Lack of Effect from Genetic Deletion of Hdac6 in a humanized mouse model of CMT2D

Abstract

Background:

Inhibition of HDAC6 has been proposed as a broadly applicable therapeutic strategy for Charcot-Marie-Tooth disease (CMT). Inhibition of HDAC6 increases the acetylation of proteins important in axonal trafficking, such as alpha-tubulin and Miro, and has been shown to be efficacious in several preclinical studies using mouse models of CMT.

Aims:

Here we sought to expand on previous preclinical studies by testing the effect of genetic deletion of Hdac6 on mice carrying a humanized knockin allele of Gars1, a model of CMT type 2D.

Methods:

Gars1^ΔETAQ mice were bred to an Hdac6 knockout strain, and the resulting offspring were evaluated for clinically relevant outcomes.

Results:

The genetic deletion of Hdac6 increased alpha-tubulin acetylation in the sciatic nerves of both wild type and Gars1^ΔETAQ mice. However, when tested at five-weeks-of-age, the Gars1^ΔETAQ mice lacking Hdac6 showed no changes in body weight, muscle atrophy, grip strength or endurance, sciatic motor nerve conduction velocity, compound muscle action potential amplitude, or peripheral nerve histopathology compared to Gars1^ΔETAQ mice with intact Hdac6.

Interpretation:

Our results differ from those of two previous studies that demonstrated benefit of the HDAC6 inhibitor Tubastatin A in mouse models of CMT2D. While we cannot fully explain the different outcomes, our results offer a counterexample to the benefit of inhibiting HDAC6 in CMT2D, suggesting additional research is necessary.

Introduction:

Charcot-Marie-Tooth disease (CMT) comprises a diverse collection of hereditary sensory and motor neuropathies (Saporta and Shy, 2013). Cumulatively, CMT affects approximately 1 in 2500 people, but mutations in at least 100 different genes in the human genome can cause CMT (Laura, et al., 2019; Skre, 1974; Timmerman, et al., 2014). Therefore, after a few common subtypes are accounted for, such as CMT1A, CMT2A, and CMT1X, individual subtypes become quite rare (DiVincenzo, et al., 2014). In addition to this genetic diversity, there are also differences in the pathophysiology and clinical diagnostic distinctions. Some forms of CMT primarily affect peripheral myelinating Schwann cells (type 1, or demyelinating CMTs), leading to reduced nerve conduction velocities, whereas other forms lead to degeneration of the peripheral sensory and motor axons themselves (type 2, or axonal CMTs), often due to specific deficits in neuronal cell biology. Nonetheless, CMT patients have shared clinical issues, including weakness, fatigue, and muscle atrophy, as well as sensory deficits, which are typically most pronounced in the distal extremities, indicating a length dependent neuropathy (Saporta and Shy, 2013).

The 100+ genes associated with CMT encode proteins involved in numerous cellular activities, such as mitochondrial function, axonal transport, endosome/lysosome trafficking, and protein synthesis (Fridman, et al., 2015). These diverse cellular functions likely indicate that there are many paths to peripheral axon dysfunction and degeneration, which should not be surprising given the extraordinary specialization of these very long cell processes. This also calls into question whether there will be a single therapy that is effective across all or many CMT subtypes. However, even if therapies are not addressing the core function of the mutated gene, there may be generally beneficial approaches that could apply to multiple forms of CMT. Such broadly beneficial approaches include the expression of neurotrophins in the periphery to maintain axon integrity (Sahenk and Ozes, 2020), the inhibition of SARM1 to prevent axon degeneration (Moss and Hoke, 2020), or the inhibition of HDAC6 to enhance alpha-tubulin acetylation and axonal transport (Rossaert and Van Den Bosch, 2020).

