|Home | About | Journals | Submit | Contact Us | Français|
The very limited ability to regenerate axons after injury in the mature mammalian central nervous system (CNS) has been partly attributed to the growth restrictive nature of CNS myelin. Oligodendrocyte myelin glycoprotein (OMgp) was identified as a major myelin-derived inhibitor of axon growth. However, its role in axon regeneration in vivo is poorly understood. Here we describe the generation and molecular characterization of an OMgp allelic series. With a single gene targeting event and Cre/FLP mediated recombination, we generated an OMgp null allele with a LacZ reporter, one without a reporter gene, and an OMgp conditional allele. This allelic series will aid in the study of OMgp in adult CNS axon regeneration using mouse models of spinal cord injury. The conditional allele will overcome developmental compensation when employed with an inducible Cre, and allows for the study of temporal and tissue/cell type-specific roles of OMgp in CNS injury-induced axonal plasticity.
Following development, axons in the mammalian central nervous system (CNS, including the brain and the spinal cord) have largely lost the ability to regenerate in response to injuries (Ramón y Cajal, 1928). Spinal cord injury patients suffer permanent functional deficits due to this inability for axon regeneration in adult CNS, which is in contrast with the robust regenerative ability of axons in the developing CNS or in the adult peripheral nervous system (PNS). The inhibitory action of myelin on axon growth has been hypothesized as one primary reason for this loss of regenerative ability (Berry, 1982; Schwab and Bartholdi, 1996). Several myelin-derived inhibitory proteins have been identified (Filbin, 2003; Yiu and He, 2006). Oligodendrocyte Myelin glycoprotein (OMgp) was identified as a major myelin-derived inhibitor that possesses potent inhibitory activity on neurite outgrowth in vitro (Kottis et al., 2002; Wang et al., 2002). Together with two other myelin inhibitors, Nogo and myelin-associated glycoprotein (MAG), it signals through at least two receptors, NgR1 and PirB, to effect the inhibitory action of myelin (Atwal et al., 2008; Yiu and He, 2006). However, the role of OMgp in axon regeneration in vivo is not well understood. In a published report, OMgp deficient mice were found to exhibit slightly enhanced axon regeneration in the ascending dorsal column sensory tract and the descending raphespinal serotonergic tract but not in the corticospinal tract (Ji et al., 2008). It was unknown whether even this modest effect was caused directly by OMgp deletion, or indirectly by the effect of OMgp deletion on the expression of other genes since injury-induced upregulation of Nogo was abolished in this OMgp mutant (Ji et al., 2008).
In contrast, compensatory gene expression changes have been reported in other germline mutants in the myelin inhibition pathway. Two distinct mutant lines of NgR1 exhibit upregulated Nogo mRNA expression (Kim et al., 2004; Zheng et al., 2005). A MAG mutant exhibits upregulated NgR1 expression (Pernet et al., 2008). Thus, mutations in either an inhibitory ligand or receptor in the myelin inhibition pathway can lead to changes in the expression of other genes in the pathway, which may mask the effect of gene deletion (in the case of the NgR1 or MAG mutants) or complicates the interpretation of the results (in the case of the OMgp mutant). This is particularly a challenge for genetic studies of axon regeneration because the most relevant stage to examine the role of these molecules is the adult stage, when any developmental compensation or other gene expression changes have already occurred in germline mutants. Therefore, inducible gene deletion with the conditional knockout technology is highly desirable for studying these genes in CNS axon regeneration (Zheng et al., 2006). Furthermore, OMgp is expressed by both neurons and glial cells (Habib et al., 1998; Huang et al., 2005; Wang et al., 2002), and a conditional allele will allow for the study of its tissue/cell type-specific function.
In order to establish an inducible and tissue/cell-type specific OMgp gene deletion system, we generated a conditional allele of OMgp. From the same gene targeting event, we also generated two additional alleles: a null allele with a LacZ reporter gene and a clean null allele without any reporter gene. We had two primary considerations when generating this allelic series: 1) minimizing disruption to OMgp expression in the conditional mutants prior to Cre-mediated deletion (Kwan, 2002); and 2) minimizing disruption to the expression of NF1, which when mutated causes Neurofibromatosis type I in humans, since the OMgp gene is embedded in an intron of the NF1 gene (Viskochil et al., 1991; Vourc'h and Andres, 2004).
