Glycyl-tRNA sequestration is a unifying mechanism underlying GARS1-associated peripheral neuropathy

Abstract

Dominantly inherited mutations in eight cytosolic aminoacyl-tRNA synthetase genes cause hereditary motor and sensory neuropathy, characterized by degeneration of peripheral motor and sensory axons. We previously identified a pathogenic gain-of-toxic function mechanism underlying peripheral neuropathy (PN) caused by heterozygous mutations in the GARS1 gene, encoding glycyl-tRNA synthetase (GlyRS). Specifically, PN-mutant GlyRS variants sequester tRNA^Gly, which depletes the cellular tRNA^Gly pool, leading to insufficient glycyl-tRNA^Gly available to the ribosome and consequently ribosome stalling at glycine codons. Given that GlyRS functions as a homodimer, a subset of PN-GlyRS mutations might alternatively cause peripheral neuropathy through a dominant negative loss-of-function mechanism. To explore this possibility, we here generated three novel PN-GlyRS Drosophila models expressing human PN-GlyRS (hGlyRS) variants that do not alter the overall GlyRS protein charge (S211F and H418R) or the single reported PN-GlyRS variant that renders the GlyRS protein charge more negative (K456Q). High-level expression of hGlyRS-K456Q did not induce peripheral neuropathy and the K456Q variant does not affect aminoacylation activity, suggesting that K456Q is not a pathogenic mutation. Expression of hGlyRS-S211F or hGlyRS-H418R in Drosophila did induce peripheral neuropathy and de novo protein synthesis defects. Genetic and biochemical evidence indicates that these phenotypes were attributable to tRNA^Gly sequestration rather than a dominant negative mechanism. Our data identify tRNA^Gly sequestration as a unifying pathogenic mechanism underlying PN-GlyRS. Thus, elevating tRNA^Gly levels may constitute a therapeutic approach for all PN-GlyRS patients, irrespective of their disease-causing mutation.

Graphical Abstract

Open in new tabDownload slide

Introduction

Dominantly inherited mutations in eight distinct cytosolic aminoacyl-tRNA synthetase (aaRS) genes give rise to Charcot-Marie-Tooth (CMT) peripheral neuropathy [1]. AaRSs are the enzymes that covalently link amino acids to the 3′ end of their cognate tRNA, a reaction referred to as aminoacylation, representing the essential first step of protein biosynthesis [2–4]. Heterozygous mutations in GARS1, the gene encoding glycyl-tRNA synthetase (GlyRS), give rise to an axonal form of CMT, type 2D (CMT2D) [1, 5]. CMT is characterized by selective degeneration of peripheral motor and sensory axons, leading to progressive muscle weakness and wasting, steppage gait, sensory dysfunction, and often foot deformities [6, 7]. In addition to CMT2D, heterozygous GARS1 mutations have been linked to related peripheral neuropathies characterized by selective motor axonal degeneration and lacking sensory involvement, including distal hereditary motor neuropathy type V (dHMN-V) [1, 5] and infantile spinal muscular atrophy (iSMA) [8, 9]. Here, we refer to these GARS1-associated peripheral neuropathies as PN-GlyRS.

We have previously shown that six CMT-causing mutations in GlyRS or tyrosyl-tRNA synthetase (TyrRS) all inhibit protein synthesis in Drosophila motor and sensory neurons, uncovering the inhibition of global protein synthesis as a unifying pathogenic mechanism underlying CMT-aaRS [10]. Consistent with this notion, inhibition of protein synthesis was observed in motor neurons of two PN-GlyRS mouse models [11]. More recently, we identified the detailed molecular mechanism underlying PN-GlyRS: PN-GlyRS variants are still able to bind tRNA^Gly but fail to release it or only release at a very slow pace. This sequestration of tRNA^Gly by PN-GlyRS variants depletes the cellular pool of free tRNA^Gly available for aminoacylation by the wild type (WT) GlyRS protein in heterozygous PN-GlyRS patients. This results in insufficient production and supply of aminoacylated glycyl-tRNA^Gly to the ribosome, which triggers ribosome stalling at glycine codons [12]. Ribosome stalling in turn induces the integrated stress response (ISR) through activation of the eIF2α kinase GCN2 [13–15], which in PN-GlyRS and CMT-TyrRS mouse models occurs selectively in motor and sensory neurons, the affected cell types in CMT [11]. Consistent with this mechanism, transgenic overexpression of tRNA^Gly replenishes the cellular tRNA^Gly pool and rescues defective protein synthesis, prevents ISR activation, and rescues peripheral neuropathy phenotypes in Drosophila and mouse PN-GlyRS models [12].

Here, we sought to evaluate whether all PN-GlyRS mutations cause disease through tRNA^Gly sequestration, what can be classified as a gain-of-toxic-function mechanism, or alternatively, whether some PN-GlyRS mutations cause disease through partial loss of aminoacylation activity. Indeed, similar to tRNA^Gly sequestration, a substantial reduction of tRNA^Gly aminoacylation activity may also lead to insufficient cellular levels of aminoacylated glycyl-tRNA^Gly and consequently ribosome stalling at glycine codons and GCN2-mediated ISR activation (Fig. 1A). Thus, although the primary molecular defect would be different, namely tRNA^Gly sequestration versus reduced aminoacylation activity, the downstream molecular pathway that triggers peripheral neuropathy (ribosome stalling and ISR activation) would be the same. Nevertheless, from a therapeutic perspective, the question whether some of the PN-associated GlyRS mutations may act through partial loss of aminoacylation activity is of key importance. Indeed, for tRNA^Gly sequestration, elevating cellular tRNA^Gly levels would constitute the optimal therapeutic approach, whereas for mutations resulting in partial loss of GlyRS aminoacylation activity, increasing the level of WT GlyRS protein would be the appropriate therapeutic strategy (Fig. 1A).

Human and mouse genetic evidence provide strong arguments against haplo-insufficiency as the mechanism underlying PN-GlyRS [1, 16, 17]. However, because GlyRS functions as a homodimer, in heterozygous PN-GlyRS patients ∼25% of GlyRS dimers will be WT homodimers, ∼25% mutant homodimers, and ∼50% WT-mutant heterodimers, assuming an equal GARS1 allelic expression and unaffected GlyRS dimerization. For PN-GlyRS mutations that result in a full loss of aminoacylation activity, this raises the possibility that a dominant negative mechanism may lead to an overall reduction of tRNA^Gly aminoacylation activity by 75%, because only WT homodimers would be able to aminoacylate tRNA^Gly [18].

Here, we evaluated the hypothesis that a subset of PN-GlyRS mutations may induce peripheral neuropathy through a dominant negative effect on tRNA^Gly aminoacylation. The plausibility of a dominant negative mechanism underlying some PN-GlyRS mutations is suggested by our finding that knock-down of endogenous Drosophila GlyRS (dGlyRS) in motor or sensory neurons is sufficient to induce peripheral neuropathy-related phenotypes. Furthermore, genetic experiments using three previously characterized PN-GlyRS Drosophila models [10] reveal that for two mutations that result in severely reduced or complete loss of aminoacylation activity (G240R and G526R), a dominant negative effect on aminoacylation activity might contribute to peripheral neuropathy, in addition to tRNA^Gly sequestration.

