An NAD+/NMN balancing act by SARM1 and NMNAT2 controls axonal degeneration

Abstract

Axonal degeneration is controlled by the TIR domain NADase SARM1. In this issue of Neuron, Figley et al. (2021) reveal a key regulatory mechanism that controls SARM1’s enzymatic activity, providing insight into how NAD+ biosynthesis by the NMNAT2 enzyme protects axons, and a new therapeutic path to tune SARM1 activity.

Mechanisms to destroy and clear neuronal processes are important for the ability of nervous systems to adapt to damage, stress, and viral infection. Studies of the degeneration of injured axons have unraveled a cell-autonomous axon death pathway whose mechanism is distinct from apoptosis (reviewed in Coleman and Höke, 2020). “Wallerian degeneration,” named after its first description by Augustus Waller in 1851, appears to be intimately linked with the synthesis and breakdown of nicotinamide adenine dinucleotide (NAD+), a key coenzyme for metabolism and cellular respiration.

The first hint of this pathway came from the fortuitous discovery of a spontaneous mutation in the C57BL/6/Ola strain background that enabled prolonged survival (up to several months) of damaged axonal stumps severed from their cell body. Characterization of the “Wallerian degeneration Slow” (WldS) mutation revealed that the protection is conferred by the ectopic localization of the NAD+ biosynthesis enzyme NMNAT1 to axons. Later work by Michael Coleman’s group found that a cytoplasmic version of the enzyme, NMNAT2, is rapidly turned over in axons. Injury and/or disruption of axonal transport interrupts the supply of NMNAT2 to distal axons (Figure 1), and its loss leads to degeneration (of axons and/or cells, depending on the extent of NMNAT2 loss). NMNAT2 is therefore considered to function as a critical “protective” factor for axons.

Figure 1. NAD+ and its precursor NMN compete for the same binding site to switch SARM1 between locked and activated states.

NAD+ binding to a regulatory site in SARM1’s ARM domain stabilizes an ARM-TIR interaction that locks the TIR domains in an inactive state. The NAD+ precursor NMN binds to this same site but switches the ARM conformation to disrupt the ARM-TIR lock, freeing the TIR domains to associate into an activated NADase enzyme. In healthy axons, the presence of NMNAT2 maintains NAD+ levels while consuming NMN as a reactant, favoring the NAD+-bound locked state. In contrast, in distal axons disconnected from the cell body by injury and/or disrupted axonal transport, reduced NAD+ biosynthesis favors NMN binding to unlock the NADase enzyme. NAD+ breakdown in the absence of NMNAT to reduce the NMN/NAD+ ratio leads to unmitigated NADase activity by activated SARM1, which leads to metabolic catastrophe and axonal degeneration. The SARM1 cartoon is adapted from Sporny et al. (2020).

But how does a NAD+ biosynthesis enzyme protect axons from degeneration? And why are axons so sensitive to its loss? Exciting work over the past 10 years has identified the sterile-α and Toll/interleukin-1 receptor (TIR) domain-containing protein 1 (SARM1) as an essential executioner of axonal degeneration and death due to NMNAT loss. TIR domains are found in proteins across most kingdoms of life. They are commonly present in proteins that function in innate immunity, where the ability of TIR domains to self-associate facilitates protein interactions. Strikingly, work in 2017 by the Milbrandt and DiAntonio groups discovered that the self-associated TIR domain of SARM1 (as well as TIR domains in other proteins) has enzymatic activity as a NADase (Essuman et al., 2017). By cleaving NAD+ (into nicotinamide and ADP-ribose), SARM1’s enzymatic activity leads to rapid depletion of NAD+ to the point of metabolic catastrophe. By implicating a new enzyme in the execution of axonal degeneration, these findings highlight the existence of an active self-destruction pathway that can accelerate what might otherwise be a passive process of metabolic rundown in distal axons removed from their cell bodies.

As an enzyme that is required for axonal death but lacking other strong loss-of-function phenotypes on its own, SARM1 has become an attractive therapeutic target (Krauss et al., 2020). A key question is to understand how it is normally restrained in healthy axons, and the molecular conditions that lead to its activation (Coleman and Höke, 2020). An N-terminal Armadillo/HEAT Motif (ARM) domain plays an important autoinhibitory role by preventing dimerization of the TIR domains, which is essential for their enzymatic activity. A flurry of recent structural studies has delineated SARM1 in its inactivated state, which assembles into an octomeric ring (Bratkowski et al., 2020; Jiang et al., 2020; Sporny et al., 2020). These studies have revealed a compelling role for NAD+ in stabilizing the inactive state of the enzyme: NAD+ binds to an allosteric site within the ARM domain, which facilitates the “lock” interactions between the ARM and TIR domains (Figure 1).

It is attractively logical that SARM1 should be regulated by NAD+; however, the exact relationship between NAD+ levels and axon degeneration has remained a cloudy area in the field (reviewed in Coleman and Höke, 2020). While ectopic induction of SARM1’s NADase activity is sufficient for degeneration and cell death, manipulations that indirectly reduce NAD+ levels, by inhibiting the synthesis of the NAD+ precursor NMN, lead to protection rather than degeneration of injured axons. Hence, moderate NAD+ reduction can be uncoupled from degeneration, and it has been proposed that rather than—or in addition to—loss of NAD+, buildup of the NMN precursor leads to axonal degeneration. Indeed, ectopic elevation of NMN can potently stimulate axonal degeneration, and recent studies suggest that this is due to direct activation of SARM1 by NMN (Zhao et al., 2019). However, conditions that elevate NMN do not always lead to degeneration (Sasaki et al., 2016). Of note, in these conditions NAD+ levels are also high.