The overall strategy for gene targeting and Cre/FLP mediated recombination is illustrated in Fig. 1a. The murine OMgp gene has two exons, with the second exon containing all coding sequence (Mikol et al., 1993). We constructed a targeting vector with two Frt sites flanking a reporter/selection cassette, two loxP sites flanking (i.e. floxing) the second exon, and a TK negative selection marker (Fig. 1a). The two Frt sites flank a cassette containing a splice acceptor (SA), a LacZ reporter gene and a Neomycin resistance gene (Neo), such that the cassette can be deleted with the FLP recombinase. The SA-LacZ reporter element would be used to track OMgp expression. The loxP sites were placed in regions of the genomic DNA that are not highly conserved between the mouse and human genomes, and thus less likely to disrupt regulatory sequences (but see below). Cre-mediated deletion of exon 2 is expected to give rise to a null mutation.
Correct gene targeting was detected by Southern blot analysis with a 3' external probe (Fig. 1b) and verified with a 5' internal probe (Fig. 1c). The wild type allele and the targeted allele (OMgpLacZNeoFlox) could be distinguished by PCR using a primer pair flanking the 3' loxP site (Fig. 1d). The targeted allele was transmitted through the germline and bred to a FLP deleter mouse (ACTB:FLPe) (Rodriguez et al., 2000) to obtain the conditional allele, or to a Cre deleter mouse (EIIA-Cre) (Lakso et al., 1996; Xu et al., 2001) to obtain the null allele with the LacZ reporter gene (OMgpLacZNeoNull) (Fig. 1a, e, f). Both the targeted mutant and the conditional mutant were genotyped using the primer pair flanking the 3' loxP site as in Fig. 1d, but they could be distinguished using primer pairs flanking the 5' loxP site (f5 and r4, or a2 and r4 in Fig. 1a, data not shown). The conditional allele was then bred further to the EIIA-Cre line to obtain the clean null allele without the LacZ reporter gene (OMgpCleanNull) (Fig. 1a, g). Because the presence of a selectable marker, especially one with the bi-directional PGK promoter, may interfere with nearby gene expression (Manley et al., 2001; Olson et al., 1996; Scacheri et al., 2001), deleting the selection cassette (PGK-Neo-bpA) would be expected to minimize any disruption to NF1 gene expression in the OMgpCleanNull mutants.
To determine whether the LacZ knockin reporter is functional, we stained adult brain and spinal cord sections with X-Gal for β-galactosidase activity in the OMgpLacZNeoNull mutants (Fig. 2). β-galactosidase activity was detected in the cerebral cortex and corpus callosum at high levels, and, among other brain regions, in the cerebellum and superior colliculus at lower levels (Fig. 2a). In contrast to the proposed role of OMgp as a white matter inhibitor of axon regeneration, robust β-galactosidase activity was detected in the gray matter but not in the white matter of the adult spinal cord in either heterozygous or homozygous mutants (Fig. 2b-f). This suggests that the expression level of OMgp in oligodendroglia is much lower than that in spinal neurons. This observation is consistent with previous immunohistochemical data indicating that OMgp expression in the adult mouse spinal cord is predominantly in the gray matter (Habib et al., 1998). mRNA in situ hybridization data from the Allen Brain Atlas also shows strong OMgp expression in the gray matter of adult mouse spinal cord, whereas a lower level of OMgp transcript is detected in the white matter (http://www.brain-map.org, with OMgp referred to as OMG). The detection of OMgp transcript (Allen Brain Atlas) but not β-galactosidase reporter activity (this study) in the spinal cord white matter likely reflects a higher sensitivity of mRNA in situ hybridization compared with X-Gal staining of the LacZ reporter gene product. Because the proposed role for OMgp in axon regeneration is primarily based on its expression in oligodendrocytes or oligodendrocyte-like cells (Huang et al., 2005; Wang et al., 2002), these observations indicate that it will be important to distinguish the role of OMgp in neurons from that in oligodendroglia in future.