However, newly generated Drosophila PN-GlyRS models with a priori lower likelihood to sequester tRNA^Gly strongly argue for tRNA^Gly sequestration as a unifying mechanism underlying PN-GlyRS, and against a dominant negative mechanism. Indeed, expression of two PN-GlyRS variants that do not alter overall GlyRS protein charge (S211F and H418R) in Drosophila induced peripheral neuropathy phenotypes and inhibited global protein synthesis, highly reminiscent of our previously characterized PN-GlyRS models [10]. Using Drosophila genetic experiments and biochemical assays, we show that tRNA^Gly sequestration by these newly characterized hGlyRS variants triggers peripheral neuropathy phenotypes. Finally, even high-level expression of the only reported PN-GlyRS variant that removes a positive charge from the GlyRS protein (K456Q) does not induce peripheral neuropathy phenotypes, and hGlyRS-K456Q displays WT-like aminoacylation activity, suggesting that K456Q may not be a pathogenic mutation. Together, our results indicate that tRNA^Gly sequestration is a unifying pathogenic mechanism underlying PN-GlyRS. This implies that elevating tRNA^Gly levels may constitute a therapeutic approach for all PN-GlyRS patients, irrespective of their specific mutation.

Materials and methods

Drosophila genetics

Flies were housed in a temperature-controlled incubator with 12:12 h on/off light cycle at 25°C. OK371-GAL4 (Bl 26160), ppk-GAL4 (Bl 32079), actin5C-GAL4 (Bl 4414), and Sca-GAL4 (Bl 6479) were obtained from the Bloomington Drosophila Stock Center (Bloomington, IN, USA). The UAS-dGlyRS line [19] was kindly provided by L. Luo. The UAS-hGlyRS-WT, UAS-hGlyRS-E71G, UAS-hGlyRS-G240R, and UAS-hGlyRS-G526R lines were previously described [10]. For imaging of Class IV multidendritic sensory neurons, ppk-GAL4 was recombined with ppk-CD4-tdGFP [20]. dGlyRS-RNAi-1 corresponds to NIG 6778R-1, and dGlyRS-RNAi-2 corresponds to VDRC KK 106748. These transgenic RNAi lines were obtained from the National Institute of Genetics (NIG, Japan) and the Vienna Drosophila Resource Center (VDRC, Vienna, Austria), respectively. tRNA^Gly-GCC BAC and tRNA^Gly-GCC scramble transgenic lines were previously described [12]. For ubiquitous expression of the UAS-hGlyRS transgenes from the adult stage onwards, the GAL80 target system was used [21]. Fly crosses were cultured at 18°C and 1-day-old adult flies carrying the tub-GAL4, tub-GAL80^ts, and UAS-hGlyRS transgenes were shifted to 29°C to induce transgene expression.

Generation of hGlyRS transgenic lines

For generation of UAS-hGlyRS-S211F, UAS-hGlyRS-H418R, and UAS-hGlyRS-K456Q transgenic lines, full-length human GARS1 cDNA was purchased from Origene (Rockville, MD, USA) and subcloned into the pGEM^®-T vector. Mutations were introduced via site-directed mutagenesis into the human GARS1 cDNA using the Q5^® Site-Directed Mutagenesis Kit (New England Biolabs, Ipswich, MA, USA). Sequences of primers used for cloning and mutagenesis of hGARS1 cDNA are listed in Supplementary Table S1. Next, hGARS1 cDNA was cloned into the pUAST-AttB transformation vector. Constructs were sequence verified and injected into Drosophila embryos by Genetivision (Houston, TX, USA) for site-specific PhiC31-mediated insertion in VK37 or VK31 landing sites for second or third chromosome insertions, respectively [22, 23]. Similar to the previously generated UAS-hGlyRS transgenes [10], the newly generated UAS-hGlyRS transgenes enable expression of either the cytosolic or mitochondrial hGlyRS isoform, depending on the translation start site used.

Adult offspring frequency

For the determination of adult offspring frequency, the number of flies eclosing was counted for each genotype. A strong ubiquitous (actin5C-GAL4) driver was used to express the transgenes and flies were kept at a 12 h on/off light cycle at 25°C. Eclosed flies were counted daily. At least 77 flies per genotype were counted.

Scoring of scutellar bristle phenotypes

Bristle phenotypes were induced by expression of dGlyRS-RNAi-2 in sensory organ precursor cells using the Sca-GAL4 driver, and we evaluated whether these bristle phenotypes could be rescued by simultaneous expression of either hGlyRS-WT or hGlyRS-K456Q. For quantification of bristle phenotypes, anterior and posterior scutellar bristles were examined and the number of normal, small or missing bristles was scored and the relative percentages were calculated.

Analysis of protein synthesis by in vivo FUNCAT

For FUNCAT analysis of de novo protein synthesis [10, 24, 25], 24 h egg collections were performed and animals were raised on Jazz-Mix Drosophila medium (Fisher Scientific) at 25°C. Seventy-two hours after egg laying, larvae were transferred to 4 mM azidonorleucine (ANL)-containing medium for 48 h (Iris Biotech, HAA1625). Larval central nervous system (CNS) was dissected in ice-cold HL3 solution and fixed in 4% paraformaldehyde (PFA) for 30 min at room temperature (RT). After fixation, the CNSs were washed 3 × 10 min with PBST (1 x PBS pH 7.2 containing 0.2% Triton X-100) and 3 × 10 min with PBS pH 7.8. Metabolically labeled proteins were tagged by ‘click chemistry’ (Copper-Catalyzed (3 + 2)-Azide-Alkyne-Cycloaddition Chemistry (CuAAC)) using the fluorescent TAMRA tag. The FUNCAT reaction mix was assembled in a defined sequence of steps. Triazole ligand (200 μM, Sigma-Aldrich, 678937), TAMRA-alkyne tag (0.2 μM, Invitrogen, T10183), TCEP solution (400 μM, Sigma-Aldrich, C4706) and CuSO4 solution (200 μM, Sigma-Aldrich, C8027) were added to PBS pH 7.8. After each addition the solution was mixed thoroughly for 10 sec and at the end for 30 sec using a high-speed vortex. Larval CNSs were incubated with 500 μL of FUNCAT reaction mix overnight at 4°C on a rotating platform. The next day, the CNSs were washed 3 times with PBS-Tween (1x PBS pH 7.2 containing 1% Tween-20) and three times with PBST for 10 min. Finally, larval CNSs were mounted in VectaShield mounting medium (Biozol, VEC-H-1000-CE) and stored at 4°C until imaging using a Leica SP8 laser scanning confocal microscope. For image acquisition, identical confocal settings were used for all samples of a given experiment. Fluorescence intensities were quantified using ImageJ/FIJI software (National Institutes of Health). Mean intensity of all cells within one motor neuron cluster from each ventral nerve cord were used as single data points for statistical analysis.