HDAC6 is an atypical deacetylase. The protein contains two catalytic domains and is localized primarily in the cytoplasm where it modifies proteins like alpha-tubulin and MIRO, an adapter protein for mitochondrial transport (Grozinger, et al., 1999; Hubbert, et al., 2002; Kalinski, et al., 2019; Verdel and Khochbin, 1999). The rationale for targeting HDAC6 in diseases such as CMT is that as axons become compromised, tubulin acetylation decreases. This impairs axonal transport, compromising the distal axon even more and setting up a positive feedback loop. Inhibiting HDAC6 decreases tubulin deacetylation (i.e. increases tubulin acetylation) and therefore slows this process. Improving the acetylation of MIRO could be similarly beneficial, improving mitochondrial transport in axons (Kalinski, et al., 2019). Thus, even though HDAC6 inhibitors are likely targeting a downstream consequence of diseases such as CMT, there is a basis for their mechanism of action in CMT and possibly other acquired neuropathies, such as diabetic neuropathy, for which a clinical trial has been performed (NCT03176472, clinicaltrials.gov), or even diseases such as ALS.

Inhibition of HDAC6 has been shown to be beneficial in several mouse models of CMT, including Hspb1/CMT2F, Mfn2/CMT2A, PMP22/CMT1A and Gars1/CMT2D (Benoy, et al., 2018; d’Ydewalle, et al., 2011; Ha, et al., 2020; Mo, et al., 2018; Picci, et al., 2020). In studies of Mfn2/CMT2A mice, similar results were obtained with pharmacological inhibition and genetic deletion of Hdac6 (Picci, et al., 2020). Here we focus on Gars1/CMT2D, which has been the subject of more studies than the rest. Dominant mutations in glycyl-tRNA synthetase (GARS1) cause peripheral neuropathy in both humans and mice (Antonellis, et al., 2003; Seburn, et al., 2006). Interestingly, dominant mutations in at least 5 other tRNA synthetase genes also cause forms of CMT (He, et al., 2023; Jordanova, et al., 2006; Latour, et al., 2010; Tsai, et al., 2017; Vester, et al., 2013), making the tRNA synthetases the largest gene family associated with forms of CMT and suggesting both a shared pathogenic mechanism and the possibility of shared benefit from a treatment such as HDAC6 inhibition. Two independent studies showed that treating mice with the HDAC6 inhibitor Tubastatin A mitigated the neuropathy phenotype in two mouse-specific Gars1 mutations that cause neuropathy (Benoy, et al., 2018; Mo, et al., 2018). In addition, Mo et al. showed an aberrant interaction between mutant forms of GARS1 protein and HDAC6, possibly providing a more direct molecular mechanism for this therapeutic approach (Mo, et al., 2018). Furthermore, using motor neurons derived from induced pluripotent stem cells engineered to carry a dominant GARS1 mutation, deficits in axonal transport were corrected by treatment with an HDAC6 inhibitor (Smith, et al., 2022). Taken together, these preclinical studies are encouraging regarding the potential for HDAC6 inhibitors to show benefit in CMT2D. However, here we re-examined the potential efficacy of HDAC6 inhibitors using a different strategy in which we crossed Hdac6 knockout mice to a humanized mouse model of CMT2D that carries a severe, early-onset patient allele of GARS1, the ΔETAQ allele (Morelli, et al., 2019). We find that genetic deletion of Hdac6 in the Gars1^ΔETAQ background does not change the onset or severity of neuropathy, offering a counterexample to previous studies.

Materials and Methods:

Mice:

All animal experiments were conducted in compliance with the Guide on the Care and Use of Laboratory Animals and were reviewed and approved by the Institutional Animal Care and Use Committee of The Jackson Laboratory. The Hdac6 knockout strain was produced at JAX and is officially designated C57BL/6J-Hdac6^em2Lutzy/J, strain #029318, MGI: 6118965. This allele deletes exons 2–4, with exon 2 containing the translation initiation codon. In internal validation studies, the mice were shown to not produce protein by western blot analysis of multiple tissues, including brain, liver, spleen, and cardiac and skeletal muscle. This result has been independently reproduced in at least three additional publications (Osseni, et al., 2020; Wang, et al., 2020a; Wang, et al., 2020b). The Gars1^ΔETAQ mice have been previously described (Morelli, et al., 2019), and are officially designated Gars1^em2Rwb, MGI: 6393282. Mice were maintained under standard housing conditions with food and water provided ad libitum and a 14:10 light:dark cycle.

