Dystrophin deficiency alters the muscle NAD metabolome and energy producing pathways
To assess the biochemical adaptations of muscle to dystrophin deficiency, we analyzed the gastrocnemius muscles of MDX mice before and after eccentric challenge using an untargeted metabolic profiling platform. The platform utilized a combination of GC/MS and LC/MS in positive and negative ion mode to identify 762 chemical entities, of which 552 were annotated as either polar metabolites or lipids. Principal component analysis of the metabolic profiles demonstrated a clear clustering of samples by time and genotype (Figure 1A). Furthermore, the MDX samples showed a well-defined chronological progression following damage. The clustering of samples observed 2 hours post-damage showed distinct separation from the other time points, suggesting a period of acute crisis in the muscles. This was the only time at which the majority of significantly altered metabolites appeared depleted when compared to unchallenged controls (Figure 1B), potentially indicating rapid degradation, release from the tissue, or a synthetic bottleneck. The pattern of the post-injury sample groups from days 2-7 reflected a transition from the acute response to a repair and recovery stage. By post-injury day 14, the samples had a chemical phenotype closely resembling that of the uninjured MDX muscles (Figure 1A-B).
Comparing the metabolic profiles of MDX mice to wildtype (WT) mice at baseline, we identified several biochemical pathways with disproportionate impact on the muscle. Among the highest confidence pathways (p<10^-5 and Pathway Impact> 0.5) were those relating to the biosynthesis and metabolism of amino acids, such as arginine, phenylalanine, tyrosine, and tryptophan, as well as that of nicotinate and nicotinamide metabolism (Figure 1C). Comparing the Pathway Impact scores from the acutely injured (2 hr) to uninjured MDX muscle, we were surprised to find glycolysis as the most responsive pathway (Figure 1D). Additional pathways relating to energy production or storage, including those of fatty acid synthesis, pyruvate metabolism, and the TCA cycle, were also injury-responsive.
Upon examining the specific metabolites altered at baseline in the MDX muscle, we found a strikingly lower abundance of NAD than almost any other metabolite (Figure 1E,F). Consistent with the pathway analysis, several other nicotinamide-containing metabolites and glycolytic intermediates also showed highly variable abundance. Primary NAD deficiency in mouse muscle and cultured myotubes has been shown to restrict glycolytic flux at the level of GAPDH, resulting in a characteristic buildup of intermediates in the pentose phosphate pathway (8,18). Consistent with this model, we observed a significant increase in ion counts for glucose-6-phosphate (G6P), dihydroxyacetone phosphate (DHAP), ribose-5-phosphate, and a positive trend in sedoheptulose-7-phosphate (S7P) in MDX muscle (Figure 1E,G). Additionally, we noted an increase in several poly-cationic species of polyamines, known to be derived from arginine (Figure 1E,H). Collectively, this pattern indicates a metabolic shift in MDX muscle at steady state, partially stemming from the loss of the metabolic co-factor, NAD.
The robust response of the global metabolome at 2 hours post-eccentric challenge led us to investigate NAD-related metabolites at this and subsequent time points. As far as the platform could resolve, NAD itself did not appear to respond to the challenge (Figure 1F). However, Nam was acutely depleted by more than one third after two hours, presumably restricting any residual activity of the NAD salvage pathway, and gradually returned to baseline after four days. Nam homeostasis was further altered by a doubling in the levels of 1-me-Nam in the two days following injury, which only normalized after 14 days. Evidence of a further constriction in glycolysis also emerged post-injury: levels of DHAP, immediately upstream of GAPDH, acutely increased in an opposing pattern to that of lactate, a glycolytic end-product (Figure 1G). The most striking indicator of the repair phase was the appearance of polyamines, including spermine, spermidine, putrescine, and N-acetylputrescine, which were elevated in the days following injury (Figure 1H and Supplemental Table 1). This class of biomolecules serves as a general marker of cellular proliferation and is required for both myocyte differentiation and alternative macrophage activation (19,20). Consistently, in the case of NAD-related metabolites, glycolytic intermediates, and polyamines, eccentric injury amplified the disparities between MDX and WT muscle.
