Charcot-Marie-Tooth-linked mutant GARS is toxic to peripheral neurons independent of wild-type GARS levels

Abstract

Charcot-Marie-Tooth disease type 2D (CMT2D) is a dominantly inherited peripheral neuropathy caused by missense mutations in the glycyl-tRNA synthetase gene (GARS). In addition to GARS, mutations in three other tRNA synthetase genes cause similar neuropathies, although the underlying mechanisms are not fully understood. To address this, we generated transgenic mice that ubiquitously over-express wild-type GARS and crossed them to two dominant mouse models of CMT2D to distinguish loss-of-function and gain-of-function mechanisms. Over-expression of wild-type GARS does not improve the neuropathy phenotype in heterozygous Gars mutant mice, as determined by histological, functional, and behavioral tests. Transgenic GARS is able to rescue a pathological point mutation as a homozygote or in complementation tests with a Gars null allele, demonstrating the functionality of the transgene and revealing a recessive loss-of-function component of the point mutation. Missense mutations as transgene-rescued homozygotes or compound heterozygotes have a more severe neuropathy than heterozygotes, indicating that increased dosage of the disease-causing alleles results in a more severe neurological phenotype, even in the presence of a wild-type transgene. We conclude that, although missense mutations of Gars may cause some loss of function, the dominant neuropathy phenotype observed in mice is caused by a dose-dependent gain of function that is not mitigated by over-expression of functional wild-type protein.

Figure 1. The CAG-GARS transgenic construct is robustly expressed in tissues and cell types affected by CMT.

(A) The wild-type GARS transgene construct contains a CAG promoter, followed by an intron, and the full-length GARS cDNA, including the N-terminal mitochondrial localization signal, with flanking 5′ and 3′ UTR sequence. These elements were between two insulator sequences (INS). (B) Immunoblotting with anti-GARS confirmed that transgenes A and D express at similar levels, both above the level of FVB/N wild-type mice in spinal cord and sciatic nerve. Anti-β actin was used as a loading control. (C) Confocal images of teased fiber preparations from sciatic nerve stained with neurofilament and GARS antibodies from one-month-old animals. All three red channel images (GARS staining) were matched for laser intensity, gain, and the Z depth of field (scale bars are 40 µm). To confirm that transgenes A and D were overexpressed in these tissues, exposure-matched analysis was performed on FVB/N control mice (n = 3), TgA mice (n = 3) and TgD mice (n = 3). F1 mice with the transgenes overexpress GARS in sciatic nerve and spinal cord. We crossed Nmf249/+ mice with WT; TgD mice and analyzed F1 progeny in all of the four resulting genotypes. Western blot was used to evaluate expression in the spinal cord (D) and in the sciatic nerve (E). On both wild-type and mutant backgrounds, the transgene overexpresses wild-type GARS in these tissues. F) Aminoacylation activity for alanyl tRNA synthetase (AARS) was not different between wild type (black) and WT;TgA (red) spinal cord homogenates. G) Spinal cord homogenates from WT;TgA mice showed much higher (>10 fold) aminoacylation activity when assayed for GARS activity. Three wild type and four TgA littermates were tested at nine weeks of age.

Figure 2. Neither motor nor sensory axon loss in Nmf249/+ mice is abated by transgenes A or D.