The new study by Figley, Gu, Nanson, Shi, and colleagues in this issue of Neuron (Figley et al., 2021) establishes an elegant solution to this long-standing puzzle by elucidating how NMN leads to activation of SARM1’s enzymatic activity. In addition to more cryoelectron microscopy structures of the full-length inactive enzyme, Figley and colleagues present a 1.7 Å crystal structure of the ARM domain alone (from Drosophila dSarm) bound to NMN. Strikingly, this structure shows NMN occupying the same binding pocket as NAD+ but conferring an altered conformation of the ARM domain. This change is predicted to disrupt the ARM-TIR lock, disrupting the interactions of the outer ring (Sporny et al., 2020), freeing the TIR domain to dimerize into an enzymatically active NADase (Figure 1).

While the complete structure of the activated enzyme is not yet resolved, the new view of structural changes in the ARM domain leads to a compelling model that NMN and NAD+ compete to switch SARM1 between active and inactive states through binding to a common allosteric site. Figley et al. (2021) tested this model with manipulations that independently alter NMN or NAD+ levels in cultured primary neurons, following the production of SARM1’s unique enzymatic product cyclic ADP-ribose (cADPR) by mass spectrometry and through flux assays with heavy-labeled nicotinamide. They found that manipulations that alter the ratio of NMN/NAD+, by either elevating NMN levels or decreasing NAD+, can equally trigger SARM1 NADase activity. Further work with purified SARM1 in vitro demonstrated that the NAD+ and NMN can directly compete for regulation of SARM1’s enzymatic activity (Figley et al., 2021).

The insight that the ratio of NMN/NAD+, rather than either nucleotide alone, controls SARM1 activation provides a solution to the mystery of how SARM1 is controlled by NMNAT2. The conversion of the NMN precursor to NAD+ by this enzyme tidily promotes a low NMN/NAD+ ratio, which favors SARM1 inhibition (Figure 1). It is perhaps not an accident that ATP is also a substrate for NAD+ synthesis by NMNAT2, linking SARM1 inhibition to sufficient levels of ATP, as well as NMNAT2, in axons. NMNAT2 appears to have a short half-life in axons and must be continuously supplied by transport from the cell body. Injury or other methods that disrupt axonal transport, such as the presence of neuropathy-inducing chemotherapy agents that disrupt axonal cytoskeleton, leads to a reduction of Nmnat2 in distal axons (Figure 1). The reduction in NAD+ synthesis should lead to a higher NMN/NAD+ ratio, favoring the activation of SARM1’s NADase activity. If the distal axon lacks the ability to resynthesize NAD+, then SARM1’s NADase activity remains unchecked and drives an accelerated loss in NAD+, which should ultimately lead to metabolic catastrophe and degeneration.

In contrast to damaged neurons, Figley et al. (2021) present further analysis of the relationship between SARM1 activation and NMNAT2 function in healthy cultured primary neurons pulsed with nicotinamide riboside (NR) (while overexpressing the NMN-synthesizing NR kinase [NRK1]). While this popular nutritional supplement is used for an end result of increasing cellular NAD+, the pulse first leads to elevated NMN levels, which activates SARM1’s NADase activity, and dampens the ability of NR to increase total NAD+ levels. Eventually (after 24 h) NAD+ levels do increase, presumably due to NMNAT function and re-silencing of the SARM1 NADase, which can explain why neurons pretreated with NR together with NRK1 show an increased resilience to degeneration (Figley et al., 2021; Sasaki et al., 2016). These observations emphasize that conditions that activate SARM1’s NADase activity do not necessarily represent a point of no return. Instead, healthy cells (which should have NAD+ biosynthesis capacity) can tolerate some level of SARM1 NADase activity, and it is indeed detectable at modest levels in healthy neurons (Sasaki et al., 2016). This basal activity may be important for signaling functions of SARM1 outside of degeneration. Multiple studies in different model organisms have detailed roles for SARM1 and its orthologs in regulating MAP kinase signaling pathways relevant for cell fate, neuronal plasticity, and innate immunity in both neurons and glial cells. Whether (and how) SARM1’s NADase activity and its regulation by NMN/NAD+ ratio influence its other signaling functions is an interesting area for future investigation.

While there are likely additional layers of SARM1’s regulation to uncover, the current insight further increases the attractiveness of SARM1 as a target to inhibit axonal loss in neuropathies associated with chemotherapy, diabetes, mitochondrial dysfunction, and potentially neuro-degenerative diseases such as ALS and FTD (Krauss et al., 2020). Recent work has shown that SARM1 inhibition at a time point after mitochondrial damage has occurred is sufficient to rescue these axons from degenerating (Hughes et al., 2021). The finding that SARM1’s enzymatic activity may be tuned by altering NMN/NAD+ ratios offers a rich new platform of potential methods to regulate SARM1, with hopes toward rescuing axons and synapses in different translational settings.