., 2000). Therefore we performed axonal counts on sciatic nerves from wild-type and Egr2I268N/I268N mice. There was no difference in the number of sciatic axons (> 1µm) in wild-type vs. Egr2I268N/I268N mice at either P14 (before onset of severe weakness) or P18 (at which time the hindlimbs are essentially paralyzed) (Fig 8A). Given that the sciatic nerve is a proximal site and predominantly sensory in mice, we examined the innervation of neuromuscular junctions (NMJs) in the extensor digitorum longus (EDL) and diaphragm muscles in P18 Egr2I268N/I268N mice. Egr2I268N/+ mice were crossed with Thy1-YFP transgenic mice (Feng et al., 2000), and Egr2I268N/I268N:Thy1-YFP mice were generated to visualize motor axons, with co-staining for α-bungarotoxin to visualize post-synaptic acetylcholine receptors. There was no evidence of axon degeneration, retraction or unoccupied NMJs in Egr2I268N/I268N mice despite the fact that the EDL is essentially paralyzed at P18 (Fig 8B). Interestingly, NMJs from Egr2I268N/I268N mice showed extensive sprouting at P18 (Fig 8E, F) which is never observed in wild-type mice (Fig 8D), nor in Egr2I268N/I268N at P14 prior to the onset of severe weakness (data not shown). NMJ sprouting is known to occur after blockage of action potentials with tetrodotoxin (Brown and Ironton, 1977), in the med/med mice which have loss of Nav1.6 function (Duchen, 1970; Burgess et al., 1995), or inhibition of presynaptic NMJ function with botulinum toxin (Pestronk and Drachman, 1978). It is believed to be mediated by secreted factors from muscle fibers in which electrical activity is lost due to structural or functional (i.e. tetrodotoxin) denervation. These data strongly support that the rapidly progressive weakness in Egr2I268N/I268N mice is due to conduction failure either at proximal sites along the nerve, or at the NMJ itself.
In this study we have described the development of an animal model of human CMT4E, due to the recessively inherited I268N mutation in Egr2. We found that this mutation leads to complete loss of interaction between Egr2 and the Nab transcriptional coregulatory proteins, and produces severe congenital hypomyelination in mice, clearly demonstrating that the Egr2-Nab complex is the critical transcriptional regulator of Schwann cell development. Furthermore we found that the weakness and premature death in these animals is due to conduction failure rather than axonal loss, confirming a second important mechanism that results in weakness and sensory loss which may play a prominent role in congenital hypomyelinating neuropathy.
Transcription factors typically function as components of large multi-molecular complexes to regulate the complex gene regulatory programs necessary for proper cellular development and maintenance of differentiation. This study, together with many others, supports the important role of Egr2 as a transcriptional regulator of myelin development (Jessen and Mirsky, 2002), but furthermore strongly confirms that it is the Egr2-Nab complex, rather than Egr2 alone, which is critical for myelin development (Le et al., 2005a). Using in vitro reporter assays Nab proteins were first posited to act as corepressors of Egr function (Russo et al., 1995; Svaren et al., 1996), or alternatively coactivators depending on the promoter context (Warner et al., 1999; Sevetson et al., 2000). Simply by disrupting the interaction between Egr2 and Nab proteins with the human disease mutation I268N we were able to largely recapitulate the hypomyelinating phenotype of either the Egr2 hypomorph, or the Nab1/2 null mice (Le et al., 2005b). Schwann cells in the Egr2I268N/I268N mice largely remain in a promyelinating phase, where they continue to express the Pou3f1 transcription factor and proliferate. Furthermore, all of the Egr2-Nab target genes investigated were expressed at low levels, supporting that although Nab proteins frequently function as transcriptional corepressors on artificial promoter systems, in myelin development Nab proteins primarily function as coactivators.
In contrast to the role of Egr2-Nab complexes in nerve development, we observed the unexpected finding that there are abnormalities in lower hindbrain development in mice with either one or two I268N alleles, and the frequency of these defects appeared to be dose dependent (i.e. ~15% with one I268N allele, and ~30% with two I268N alleles). This raises several interesting points: first, given that none of the Egr2I268N/+ mice die prematurely despite the fact that 15% have similar hindbrain defects, abnormal hindbrain function is not the likely cause of lethality in Egr2I268N/I268N mice; second, the dose responsive nature of the hindbrain defect suggests that Nab transcriptional repression plays a predominant role in determining this phenotype. This then explains the dominant inheritance of this endophenotype – presumably Nab proteins are repressing Egr2-dependent transcription of certain genes involved in hindbrain development similar to that observed in zebrafish (Mechta-Grigoriou et al., 2000). Nab repression of Egr2 activity is partly lost in the Egr2I268N/+ mice, and completely lost in Egr2I268N/I268N mice, leading to the dose response relationship. Interestingly a similar observation was made in a recent report of another Egr2 knockin mutation (Egr2I268F) which found persistent expression of Egr2 target genes in developing hindbrain suggesting gain of function of the Egr2I268F allele (Desmazieres et al., 2008).