Semi-thin sections of motor and sensory branches of the femoral nerve collected from one month old mice were stained with Toluidine Blue. Representative images are shown in (A-L)(scale bar in J is 100 µm and higher magnification insets are 30×30 µm). (M) Myelinated axons were counted in femoral nerves. Axon counts in the motor nerves were decreased from 587±32 (n = 12) in wild type mice to 435±26 (n = 9) in Nmf249/+ (p<0.001). Wild type mice with TgA and TgD had motor axon counts of 579±29 (n = 11) and 614±22, respectively and were not statistically different from counts in wild type mice. Nmf249/+ mice with TgA had motor axon counts of 441±43 (n = 5) and with TgD had motor axon counts of 438±52 (n = 5), which were not significantly different compared to Nmf249/+ mice. Mice with the mutant allele all had significantly reduced motor axon counts when compared to any of the wild type groups (p<0.001). Myelinated axon numbers in the sensory nerves were significantly (p<0.001) decreased from 876±83 (n = 12) for wild type mice to 654±94 (n = 9) for Nmf249/+. Wild type mice with TgA and TgD had sensory axon counts of 842±64 (n = 11) and 885±89 (n = 6), respectively, and were not statistically different from wild type mice. Nmf249/+ mice with TgA had motor axon counts of 730±118 (n = 5), Nmf249/+ mice with TgD had motor axon counts of 758±77 (n = 5). Transgenes A and D did not significantly alter sensory axon number compared to Nmf249/+ alone, but were somewhat intermediate with Nmf249/+; TgA mice being modestly reduced relative to wild type controls (p<0.05), whereas TgD mice were not different from controls. (N) A cumulative histogram shows the distribution of axon diameters in all six genotypes and demonstrates that there is a shift to smaller axons in the mutant genotypes. There is an obvious clustering of WT, WT; TgA, and WT; TgD mice and Nmf249/+, Nmf249/+; TgA, and Nmf249/+; TgD mice.

Figure 3. Wild-type over-expression does not improve motor nerve conduction and body weight in Nmf249/+ mice.

(A) The Nmf249/+ allele significantly reduces nerve conduction velocity (NCV) in one-month-old mice from 33.9±6.7 m/s (n = 8) measured in control wild-type animals to 7.8±1.7 m/s (n = 7) (p<0.001). Nmf249/+; TgA had nerve conduction of 10.2±1.6 m/s (n = 5) and Nmf249/+; TgD had velocities of 12.3±3.0 m/s (n = 3), which were not significantly different than Nmf249/+. Compared to wild-type, all three mutant genotypes had significantly reduced NCVs (p<0.001). WT; TgA and WT; TgD NCVs were not significantly different from wild-type mice: 32.0±5.4 (n = 8) and 29.3±2.8 (n = 6), respectively. (B) As an indicator of overall health, body weight was measured at P30. Wild-type mice weighed 18.1±2.6 g (n = 8) and were not significantly different from WT; TgA mice, which weighed 19.5±1.8 g (n = 6), and WT; TgD mice that weighed 19.0±1.6 g (n = 6). Compared to wild type mice, Nmf249/+ mice were significantly lighter, weighing 10.1±1.4 g (n = 8)(p<0.001). Nmf249/+; TgA and Nmf249/+; TgD mice were not significantly different than nontransgenic mutant mice with a mean weight of 11.6±0.8 g (n = 5) and 12.9±0.5 g (n = 3) respectively.

Figure 4. Neuromuscular junction morphology and occupancy.

Neuromuscular junctions (NMJs) from plantaris muscles were collected from one-month-old mice and visualized with antibodies labeling neurofilament (NF) and synaptic vesicle protein 2 (SV2) (green), and α-bungarotoxin, which binds acetylcholine receptors (AChRs) (red). NMJs from (A) wild type, (B) WT; TgA, and (C) WT; TgD mice were fully occupied by presynaptic terminals that completely overlap signals from postsynaptic markers with a complex, pretzel-like morphology. D) Nmf249/+, (E) Nmf249/+; TgA, and (F) Nmf249/+; TgD mice have NMJs with regions of post-synaptic staining that are not innervated (arrowheads) and some post-synaptic NMJs have no associated nerve (arrows). (G) The occupancy of NMJs was quantified by scoring as fully innervated, partially denervated or fully denervated, based on the overlap between pre- and post-synaptic staining. 96.7±2.1% of wild-type NMJs were fully innervated, while only 19.6±9.3% were fully innervated in Nmf249/+ mice (p<0.001). No significant change was seen between mutant mice and mutant mice with transgenes A and D, where 14.5±5.9% and 25.5±12.8% of NMJs were fully innervated, respectively. All scale bars are 20 µm.

Figure 5. Wire hanging deficits in C201R/+ mice are not improved by over-expression of GARS.