Analysis of protein synthesis by in vivo THRONCAT

For THRONCAT analysis of de novo protein synthesis [26], 24 h egg collections were performed and larvae were raised on Jazz-Mix Drosophila medium (Fisher Scientific) at 25°C. Seventy-two hours after egg-laying, larvae were transferred to medium containing 4 mM ß-ethynylserine (ßES) for 48 h. Larval CNS was dissected in ice-cold HL3 solution and fixed in 4% PFA for 30 min at RT. After fixation, the CNSs were washed 3 × 10 min with PBST (1x PBS pH 7.2 containing 0.2% Triton X-100) and 3 × 15 min with PBS pH 7.2. CuAAC ‘click chemistry’ was performed using the fluorescent TAMRA tag and the reaction mix was assembled as follows: Tris-hydroxypropyltriazolylmethylamine (THPTA, Sigma-Aldrich, 762 342) ligand (200 μM), TAMRA-azide tag (0.2 μM, Invitrogen, T10182), CuSO4 solution (4 mM), and sodium ascorbate solution (40 mM, Sigma-Aldrich, A7631) were added to PBS pH 7.2. After each addition the solution was mixed thoroughly for 10 s and at the end for 30 s using a high-speed vortex. CNSs were incubated with 500 μL of the reaction mix overnight at 4°C on a rotating platform. The next day, CNSs were washed three times with PBS-Tween (1 x PBS pH 7.2 containing 1% Tween-20) and three times with PBST for 15 min. Next, samples were blocked for 1 h in blocking buffer (10% goat serum in PBST) at RT before incubating with antibodies against hGlyRS (rabbit polyclonal; 1:200; Proteintech, 15831-AP) in blocking buffer at 4 °C overnight on a rotating platform. The next day, CNSs were washed three times with PBST for 15 min before incubation with Alexa Fluor™ 647-conjugated goat-anti-rabbit IgG antibodies (1:500; Thermo Fisher, A27040) in blocking buffer for 3 h at RT. After three washes with PBST for 15 min, CNSs were mounted in VectaShield mounting medium (Biozol, VEC-H-1000-CE) and stored at 4°C until imaging using a Leica SP8 laser scanning confocal microscope. For image acquisition, identical confocal settings were used for all samples of a given experiment. Fluorescence intensities were quantified using ImageJ/FIJI software (National Institutes of Health). Mean intensity of all cells within one motor neuron cluster from each ventral nerve cord was determined, as well as the mean intensity of five cells in close proximity of the motor neuron cluster that do not express hGlyRS transgenes. The mean intensity of the motor neuron cluster was normalized to the mean intensity of the five surrounding cells and used as single data point for statistical analysis.

Analysis of larval muscle innervation

To analyze synapse length on muscle 8, third instar larvae were dissected in HL3 buffer and fixed in Bouin's for 3 min at RT. After permeabilization and blocking (10% goat serum in PBST), immunostaining was performed with anti-Discs large 1 (mouse monoclonal; 1:200; DSHB 4F3), to visualize the postsynaptic side of the neuromuscular synapse. Images were taken of muscle 8 in abdominal segment 5 using a Leica SP8 laser scanning confocal microscope with 20x Plan-Apochromat objective (0.8 NA). Confocal sections were taken every 2 μm covering the entire 3D volume of the NMJ. Using Fiji software, maximum intensity projections of all images in a z-stack were used to measure the synapse length, using the ‘segmented line’ tool to draw a line through the middle of every branch of the NMJ. When multiple NMJ branches were present, the sum of the length of individual branches was used as synapse length. The mean of the length of the left and right NMJ on muscle 8 was used as one data point for statistical analysis. For innervation status of muscle 24, the presence of the NMJ on muscle 24 in abdominal segment 6 on both sides of the larvae was assessed and counted as present (one or both sides) or absent (neither side).

Automated negative geotaxis assay to analyze motor performance

To evaluate motor performance, newly enclosed male or virgin female flies were collected and kept in groups of ten flies per vial until measurements. Flies were not anesthetized 24 h before measurements. Seven-day-old flies were flipped to test tubes and their motor performance was analyzed in a rapid iterative negative geotaxis assay (RING-assay) as previously described [10]. Five or six iterative movies were recorded using a Nikon Z50 camera. A python-based script was used to analyze every movie by tracking the position of each fly in each frame and tracking all flies, creating tracks and allowing to obtain the climbing speed (mm/s) over the course of 5 s. The speed of all flies within one tube was averaged. The three movies with the highest number of moving flies detected were then averaged and used as one data point for the respective genotype. Movies with less than three moving flies were excluded from analysis.

Analysis of dendritic morphology of class IV multidendritic sensory neurons

Imaging of class IV multidendritic sensory neuron morphology was performed on living third instar larvae. Microscopy slides were prepared with two coverslips glued on as spacers with conventional superglue. Larvae were thoroughly washed in tap water and single chilled larvae were immobilized under a coverslip, which itself was glued onto the spacing coverslips to produce equal spacing and pressure on the larvae. Only dorsal class IV da neurons of larval segment A1 were imaged using a Leica SP8 laser scanning confocal microscope. For quantification, maximum intensity projections were processed with Adobe Photoshop, rotated and digitally placed under a transparent grid containing a crosshair. The crosshair was used to fit the grid onto the cell body of the class IV da neuron in order to divide the dendritic tile into four quadrants. The posterolateral quadrant of the imaged da neuron was chosen for quantification of dendritic coverage, as least imaging artifacts and interference of denticle bands were encountered in this area. For quantification within the posterolateral quadrant, a sector of 200 μm to the lateral and 280 μm to the posterior was defined, equaling 56 000 μm² divided into 560 squares of 8 × 8 pixels each. Presence versus absence of a piece of dendrite in the defined boxes was scored and calculated as measure of relative dendritic coverage.

Western blot analysis

For western blot analysis, tub-GAL4, tub-GAL80^ts was used to express the UAS-hGlyRS transgenes from the adult stage onwards. Flies were raised at 18°C and 1–3-day-old adult flies carrying the tub-GAL4, tub-GAL80^ts and UAS-hGlyRS transgenes were shifted to 29°C to induce transgene expression. After five days of transgene expression, flies were snap frozen in liquid nitrogen and fly heads were collected. Protein extracts were prepared from 30 fly heads per sample by homogenization in extraction buffer (50 mM Tris/HCl pH 7.4, 150 mM KCl, 0.25 M sucrose, 5 mM MgCl2, and 0.5% Triton X-100 containing 1 U complete mini protease inhibitor cocktail (Roche)). Next, samples were centrifugated at 4°C for 30 min at 13 000 x g to pellet cell debris. Samples were diluted to a concentration of 100 μg/mL. 20 μL per sample was separated on a 10% SDS-PAGE gel and transferred to a polyvinylidene difluoride (PVDF) membrane using the Trans-Blot Turbo Mini 0.2 μm PVDF Transfer Pack (Bio-Rad) for 10 min at 25 V, 2.5 A on a Trans-Blot^® Turbo Transfer System (Bio-Rad). The blotted membranes were blocked in 5% milk in TBS-T (20 mM Tris Base pH 7.6, 140 mM NaCl, 0.1% Tween-20) for 1 h at RT before overnight incubation at 4°C with antibodies against hGlyRS (rabbit polyclonal; 1:1000; Proteintech, 15831-AP) or histone H3 (rabbit polyclonal; 1:10,000; Abcam, Ab1791). Membranes were subsequently incubated with anti-rabbit IgG coupled to horseradish peroxidase (goat polyclonal; 1:10,000; Invitrogen, 31460) for 1 h at RT to visualize proteins. Membranes were developed with enhanced chemiluminescence (ECL™ Prime Western Blotting System, Cytiva, RPN2232) and scanned on a Bio-Rad Universal Hood III Imager (Bio-Rad). Band intensities were quantified using ImageJ/FIJI (National Institutes of Health) by drawing a rectangle around the specific band, obtaining the peak pattern of the rectangle and measuring the area under the peak that represents the specific band.

Isolation of total RNA and quantitative real-time PCR analysis

Quantification of dGlyRS transcript levels

dGlyRS-RNAi or dGlyRS transgenes were expressed with a ubiquitous driver (actin5C-GAL4). Three (dGlyRS-RNAi) or five (dGlyRS) third instar larvae were pooled per sample and total RNA was extracted using Trizol reagent (Life Technologies) and Direct-zol RNA Microprep kit (R2062, Zymol) according to the manufacturer's instructions. Samples were DNase treated and 2500 ng RNA was subsequently used for reverse transcription to obtain cDNA, using the SensiFAST cDNA synthesis kit (BioLine). cDNA samples were then used as templates for real-time PCR assays performed on a Bio-Rad CFX system using the SensiFAST SYBR No-ROX Kit (BioLine). Beta-actin (dGlyRS-RNAi) or alpha-tubulin (dGlyRS) mRNA was used as reference gene for normalization. Experiments included no-reverse transcriptase and no-template controls for each primer pair. Primer sequences can be found in Supplementary Table S2. Data were analyzed using the ΔΔCt calculation method.

