Inhibition of the integrated stress response rescues a histidyl-tRNA synthetase variant associated with Charcot–Marie–Tooth disease

Abstract

Cytoplasmic histidyl-tRNA synthetase (HARS1) is an essential protein in translation, ligating histidine to its cognate tRNA^His. HARS1 is one of several aminoacyl-transfer RNA (tRNA) synthetases associated with Charcot–Marie–Tooth disease, an axonal peripheral neuropathy. Advances in genetic testing identify many variants of uncertain significance. We characterize a novel heterozygous allele in HARS1, c.1200G > T (p.Leu400Phe) with a familial inheritance pattern of peripheral neuropathy. Using a humanized yeast model and biochemical assays, we determined that HARS-L400F causes HARS aggregation, reduced thermal stability, and a temperature-dependent reduction of aminoacylation activity. In humanized yeast, L400F leads to a pronounced growth defect, especially at elevated temperatures. Contrary to previously described pathogenic HARS alleles, cognate amino acid or tRNA substrate supplementation does not ameliorate the growth defect. We show that, in a humanized yeast model, HARS L400F leads to the activation of the integrated stress response (ISR) and upregulated chaperone expression. The yeast growth phenotype can be rescued by inhibition of the ISR using a specific inhibitor of general control non-depressible 2 (GCN2) kinase, opening a novel therapeutic avenue for pathogenic HARS1 alleles that do not respond to substrate supplementation.

Graphical Abstract

Graphical Abstract.

Introduction

Charcot–Marie–Tooth (CMT) disease, a genetically heterogenous disorder, affects the peripheral nervous system, resulting in motor and sensory deficits [1]. Electrophysiological analysis divides CMT into two major subtypes: CMT1, the demyelinating form characterized by uniformly slowed nerve conduction velocity (NCV; < 38 m/s), and CMT2, the axonal type that presents with reduced compound muscle action potential and normal to decreased NCV [1, 2]. A third subtype, intermediate CMT (CMTI), features both axonal degeneration and demyelination [2]. Clinical manifestations of CMT across all subtypes typically include distal muscle weakness and atrophy, hammertoes, gait abnormalities, balance problems, chronic limb pain, and decreased sensation to pain and temperature [3]. At present, there is no disease-specific treatment available for CMT [1]. With over 150 genes implicated, CMT can be inherited in an autosomal dominant, autosomal recessive, or X-linked manner [4–6]. Many CMT-linked genes play roles in crucial cellular processes, like protein translation, intracellular transport, and mitochondrial function [7]. Among the largest gene families implicated in CMT are the aminoacyl-tRNA synthetases (aaRSs). To date, around 60 pathogenic alleles in seven aaRSs have been associated with CMT, including alanyl-, methionyl-, glycyl-, tryptophanyl-, tyrosyl-, histidyl-, and asparaginyl-tRNA synthetases [8].

In eukaryotes, 20 standard amino acids are incorporated into proteins, with each proteinogenic amino acid having its own designated aaRS [9]. As essential enzymes, aaRSs function by covalently linking their cognate transfer RNAs (tRNAs) with the cognate amino acid via a transesterification reaction [9, 10]. Once charged with their respective amino acids, the aminoacylated tRNAs engage with the ribosomes to facilitate the incorporation of amino acids into the nascent polypeptides [9]. Thus, aaRS-associated CMT disease phenotypes are thought to arise due to normal cellular protein synthesis or protein quality control mechanisms being compromised over time; however, the exact underlying molecular mechanism driving these consequences can vary between different aaRS variants [8]. Loss-of-function aaRS mutations may result in decreased aminoacylation function or reduced aaRS stability and structural integrity. Conversely, in a gain-of-function manner, certain pathogenic aaRS alleles can mischarge the tRNAs, leading to misincorporation of the non-cognate amino acids at the cognate codons [8]. The human histidyl-tRNA synthetase (HARS) is responsible for ligating histidine onto tRNA^His [11]. For most aaRSs, the tRNA anticodon serves as a major identity element, however, cognate tRNA^His is identified by HARS through a unique 5′ G-1 residue that is added post-transcriptionally by the guanylyl-transferase (Thg1) [12]. Cytoplasmic HARS (HARS1) functions as a homodimer, and its structure is comprised of a WHEP-TRS domain, a catalytic domain, and a tRNA-binding domain (Fig. 1) [8]. Pathogenic variants in the HARS gene are associated with axonal CMT-type 2W (CMT2W), with an autosomal dominant mode of inheritance [7], as well as the homozygous variant HARS Y454S causing Usher syndrome [13].

Figure 1.

A novel HARS variant L400F is localized near the tRNA-binding site and dimer interface. (A) Pedigree and HARS variants. Filled symbols indicate individuals heterozygous for the L400F HARS variant; squares indicate males, and circles, females. Short horizontal lines indicate individuals who had targeted variant testing. Arrow indicates the proband. (B) Domain structure of HARS1. The codon encoding Leu400 is localized in the linker region between tRNA binding and catalytic domains. (C) Crystal structure of the human HARS dimer (PDB: 6O76) with monomer 1 in green and monomer 2 in orange. The human HARS1 structure was super imposed with the structure of Thermus thermophilus HARS (PDB: 4RDX). The protein domain of T. thermophilus was omitted for clarity and the tRNA is shown in purple and ATP and histidine in light blue. (D) Close up of L400 from panel (C). L400 is localized in an alpha-helix near the active site, dimer interface, and tRNA-binding domain. Amino acids within 5 Å radius are shown in light blue.