The severity of motor dysfunction in C201R/+ mice was quantified using a wire hang test of grip strength. The mice were placed on a wire grid, which was then inverted and the latency to fall was determined. The test was stopped at 60 seconds. Each mouse was evaluated in three trials with a 30 sec inter-trial interval. The means ± SD of the trials are plotted, and each point represents an individual animal. This test was performed at 3 weeks (A, C) and 6 weeks (B, D). Some wild-type mice were noncompliant and failed to hang for the full 60 seconds. Mutants were never able to finish the task. Neither transgene improved the behavioral deficits in C201R/+ mice at 3 or 6 weeks of age. (E, F) Nerve conduction velocities (NCVs) were also significantly reduced in this milder CMT2D mouse model and this was not improved by either transgene. NCVs were measured at 12 weeks for mice crossed with transgene A and at 8 weeks for mice crossed with transgene D. (E) In the cross between C201R/+ mutant mice and transgene A the NCVs were: 36.1±5.4 for wild type (n = 6), 32.9±4.9 for WT; TgA (n = 6), 21.1±9.4 for C201R/+ (n = 5), and 20.7±3.8 for C201R/+; TgA mice (n = 5). (F) In the cross between C201R/+ mutant mice and transgene D, the NCVs were: 38.9±3.5 for wild type (n = 2), 33.9±7.3 for WT; TgD (n = 4), 17.3±1.9 for C201R/+ (n = 5), and 22.6±2.5 for C201R/+; TgD mice (n = 6). Transgenes A and D did not significantly improve NCVs in C201R/+ mutant mice.

Figure 6. Transgenes A and D restore viability to C201R/XM256 mice.

Motor and sensory branches of the femoral nerves of rescued C201R/XM256; Tg mice were isolated at 8 weeks of age and compared to littermate wild-type, XM256/+, and C201R/+ control mice. Toluidine blue stained sections show that motor and sensory nerves from wild type (A, B) and XM256/+ mice (C, D) control mice are qualitatively similar to nerves from C201R/+ (E, F), C201R/XM256; TgA (G, H), and C201R/XM256; TgD (I, J) mice. (K) Functionally, nerve conduction velocities in wild type and XM256/+ are not significantly different (35.2±3.4 (n = 7), and 35.5±4.2 (n = 4), respectively). However, as anticipated, values are reduced in C201R/+ mice (20.5±1.9 (n = 6), p<0.001). Conduction velocities in rescued C201R/XM256 mice were not significantly different than C201R/+ mice, but still significantly reduced compared to wild type controls (p<0.001). Values were: C201R/XM256;TgA = 25.7±0.9 (n = 4) and C201R/XM256;TgD = 20.9±2.7 (n = 3). (L) Axon numbers in the motor branch of the femoral nerve were as follows: wild type = 575±17 (n = 17), XM256/+ = 573±34 (n = 6), C201R/+ = 566±59 (n = 6), C201R/XM256;TgA = 564±14 (n = 6), C201R/XM256;TgD = 546±17 (n = 4). There were no significant differences in these values. Axon numbers in the sensory branch of the femoral nerve were also unchanged between genotypes as follows: wild type = 876±50, XM256/+ = 859±32, C201R/+ = 830±89, C201R/XM256;TgA = 822±49, C201R/XM256;TgD = 803±42. Animal numbers are the same as for motor axons except for wild type, where n = 7.

Figure 7. Neuromuscular junctions in rescued mice are similar to those in C201R/+ mice.

Neuromuscular junctions (NMJs) in wild-type (not shown) and XM256/+ (A) mice are mature pretzel-like structures. B) While most NMJs in C201R/+ mice are fully innervated, some NMJs are partially denervated or fully denervated. In rescued C201R/XM256; TgA (C), and C201R/XM256; TgD (D) mice, NMJ morphology is similar to that of C201R/+mice, with regions of denervation in some NMJs with neighboring NMJs that are fully innervated. NMJ occupancy was evaluated in these mice to quantitatively determine the severity of the phenotype. (E) 98.5±1.5% (n = 6) of NMJs were fully occupied in wild-type mice and 95.5±1.4% (n = 6) in XM256/+ mice. These were both significantly higher than the 57.0±12.0% (n = 6) of NMJs that were fully occupied in C201R/+ mice (p<0.001). The percentage of fully innervated NMJs was 67.0±14.3% (n = 6) in rescued C201R/XM256; TgA mice and 64.0±9.8% (n = 4) in C201R/XM256; TgD mice, and was not signficantly different from C201R/+ mice.