Immunoassay:

Sciatic nerves were collected from 5-week-old mice and snap frozen. Samples were then thawed on ice, rinsed with cold PBS, and homogenized in 200 μL of lysis buffer consisting of chilled T-Per (Thermo Scientific 78510) containing Halt protease and phosphatase inhibitor cocktail (1X, Thermo Scientific 78442) and deacetylase inhibitor cocktail (1X, APExBIO K1017) in 0.5mL Dounce homogenizers. Once homogenized, lysate was transferred to an Eppendorf tube and centrifuged at max speed at 4°C for 15 minutes. Supernatant was kept and used for BCA Assay per the kit protocol (Pierce BCA Protein Assay Kit, Thermo Scientific PI23227).

Capillary immunoassays were performed using the Jess Simple Western system following the manufacturer’s protocols. Samples were prepared at 0.1 mg/ml final concentration and run on 12–230 kDa Separation Modules (Bio-Techne SM-W004). Detection was performed using the Anti-Rabbit Detection Module (Bio-Techne DM-001) and Total Protein Detection Module (Bio-Techne DM-TP01). Antibodies used were rabbit anti-alpha-tubulin (Cell Signaling Technology #2125; 1:1000), rabbit anti-acetyl-alpha-tubulin (Cell Signaling Technology #5335; 1:5000), rabbit anti-GAPDH (Cell Signaling Technology #2118; 1:1000) and the system control primary antibody (rabbit, Bio-Techne 042–196). Analysis was based on the manufacturer’s protocol, and in pilot experiments GAPDH normalization yielded similar and more consistent results than total protein, so all results presented here use GAPDH normalization. Ratios of protein concentration were then calculated using Microsoft Excel and data were plotted for visualization using GraphPad Prism.

Wire hang:

As a behavioral measure of grip strength and endurance, we utilized the inverted wire hang test (Gomez, et al., 1997; Rafael, et al., 2000; Spaulding, et al., 2016). In this assay, mice were placed on an inverted wire grid for a maximum trial time of one minute, and latency for mice to fall was measured. Mice were tested in three successive trials with a 30- to 60-second break in between trials and the average time-to-fall was reported. Mice were not trained prior to the initial test session.

Neurophysiology:

Nerve conduction velocity of motor axons of the sciatic nerve was measured (Burgess, et al., 2010; Morelli, et al., 2017). Mice were anaesthetized with 2% isoflurane and placed on a thermostatically regulated heating pad to maintain body temperature. Action potentials were produced by placing stimuli proximally at the sciatic notch or distally at the ankle. The latencies to elicit a compound muscle action potential (CMAP) recorded in the thenar muscles of the hind paw following proximal and distal stimuli were determined. NCV was calculated as [conduction distance/(proximal latency-distal latency)]. Amplitude of the CMAP was also measured.

Histopathology:

Nerve histology was performed as previously described (Burgess, et al., 2010). In brief, the femoral motor and sensory nerves were fixed in place using a drop of 2% paraformaldehyde, 2% glutaraldehyde in a 0.1M cacodylate buffer. The nerves were then dissected free, post-fixed overnight in the same solution and rinsed into PBS. Samples were dehydrated and plastic embedded for sectioning. Thick sections (0.5 μm) were placed on slides and stained with toluidine blue for light microscopy. Images were collected on a Nikon Eclipse 600 microscope with a 40X objective. Axon number and size were quantified using ImageJ.

Statistics:

Statistical tests were performed using GraphPad Prism software. Comparisons across four genotypes were tested using an ordinary one-way ANOVA with multiple comparisons and lack of significance was confirmed in pairwise comparisons using a Student’s t test. Differences in the cumulative frequency distribution of axon areas were tested using nested one-way ANOVA with multiple comparisons and lack of significance was confirmed in pairwise comparisons using a nested t test. Differences were considered significant with a p value < 0.05.