Here, we characterize a novel heterozygous HARS-L400F variant of uncertain significance (VUS). We found that the L400F protein is thermally labile and displays reduced dimerization in vitro. A HARS-L400F humanized yeast model revealed a significant growth deficiency in cells expressing L400F at 30°C and 37°C. At 40°C, yeast cells expressing L400F mutant showed a temperature-sensitive lethal phenotype, characterized by a complete loss of viability. In this model, histidine and tRNA^His supplementation did not rescue L400F-associated phenotype. Our data further showed that L400F induces the activation of the integrated stress response (ISR), and the yeast growth defect was restored with the inhibition of general control non-depressible 2 (GCN2), opening new therapeutic approaches for CMT treatment.

Materials and methods

DNA sequencing

The proband was tested at Invitae using the CMT multigene panel of 57 genes (Invitae, 2023), with further testing through the Invitae comprehensive neuropathies panel of 102 genes. Family members were tested using targeted variant testing.

Yeast strain and growth conditions

Humanized HARS yeast model was generated previously [14]. BY4742 was used as a haploid S. cerevisiae control strain. Yeast strains were grown on yeast extract–peptone–dextrose (YPD) (10 g/l yeast extract, 20 g/l peptone, and 20 g/l dextrose) or synthetic defined (SD) (6.7 g/l yeast nitrogen base, 2% glucose, 60 mg/l L-isoleucine, 20 mg/l L-arginine, 40 mg/l L-lysine, 60 mg/l L-phenylalanine, 10 mg/l L-threonine, 10 mg/l L-methionine, 10 mg/l adenine hemisulfate salt, 60 mg/l L-leucine, 40 mg/l L-tryptophan, 20 mg/l L-histidine, and 20 mg/l uracil) liquid media. SD media was supplemented with 20 g/l agar for solid medium growth. Histidine concentration was increased to 200 mg/l for high histidine growth conditions. GCN2 inhibitor (GCN2iB, Cayman Supports Science Item No. 35897) was supplemented in the media at a concentration of 10 nM. BY4742 was grown in YPD media. Humanized yeast strains were either grown in SD media lacking L-leucine (SD-leu^-) to select for plasmid expressing either HARS-wild type (WT) or L400F or SD media lacking L-leucine and uracil (SD-ura^-/leu^-) to select for plasmids expressing either HARS WT or L400F and tRNA^His. For growth curves, yeast cultures were normalized to an optical density (OD) of 0.1 at 600 nm and grown at 30°C , 37°C, or 40°C in 96 well plates for up to 48 h.

Plasmids

Human HARS WT and L400F were expressed as YFP fusion proteins in yeast from p425-GPD ccdB (LEU2), as described previously [14]. tRNA^His was cloned into p426-GPD ccdB for yeast expression, as described previously [15]. For biochemical assays, HARS WT and L400F was cloned into a pDEST-527 with a 6x-His tag and transformed into Escherichia coli BL21C + strain. Recombinant proteins were purified using the Ni-NTA chromatography, as described previously [14].

The L400F mutation was generated using QuikChange site-directed mutagenesis into the WT HARS open reading frame using two primers: forward (5′-GGAACAGAGACTAGAGGCTTTT GAGGAGAAGATACGGACCACG-3′) and reverse (5′-CGTGGTCCGTATCTTCTCCTCA AAAGCCTCTAGTCTCTGTTCC -3′). The open reading frames were verified by DNA sequencing at Azenta Life Sciences (Azenta Life Sciences, Chelmsford, MA, USA).

Sedimentation assay and western blotting

Sedimentation assays were adapted from Shiber et al. and carried out as previously described [15, 16]. Briefly, yeast cultures were normalized to OD of 1 at 600 nm and lysed (100 mM Tris, pH 7.5, 200 mM NaCl, 1 mM ethylenediaminetetraacetic acid, 1 mM dithiothreitol (DTT), 5% glycerol, 0.5% Triton-X, 50 mM N-ethylmaleimide, 2 mM phenylmethylsulfonyl fluoride, 3.2 mg/ml protease inhibitor) before being separated by centrifugation into soluble and insoluble protein fractions. Equal volumes of each fraction were loaded on to 12% Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) gels in at least three biological replicates and subsequently either stained with coomassie or transferred onto a polyvinylidene fluoride membrane for Western blotting. For blots with GCN2iB, only the whole-lysate fractions were used. Anti-GFP antibody (Abcam, Cambridge, UK, ab183734) was used at a ratio of 1:1000 to quantify HARS-YFP abundance. Antibodies for eIF2α (Invitrogen, Waltham, MA, USA AHO0802) and phospho-eIF2α (Abcam, Cambridge, UK, ab32157) were used at a ratio of 1:1000. For loading control, anti-PGK1 (Cell Signalling, Danvers, MA, USA 68540) was used at a ratio of 1:1000. IRDye® 800CW Donkey anti-Rabbit IgG and 680CW Goat anti-Mouse (LI-COR Biotechnology, Lincoln, NE, USA, 926-32214, 926-68070) were used as secondary antibodies at 1:5000 dilutions. Quantifications of the soluble versus insoluble protein fractions and the blots were carried out using ImageJ.