Quantification of hGlyRS transcript levels

hGlyRS transgenes were expressed with a ubiquitous driver from the adult stage onwards (tub-GAL4, tub-GAL80^ts). Heads of 30 1–3-day-old adult flies were collected and snap frozen in liquid nitrogen. Heads were ground to a powder with a pestle and total RNA was extracted using Trizol reagent (Life Technologies) and Direct-zol RNA Microprep kit (R2062, Zymol) according to the manufacturer's instructions. Around 500 ng RNA was used for cDNA synthesis using the SensiFAST cDNA synthesis kit (BioLine). cDNA samples were then used as templates for real-time PCR assays performed on a Bio-Rad CFX system using the SensiFAST SYBR No-ROX Kit (BioLine). Alpha-tubulin mRNA was used as a reference gene for normalization. Experiments included no-reverse transcriptase and no-template controls for each primer pair. Primer sequences can be found in Supplementary Table S2. Data were analyzed using the ΔΔCt calculation method.

In vitro aminoacylation assay

In vitro transcription of Drosophila tRNA^Gly-GCC and aminoacylation with glycine was performed as described [27]. Briefly, tRNA^Gly-GCC and the respective recombinant hGlyRS variant (WT or K456Q) were mixed in 10:1 molar ratio (10 μM:1μM) and incubated for 15 min at 37°C in 60 mM Tris-HCl pH 7.5, containing 10 mM KCl, 12.5 mM MgCl₂, 1 mM spermidine, 1.5 mM DTT, 1.5 mM ATP, 1 mM glycine and 0.5 μL SUPERase·In™ RNase Inhibitor (Invitrogen) in a total reaction volume of 20 μL. The reaction was terminated by dilution up to 300 μl with 0.3 M NaOAc pH 4.5, containing 10 mM EDTA. The tRNA was precipitated with 1 volume isopropanol and resuspended in 10 mM NaOAc pH 4.5 (note that the ester bond is stable in acidic pH).

To distinguish between aminoacylated and uncharged tRNA species, we used the splinted ligation reaction that generates amino acid-bridged chimeric RNA molecules to a 5′-phosphorimidazolide-activated oligoribonucleotide [27], thereby following the reaction scheme and using the same RNA:DNA oligonucleotides used to extend the 5′ and 3′ ends of the tRNA as described [28]. The oligonucleotide with 5′-phosphorimidazolide ligates preferably aminoacyl-tRNA entities with more than three orders of magnitude higher efficacy than the non-amonoacylated tRNAs. tRNA^Gly-GCC recovered after the aminoacylation reaction was incubated with the 5′-phosphorimidazolide oligonucleotide for 30 min at 25°C in 100 mM MES buffer pH 5.5, containing 2.5 mM MgCl₂, in the presence of 50 mM 1-(2-hydroxyethyl)imidazole (HEI) pH 6.5 catalyst, followed by an extended incubation at 4°C overnight. The different tRNA species were separated on a 10% denaturing acidic PAGE (8M urea and 0.1 M NaOAc pH 5) and visualized by SYBR^TM Gold Nucleic Acid Stain and quantified using ImageJ software. The percentage of the aminoacyl-tRNA was determined as the ratio to the total (aminoacyl-tRNA and non-aminoacyted tRNA) and data are represented in Fig. 3K as aminoacylation activity relative to hGlyRS-WT (100%).

Expression and purification of recombinant hGlyRS proteins

Two PN-GlyRS variants (S211F and H418R) were expressed with two tags (6xHis and SUMO tag) in E. coli Rosetta strain and purified as described previously for other PN-GlyRS variants [12]. Briefly, crude extracts were incubated with HisPur Ni-NTA resin (Thermo) for 30 min at 4°C, followed by multiple washing steps with 20 mM HEPES-NaOH pH 7.4, containing 500 mM NaCl and 10 mM imidazole, and two washing steps with imidazole concentration increased to 20 mM. The PN-GlyRS variants were dissolved from the Ni-NTA resin by cleavage after the SUMO tag with 0.5 mg/mL of ULP overnight at 4°C. The released PN-GlyRS variants in the supernatant were concentrated and the dimer and monomer forms were fractionated using size-exclusion chromatography on Superdex 200.

Determination of K_on and K_off values

The K_on values for the cognate tRNA^Gly-GCC were determined by monitoring the quenching of intrinsic PN-GlyRS tryptophan residues as described previously [12]. tRNA^Gly-GCC was generated via in vitro T7 transcription and purified via 10% polyacrylamide gel electrophoresis. Briefly, 750 nM of each PN-GlyRS variant was incubated with different tRNA concentrations, ranging from 0 to 1 μM, in 25 mM sodium acetate buffer pH 6.0 containing 10 mM MgCl₂, 5 mM DTT, in a final volume of 50 μL. The Trp emission was recorded at 350 nm (excitation 280 nm) at 37°C on a TECAN Spark plate-reader. Binding curves were recorded in multiple replicates and fitted to exponential decay functions and quantified with OriginPro.

For determination of the K_off values, to each of the above reactions containing different (0.1–1μM) tRNA^Gly concentrations, 1 mM glycine and 1 mM ATP were added. Fluorescence spectra (emission at 350 nm and excitation at 280 nm) were recorded over 5 min, step 1 s. Spectra were fitted in OriginPro to an exponential function.

Statistics

All data are presented as mean ± standard error of the mean (SEM) and differences were considered significant when P< 0.05. Animals or samples were assigned to the various experimental groups based on their genotype. For a given experiment, samples from the different experimental groups were processed in parallel and analyzed in random order.

Before analysis, a Robust regression and Outlier removal method (ROUT) was performed to detect statistical outliers. This nonlinear regression method fits a curve that is not influenced by outliers. Residuals of this fit were then analyzed by a test adapted from the False Discovery Rate approach, to identify any outliers. All data points that were considered outliers were excluded from further data analysis. Normality and homoscedasticity of all data sets was assessed by a Shapiro–Wilk and Brown-Forsythe (F-test for t-tests) test, respectively. Subsequent statistical tests were only performed if all assumptions were met.

For comparison of normally distributed data of more than two groups, one-way ANOVA followed by Sidak’s or Tukey’s multiple comparisons post hoc test was used, provided that the SDs were not significantly different between groups (equal variance, evaluated by Brown-Forsythe test and Bartlett’s test). For comparison of normally distributed data of more than two groups with unequal variance, Brown-Forsythe and Welch ANOVA followed by Dunnett’s T3 multiple comparisons post hoc test was used. For comparison of not normally distributed data of more than two groups, Kruskal–Wallis test with Dunn's multiple comparisons test was used. For comparison of normally distributed data from two groups with equal variance, unpaired t-test was used. Fisher's exact test was used to analyze offspring frequency data, innervation status of larval muscle 24, and scutellar bristle phenotpes.