Results:

The Hdac6 gene is on the X chromosome in mice. Therefore, we bred a Gars1^ΔETAQ male to a heterozygous Hdac6 knockout (Hdac6^KO/+) female. The Hdac6^ko mice carry a deletion of exons 2–4 of the Hdac6 gene and do not produce detectable protein, as shown previously (Osseni, et al., 2020; Wang, et al., 2020a; Wang, et al., 2020b). We analyzed male offspring of this cross, half of which carried the dominant Gars1 mutation and half of which carried the Hdac6 knockout allele. The Gars1^ΔETAQ mice have an early onset peripheral neuropathy, with changes evident by a few weeks of age (Morelli, et al., 2019). We therefore performed all analyses at 5 weeks of age. To evaluate the effect of the Hdac6 knockout on alpha-tubulin acetylation, we used a digital capillary western blot to compare total alpha-tubulin and acetylated alpha-tubulin in the sciatic nerves of mice of all four genotypes [1) WT;WT, 2) WT;Hdac6^KO, 3) Gars1^ΔETAQ;WT, and 4) Gars1^ΔETAQ;Hdac6^KO] using GAPDH as an internal standard (Figure 1). In both wild type and Gars1^ΔETAQ backgrounds, the loss of Hdac6 led to a significant (~2 fold) increase in acetylated alpha-tubulin. This was anticipated and indicates that our genetic approach does indeed increase the acetylation of HDAC6 targets in peripheral nerve, although we do not understand the increased variability in tubulin acetylation levels in the Hdac6^KO background compared to the tightly consistent levels seen in wild-type and Gars1^ΔETAQ mice. The Gars1^ΔETAQ samples did not show a significant decrease in acetylated alpha-tubulin compared to wild-type when tested by ANOVA including all four genotypes, despite this result being shown previously (Benoy, et al., 2018; Mo, et al., 2018; Smith, et al., 2022). However, using a t test in a pairwise comparison of Gars1^ΔETAQ;WT vs WT;WT yielded a statistically significant result (16.5 ± 4.3% decrease in Gars1^ΔETAQ, p=0.003). A stronger decrease in alpha-tubulin acetylation the Gars1^ΔETAQ mice may have been found if we had examined older mice or more distal nerves than the sciatic. Previous studies suggest that GARS1 protein levels are unchanged or perhaps slightly elevated in the dominant mutant alleles (see for example (Motley, et al., 2011)); however, here we did not confirm whether GARS1 levels change in the absence of Hdac6.

Figure 1:

Digital western blotting of alpha tubulin and acetylated alpha-tubulin. Sciatic nerve samples from mice of the indicated genotypes (n=6 per group) were analyzed by digital western for the ratio of acetylated to total tubulin, normalized to GAPDH. Data were analyzed by ordinary one-way ANOVA with Tukey’s multiple comparisons test. The more sensitive t-test was also applied to the pairwise comparison Gars1^ΔETAQ;WT vs WT;WT, yielding a statistically significant result (p=0.003).

Given that our analyses include four genotypes, we used one-way ANOVA to test for significance, as in the results presented in figure 1. As described above, however, the more sensitive t-test was needed to detect a significant decrease in tubulin acetylation in Gars1^ΔETAQ sciatic nerve. In all results presented in figures 2–4, the Gars1^+/+ and Gars1^ΔETAQ mice differed significantly when analyzed by ANOVA, but the Hdac6 genotype had no effect in either the Gars1^+/+ or Gars1^ΔETAQ mice. To perform a more targeted pairwise comparison, we also used a t test to ask whether Gars1^ΔETAQ mice differed with and without the Hdac6^KO allele. Nested ANOVA and t test were used for the axon area distributions in figure 4C. We did not find differences between Gars1^ΔETAQ and Gars1^ΔETAQ without Hdac6, even using the pairwise t tests.