HARS purification and protein biochemistry

Recombinant HARS WT and L400F were purified using the Ni-NTA column as 6×-His-tag fusion proteins, as previously described [14]. Thermal shift assay was carried out with and without the substrates (histidine and in vitro transcribed full length tRNA^His) using the Protein Thermal Shift^TM Dye Kit (Thermo Fisher Scientific Waltham, MA, USA, 4461146) according to manufacturer’s instructions, as described before [14]. Melting temperatures (T_m) of the recombinant proteins were determined using the Thermal Shift^TM software (Thermo Fisher Scientific). Size-exclusion chromatography was performed using the Superdex 200 Increase 10/300 column (Cytiva Marlborough, MA, USA, 28-9909-44), as previously described [7]. To confirm the purification of the recombinant HARS proteins, samples were loaded on to 12% SDS–PAGE gels and blotted for 6x-His tag (Invitrogen, MA1-21315). HARS aminoacylation assays were carried out using recombinant proteins and our previously published protocols [17, 18]. HARS proteins were either measured at room temperature, or after a 10-min incubation at 42 °C.

Proteomics analysis by mass spectrometry

Samples were reduced with 10 mM DTT, alkylated with 20 mM iodoacetic acid, acetone-precipitated, and then digested overnight with MS-grade trypsin (Promega, Wisconsin, USA). Digested peptides were desalted with in-house made C18 spin columns, then dried and resuspended in 12 µl of 0.1% formic acid.

MS/MS data acquired on Thermo Orbitrap Fusion Lumos for each sample were carried out in data-dependent acquisition mode with a cycle time of 3 s. In the first round, MS1 scan data were obtained at 120 000 resolution (at 400 m/z) with a mass range of 400–1600 m/z. The automatic gain control (AGC) was set to standard, with a maximum ion injection time of 50 ms. The radio frequency lens was set to 30%. The charge state filter was set to 2–8, and the dynamic exclusion was set to 15 s. Isolation for MS2 scans was performed in the quadrupole, with an isolation window of 0.7 Da. MS2 scan data were acquired at a resolution of 15 000 m/z in the orbitrap, with an AGC target of 50 000 and a maximum ion injection time of 22 ms. The scan range of MS2 was also set to auto. Higher-energy collisional dissociation (fixed normalized collision energy: 30%) was used for generating MS2 spectra, with the number of microscans set to 1.

MS raw files were processed using PEAKS Studio 12 (Bioinformatics Solutions Inc., Ontario, Canada). The data were searched against the reviewed Saccharomyces cerevisiae UniProt database. Precursor ion mass error tolerance was set to 10 ppm, and fragment ion mass error tolerance was set to 0.02 Da. Semi-specific cleavage with trypsin was selected with a maximum of two missed cleavages. A fixed modification of carbamidomethylation (+57.02 Da) on cysteine residues and a variable modification of oxidation (+15.99 Da) on methionine were specified. A 1% false discovery rate was set for the database search. After the database search, mutations were identified by variable PTM search using the monoisotopic mass shift corresponding to the amino acid substitutions His > Tyr (+26.00 Da), Tyr > His (−26.00 Da), His > Asp (−22.03 Da), Asp > His (+22.03 Da), His > Gln (−9.00 Da), Gln > His (+9.00 Da), His > Glu (−8.02 Da), and Glu > His (+8.02 Da). Only confident mutations supported by adjacent b- and y-ion pairs with relative fragment ion intensities >1% are reported.

Ethics statement

All procedures followed were in accordance with the ethical standards of the Helsinki Declaration. The study was approved by the institutional ethical review board (Western University Health Science Research Ethics Board, REB#126312). We have received consent for participation from every individual whose data were included.

Statistical analysis

All statistical significance was determined by unpaired t-tests to compare the means and standard deviations of the control data and the experimental data set, or a one-way analysis of variance test for multiple comparisons. Significance levels are indicated using asterisks (****P < .0001, ***P < .001, **P < .01, *P < .05, ns = statistically non-significant).

Results

Case presentation

A 60-year-old male presented with progressive numbness and tingling in his hand and feet, chronic foot and back pain, and foot deformities. He had difficulties with balance, especially on uneven ground, and tended to fall easily. He had easy fatiguability and tinnitus since adolescence and hearing loss since his 40s, requiring hearing aids. Vision was good apart from presbyopia, with no history of nyctalopia. He had developed dysphagia, especially with solids, but no history of aspiration. For the preceding 4 years, he had been experiencing daily throbbing headaches or a sharp stabbing pain on bending forward, coughing, or sneezing. There was no history of tremor or change in his handwriting. He experienced nocturnal sweating.

On clinical examination, he had intrinsic foot muscle atrophy with evidence of pes cavus and “inverted champagne bottle” appearance of the lower legs with no significant atrophy of his arms or proximal arms and legs. Electrodiagnostic testing demonstrated a primarily axonal motor and sensorineural peripheral polyneuropathy affecting the lower extremities. Magnetic resonance imaging of head was negative. Computed tomography scan imaging of both ankles revealed severe bilateral ankle osteoarthritis with marked talar tilting within the tibial plafond and total loss of articular surface in the joint spaces.

Family history: The proband’s father (Fig. 1A, I-2) had similar foot deformities, attributed to “rheumatic fever,” but had never sought medical attention, and had 10 siblings, none of whom were symptomatic. The proband’s 52-year-old sister (Fig. 1A, II-5) had chronic neuropathic pain in her feet, was tested on a CMT panel of genes, and was apparently positive for the same HARS1 variant. Two brothers and another sister were asymptomatic and did not carry the variant. The proband’s 33-year-old daughter carries the variant and is asymptomatic as of to date (Fig. 1A, III-1).