Results

Endogenous dGlyRS knock-down inhibits protein synthesis and induces peripheral neuropathy phenotypes

To understand whether a loss-of-function scenario is a plausible mechanism for PN-GlyRS, we first evaluated whether a substantial reduction of dGlyRS levels, which consequently reduces tRNA^Gly aminoacylation, is sufficient to inhibit protein synthesis and induce peripheral neuropathy-related phenotypes. To do so, we used two independent transgenic RNAi lines that allow for knock-down of dGlyRS. To evaluate the efficiency of the dGlyRS-RNAi lines, we expressed the dGlyRS-RNAi transgenes ubiquitously (actin5C-GAL4), and quantified dGlyRS transcript levels in third instar larvae by quantitative real-time PCR. The two dGlyRS-RNAi lines reduced dGlyRS transcript levels by ∼69% and ∼65%, respectively (Fig. 1B). Due to the lack of a dGlyRS antibody, it was not possible to determine the corresponding endogenous dGlyRS protein levels. Nevertheless, the reduction in dGlyRS transcript levels may be representative for a dominant negative scenario, in which the aminoacylation-active dGlyRS pool is reduced by up to 75%.

We next evaluated the effect of dGlyRS knock-down in motor neurons (OK371-GAL4), because motor neurons are the affected cell type in all GARS1-associated peripheral neuropathies. In vivo cell-type-specific FUNCAT [24, 25] revealed a marked inhibition of nascent protein synthesis in motor neurons by ∼52% and ∼56% for the two independent dGlyRS-RNAi lines (Fig. 1C). These data suggest that dGlyRS knock-down substantially inhibits tRNA^Gly aminoacylation, leading to depletion of cellular glycyl-tRNA^Gly levels and consequently inhibiting protein synthesis. To determine whether dGlyRS knock-down in motor neurons can trigger peripheral neuropathy-like phenotypes, we evaluated motor performance of 7-day-old adult flies using an automated negative geotaxis assay [10]. This assay revealed that both dGlyRS-RNAi lines induced a significant reduction in climbing speed (Fig. 1D). Finally, we evaluated the effect of dGlyRS knock-down on sensory neuron morphology, given that sensory neurons are the other affected cell type in GARS1-associated CMT (CMT2D). Selective knockdown of dGlyRS in class IV multidendritic sensory neurons in the body wall of third instar larvae (ppk-GAL4) significantly reduced their dendritic coverage for both dGlyRS-RNAi lines (Fig. 1E and F). Together, these data indicate that a substantial reduction of dGlyRS levels is sufficient to inhibit global protein synthesis and induce peripheral neuropathy-related phenotypes.

Effect of dGlyRS overexpression on peripheral neuropathy phenotypes and protein synthesis defects in PN-GlyRS Drosophila models

We previously generated and characterized PN-GlyRS Drosophila models that allow transgenic expression of three human PN-GlyRS mutant proteins: E71G, G240R, and G526R [10]. While we previously showed that each of these three hGlyRS variants sequester tRNA^Gly [12], we sought to explore whether a dominant negative effect may constitute an additional pathogenic mechanism. This is particularly relevant for the G240R and G526R mutations, which severely reduce tRNA^Gly aminoacylation activity [10, 29–31]. In contrast, hGlyRS-E71G retains aminoacylation activity [10, 29, 30] (Supplementary Table S3). In case of a dominant negative scenario, increasing WT dGlyRS protein levels should increase the fraction and number of WT dGlyRS dimers and therefore would be expected to rescue the PN-GlyRS phenotypes. We therefore evaluated the effect of transgenic dGlyRS overexpression on inhibition of protein synthesis and peripheral neuropathy phenotypes in our PN-GlyRS Drosophila models. Ubiquitous (actin5C-GAL4) expression of the dGlyRS transgene in third instar larvae resulted in a 12-fold increase in dGlyRS mRNA levels (Fig. 2A), confirming robust transgene expression. As expected, in vivo cell-type specific FUNCAT revealed inhibition of protein synthesis induced by expression of hGlyRS E71G, G240R or G526R variants in third instar larval motor neurons (Fig. 2B and C). Simultaneous overexpression of dGlyRS did not modify the protein synthesis defect induced by hGlyRS-E71G, but partially rescued the inhibition of translation induced by hGlyRS G240R and G526R (Fig. 2B and C). Similarly, the adult motor performance defect induced by hGlyRS-E71G expression in motor neurons was not altered by simultaneous dGlyRS overexpression, while adult motor deficits induced by hGlyRS-G240R and hGlyRS-G526R expression were partially rescued by dGlyRS co-overexpression (Fig. 2D).

Next, we investigated third instar larval neuromuscular junction (NMJ) phenotypes induced by expression of hGlyRS-G240R and hGlyRS-G526R in motor neurons [10]. Expression of hGlyRS-G240R or hGlyRS-G526R in motor neurons frequently resulted in denervation of larval muscle 24, and this denervation phenotype was mitigated by simultaneous dGlyRS overexpression. This effect reached statistical significance for hGlyRS-G240R but not for hGlyRS-G526R (Fig. 2E). Furthermore, the neuromuscular synapse length on larval muscle 8 was significantly reduced by expression of hGlyRS-G240R and hGlyRS-G526R in motor neurons. This defect was slightly but significantly improved by simultaneous dGlyRS overexpression for the hGlyRS-G526R variant but not for hGlyRS-G240R (Fig. 2F and G). Finally, selective expression of each of the three hGlyRS mutants in class IV multidendritic sensory neurons significantly reduced their dendritic coverage. Simultaneous dGlyRS overexpression did not rescue this sensory neuron morphology defect. Rather, for the E71G and G526R mutants, dGlyRS co-overexpression slightly but significantly enhanced the sensory neuron morphology defect (Fig. 2H and I). Together, these data indicate that the hGlyRS-E71G variant does not act through a dominant negative mechanism. In contrast, at least in motor neurons, the protein synthesis defect and peripheral neuropathy phenotypes induced by the G240R and G526R variants may in part be attributable to a dominant negative mechanism, in addition to tRNA^Gly sequestration.

Generation and characterization of novel Drosophila PN-GlyRS models

Of the 35 human PN-GlyRS mutations reported to date, 20 (57%) add net positive charge to the GlyRS protein (by replacing a negatively charged amino acid by a neutral or positively charged amino acid, a neutral residue by a positively charged one or by deleting a negatively charged residue), 14 do not change the GlyRS protein charge, and only 1 mutation increases the net negative charge by replacing a positively charged residue by an uncharged residue (K456Q) (Fig. 3A). Coincidentally, each of the three GlyRS mutations that we previously modelled in Drosophila increase the net positive charge to the protein (E71G, G240R, and G526R). Although we previously showed that addition of positive charge to the GlyRS protein is not required for a mutation to induce tRNA^Gly sequestration in vitro [12], increasing the net positive charge is likely to promote tRNA^Gly sequestration. We therefore reasoned that if some PN-GlyRS mutations would act through a dominant negative mechanism, mutations that do not alter the net GlyRS charge or even increase the net negative charge are the most likely candidates. Hence, we generated three novel PN-GlyRS Drosophila models. We selected the S211F and H418R mutations, because (i) they do not change the net GlyRS charge, (ii) they clearly segregate with disease in families [32, 33] and (iii) they result in loss of aminoacylation activity [29, 34, 35], which is a prerequisite for a dominant negative mechanism (Supplementary Table S3). In addition, we selected the K456Q mutation [35, 36] because this is the only mutation reported that renders the overall GlyRS charge more negative (Fig. 3A, Supplementary Table S3). After introduction of the respective mutations, we used site-specific transgenesis to integrate the hGlyRS transgenes in landing sites on the second and third chromosomes of Drosophila. To maximize the comparability between PN-GlyRS models, the same genomic landing sites were used as for our initial PN-GlyRS Drosophila models. Similar to the previous PN-GlyRS Drosophila models [10], the newly generated UAS-hGlyRS transgenes allow for expression of the cytosolic and mitochondrial hGlyRS isoforms, depending on the translation start site used.