Figure 2:

Neuromuscular phenotype of Gars1;Hdac6 mice. A) Body weight, B) muscle weight to body weight ratio (a measure of muscle atrophy), and C) motor performance as assayed by the average latency to fall from an inverted wire grid. For all measures, Gars1^ΔETAQ mice had values that were significantly less than those of Gars1^WT mice, regardless of Hdac6 genotype. In contrast, genetic ablation of Hdac6 had no effect in Gars1^ΔETAQ on any measure, even when the more sensitive t test was used to compare Gars1^ΔETAQ mice with and without Hdac6. Data are shown as mean ± S.D. with points representing individual mice with the following numbers per group: WT=10, Hdac6^KO=11, Gars1^ΔETAQ=15 and Gars1^ΔETAQ;Hdac6^KO=17.

Figure 4:

Femoral motor nerve histopathology of Gars1;Hdac6 mice. A) Images of cross-sections of femoral motor nerve, showing the smaller overall nerve area of Gars1^ΔETAQ mice, regardless of Hdac6 genotype. B) Number of myelinated axons in the motor branch of the femoral nerve. The Gars1^ΔETAQ mutation causes a decrease in the number of axons, and Hdac6 ablation has no effect. C) Area of myelinated axons in the motor branch of the femoral nerve, shown by the cumulative frequency distribution. The Gars1^ΔETAQ mutation causes a decrease in axon area, and Hdac6 ablation has no effect. Data in (B) are shown as mean ± S.D. with points representing individual mice with the following numbers per group: WT=10, Hdac6^KO=11, Gars1^ΔETAQ=15 and Gars1^ΔETAQ;Hdac6^KO=17. Scale bar in (A) = 50μm for all images.

The Gars1^ΔETAQ mice have a reduced body weight by 5 weeks-of-age, and this was not affected by the loss of Hdac6 (Figure 2A). The reduced body weight is, at least in part, attributed to muscle atrophy. This is reflected in a reduction in the ratio of the weight of the triceps surae muscles to total body weight (MW:BW ratio). This measure of muscle atrophy also was not altered by the loss of Hdac6 (Figure 2B). We also used the wire hang test to assess grip strength and endurance. The Gars1^ΔETAQ perform very poorly in this test and hang on less than 20 seconds before falling, whereas wild type mice typically hang on for the full one-minute duration of the test. The deletion of Hdac6 did not improve the performance of the Gars1^ΔETAQ mice in this task (Figure 2C).

We also assessed the effects of deleting Hdac6 using neurophysiology. The sciatic motor nerve conduction velocity is markedly reduced in the Gars1^ΔETAQ mice, as we have seen in other Gars1/CMT2D mouse models (Achilli, et al., 2009; Seburn, et al., 2006). The genetic deletion of Hdac6 did not improve nerve conduction velocity in the Gars1^ΔETAQ mice (Figure 3A). In contrast, other treatments such as inhibiting GCN2 and the integrated stress response or allele-specific knockdown of mutant Gars1 mRNA using gene therapy approaches were able to significantly improve this outcome (Morelli, et al., 2019; Spaulding, et al., 2021). The amplitude of the compound muscle action potential in the thenar muscles of the foot following stimulation of the nerve at the ankle is also reduced in the Gars1^ΔETAQ mice, consistent with denervation and/or failures in synaptic transmission at the neuromuscular junction. Deletion of Hdac6 had no effect on CMAP amplitude (Figure 3B).

Figure 3:

Neurophysiology of Gars1;Hdac6 mice. A) Sciatic motor nerve conduction velocity and B) compound muscle action potential amplitude were both decreased in Gars1^ΔETAQ mice compared to Gars1^WT, regardless of Hdac6 genotype. As in Figure 2, genetic ablation of Hdac6 had no effect in Gars1^ΔETAQ on either measure, even when the more sensitive t test was used. Data are shown as mean ± S.D. with points representing individual mice with the following numbers per group : WT=10, Hdac6^KO=11, Gars1^ΔETAQ=15 and Gars1^ΔETAQ;Hdac6^KO=17.

Finally, we examined peripheral nerves by histopathology. The motor branch of the femoral nerve, which innervates the quadriceps muscles of the thigh, shows loss of myelinated axons in the Gars1^ΔETAQ mice, and the remaining axons are smaller. Neither axon number nor axon size was improved with the deletion of Hdac6 (Figure 4A–C).