Both the homozygous OMgpFlox and homozygous OMgpCleanNull mutants were viable, fertile and morphologically indistinguishable from their wild type littermates. We next examined the expression of OMgp and NF1 in homozygous OMgpFlox and OMgpCleanNull mutants by Western blot analysis on total spinal cord extracts (Fig. 3). As expected, homozygous OMgpCleanNull mutants did not express OMgp protein (Fig. 3a). However, the level of OMgp protein in homozygous OMgpFlox mice is reduced, to ~35% of the wild type levels (Fig. 3a, b). This indicates that even though we followed all possible steps to minimize the possibility of altered gene expression in the OMgp conditional mutants prior to gene deletion (Kwan, 2002), the expression of OMgp was affected by the limited sequence changes at the loxP insertion sites flanking exon 2. This may reflect the fact that the sequence of the single intron in OMgp is highly conserved between mice and humans (Mikol et al., 1993).
Importantly, the expression of NF1 protein remained at the same level in homozygous OMgpCleanNull mutants as in wild type controls (Fig. 3c, d). There is a trend for a slight increase of NF1 expression in homozygous OMgpFlox mutants that did not reach statistical significance (Fig. 3c, d). In this regard, we noted that a previously published OMgp null allele carries a GFP-Neo fusion gene, and it is not known whether NF1 expression is altered in this mutant (Ji et al., 2008). Thus, our OMgpCleanNull allele may offer a unique advantage when assessing the baseline phenotype of an OMgp null mutant by avoiding any confounding changes in NF1 expression.
To determine whether inducible knockout can be achieved, we bred the OMgp conditional allele (OMgpFlox) to an inducible Cre line with a ubiquitous promoter, CAGGCre-ER™ (Hayashi and McMahon, 2002). The Cre activity from this transgene is dependent on the presence of tamoxifen (TM), which binds to the mutated form of estrogen receptor (ER), leading to the nuclear translocation and activation of the CreER fusion protein (Feil et al., 1996). Mice homozygous for the conditional allele and carrying the CAGGCre-ER™ transgene were analyzed for OMgp protein levels by Western blot analysis following tamoxifen treatment (Fig. 4). Two weeks following the completion of tamoxifen treatment, there was already a significant reduction of OMgp in homozygous OMgpFlox mice with the CAGGCre-ER™ transgene when compared with homozygous OMgpFlox littermate controls without the CAGGCre-ER™ transgene. Four weeks following tamoxifen treatment, OMgp levels in the inducible knockout mice reduced to ~8% of that in littermate controls (Fig. 4), or ~3% of wild type levels. Thus, even though mice homozygous for the OMgpFlox allele exhibited a lower level (~35%) of baseline OMgp protein expression than wild type mice, inducible gene deletion is still effective with this conditional allele. Provided that ~35% of the OMgp protein is sufficient for its normal function and does not cause significant compensatory expression of other genes, the lower baseline level of OMgp in homozygous OMgpFlox mice may even be advantageous for conditional gene deletion: since there is less OMgp protein to start with, even an incomplete conditional gene deletion may be sufficient to disrupt the normal function of OMgp.
In summary, we generated an allelic series for the gene encoding OMgp, a major myelin-derived inhibitor of axon growth. The clean null allele, OMgpCleanNull, will be useful in analyzing the baseline phenotype of an OMgp null mouse without significantly affecting the expression of NF1. The OMgpLacZNeoNull allele will be a useful reporter line to track OMgp gene expression with the LacZ knockin reporter. The conditional allele, OMgpFlox, will be useful in analyzing OMgp function in axon plasticity in adult CNS with minimized compensatory gene expression, in establishing the temporal roles of OMgp in different phases post-injury, and in investigating the tissue/cell-type specific function of OMgp.