Laboratory investigations: The proband was tested on a CMT multigene panel of 57 genes (Invitae, 2023) and was found to be heterozygous for a missense variant in the HARS1 gene, consisting of a c.1200G > T (pLeu400Phe) substitution (NM_002109.5). Further testing through the Invitae comprehensive neuropathies panel of 102 genes revealed the same HARS1 variant with no other sequence variants. Targeted variant testing revealed that the proband’s father, symptomatic sister, and asymptomatic daughter all carry the same HARS variant (Fig. 1).

HARS L400F causes HARS protein aggregation and reduces protein stability

The L400F mutation is localized in an alpha helix in the linker region between the HARS catalytic domain and the tRNA-binding domain (Fig. 1B–D). A phenylalanine residue in this position introduces a large aromatic residue that may disrupt helix formation and protein folding. No tRNA-bound human HARS structure is deposited in the protein database, and we therefore generated a human HARS-tRNA model by aligning the Thermus thermophilus HARS structure in complex with tRNA (pdb 4RDX) with the human HARS dimer structure (pdb 6O76). This model shows that HARS L400 is in close proximity to the modeled tRNA-binding site (Fig. 1C and D) and could disrupt tRNA binding by the HARS dimer or impair HARS dimerization, which is essential for catalytic activity [19]. We therefore purified recombinant HARS WT and L400F protein variants and characterized the thermal stability and dimerization of HARS WT and L400F proteins (Fig. 2). Thermal stability assays showed a significant reduction in HARS L400F melting temperature, where the WT protein showed a melting temperature of 61°C, and the L400F protein showed a melting temperature of 54°C. Addition of histidine, tRNA^His, or a combination of the two substrates did not rescue protein stability.

Figure 2.

Biochemical properties of HARS L400F. (A) Representative melting curve of recombinant HARS WT and L400F. (B) The melting temperature of HARS L400F is significantly reduced compared to WT HARS and is not stabilized by individual substrates tRNA^His or histidine (His), or a combination of both substrates. (C) Aminoacylation assay of WT and L400F HARS at room temperature or after a 10-min incubation oat 42°C. (D) Size-exclusion elution chromatogram and (E) quantification of the area under the curve for peaks corresponding to elution profiles of the monomer, dimer, and aggregated HARS of HARS WT (gray), HARS L400F (red), and a 1:1 mixture of the proteins (pink) shows impaired dimer formation and increased aggregation in L400F. All error bars indicate one standard deviation. *P < .05, **P < .01, ***P < .001, ****P < .0001, ns = not significant.

Next, we assessed the dimerization status of the recombinant proteins at 4 °C using size-exclusion chromatography and compared elution volumes to protein standards. We found that WT HARS mainly existed as a dimer (72% dimer, 22 % aggregates, 6% monomer; Fig. 2C and D), while L400F showed significantly increased aggregation (46 % dimer, 40 % aggregates, 14% monomer; Fig. 2C and D). The 1:1 mixture shows intermediate ratios that are closer to the L400F variant in distribution (57 % dimer, 32 % aggregates, 11% monomer; Fig. 2C and D). Together, these data indicate that HARS L400F is significantly less stable than the WT protein and is prone to aggregation, which persists in heterodimers.

Finally, we assessed recombinant protein aminoacylation activity at both room temperature and after a short 10-min incubation time at 42 °C. No significant defect of L400F aminoacylation was observed at room temperature, but a significant 1.4-fold reduction in aminoacylation was observed under heat stress, which is consistent with a thermolabile protein (Fig. 2C).

HARS L400F shows reduced growth in a humanized yeast model that is exacerbated at 40 °C

To test the effect of the L400F HARS allele on the translation machinery in a eukaryotic cell, we generated a humanized yeast (S. cerevisiae) model, where haploid yeast lacks the endogenous histidyl-tRNA synthetase (HTS1) but is complemented with either a WT or mutant (L400F) HARS allele. We have previously used humanized yeast models to characterize pathogenic HARS alleles [7, 8, 14, 15]. We assessed doubling time as a proxy of maximal growth rate and metabolic activity, and the area under the curve as a measure of overall growth and survival. At 30 °C, the L400F yeast showed a mild but significantly reduced growth phenotype, with both increased doubling time and decreased area under the curve (Fig. 3A–C). Since our biochemical analysis showed that L400F is thermolabile, we also assessed growth at 40 °C and found that L400F yeast are unable to survive at elevated temperatures (Fig. 3D).

Figure 3.

Humanized yeast model for HARS L400F reveals a growth phenotype that is exacerbated at 40 °C and causes HARS aggregation. (A) Growth curve of ΔHTS1 yeast expressing either WT HARS or L400F at 30 °C, and quantification of (B) doubling time and (C) area under the curve. (D) Growth curves of yeast expressing WT and L400F at 40 °C, or (E) two plasmid encoded copies of WT HARS (WT/WT) or WT and L400F HARS (WT/L400F) at 30 °C. (F, G) GFP western blots, and (H, I) quantification of blots visualizing HARS abundance in the whole-cell lysate (W), soluble protein fraction (S) and insoluble protein fraction (P). Error bars represent one standard deviation from at least triplicate growth curves. *P < .05, **P < .01, ***P < .001.

Since our biochemical analysis showed increased protein aggregation of HARS L400F, we tested if the HARS WT or L400F aggregates in the yeast system. We separated the soluble from the insoluble fraction of yeast grown at 30 °C and blotted for YFP-HARS in the whole cell extract (W), soluble fraction (S), and insoluble fraction (P). We found that the ratio of soluble to insoluble HARS in HARS L400F-expressing yeast is significantly smaller, showing an eight-fold reduced ratio, which indicates HARS aggregation in cells (Fig. 3F and H). This demonstrates that the HARS L400F protein aggregates in yeast and causes a growth defect that is lethal at 40°C.