Discussion:

In this study, we revisited the potential benefit of inhibiting HDAC6 as a treatment for CMT2D, caused by dominant GARS1 mutations. We used an approach that differs from previous studies in two major ways. First, we used genetic deletion of Hdac6 on the X chromosome of mice instead of using a pharmacological inhibitor such as Tubastatin A. Second, we used a mouse model of CMT2D carrying a small internal deletion in exon 8 of GARS1 that was identified in a patient with severe early onset motor neuropathy (Morelli, et al., 2019). Previous studies used alleles of mouse Gars1 that cause peripheral neuropathy, but the amino acids changed (P278KY and C201R) are different than known human disease variants (Benoy, et al., 2018; Mo, et al., 2018). However, in other studies, the Gars1^ΔETAQ mice have behaved very similarly to other alleles (Morelli, et al., 2019; Spaulding, et al., 2021). Unlike previous studies (Benoy, et al., 2018; Mo, et al., 2018), we found no benefit from eliminating Hdac6 in the Gars1^ΔETAQ background by any of our outcome measures (body weight, muscle atrophy, motor performance, sciatic motor nerve conduction velocity, compound muscle action potential amplitude, and peripheral nerve histopathology, figures 2–4) despite robust elevation of acetylation of alpha-tubulin in peripheral nerves following Hdac6 deletion (figure 1). Below, we will discuss differences in these experiments and possible interpretations.

In the study by Mo et al., Gars1^P278KY (also called P234KY due to differences caused by the inclusion/exclusion of a mitochondrial targeting sequence in the amino acid count) were used (Mo, et al., 2018). These mice are similar to, but slightly more severe than, the Gars1^ΔETAQ mice. Decreases in alpha-tubulin acetylation (~3-fold) were seen as early as one week postnatally in peripheral nerves of Gars1^P278KY mice, while we saw smaller decreases in alpha-tubulin acetylation in the Gars1^ΔETAQ mice compared to wild-type controls at 5 weeks of age. This may reflect a more pronounced change in tubulin acetylation in the P278KY allele, and we may have found more pronounced changes in acetylation if we have aged the ΔETAQ allele longer, or had looked in more distal nerves, as seen in Mfn2 mice modeling CMT2A, for example (Picci, et al., 2020). In Mo et al., axonal transport of quantum-dot labeled Nerve Growth Factor (NGF) was decreased in cultured DRG neurons from the Gars1^P278KY mice, and this was corrected by treatment with the HDAC6 inhibitor Tubastatin A (Mo, et al., 2018). Similar results for mitochondrial transport have been reported in human iPSC-derived neurons engineered to carry the same mutation (Smith, et al., 2022). Additionally, retrograde axonal transport of signaling endosomes labeled with a fluorescent tetanus toxin heavy chain is restored in vivo in Gars1/CMT2D mice by other methods such as treatment with the neurotrophin BDNF (Sleigh, et al., 2023). We did not evaluate axonal transport in this study, and while the improvements in axonal transport observed in multiple previous studies are encouraging (Mo, et al., 2018; Sleigh, et al., 2023; Smith, et al., 2022), improved axonal transport has yet to be causally tied to improved clinically relevant outcomes. In in vivo studies, treatment of the Gars1^P278KY mice with Tubastatin A starting at 35 days of age and lasting 2 weeks led to a modest improvement in hind limb clasping, rotarod performance, and gait parameters, but the mice still performed well below wild type littermates (Mo, et al., 2018). When the same Gars1^P278KY mice are bred to a Gcn2 knockout strain, the double mutants perform comparably to littermate controls (Spaulding, et al., 2021), so the effects of Tubastatin A are comparatively modest, but Tubastatin A treatment was started after the onset of a phenotype and only lasted two weeks, so the comparison is complicated.