A novel synthetic CD38 antagonist increases NAD in multiple tissues
Our group previously reported a series of novel chemical entities, which potently inhibit the constitutive NAD-degrading enzyme, CD38. Related screening efforts yielded the imidazoquinoline dubbed GSK978A, which exhibited 10-fold higher potency against mouse recombinant CD38 than human enzyme (Figure 2A). This potency is orders of magnitude greater than that of natural products, such as quercetin, and approximates that of “78c”, the best-studied synthetic CD38 inhibitor to-date (11,14). Yet GSK978A was more soluble and outperformed 78c in a chromosomal stability test of genotoxicity, indicating improved suitability for long-term administration (data not shown). GSK978A was also predicted to have low intrinsic clearance in several small animal preclinical species, including mice and rats, but not larger cynomolgus monkeys (Figure 2B). As literature suggested a role for CD38 in the prevention of diet-induced obesity (21), the drug metabolism and pharmacokinetic characterization of the quinoline series was originally performed in obese WT mice. To confirm the slow clearance kinetics in vivo, we administered a single intermediate oral dose of 10 mg/kg and found the compound still detectable in the blood after 24 hours (Figure 2C). We next performed a pharmacokinetic analysis of tissues sampled 2 hours after oral doses from 1-30 mg/kg. At these doses, the compound was identified within liver, gastrocnemius, adipose, and brain tissues at exposures that well exceeded the IC50 of ~12 ng/mL (Figure 2D). Accordingly, the NAD content was found to be significantly elevated in the liver, muscle, and brain when normalized to an internal analytical standard (Figure 2E). Despite high exposure, the NAD recovery from adipose was low and NAD changes were not significant. However, in most tissues, peak NAD elevation of at least 30% was achieved at a dose of 3 mg/kg (Figure 2F).
To assess the potential pharmacodynamics of GSK978A in an eccentric challenge model, an acute study was performed in MDX mice following 5 days of dosing at 3 mg/kg. Dietary NR, which has been suggested to improve the performance of MDX muscle (10), was included as a comparator. Within 24 hours of eccentric challenge, tetanic strength was lessened by >50% in all treatment groups, despite preservation of mass in the largest affected gastrocnemius muscles (Figure 3A-C). Interestingly, the challenged muscles also showed NAD depletion compared to the contralateral side, indicating that the muscle NAD pool does acutely respond to lengthening contractions (Figure 3D). Mice treated with GSK978A, but not NR, showed a trend toward protection from this effect, though it could not be attributed to specific NAD elevation in either limb (Figure 3E). As a biomarker of muscle repair, total muscle polyamines showed clear elevation in the injured limbs with a trend toward protection by GSK978A, especially when polyamines were normalized to NAD content (Figure 3F-H). These results suggested that a longer treatment regimen might be necessary to provide functional improvements to MDX mice.
Chronic NAD repletion does not provide functional protection from repetitive eccentric challenges
We next designed a long-term study with chronic administration of GSK978A or NR to longitudinally assess the physiology of MDX mice during three stages distinct stages: growth, recovery from an eccentric challenge, and recovery from a repeated challenge (Figure 4A). We reasoned that this design would model the efficacy requirements of boys diagnosed with DMD. Beginning at 7-9 weeks of age, during a period of rapid growth and peak muscle necrosis (22), MDX mice were randomized by hindlimb contractility and body weight, with a WT group included as a positive control for recovery. Over the course of 20 weeks, hindlimb weakness and increased CK release persisted in the MDX mice compared to WT controls (Figure 4B-C). The pattern of hindlimb strength and serum creatine kinase (CK) release generally trended downward as MDX animals reached maturity, but remained unchanged in both compound-treated groups, compared to the vehicle-treated controls. MDX mice also accumulated lean mass steadily over the course of the study, reflecting characteristic hypertrophy, in a manner that was treatment-independent (Figure 4D). At the study conclusion, gastrocnemius muscles from the challenged MDX limbs were found to be ~15% less massive than the contralateral side in all treatment groups, reflecting an inability to fully regenerate injured fibers that was not observed in the WT. Surprisingly, contralateral muscles tended to be largest in mice treated with GSK978A (Figure 4E-F), which may be a consequence of altered gait mechanics to favor the contralateral side, as it was not reflected in total lean mass.