hGlyRS-K456Q does not induce peripheral neuropathy phenotypes in Drosophila and retains aminoacylation activity

As a first step, we evaluated whether expression of the new hGlyRS variants in an otherwise WT dGlyRS background would induce peripheral neuropathy phenotypes and inhibit protein synthesis. While the expression of hGlyRS-S211F and hGlyRS-H418R clearly induced peripheral neuropathy phenotypes and inhibited protein synthesis (see below and Fig. 4), this was not the case for hGlyRS-K456Q. Whereas ubiquitous expression (actin5C-GAL4) of hGlyRS-S211F and hGlyRS-H418R transgenes induced developmental lethality, ubiquitous expression of either one or two copies of hGlyRS-K456Q transgene did not reduce adult offspring frequency (Fig. 3B and C). Furthermore, expression of one or two copies of hGlyRS-K456Q transgenes in motor neurons did not induce adult motor performance deficits (Fig. 3D). Moreover, even high-level expression of hGlyRS-K456Q in class IV multidendritic sensory neurons did not induce dendritic morphology defects (Fig. 3E). Finally, high-level expression of hGlyRS-K456Q did not affect de novo protein synthesis in larval motor neurons (Fig. 3F). To ensure that the lack of peripheral neuropathy phenotypes following hGlyRS-K456Q expression is not attributable to reduced hGlyRS-K456Q protein levels, we performed western blotting using heads of flies expressing hGlyRS transgenes ubiquitously from the adult stage onwards, to circumvent developmental lethality (tubulin-GAL80^ts; tubulin-GAL4, see Materials and Methods for details). This experiment revealed that hGlyRS-K456Q levels were comparable to hGlyRS-WT levels, whereas hGlyRS-S211F and hGlyRS-H418R protein levels were significantly reduced (Fig. 3G and H). The reduced hGlyRS-S211F and hGlyRS-H418R protein levels are likely due to inhibition of global protein synthesis induced by these PN-GlyRS mutants (see below and Fig. 4D and E), because the presence of PN-GlyRS mutations did not affect hGlyRS transcript levels (Fig. 3I). Collectively, our data indicate that even high-level expression of hGlyRS-K456Q did neither affect protein synthesis, nor induced peripheral neuropathy in Drosophila, indicating that the K456Q variant does not act through a gain-of-toxic function mechanism.

Because the hGlyRS transgenes were expressed in a WT dGlyRS background, we wanted to exclude that hGlyRS-K456Q causes peripheral neuropathy in human CMT patients through a loss-of-function mechanism. We therefore evaluated the effect of the K456Q mutation on tRNA^Gly aminoacylation activity, using both an in vitro aminoacylation assay and in vivo genetic complementation experiments. The in vitro aminoacylation assay with recombinant hGlyRS proteins revealed that hGlyRS-K456Q displays tRNA^Gly aminoacylation activity similar to that of hGlyRS-WT (Fig. 3J,K). Knock-down of dGlyRS in sensory organ precursor cells (Sca-GAL4> UAS-dGlyRS-RNAi-2) resulted in shorter or missing scutellar bristles, which are long in control animals (Fig. 3L). Simultaneous expression of hGlyRS-WT fully rescued these bristle phenotypes (Fig. 3M, Supplementary Fig. S1), indicating that the dGlyRS and hGlyRS are functional homologs. Consistent with the WT-like aminoacylation activity, expression of hGlyRS-K456Q also fully rescued dGlyRS-RNAi induced bristle phenotypes (Fig. 3M, Supplementary Fig. S1). Finally, the climbing defect induced by dGlyRS knock-down in motor neurons (OK371-GAL4) was also fully rescued by simultaneous expression of either hGlyRS-WT or hGlyRS-K456Q (Fig. 3N). Together, our data show that hGlyRS-K456Q exhibits WT-like tRNA^Gly aminoacylation activity and indicate that the K456Q mutation may not be pathogenic for PN-GlyRS (see discussion).

Peripheral neuropathy phenotypes induced by hGlyRS-S211F and hGlyRS-H418R are not attributable to a dominant negative mechanism

We next asked whether a dominant negative mechanism may underlie peripheral neuropathy phenotypes induced by hGlyRS-S211F and hGlyRS-H418R expression. Therefore, we evaluated the effect of dGlyRS co-overexpression, because increasing dGlyRS levels is expected to induce a phenotypic rescue in case of a dominant-negative mechanism. Expression of hGlyRS-S211F or hGlyRS-H418R in motor neurons reduced the climbing speed of adult flies in the negative geotaxis assay by ∼50% (Fig. 4A). Strikingly, simultaneous dGlyRS overexpression did not rescue this motor performance defect. Expression of hGlyRS-S211F and hGlyRS-H418R in motor neurons also reduced the neuromuscular synapse length on larval muscle 8 by ∼41% and ∼49%, respectively (Fig. 4B). Again, simultaneous dGlyRS overexpression did not significantly rescue this NMJ morphology defect. Selective expression of hGlyRS-S211F and hGlyRS-H418R in class IV multidendritic sensory neurons reduced their dendritic coverage by ∼35% and ∼46%, respectively (Fig. 4C). Consistent with the results in motor neurons, simultaneous dGlyRS overexpression did not rescue the sensory neuron morphology defect. For hGlyRS-S211F, dGlyRS co-overexpression rather aggravated the sensory neuron morphology defect (Fig. 4C).

Finally, we used THRONCAT, a recently developed method that allows for metabolic labelling of newly synthesized proteins with the non-canonical amino acid β-ethynylserine by the endogenous threonyl-tRNA synthetase [26] (see “Materials and methods” for details). With this method, we observed that expression of the previously characterized hGlyRS-G240R variant in motor neurons reduced de novo protein synthesis by ∼19% (Supplementary Fig. S2). Similarly, expression of hGlyRS-S211F or hGlyRS-H418R in motor neurons reduced protein synthesis by ∼17% and ∼18%, respectively (Fig. 4D,E). Importantly, dGlyRS co-overexpression did not rescue this protein synthesis defect (Fig. 4D and E). In all, these data show that the peripheral neuropathy phenotypes induced by hGlyRS-S211F and hGlyRS-H418R expression are not attributable to a dominant negative mechanism.

Transgenic tRNA^Gly overexpression rescues peripheral neuropathy induced by hGlyRS-S211F and hGlyRS-H418R expression

We next investigated whether tRNA^Gly sequestration may be the pathogenic mechanism underlying PN-GlyRS caused by S211F and H418R mutations. To do so, we evaluated the effect of transgenic tRNA^Gly overexpression on peripheral neuropathy phenotypes and the mRNA translation defect induced by hGlyRS-S211F and hGlyRS-H418R. We used two previously described tRNA^Gly-GCC transgenic lines, a line that contains a BAC transgene with five tRNA^Gly-GCC genes that was inserted in the second and the third chromosome (overall resulting in 10 extra tRNA^Gly-GCC gene copies), and a tRNA^Gly-GCC ‘scramble’ transgenic line that contains 10 tRNA^Gly-GCC genes and results in higher level tRNA^Gly-GCC overexpression (∼13% overexpression for tRNA^Gly-GCC BAC and ∼30% for tRNA^Gly-GCC scramble) [12]. Interestingly, the adult motor performance deficit induced by expression of hGlyRS-S211F or hGlyRS-H418R in motor neurons was rescued by both the tRNA^Gly-GCC BAC and tRNA^Gly-GCC scramble transgenes, and the degree of rescue was more pronounced for higher tRNA^Gly overexpression levels (Fig. 5A and B). Consistent with this result, the reduced neuromuscular synapse length on larval muscle 8 induced by expression of hGlyRS-S211F or hGlyRS-H418R in motor neurons was significantly rescued by tRNA^Gly overexpression (Fig. 5C). Furthermore, the sensory neuron dendritic morphology defect induced by hGlyRS-S211F or hGlyRS-H418R expression in class IV multidendritic sensory neurons was significantly rescued by tRNA^Gly overexpression (Fig. 5D). Finally, THRONCAT revealed that the reduced protein synthesis following hGlyRS-S211F or hGlyRS-H418R expression in motor neurons is significantly rescued by tRNA^Gly overexpression (Fig. 5E and F). Collectively, these data indicate that also for hGlyRS-S211F and hGlyRS-H418R mutations, tRNA^Gly sequestration is the underlying pathogenic mechanism.