Figure 8. Increased dosage of mutant Gars increases the severity of neuropathy, but wild-type over-expression only improves viability in homozygous C201R mice.

The motor branch of the femoral nerve was used to assess the severity of the neuropathy in homozygous and compound heterozygous mice at post-natal day 17. Motor nerves from wild-type (A), C201R/+(B), and Nmf249/+ (C), were compared to nerves from C201R/C201R; TgD (D), C201R/Nmf249 (E), and C201R/Nmf249; TgD (F), which were smaller. The scale bar (shown in C) is 50 µm. The gross observation that the motor nerves are smaller in the homozygous and compound heterozygous mutant mice was confirmed quantitatively. (G) Motor nerve count averages were as follows: 543±24 for wild type mice (n = 8), 533±20 for C201R/+ mice (n = 7), 461±36 for Nmf249/+ mice (n = 5), 246±58 for C201R/C201R; TgD mice (n = 9), 233±18 for C201R/Nmf249 mice (n = 4), and 236±42 for C201R/Nmf249; TgD mice (n = 5). C201R/C201R; TgD, C201R/Nmf249, and C201R/Nmf249; TgD mice had fewer motor axons than Nmf249/+ mice (p<0.001 for all three genotypes). There were no significant differences among the three genotypes that are homozygous or compound heterozygous for Gars mutation. (H) Body weights were as follows: wild type = 10.7±1.4 g (n = 7), C201R/+ = 9.7±1.5 (n = 8), Nmf249/+ = 8.8±2.3 (n = 5), C201R/C201R; TgD = 3.8±0.7 (n = 9), C201R/Nmf249  =  7.1±1.4 (n = 4), and C201R/Nmf249; TgD = 5.6±2.5 (n = 5). C201R/C201R; TgD, and C201R/Nmf249; TgD mice weighed less than C201R/+ mice (p<0.001, p<0.01, respectively). C201R/Nmf249 mice weighed more than C201R/C201R; TgD mice (p<0.05) and were not significantly different from C201R/+ mice. Plantaris muscles were removed from P17 pups for neuromuscular junction imaging (I-O). Green pre-synaptic staining is from NF and SV2 antibodies, red post-synaptic staining is α-bungarotoxin. Nearly mature NMJs were seen in wild-type animals (I). As shown in Figure 3 and Figure 7, C201R/+ (J) and Nmf249/+ (K) have abnormal NMJ morphology with areas of denervation in some NMJs and others that are fully denervated. More severe signs of denervation were seen in C201R/C201R; TgD, C201R/Nmf249, and C201R/Nmf249; TgD mice. There was a wide range of dysmorphic NMJ pathology in C201R/C201R; TgD mice. Some sections had mostly denervated NMJs (arrows) (L), while other areas had some partially innervated junctions (arrowheads) (M). A similar mix of morphologies was seen with C201R/Nmf249 mice (N) and C201R/Nmf249; TgD mice (O). The scale of I-O is indicated by the 28 µm bar in O. Scoring of neuromuscular junctions was performed to quantitatively compare the proportion of NMJs that were fully innervated (green), partially innervated (yellow), and denervated (red) (P). In Nmf249/+ mice, 14.2±3.0% of NMJs were denervated. Significantly more junctions were fully denervated in C201R/C201R; TgD mice where 56.2±7.5% of NMJs were denervated, and in and C201R/Nmf249; TgD mice where 38.2±8.0% of NMJs were denervated (p<0.001 for both comparisons). Compared to C201R/+ mice, where 2.8±2.4% of NMJs were denervated, significantly more were denervated in C201R/Nmf249 mice where 26.8±8.7% of junctions were denervated, but this was not statistically different from the Nmf249/+ mice. Compared to both C201R/Nmf249 and C201R/Nmf249; TgD mice, C201R/C201R; TgD mice had a greater proportion of denervated NMJs (p<0.05, p<0.001, respectively).