The study by Benoy et al. reached similar conclusions to Mo et al. The Benoy et al. study used Gars1^C201R mice, which have a milder phenotype than either ΔETAQ or P278KY (Benoy, et al., 2018). These mice also showed decreased alpha-tubulin acetylation selectively in sciatic nerve, but this was assayed in one year old animals, consistent with the idea that here we may have seen greater changes in tubulin acetylation in the ΔETAQ mice in older animals. Decreased axonal transport of mitochondria in cultured DRG neurons was again observed, and was corrected by Tubastatin A. In in vivo studies, the Gars1^C201R mice were treated with Tubastatin A at 4 months of age for 40 days. This led to an improvement in grip strength, performance in a turning grid test, CMAP amplitude, sensory nerve action potential (SNAP) amplitude and innervation at the neuromuscular junction. Rotarod performance and performance in the wire hang test were not significantly improved. The values for Gars1^C201R mice after Tubastatin A treatment are still well below those of control mice reported previously in the same study, so like Mo et al., the effect size was modest.

In our study, we saw no improvement with the genetic deletion of Hdac6. Our studies were well-powered, with 15 Gars1^ΔETAQ mice compared to 17 double Gars1^ΔETAQ:Hdac6^KO mutants, for example. Our outcome measures were largely quantitative values that did not depend on subjective judgement from the tester and these outcomes have been responsive to treatment in other preclinical studies (Morelli, et al., 2019; Spaulding, et al., 2021). The tester was not technically blinded to genotype during data collection or analysis, but this lack of effect was also not the anticipated result when these studies were initiated, given the previous results with Tubastatin A in other mouse models of CMT2D. Some of our tests were the same as those used by Benoy et al., including wire hang (no improvement in either study) and CMAP amplitude (improved in Benoy et al., but not here). Similarly, Benoy et al. did not see an improvement in rotarod performance, whereas Mo et al. reported an improvement. Therefore, previous findings were also not completely consistent.

A criticism of genetic approaches such as the deletion of Hdac6 used here instead of pharmacological approaches is that the intervention is present from conception and thus is effectively started before disease onset, which is less clinically relevant. This is true, but should have worked in our favor, and instead, we saw no improvement in the Gars1^ΔETAQ mice with the deletion of Hdac6. An additional criticism of genetic approaches is the possibility of compensation by related genes. As HDAC6 is an atypical deacetylase, this is less likely, and we found elevated levels of tubulin acetylation in the sciatic nerves of Hdac6^KO mice (figure 1), indicating that any compensation by other deacetylases is incomplete. The only other type IIb HDAC is HDAC10, making it the most likely to compensate, and it is indeed expressed in alpha motor neurons in the mouse spinal cord, albeit at an mRNA level roughly half that of Hdca6 (https://seqseek.ninds.nih.gov/spinalcordinjury) (Alkaslasi, et al., 2021). In contrast, pharmacological inhibitors carry the possible caveat of off-target effects. Whether the benefits of Tubastatin A in CMT2D mice are due to lack of specificity for HDAC6 is unknown, although Tubastatin A has been shown to also significantly inhibit HDAC10 as well as sirtuin deacetylases, any of which may be partially redundant with and compensatory for HDAC6 (Choi, et al., 2019; Ptacek, et al., 2023). Additional tests with other HDAC6 inhibitors may clarify this point. Alternatively, inhibitors could be tested in an Hdac6^KO background, where they would be anticipated to have no additional effect, to establish their specificity.

Thus, it is unclear why our genetic deletion model did not reproduce the effects reported in previous studies of Gars1/CMT2D mice that used HDAC6 inhibitors. Perhaps the effects of Hdac6 deletion would have been more pronounced on disease progression and we would have observed differences if we had aged the animals longer, but Mo et al. reported effects after just two weeks of treatment and Benoy et al. after 40 days. Our results do clearly indicate that the reported interaction between mutant forms of GARS1 protein and HDAC6 is not a causal part of the disease mechanism, as the complete removal of Hdac6 did not alter the disease course at our five-week time point. Our results offer a counterexample to previous studies on the efficacy of inhibiting HDAC6 as a treatment for CMT2D and suggest that additional research is necessary to validate HDAC6 as a therapeutic target for this disease and to clarify the magnitude of the anticipated benefit.

Data availability:

All data relevant to this study are presented in the paper. Raw data are available upon reasonable request by contacting the authors.