Biochemical evidence for tRNA^Gly sequestration by hGlyRS S211F and H418R variants

To complement our Drosophila genetic data, we performed in vitro experiments to evaluate tRNA^Gly sequestration by hGlyRS S211F and H418R variants. We purified recombinant hGlyRS-WT, hGlyRS-S211F, and hGlyRS-H418R proteins and evaluated the effect of the mutations on hGlyRS dimer formation by size exclusion chromatography. Whereas hGlyRS-WT migrated predominantly as a dimer (dimer:monomer (D:M) ratio 9.1 : 0.9), hGlyRS-S211F partitioned predominantly as monomer (D:M ratio 3.8 : 6.2). For hGlyRS-H418R, the dimer was the predominant form but to a much lower fraction than hGlyRS-WT (D:M ratio 7.0 : 3.0) (Fig. 6A). We next used the hGlyRS dimer fraction to determine the association (K_on) and release (K_off) kinetic constants for tRNA^Gly-GCC to hGlyRS-WT, hGlyRS-S211F and hGlyRS-H418R. Both hGlyRS-S211F and hGlyRS-H418R variants were able to bind tRNA^Gly, albeit with a reduced affinity (∼3-fold and ∼5-fold fold lower K_on values than WT, respectively, Fig. 6B). Importantly, these variants displayed markedly slower tRNA^Gly release kinetics (K_off), with more than 90% of traces showing no tRNA^Gly release within the experimental time frame (Fig. 6B). These data are consistent with tRNA^Gly sequestration by hGlyRS-S211F and hGlyRS-H418R variants.

Discussion

For many genetic diseases, multiple disease-causing mutations in the same gene have been identified, often distributed throughout the coding region and spread across diverse functional domains. For the vast majority of these diseases, it is unclear whether all mutations cause disease through the same molecular mechanism. Here, we addressed this question for peripheral neuropathy associated with mutations in GARS1, encoding GlyRS. Previously, we provided genetic and biochemical evidence for seven PN-GlyRS mutations, that sequestration of tRNA^Gly and consequent depletion of the cellular tRNA^Gly pool is the underlying molecular mechanism [12]. The insufficient availability of tRNA^Gly substrate for aminoacylation by WT GlyRS, derived from the WT GARS1 allele in heterozygous patients, leads to insufficient glycyl-tRNA^Gly to be used by the translating ribosome. This in turn triggers ribosome stalling at glycine codons [12] and activation of the ISR through the eIF2α kinase GCN2 [11]. This mechanism also provides an explanation for the protein synthesis defect observed in PN-GlyRS Drosophila and mouse models [10, 11].

Here, we evaluated whether all known PN-GlyRS variants cause disease through tRNA^Gly sequestration, or alternatively, whether a subset of these mutations may trigger peripheral neuropathy through a dominant negative effect on GlyRS aminoacylation activity. Indeed, because GlyRS functions as a homodimer with a single tRNA^Gly interacting with both GlyRS monomers [37], a GlyRS dimer consisting of one WT and one mutant monomer may fail to aminoacylate tRNA^Gly, provided that the mutation results in loss of aminoacylation activity but does not inhibit dimerization with WT GlyRS. In such a dominant negative scenario, the aminoacylation-active GlyRS population, which consists of WT GlyRS homodimers, may be reduced by 75%. In turn, 75% reduction of tRNA^Gly aminoacylation capacity may substantially lower the levels of glycyl-tRNA^Gly available for translation, thus potentially triggering peripheral neuropathy. In Drosophila PN-GlyRS models, a dominant negative mechanism would require dimerization between endogenous dGlyRS and transgenic hGlyRS, which we previously demonstrated to occur [10]. The plausibility of a dominant-negative mechanism underlying PN-GlyRS was suggested by our finding that reducing endogenous dGlyRS levels by transgenic RNAi-mediated knock-down is sufficient to induce peripheral neuropathy phenotypes and to substantially inhibit de novo protein synthesis (Fig. 1).

To evaluate a possible dominant negative mechanism and because all PN-GlyRS mutations in Drosophila and mouse models generated thus far increase the net positive charge of the GlyRS protein [10, 16, 38–40], we decided to generate three novel Drosophila PN-GlyRS models for two mutations that do not alter GlyRS protein charge (S211F and H418R) [32, 33], and the single GlyRS mutation described that reduces positive charge (K456Q) [35, 36]. The S211F and H418R mutations were previously reported to result in severely reduced tRNA^Gly aminoacylation activity [29, 34], whereas the effect of the K456Q mutation on aminoacylation activity has not been reported. Very similar to our previously reported Drosophila PN-GlyRS models [10], expression of hGlyRS-S211F or hGlyRS-H418R in motor and sensory neurons induced peripheral neuropathy phenotypes and substantially inhibited de novo protein synthesis in motor neurons (Fig. 4 and 5). This phenotypic similarity supports the notion of a similar underlying pathogenic mechanism.

To evaluate whether either a dominant negative mechanism or tRNA^Gly sequestration may underlie PN-GlyRS caused by S211F and H418R mutations, we studied the effect of transgenic overexpression of WT dGlyRS or tRNA^Gly, respectively. For both hGlyRS-S211F and hGlyRS-H418R variants, dGlyRS overexpression did not rescue peripheral neuropathy or inhibition of protein synthesis, providing a strong argument against a dominant negative mechanism (Fig. 4). This result is consistent with the finding that transgenic overexpression of WT hGlyRS did not rescue peripheral neuropathy phenotypes in two PN-GlyRS mouse models [17]. In contrast, tRNA^Gly overexpression dosage-dependently rescued both peripheral neuropathy phenotypes and protein synthesis (Fig. 5), providing genetic evidence for depletion of the cellular tRNA^Gly pool due to tRNA^Gly sequestration as the underlying pathogenic mechanism. In addition, we provide biochemical evidence of tRNA^Gly sequestration by hGlyRS-S211F and hGlyRS-H418R variants (Fig. 6). Thus, addition of positive charge is not required for a PN-GlyRS mutation to induce tRNA^Gly sequestration. This is consistent with our previous finding that the hGlyRS-L129P and hGlyRS-E279D mutations, which do not alter protein charge, clearly induce tRNA^Gly sequestration in biochemical experiments [12].