Next, we generated a heterozygous yeast model, where either two WT alleles or one WT and one L400F HARS are expressed from two separate plasmids to better recapitulate the heterozygous alleles in the patient. Consistent with the mild phenotype of L400F alone, and the late onset of disease in the patient, we did not observe a growth phenotype for the heterozygous yeast (Fig. 3E). This may be due to the limitations of the yeast model as observed previously for a clearly pathogenic homozygous HARS Y454S variant that showed no growth defect in a homozygous yeast model [14]. Nontheless, when blotting for HARS abundance in the soluble and insoluble protein fractions, we found that HARS WT/L400F shows significantly increased protein aggregation (Fig. 3G and I), which further indicates that L400F causes HARS protein aggregation in the cell.

Histidine supplementation does not rescue HARS L400F

We previously showed that humanized yeast models for two pathogenic HARS variants, V155G and S356N, were rescued with histidine supplementation [14]. We therefore tested if histidine supplementation can restore growth in the HARS L400F yeast model. No rescue of the growth defect was observed at 30 °C (Fig. 4A, D, and E) or at 40 °C (Fig. 4B). We also tested if growth was improved at 37 °C, the physiological temperature for human HARS (Fig. 4C, F, and G). We did not observe a rescue of doubling time at any temperature, demonstrating that histidine supplementation does not rescue the growth defect.

Figure 4.

The humanized yeast model for HARS L400F is not rescued by histidine supplementation. Growth curves for humanized yeast expressing WT or L400F HARS at (A) 30°C or (B) 40°C or (C) 37°C show that histidine supplementation does not rescue L400F growth. Quantification of doubling time at panel (D) 30°C or (F) 37°C, and area under the curve (E) at 30°C, reveal no significant difference in L400F growth. Maximum A600 at 37°C (G) reveal reduced growth with histidine. Error bars represent one standard deviation from at least triplicate growth curves. *P < .05, **P < .01, ns = not significant.

tRNA^His supplementation does not rescue HARS L400F

tRNA supplementation has been considered an emerging therapeutic avenue for a manifold of diseases including aaRS diseases [20]. Some pathogenic HARS alleles that do not respond to histidine supplementation can be rescued by tRNA supplementation, as we previously demonstrated for pathogenic variants V133F and Y330C [14, 15]. We therefore tested if co-expression of human tRNA^His can ameliorate the growth phenotype of HARS L400F. To our surprise, tRNA supplementation did not only not rescue the phenotype, but, in contrast, exacerbated the growth defect (Fig. 5). At 30°C, HARS L400F only showed a mild growth defect, but when supplemented with tRNA^His, the viability of HARS L400F decreased drastically (Fig. 5A and B), and the doubling time increased three-fold from 4 to 10 h. Interestingly, some of this growth defect could then be rescued by supplying cells with high concentrations of histidine, reducing the doubling time in tRNA^His and His supplemented L400F to 6 h (Fig. 5A and B). At elevated temperatures, neither tRNA^His supplementation alone nor in combination with histidine was able to sustain growth in L400F, while histidine supplementation in combination with tRNA^His increased growth in WT HARS expressing cells (Fig. 5C and D).

Figure 5.

tRNA^His supplementation does not rescue HARS L400F but exacerbates the growth phenotype. Selected growth curves averaged from at least three independent strains of yeast expressing WT and L400F HARS, supplemented with 20 or 200 mg/l of L-histidine with or without human tRNA^His at (A) 30°C or (C) 40°C. (B) Doubling time of L400F strains supplemented with tRNA^His at high and regular histidine concentrations. (D) Area under the curve from growth assays performed at 40°C. (E) Western blots and (F) quantification of blots for HARS-YFP in HARS WT and L400F yeast strains and strains supplemented with human tRNA^His. (G) Sedimentation assay and (H) quantification of WT and L400F strains at 30°C or 40°C, and at 30°C supplemented with tRNA^His. (W, whole cell extract; S, soluble proteins; and P, insoluble proteins). (I) Sedimentation assay blots and (J) quantification of blots for HARS-YFP in samples supplemented with tRNA^His. Error bars represent one standard deviation from at least triplicate growth curves. *P < .05, **P < .01, ***P < .001, ns = not significant.

We next analyzed whether tRNA supplementation could stabilize HARS abundance in the yeast model. When blotting for HARS-YFP abundance in yeast grown at 30°C or 40°C, we found no significant change in HARS abundance at 30°C (Fig. 5 E and F). Interestingly, tRNA^His supplementation significantly decreased overall HARS abundance at 40°C by 12-fold in WT and eight-fold in L400F yeast (Fig. 5E and F).

We next tested if the L400F HARS yeast display increased overall protein aggregation, indicative of global translation errors, as observed for other alleles [14, 15]. We carried out a sedimentation assay with tRNA-supplemented and unsupplemented cells. When comparing the soluble/insoluble proteome of L400F cells to WT HARS expressing cells, we found no change in the ratio of soluble/insoluble proteome, indicating no increase in aggregation of insoluble proteins at 30°C or 40°C. We also did not observe an increase in overall protein aggregation in response to tRNA supplementation or increased temperature, indicating that tRNA supplementation does not exacerbate the insoluble protein aggregation (Fig. 5G and H). We next blotted for HARS abundance in the soluble and insoluble protein fractions and found that L400F cells showed a greater fraction of HARS proteins in the insoluble fraction with or without tRNA supplementation. This demonstrates that HARS L400F aggregates in yeast even in the presence of additional tRNA^His (Figs 3E and 5I, J). Overall, these data show that tRNA^His supplementation does not rescue HARS L400F aggregation and exacerbates the growth phenotype in yeast.