A replacement gene targeting vector was constructed with a 3.9 kb left arm homology, a Frt-flanked SA-LacZ-Neo reporter gene and positive selection cassette, a loxP site, a 2.7 kb middle arm homology, a second loxP site (in the same orientation as the first one), a 4.1 kb right arm homology, and a herpes simplex virus thymidine kinase gene (TK) as the negative selection marker. The selection cassette and the first (5′) loxP site replaced a 97 bp sequence in the intron that is not conserved between the mouse and human genomes. The second (3') loxP site is marked with a BglII site so that the conditional allele (resulting from a crossover at the left arm and a crossover at the right arm) can be distinguished from the non-conditional allele (resulting from a crossover at the left arm and a cross over at the middle arm) as well as the wild type allele on a Southern blot with BglII digest hybridized with the 3' external probe: wild type, 10.2 kb; conditional allele, 5.9 kb; non-conditional allele, 13.7 kb (data not shown). The homology arms were amplified from isogenic 129S5 genomic DNA with a high fidelity PCR amplification kit (Roche) and sequence verified. The SA-LacZ (or SA-β-Galactosidase) cassette was a gift from Philippe Soriano (Fred Hutchinson Cancer Center) (Friedrich and Soriano, 1991). The floxed PGK-Neo-bpA cassette was a gift from Richard Behringer (M.D. Anderson Cancer Center). Sequences of primers used for constructing the homology arms or genotyping are provided as supplementary information (Supplementary Table 1).
The linearized targeting vector was electroporated into AB2.2 embryonic stem cells followed by positive selection with Geneticin (for Neo) and negative selection with FIAU (against TK). The sequence of all primers used in constructing the targeting vector, genotyping and detecting the Cre/FLP recombination products is provided as supplementary information. Targeted ES cells harboring the OMgpLacZNeoFlox allele were injected into blastocyst embryos (by the UCSD Transgenic Mouse Core). Chimeric mice were bred to C57BL/6 females for germline transmission.
3-5 month old mice were euthanized with anesthetic overdose and perfused transcardially with 4% paraformaldehyde. Brain and spinal cord were extracted and post-fixed in 4% paraformaldehyde for two hours followed by cryoprotection in 30% sucrose overnight. Specific regions of the spinal cord were embedded in OCT, frozen over dry ice and 20 μm cryosections thaw-mounted onto Superfrost Plus slides. To stain for LacZ, slides were incubated overnight at 37°C in X-gal staining solution (5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, 2 mM MgCl2, 40 mg/ml X-gal in 1× PBS), and then dehydrated through graded alcohol washes and cover-slipped.
Tamoxifen (50 mg/ml) was diluted in 90% sunflower oil and 10% ethanol and mice were injected once daily (0.125 mg/g of body weight) intraperitoneally for five consecutive days. At 2 or 4 weeks after the last injection, mice were sacrificed for analysis. Spinal cord tissue was homogenized in protein extraction buffer, centrifuged and protein concentration in supernatant quantified using Lowry assay (Bio-Rad DC Protein Assay kit). 30 μg protein samples from different OMgp mutant mice were separated in a 5% resolving gel using SDS-PAGE and then transferred onto a nitrocellulose membrane (Amersham Biosciences). Membranes were washed, blocked (5% milk in TBS-T) and incubated in the appropriate antibodies overnight at 4°C. Next day, membranes were washed and incubated in the appropriate HRP-conjugated secondary antibodies (Pierce) and detected using Supersignal West Dura ECL solution (Pierce). Antibodies for Western blot analysis: rat anti-OMgp at 1:1000 (R&D); rabbit anti-Neurofibromin 1 (or NF1) at 1:200 (Santa Cruz); mouse anti-α-Tubulin at 1:1000 (Sigma). Scanned images were quantified using ImageJ software (normalized to α-Tubulin levels) and tested for statistical significance using GraphPad Prism software.
We thank Joseph Gleeson and Silvia Evans for sharing transgenic mouse lines; Anthony Wynshaw-Boris for advice on the breeding scheme with the Cre deleter line. Supported by grants from NIH/NINDS (R01NS054734) and the Christopher and Dana Reeve Foundation (to B.Z.). Imaging support was partly provided by the UCSD Neuroscience Microscopy Shared Facility (P30NS047101). J.K.L. is supported by a Ruth L. Kirschstein NRSA Postdoctoral Fellowship (F32NS056697).