For our previously characterized PN-GlyRS Drosophila models (E71G, G240R, G526R mutations) [10], we had shown that tRNA^Gly overexpression at least partially rescues peripheral neuropathy phenotypes and the protein synthesis defect [12]. Here, we evaluated the effect of dGlyRS overexpression on these phenotypes. For hGlyRS-E71G, dGlyRS overexpression did not rescue peripheral neuropathy or protein synthesis (Fig. 2), arguing against a dominant negative mechanism. Given that the E71G mutation does not significantly affect tRNA^Gly aminoacylation activity [10, 29, 30] or GlyRS dimerization [12], this result could be expected. In contrast, G240R and G526R severely reduce aminoacylation activity [10, 30, 31, 34], which may hint to a dominant negative mechanism. Interestingly, dGlyRS overexpression partially rescued peripheral neuropathy and protein synthesis defects induced by the G240R and G526R variants. A possible explanation for this result is that in addition to tRNA^Gly sequestration, these two GlyRS variants also induce a dominant negative effect, which may, at least in part, contribute to the insufficient glycyl-tRNA^Gly production and supply to the ribosome. An alternative explanation for this observation is that WT and mutant GlyRS species may compete for binding to the limited cellular tRNA^Gly pool. In this scenario, a substantial increase in dGlyRS levels may increase the tRNA^Gly fraction interacting with and aminoacylated by WT dGlyRS, thus increasing glycyl-tRNA^Gly levels available to the translating ribosome. Consistent with this hypothesis, the fact that the G240R mutation was reported to substantially reduce hGlyRS dimerization capacity [30] argues against a dominant negative mechanism. Co-immunoprecipitation experiments in transfected cells previously suggested that hGlyRS-G526R may retain or even strengthen GlyRS dimerization capacity [30, 31], but size-exclusion chromatography revealed that recombinant hGlyRS-G526R protein predominantly occurs as a monomer [12].

Although our data show a clear correlation between the degree to which PN-GlyRS variants inhibit global protein synthesis and the severity of peripheral neuropathy phenotypes, this correlation is not perfect. Furthermore, while tRNA^Gly overexpression substantially rescues peripheral neuropathy phenotypes of all PN-GlyRS Drosophila models, the rescue is not complete. These observations may be explained by the fact that the UAS-hGlyRS transgenes allow expression of both the cytosolic and the mitochondrial hGlyRS isoforms, depending on the translation start site used. Because in higher eukaryotes, mitochondrial tRNAs are encoded by the mitochondrial genome and mitochondria typically do not import nuclear encoded tRNA [41], mitochondrial translation defects and toxicity induced by PN-GlyRS variants are likely not rescued by tRNA^Gly overexpression. Moreover, additional pathogenic mechanisms, such as axonal transport defects induced by decreased α-tubulin acetylation [42, 43], may not be fully rescued by tRNA^Gly overexpression. In spite of the possible contribution of mitochondrial toxicity induced by PN-GlyRS variants [44], the cytosolic toxicity induced by tRNA^Gly sequestration is likely a critical contributor to peripheral neuropathy phenotypes, because (i) in PN-GlyRS mouse models, transgenic tRNA^Gly overexpression fully rescues peripheral neuropathy phenotypes [12], and (ii) all of the seven other PN-associated aaRS genes exclusively encode the cytosolic aaRS, with the corresponding mitochondrial aaRSs being encoded by different genes.

Size-exclusion chromatography revealed that GlyRS-S211F predominantly occurs as a monomer, similar to some other PN-GlyRS variants (e.g. L129P, C157R, G240R) [12, 30]. We previously reported that PN-GlyRS monomers are able to bind and sequester tRNA^Gly [12], which we anticipate may also be the case for GlyRS-S211F. Consistent with this notion, peripheral neuropathy phenotypes induced by GlyRS S211F, C157R and G240R variants in Drosophila and mouse models are rescued by transgenic tRNA^Gly overexpression.

In sharp contrast to hGlyRS-S211F and hGlyRS-H418R, expression of hGlyRS-K456Q induced neither peripheral neuropathy phenotypes, nor inhibition of protein synthesis. Although the K456Q variant was previously classified as likely pathogenic [35], according to the most recent American College of Medical Genetics and Genomics (ACMG) guidelines [45], this variant would be classified as a variant of uncertain significance. Our data in Drosophila (Fig. 3) indicate that this hGlyRS variant may not be pathogenic. Because the K456Q variant is the only one of the 35 reported PN-GlyRS variants that eliminates a positive charge from the GlyRS protein, our data further suggest that mutations that render the overall GlyRS charge more negative than the WT protein may not be compatible with PN-GlyRS pathogenicity in heterozygous patients. In contrast, 4 of the 7 GARS1 mutations reported in compound heterozygous patients with biallelic GARS1 mutations remove a positive charge from the GlyRS protein [1]. These patients display systemic mitochondrial disease or multisystem developmental syndrome [44, 46–49], and these phenotypes are likely attributable to a substantial reduction of tRNA^Gly aminoacylation activity [1]. The systemic mitochondrial disease features suggest that mitochondrial mRNA translation may be predominantly affected, leading to mitochondrial respiratory chain dysfunction [44, 46, 47, 49].

Overall, our data indicate that it is highly likely that all PN-GlyRS mutations trigger peripheral neuropathy through a common tRNA^Gly sequestration mechanism. From a therapeutic perspective, this suggests that elevating tRNA^Gly levels in the affected motor and sensory neurons may constitute a viable therapeutic approach for all PN-GlyRS patients, irrespective of the disease-causing mutation.

Acknowledgements

We thank Robin Thompson for generating the plasmids for purification of the hGlyRS-S211F and hGlyRS-H418R variants, Bob Ignacio and Kimberly Bonger for providing βES for THRONCAT experiments, and the General Instruments Facility (Faculty of Science, Radboud University) for advice on image acquisition and analysis. Graphical abstract image was created in BioRender https://BioRender.com/h26s001.

Author contributions: N.M.: formal analysis [equal], investigation [lead], methodology [equal], supervision [supporting], visualization [supporting], writing - original draft [supporting], and writing - review & editing [lead]; E.F.J.S.: formal analysis [equal], investigation [lead], methodology [equal], project administration [lead], writing - original draft [supporting], and writing - review & editing [lead]; V.L.: formal analysis [equal], investigation [equal], methodology [equal], visualization [supporting], and writing - review & editing [supporting]; M.P.M.: formal analysis [supporting], investigation [equal], writing - review & editing [supporting]; M.M.: formal analysis [supporting], investigation [supporting], and writing - review & editing [supporting]; P.V.L.: formal analysis [supporting], and investigation [supporting]; C.S.: formal analysis [supporting], and investigation [supporting]; N.V.B.: formal analysis [supporting], and investigation [supporting]; Z.I.: formal analysis [equal], funding acquisition [supporting], supervision [supporting], and writing - review & editing [lead]; and E.S.: conceptualization [lead], formal analysis [lead], funding acquisition [lead], project administration [supporting], supervision [lead], visualization [lead], and writing - original draft [lead], and writing - review & editing [lead]).

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

E.S. is inventor on patent WO2021158100A1 on ‘tRNA delivery as a therapeutic approach for CMT-aaRS’ and is co-founder and shareholder of XtRNA Bio, a tRNA gene therapy company. Z.I. is inventor on several patents related to tRNA-based therapeutics and a scientific advisor for Tevard Biosciences and XtRNA Bio. All other authors declare no conflict of interest.

Funding

E.S. was supported by an ERC consolidator grant (ERC-2017-COG 770244), and funding from the Radala Foundation, ‘Stichting ALS Nederland’, AFM-Telethon, ARSLA, the ‘Prinses Beatrix Spierfonds’ (W.OR22-03), the Muscular Dystrophy Association (MDA 946876), an NWO Open Competition ENW-M grant, the EU Joint Programme – Neurodegenerative Disease Research (JPND; grant numbers ZonMW 733051075 (TransNeuro) and ZonMW 733051073 (LocalNMD)), and the Donders Center for Neuroscience. Z.I. is supported by the Deutsche Forschungsgemeinschaft DFG (IG73/21-1).

Data availability

All data are presented in the main manuscript and supporting files. Raw data and images underlying this article will be shared on request to the corresponding authors.

References

Author notes