HARS L400F perturbs the proteome and activates chaperone expression

Pathogenic alleles in human HARS were previously shown to lead to mistranslation and a perturbed proteome in yeast. To test if L400F causes mistranslation or a major shift in the cellular proteome, we carried out a proteomic analysis of HARS WT, HARS L400F, and L400F tRNA^His supplemented cells. We included the tRNA^His supplemented cells in this analysis since tRNA overexpression further exacerbates the yeast growth phenotype and may yield additional insight into the mechanism of pathogenesis. The proteomic analysis revealed changes in the proteome due to L400F expression when compared to WT (Fig. 6A and B; supplemental files 1–3). To analyze and visualize the changes in the proteome, we examined proteins that were significantly up- or downregulated by at least twofold using STRING analysis [21]. These data showed that proteins downregulated in L400F compared to WT-expressing cells were associated with amino acid biosynthesis, such as leucine and arginine biosynthesis proteins (Fig. 6C). In addition, several ribosomal subunits were downregulated that also have a role in nonsense-mediated decay, indicating a reduced expression of components of the translational machinery to compensate for reduced HARS activity (Fig. 6C). No functional enrichment was found in proteins upregulated in L400F, but many of the enriched proteins are involved in metabolic processes (Fig. 6D). To investigate why tRNA^His supplementation causes a further reduction in growth, we compared the proteomes of HARS L400F and L400F tRNA^His supplemented yeast. Proteins that were downregulated in L400F when tRNA^His was overexpressed mostly included proteins with roles in metabolic pathways, including again amino acid biosynthesis and pyrimidine biosynthesis, indicating a slowed metabolism and compensation for the loss of HARS function (Fig. 6E). No functional enrichment was found in proteins upregulated in tRNA^His supplemented yeast (Fig. 6F). We additionally carried out a search for mistranslated peptides using the PEAKS SPIDER [22] software but found no evidence of mistranslation in HARS L400F or L400F tRNA^His supplemented cells.

Figure 6.

HARS L400F leads to perturbation of the yeast proteome and increased abundance of heat shock proteins, exacerbated by tRNA^His supplementation. (A) Heat map of differentially abundant proteins identified by label-free quantification proteomic analysis between yeast expressing WT, L400F, and L400F supplemented with tRNA^His. (B) Volcano plot of proteins with significant changes between WT and L400F are shown, upregulated proteins are identified in red and downregulated proteins are identified in green. String analysis of proteins (C) downregulated in L400F compared to WT, (D) upregulated in L400F compared to WT, (E) downregulated in tRNA supplemented relative to non-supplemented L400F, and (F) upregulated in tRNA supplemented relative to non-supplemented L400F. (G) Peptide area of heat shock proteins obtained from the proteomic analysis. (H) Western blots of Hsp70 and Hsp90 and (I) quantification from WT and L400F yeast strains. Samples were grown in SD Leu^-, or SD Ura^-/Leu^- when supplemented with tRNA^His, at 30°C and 40°C, normalized to A600 = 1, lysed, and separated by 12% SDS–PAGE. PGK1 was used as a loading control for all conditions. Error bars represent one standard deviation from at least triplicate blots. *P < .05, **P < .01, ***P < .001, ns = not significant.

A response to reduced translation rates or unfolded proteins can cause an upregulation of chaperones [23, 24]. To investigate if chaperones are increased in abundance in L400F cells, we first analyzed our proteomic data specifically for a change in abundance in chaperones. In the proteomic analysis, HCH1, HSP30, and HSP26 were detected across all samples. HCH1 is a heat shock protein regulator that binds to Hsp90 and may stimulate ATPase activity, and is often upregulated in response to stress [25]. We found that HCH1 is significantly upregulated by threefold in L400F cells compared to WT cells (Fig. 6G). On the contrary, heat shock protein HSP30, the only integral plasma membrane heat shock protein with a role in energy conservation [26] is downregulated by 10-fold (Fig. 6G). The small heat shock protein 26 (HSP26), which protects unfolded proteins from aggregation is upregulated by 3.5-fold (Fig. 6G). The addition of tRNA^His to the L400F cells did not further alter heat shock protein abundance for these three proteins. Since two other heat shock proteins and major chaperones, HSP70 and HSP90, were not detected across all samples in the proteomics dataset, we used western blotting with specific antibodies to compare WT and L400F cells, supplemented and unsupplemented with tRNA^His, at 30°C and 40°C. Heat shock protein 70 (HSP70) is a molecular chaperone that plays a critical role in protein homeostasis. At 30°C, HSP70 abundance is unchanged between WT and L400F mutant but is downregulated by twofold in tRNA^His supplemented cells for both strains (Fig. 6H and I). Not surprisingly, at 40°C, HSP70 is significantly upregulated in both cell lines compared to 30°C but is also increased by twofold in the L400F cells compared to WT cells, indicating a greater stress response (Fig. 6H and I). No change in HSP70 was observed in response to tRNA^His supplementation in WT cells, whereas tRNA^His supplementation resulted in decreased HSP70 levels in L400F cells, reducing HSP70 to WT levels. Heat shock protein 90 (HSP90) is another molecular chaperone in yeast. Similar to HSP30 abundance, no difference was observed between WT and L400F cells at 30°C, while supplementation with tRNA reduced HSP90 for both strains. No changes were observed between any samples at 40°C (Fig. 6H and I). Taken together, our data indicate that three chaperones, HCH1, HSP26, and HSP70, are increased in abundance compared to WT cells, while HSP90 was unchanged and HSP30 was decreased. tRNA^His supplementation reduced some chaperone expression.