One of the remarkable aspects of hereditary disorders of myelin is that patients can have significant disorders of myelin and slow nerve conduction velocity, but not have symptoms unless there is a secondary effect on axonal or synaptic function. This is illustrated well in the case of mutations in ARHGEF10, in which a large family has been described that has diffuse slowing of NCV and demyelination, but minimal symptoms (Verhoeven et al., 2003). Secondary axonal loss appears to be the predominant mechanism by which patients develop disability from distal sensory loss and weakness in CMT1A, both in human patients (Krajewski et al., 2000) and mouse models of the disease (Sancho et al., 1999). However, conduction block (the loss of action potential propagation along an axon) is observed in some forms of CMT (particularly CMTX and CMT1B), and may also contribute to symptoms (Lewis et al., 2000; Pareyson et al., 2006). Additionally, abnormalities in the NMJ have been observed at both the structural and functional level in several animal models of CMT, although the degree to which this occurs in patients remains unclear (Yin et al., 2004; Court et al., 2008). We found that the NMJs of Egr2I268N/I268N mice show normal innervation with marked terminal sprouting that develops in concert with the onset of severe weakness. Terminal sprouting is typically observed after structural or functional denervation, and is likely mediated by muscle derived signals to ultimately produce collateralization (Tam and Gordon, 2003). Given that terminal sprouting can be observed after treatment with either tetrodotoxin (Brown and Ironton, 1977) or botulinum toxin (Pestronk and Drachman, 1978), we cannot be certain whether the sprouting in the Egr2I268N/I268N mice is due to axonal or synaptic dysfunction or both. However, the segmental defects in myelination and preserved distal CMAP amplitudes argue more for an axonal origin, i.e. conduction block. Our data support that axonal dysfunction can contribute to disability in a mouse model of inherited neuropathy, and suggests that a similar mechanism may be involved in congenital hypomyelinating neuropathy in humans as well.
Egr2I268N/I268N mice are now the second animal model produced of congenital hypomyelinating neuropathy (CHN), and is the only one in which a missense mutation found in human patients at normal gene dosage recapitulates disease, in contrast to other CHN models using knockdown of Egr2 (Topilko et al., 1994; Le et al., 2005a), or overexpression of myelin protein zero (Wrabetz et al., 2000). These models share the expected features of human CHN – severe decrease or absence in peripheral myelin formation from birth, with associated early onset or congenital weakness and sensory loss (Houlden and Reilly, 2006). However, while mice overexpressing MPZ have a similar degree of hypomyelination histologically, they live an apparently normal lifespan in contrast to mice with Egr2-Nab dysfunction, which uniformly die within 3-weeks of birth (Wrabetz et al., 2000). The origin of this difference remains unclear, but we suspect differences in genetic background may play a role, possibly influencing the conduction block and/or NMJ dysfunction seen in Egr2I268N/I268N mice. Interestingly Egr2 and Nab proteins are expressed in peripheral neurons (in contrast to MPZ), suggesting the alternative explanation that loss of Egr2-Nab function in peripheral neurons diminishes their ability to compensate for hypomyelination and continue propagating action potentials. Further studies of Egr2I268N/I268N mice on different genetic backgrounds, as well as more detailed analysis of Egr2-Nab in regulating nodal components in peripheral neurons will help to answer this question.
Despite being a relatively common disease associated with significant disability, no treatments which can slow or reverse the progression of hereditary peripheral neuropathies are currently available. Given the recent evidence that progesterone antagonists are protective in a rat model of CMT1A by modulating gene transcription (Sereda et al., 2003), a detailed understanding of transcriptional pathways altered in models such as the Egr2I268N/I268N mice may be useful in designing therapeutic interventions in the future.
Generation of Egr2-I268N knockin mice. A) Schematic diagram of Egr2 locus and targeting construct. The I268N mutation was introduced into exon 3, thereby preserving the intron-exon structure of the locus. The PGK-Neo cassette used for selection was placed in intron 2 and flanked by loxP sites. Embryonic stem cells were selected where recombination took place distal to the I268N mutation, and used to generate chimeric mice. The PGK-neo cassette was then removed by breeding to mice ubiquitously expressing the Cre recombinase under control of the actin promoter, generating the final knock-in allele (Egr2I268N) with only the small residual loxP scar in intron 2. B) Southern blot of EcoRI digested genomic DNA showing the presence of a 7.2 kb band for the wild-type Egr2 locus, and a 6.1 kb band for the recombined locus from the insertion of an EcoRI site which remains next to the loxP scar.
Hematoxylin and Eosin staining of foot pad skin from wild-type (A), Egr2I268N/I268N (B), and Nab1/2 double knockout mice (C). In contrast to the Nab1/2 null knockout mice, skin was normal in Egr2I268N/I268N mice, indicating that the effect of loss of Nab proteins in skin is independent of their interaction with Egr2. In order to investigate a possible difference between motor and sensory nerve involvement, we examined dorsal and ventral L5 nerve roots with electron microscopy. Egr2I268N/I268N mice showed severe dysmyelination in both dorsal sensory (F) and ventral motor (G) roots. Wild-type dorsal (D) and ventral (E) nerve roots are shown for comparison. Scale bars in (D, F) = 5 µm; (E,G) = 2 µm.
Wild-type litter mate mouse at postnatal day 14.
Egr2I268N/I268N mouse at postnatal day 14. Mice are normal size, have mild tremor and hindlimb weakness.
This work was supported by National Institutes of Health (NIH) Neuroscience Blueprint Core Grant NS057105 to Washington University, the Hope Center for Neurological Disorders, by NIH Grants NS040745 (J.M.), 5-T32-DA07261 (E.J.R.), 1K08NS055980 (R.H.B), and the Muscular Dystrophy Association and Children’s Discovery Institute (R.H.B.). R.H.B. holds a Career Award for Medical Scientists from the Burroughs Wellcome Fund. We thank Tatiana Gorodinsky, Nina Panchenko, Robert Schmidt, Karen Green, and Amber Neilson for technical assistance; and members of the Milbrandt and Baloh laboratories for helpful discussions and feedback on this manuscript.