Hindlimb strength was also serially assessed following eccentric challenges beginning at week 10 of treatment, to determine whether treated groups were protected from injury or recovered faster. Compared to WT controls, MDX mice showed approximately double the functional deficit within one day of eccentric challenge, despite similarly shaped tetani, but neither parameter was affected by GSK978A or NR (Figure 5A-B). Furthermore, MDX hindlimbs showed highly similar recovery kinetics between the first and second challenges, while WT controls rebounded faster after the second bout (Figure 5C-D). The absence of a protective repeated-bout effect in MDX hindlimbs may reflect that fact that adult dystrophin-deficient muscles are preconditioned to such cycles of damage and repair by activities of normal living.
CD38 inhibition significantly reverts the MDX muscle metabolome to the WT state
Given the lack of physiological protection conferred upon MDX muscle by NAD-modulating compounds, we suspected that the treatments were simply ineffective at correcting the basal metabolic imbalance that formed the basis for our rationale. To address this question, we analyzed the uninjured gastrocnemius muscles from the chronically treated mice using a second untargeted metabolomics platform, which utilized LC/MS to provide broader coverage of the negatively charged metabolome, including organic acids. This platform detected 3,415 putatively annotated polar metabolites, based on mass. When comparing the metabolites significantly altered between vehicle-treated MDX and WT groups, a reversing pattern emerged in the GSK978A-treated muscles, which was not observed in the NR-treated MDX animals (Figure 6A). This anti-correlation was confirmed with robust significance (R = -0.56, p<1E^-100) only in the GSK978A treatment group (Figure 6B-C).
Suspecting that such a dramatic reversion effect may have resulted from restoration of a pleiotropic co-factor, like NAD, we again performed a pathway analysis. Surprisingly, we found the most significant enrichment in only two pathways: purine metabolism and the pentose phosphate pathway (Figure 6D). Since the untargeted platform was limited in its ability to address the central question of whether positively charged NAD was specifically restored, we instead utilized an enzymatic NAD assay on the same tissue samples, and found only statistically non-significant elevations compared to MDX vehicle controls (Figure 6E). Though the NR-treated MDX muscles still contained significantly less NAD than WT muscle, the intermediate GSK978A group was statistically indistinguishable from vehicle-treated groups of either genotype.
Though sustained NAD elevations were not observed, we found several metabolic reversions consistent with transient restoration of NAD-dependent processes. For example, NAADH, the neutral reduced form of a suspected biomarker of NAD repletion (23) or overload, was detected only in GSK978A-treated muscles (Supplemental Table 1). Importantly, ion counts for components of the proximal glycolytic pathway, including those for glucose-6-phosphate and fructose-1,6,-bisphosphate, were largely normalized by GSK978A, as were the GAPDH substrate, glyceraldehyde-3-phosphate, and the pentose phosphate intermediate, ribose-5-phosphate (Figure 6F). We also observed discrepant influence on TCA cycle intermediates and a subtle net effect of increased ATP and decreased phosphocreatine in this group (Supplemental Table 1), potentially indicating elevated substrate level phosphorylation. Collectively, these global metabolomic profiles of dystrophin-deficient muscle highlight a restorative effect of CD38 inhibition on multiple metabolic pathways that is not recapitulated by supplementing an NAD precursor.