HARS L400F activates the integrated stress response, and GCN2 inhibition restores growth

It is well established that translation disruption induces cellular stress, triggering activation of adaptive responses like the ISR to promote energy conservation and maintain protein homeostasis [27]. In the ISR pathway, global messenger RNA translation is repressed by GCN2 kinase-dependent phosphorylation of the eukaryotic initiation factor 2 (eIF2) alpha subunit at serine 51 (Ser51). Under normal conditions, eIF2 facilitates the GTP-dependent delivery of the charged initiator tRNA^Met (Met-tRNA^Met) to the ribosome [28]. However, when eIF2α is phosphorylated, it remains bound to the guanine nucleotide exchange factor, the eukaryotic initiation factor 2B (eIF2B), preventing the exchange of GDP for GTP and the formation of eIF2-GTP-Met-tRNA_i^Met ternary complexes required for the new rounds of translation initiation [27, 29] (Fig. 7A). In addition to overall protein synthesis attenuation, the activation of ISR also leads to an increase in preferential translation of certain key transcription factors such as the mammalian ATF4 (or yeast GCN4) that are involved in the activation of various stress response genes [29, 30] (Fig. 7A).

Figure 7.

Inhibition of GCN2-mediated ISR rescues the L400F yeast growth phenotype. (A) Schematic showing the GCN2-mediated ISR pathway. Western blots for eIF2α and p-eIF2α (B) of haploid WT yeast ( BY4742) and humanized yeast models expressing WT HARS and L400F HARS, and (C) yeast treated with GCN2iB. Quantification of western blots of (D) eIF2α and (E) p-eIF2α. Growth curves of WT HARS and L400F yeast with and without GCN2iB supplementation at (F) 30°C or (G) 40°C, and quantification of the (H) doubling time and (I) area under the curve for growth curves at 30°C. Error bars represent one standard deviation from at least triplicate experiments. *P < .05, **P < .01, ***P < .001, ns = not significant.

To test if the ISR is activated in the HARS L400F-overexpressing cells, we blotted for eIF2α (subunit 1) and p-eIF2α in WT yeast (BY4742), WT HARS-expressing cells, and HARS L400F-expressing cells. We found that eIF2α levels were decreased by threefold in yeast expressing HARS-WT when compared to yeast expressing the endogenous HTS1 (BY4742). When comparing WT to L400F-expressing cells, we found a threefold increase of eIF2α abundance. When comparing the phosphorylation levels in the three yeast cell lines, we found that overall p-eIF2α were lowest in BY4742, with a significant twofold increase in HARS WT-expressing yeast compared to BY4742, and a further twofold increase in L400F expressing yeast compared to WT expressing yeast (Fig. 7B and E). It is interesting to note that the ISR is activated in the yeast expressing human HARS WT compared to endogenous yeast HTS1. This shows that though the HARS WT-expressing cells are viable, human HARS cannot completely compensate for the loss of yeast HTS1. Nonetheless, absolute phosphorylation of eIF2α was most pronounced in the HARS L400F yeast (Fig. 7E) and is accompanied by a growth defect. We next tested if inhibition of the GCN2 would lead to reduced eIF2α phosphorylation. In WT HARS-expressing cells no effect was observed, while the phosphorylation in L400F-expressing cells was significantly reduced, fully restoring WT HARS p-eIF2α levels (Fig. 7F). Overall, these data demonstrate that the ISR is activated in L400F and can effectively be inhibited by using a GCN2 inhibitor.

Subsequently, we investigated whether inhibition of the ISR by GCN2iB can rescue the yeast growth defect by measuring the growth phenotype at 30°C and 40°C. Excitingly, the increase in doubling time for L400F yeast that is apparent at 30°C is rescued by the GCN2 inhibitor, to levels indistinguishable from WT HARS-expressing cells (Fig. 7F and G). The area under the curve between GCN2iB-treated WT and L400F-expressing cells is also rescued (Fig. 7F and G). In WT HARS-expressing cells, the reduction of eIF2α phosphorylation at 30°C after GCN2 inhibition to background levels is notable but is not accompanied by a decrease in doubling time, indicating that ISR activation is always a growth rate limiting process (Fig. 7H and I). Interestingly, at 40°C, no rescue in growth was observed for L400F, and the WT HARS-expressing cells showed reduced growth, indicating that the ISR activation is required to adjust to heat stress in S. cerevisiae (Fig. 7G). Taken together, these data indicate that the activation of the ISR may exacerbate a yeast growth phenotype in L400F cells, which can be rescued by GCN2 inhibition at 30°C but not at 40°C.

Discussion

HARS L400F is a likely pathogenic variant associated with CMT2W

The diagnosis of a VUS leaves a patient without a diagnosis and often without the ability to access therapies and support. We previously established a humanized HARS S. cerevisiae model to assess if alleles described as VUS cause a growth phenotype in yeast and allow reclassification into likely benign of likely pathogenic. Human HARS complements a yeast hts1 deletion strain, albeit does not completely restore growth, which shows the limitation of humanized yeast models [8, 14]. Nonetheless, these yeast models and biochemical assays allow us to investigate the underlying disease-causing mechanisms behind pathogenic variants and to develop potential allele-specific therapeutic strategies [7, 14, 31]. We previously characterized multiple known pathogenic CMT2W-linked variants to delineate their distinct molecular consequences and identified therapies tailored to individual alleles [7, 15], including targeted histidine supplementation and tRNA therapeutics. Additionally, we identified a HARS VUS as pathogenic that could be rescued by histidine supplementation [7].

