The integrated stress response contributes to tRNA synthetase-associated peripheral neuropathy

Abstract

Dominant mutations in ubiquitously expressed transfer RNA (tRNA) synthetase genes cause axonal peripheral neuropathy, accounting for at least six forms of Charcot-Marie-Tooth (CMT) disease. Genetic evidence in mouse and Drosophila models suggests a gain-of-function mechanism. In this study, we used in vivo, cell type–specific transcriptional and translational profiling to show that mutant tRNA synthetases activate the integrated stress response (ISR) through the sensor kinase GCN2 (general control nonderepressible 2). The chronic activation of the ISR contributed to the pathophysiology, and genetic deletion or pharmacological inhibition of Gcn2 alleviated the peripheral neuropathy. The activation of GCN2 suggests that the aberrant activity of the mutant tRNA synthetases is still related to translation and that inhibiting GCN2 or the ISR may represent a therapeutic strategy in CMT.

Figure 1.. Differential gene expression in Gars/CMT2D mice.

(A) 1976 up- and 126 down-regulated ribosome associated mRNAs were found in Gars^C201R/+ motor neurons compared to Gars^+/+. (B) 603 up- and 233 down-regulated ribosome-associated mRNAs were found in Gars^P278KY/+ motor neurons compared to Gars^+/+. (C) 174 upregulated genes were found in common between Gars^C201R/+ and Gars^P278KY/+, six genes relevant to subsequent studies are noted. (D) The intersections of upregulated gene lists from whole spinal cord RNAseq in three different dominant alleles of Gars. (E) The upregulated genes shared in RiboTagging data overlap the union upregulated gene list of whole spinal cord RNAseq. (F) Ingenuity Pathway Analysis of upregulated ribosome-associated mRNAs with log FC≥1.5 and FDR<.05 in Gars^P278KY/+ motor neurons. (G) Activation of the ISR was confirmed with immunolabeling of phospho-eIF2α, which was found specifically in ventral horn motor neurons (arrowheads) in Gars^P278KY/+ mutant spinal cord (right hand panel). (H) The ratio of phospho-eIF2α labelling intensity to that of the motor neuron marker ChAT is increased in all three alleles of Gars, and in Gars^P278KY/+ compared to Gars^C201R/+, points represent individual motor neurons, n=3 mice/genotype. Volcano plots show transcripts with log FC>|1.5|; P<.05. N = 4–6 mice (in A-C) or 6 mice (3 males, 3 females in D) per genotype at 8-weeks-of-age.

Figure 2:. ATF4 target gene expression is restricted to alpha motor neurons within mutant Gars spinal cord.

(A) Motor neurons in the ventral horn of the spinal cord include alpha- (Spp1-positive) and gamma- (Spp1-negative) populations, but only Spp1-positive cells express Fgf21 in mutant mice. (B) RNAscope in situ hybridization (ISH) showed upregulation of the ATF4 target gene Fgf21 in a subset of ventral horn motor neurons in Gars^ΔETAQ/+ mice. (C) ISH in spinal cord of Gars^P278KY/+ mice. Chat (red), Gdf15 (green), Fgf21 (yellow). Some Chat-positive mutant Gars motor neurons do not show expression of disease-related transcripts (arrows) (D) Quantification of background-subtracted ISH fluorescence intensity. ***p<.001, ****p<.0001 (two-tailed Student’s T-test, N=4 mice per genotype at 8 weeks of age.) (E) Percentage of Gars^P278KY/+ motor neurons that expressed the 5 disease-associated transcripts. (F) RNAscope ISH in spinal cord with Chat and Spp1 revealed the same percentage of alpha- and gamma motor neurons in mutant mice. (G, H) Only Spp1-positive alpha motor neurons in mutant spinal cords express Fgf21. Arrows in (H) point to cells labeled with Chat that do not express Spp1 or Fgf21. N=3 mice per genotype at 8 weeks of age in F-H.

Figure 3.. Deletion of Gcn2 alleviates the Gars^P278KY/+ phenotype.

(A) Gars^P278KY/+ mice have a phenotype by 2-weeks-of-age, and actively lose peripheral axons until ~8-weeks-of-age, after which the phenotype progresses very slowly. (B) Body weight of male Gcn2^KO/KO;Gars^P278KY/+ mice is significantly increased over Gcn2^{+/+ or +/KO};Gars^P278KY/+ (one-way ANOVA, Gcn2^any denotes +/+, +/KO, or KO/KO genotype). (C) Gcn2^KO/KO;Gars^P278KY/+ mice have increased latency to fall from an inverted wire grid (one-way ANOVA). The test is stopped after 60 seconds and the mean of 3 trials is reported for each day. Body weight and wire hang analyses performed with 18–22 mice per grouped genotype. (D) Motor nerve conduction velocity of the sciatic nerve is increased in Gcn2^KO/KO;Gars^P278KY/+ mice compared to Gcn2^{+/+ or +/−};Gars^P278KY/+, and no longer different from Gars^+/+ regardless of Gcn2 genotype (one-way ANOVA). (E) The average number of motor axons in the femoral nerve of 16-week-old Gcn2^KO/KO;Gars^P278KY/+ mice is greater compared to Gcn2^{+/+ or +/−};Gars^P278KY/+. Gcn2^KO/KO;Gars^P278KY/+ axon counts are no longer different from Gars^+/+ mice regardless of Gcn2 genotype (one-way ANOVA). (F) The motor branch of the femoral nerve in Gcn2^KO/KO;Gars^P278KY/+ mice is intermediate in size compared to Gars^+/+ and Gcn2^{+/+ or +/−};Gars^P278KY/+. (G) Gcn2^+/+;Gars^+/+motor neurons show no expression of ATF4 target genes Gdf15 (green) or Fgf21 (yellow), while Gcn2^+/+;Gars^P278KY/+ motor neurons show robust upregulation. ATF4 target gene expression is eliminated in Gcn2^KO/KO;Gars^P278KY/+ motor neurons. Data points in D and E represent individual mice of both sexes.

Figure 4.. Pharmacological inhibition of GCN2 is efficacious.

(A) Study design: Cohorts of Gars^ΔETAQ/+ or littermate control (Gars^+/+) mice were treated with GCN2iB or vehicle starting at 2-weeks-of-age for three weeks, 8–9 mice per sex per genotype per treatment group were enrolled. (B) GCN2iB improved body weight in Gars^ΔETAQ/+ male mice. (C) GCN2iB improved strength and endurance in the wire hang test. (D) GCN2iB improved sciatic motor nerve conduction velocity in Gars^ΔETAQ/+ mice, although values were still reduced compared to controls. (E) Compound muscle action potential amplitude was not significantly improved over vehicle-treated mutant mice. (F) The decrement in CMAP amplitude following bouts of repetitive nerve stimulation was alleviated in Gars^ΔETAQ/+ mice with GCN2iB treatment. Data from male and female mice is pooled in C-F, although improvement was generally greater in males. (G) Expression of the ATF4 target gene B4galnt2 was reduced with GCN2iB treatment; effects were greater in males. Statistical analysis: B, C, F, G were tested with 2-way ANOVA and correction for multiple comparisons, D, E were tested with 1-way ANOVA and correction for multiple comparisons. Asterisks in B, C, F indicate significance between GCN2iB-treated Gars^ΔETAQ/+ and vehicle-treated Gars^+/+ mice, and the pairwise comparisons noted in D, E and G.