We here characterized a novel L400F variant found in a patient with hereditary peripheral neuropathy. The L400F protein is thermally unstable, as described for other HARS alleles, including dominant CMT-associated alleles V133F and V155G, as well as the recessive HARS syndrome-associated Y454S allele [14]. Interestingly, while V133F was rescued by tRNA supplementation and V155G by histidine supplementation, the L400F allele did not respond favorably to supplementation with either substrate. In line with the biochemical data, the L400F variant caused a growth defect in a humanized yeast model that was exacerbated at 40°C. HARS protein aggregation was observed for both the recombinant protein as well as in yeast. Interestingly, and contrastingly to our previously characterized variants, we found no evidence of mistranslation that we observed for other HARS alleles that were rescued by substrate supplementation [14, 15]. It is likely that the HARS L400F mutation causes HARS insufficiency, and that the phenotype can be attributed to aggregation of a sticky HARS L400F mutant, which is further supported by the upregulation of several chaperones. Other possible mechanisms underlying aaRS variant pathogenesis include tRNA sequestration [32], WT subunit poisoning [18], and mistranslation [7, 14], and dysregulated protein synthesis due to HARS insufficiency was previously suggested as the disease-causing mechanisms for some HARS variants [33].

Integrated stress response and GCN2 inhibition

The activation of the ISR has long been associated with and suggested as causative for disease progression in CMT disease [34] and other diseases such as cancer and age-associated diseases [35]. Amino acid starvation and the accumulation of uncharged tRNAs lead to ISR activation [36]. The ISR is activated in response to mutations in aaRS proteins. For example, pathogenic mutations in glutamyl-prolyl-tRNA synthetase cause altered tRNA binding or impaired enzyme function that elicit ISR, diminishing cell viability [37, 38]. In yeast models, mistranslation due to aaRS mutations was shown to chronically activate the ISR [39, 40].

Activation of the ISR was previously shown for several HARS alleles, including work in rat cells on HARS V155G, Y330C, and R137Q [33]. In mouse models for disease-associated GARS alleles, the ISR was activated due to impaired aminoacylation, and tRNA^Gly overexpression released ISR activation [41–43]. Since the growth defects in our HARS L400F model could not be rescued by substrate supplementation, we speculated that the ISR may be at least partially causative for the growth defect. Inhibition of the ISR was previously suggested as a therapeutic approach to aaRS-related synthetase mutations [43, 44]. Indeed, inhibition of GCN2 in our yeast model restored the growth defect in L400F yeast at 30°C and restored WT levels of eIF2α phosphorylation. The fact that tRNA supplementation further exacerbated the growth defect in yeast indicates that additional uncharged tRNA^His may further increase the pathogenic ISR activation. Our proteome data indicate an attenuated ISR activation, which has previously been observed in response to chronic stress [35, 45, 46]. Phosphorylated eIF2α inhibits the function of a guanine nucleotide exchange factor, eIF2B, thus down-regulates global protein synthesis while activating stress response pathways [47]. Nonetheless, we agree that the overall response in the proteome shows an attenuated stress response. In general, it is expected that the ribosomal proteins would be downregulated in response [48–50], which we found to be true in our proteomic sdata (Fig. 6). In contrast, amino acid biosynthesis is expected to be upregulated [51, 52], which we did not observe in our proteomics data. Our data showed that some amino acid biosynthesis proteins are downregulated, e.g. the arginine biosynthesis protein Arg1 (argininosuccinate synthetase) or Thr4 (threonine synthase), which catalyzes threonine biosynthesis. Most amino acid biosynthesis protein appear unchanged (see Supplemental file). We conclude that this results in an attenuated stress response, where amino acid biosynthesis is not increased. A differential activation of the ISR and incomplete pathway activation was described previously and is a common adaptation in response to chronic stress [35, 45, 46].

Pharmacological inhibition of GCN2 was previously suggested as a potential therapeutic approach for certain cancers [53]. Several GCN2 inhibitors have been tested and reviewed for specificity and efficacy in ISR inactivation [54]. GCN2iB, the inhibitor used in this study, showed high efficacy in mouse models of certain cancer cell types, like androgen-sensitive and castration-resistant mouse models of prostate cancer [55], and was effective in combination therapies [42, 56]. Other research suggests that low-dose GCN2 inhibition can have an activating effect on the ISR [57], cautioning that more research is required before employing GCN2 inhibitors as chemotherapeutics [44]. Nonetheless, these findings suggest that targeting the ISR with GCN2 inhibitors may be a worthwhile therapeutic approach in the future for patients harbouring aaRS variants that trigger stress signalling pathways and do not respond to amino acid supplementation.

Supplementary data

Supplementary data is available at NAR Molecular Medicine online.

Conflict of interest

None declared.

Funding

Ontario Ministry of Research and Innovation (ER18-14-183), Natural Sciences and Engineering Research Council of Canada (04776, 580214), Canadian Institute for Health Research (185961), and Rare Disease Models and Mechanism Network.

Data availability

The data underlying this article are available in the article and in its online supplementary material. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD067079 and 10.6019/PXD067079.