Despite diverse clinical manifestations, axonal degeneration is an underlying feature of traumatic brain injury, peripheral neuropathies, and other neurodegenerative diseases, including Alzheimer’s disease, Huntington’s disease, and Parkinson’s disease. These diseases account for extensive morbidity and mortality as there are no approved therapies. SARM1 has recently emerged as a promising therapeutic target for diseases associated with axonal degeneration because SARM1 knockdown or knockout prevents axonal degeneration and disease pathology (Geisler et al., 2016; Henninger et al., 2016; Loring and Thompson, 2020).

SARM1 was originally identified in a screen that selected for mutants that suppress injury–induced axonal degeneration (Osterloh et al., 2012). However, it was unknown how SARM1 loss prevented neurodegeneration post–injury. Subsequently, SARM1 was shown to possess enzymatic activity and catalyze the hydrolysis of NAD⁺ (Figure 1A; Essuman et al., 2017). Depletion of the NAD⁺ pool promotes degeneration, whereas inactive variants of SARM1 or supplementation with NAD⁺ precursors and minimizing NMN accumulation prevent degeneration after injury (Essuman et al., 2017; Gerdts et al., 2015; Wang et al., 2015).

Figure 1

Download asset Open asset

SARM1 consists of three domains: a HEAT/armadillo (ARM) domain, two tandem sterile alpha motif (SAM) domains, and a C–terminal toll interleukin receptor (TIR) domain that catalyzes NAD⁺ hydrolysis (Figure 1B). The TIR domain contains a BB loop (Residues 594–605), which is thought to promote self–association, and a SARM1–specific (SS) loop (Residues 622–635), which is a characteristic feature of catalytically active TIR domains; TIR domains lacking this loop do not possess enzymatic activity (Carlsson et al., 2016; Horsefield et al., 2019; Loring and Thompson, 2020; Summers et al., 2016). Additionally, E642 (within the TIR domain and adjacent to the SS loop) is thought to be a key catalytic residue based on structural alignments with other nucleotide hydrolases and mutagenesis studies (Essuman et al., 2017; Gerdts et al., 2015; Gerdts et al., 2013; Loring and Thompson, 2020). In the absence of disease, the ARM domain is thought to inhibit catalysis via an intramolecular interaction with the TIR domain as deletion of this domain results in constitutively active SARM1 (Chuang and Bargmann, 2005; Gerdts et al., 2013). This hypothesis is supported by recent cryoEM structures of the inactive enzyme, which show that SARM1 exists as an octamer with the ARM domain binding to the TIR domain (Bratkowski et al., 2020; Jiang et al., 2020; Sporny et al., 2020). Mutations that disrupt the TIR–ARM interface or delete the ARM domain entirely, however, only activate SARM1 by ~1.8- to 2.4-fold (Jiang et al., 2020). Thus, the fully active conformation/state of the TIR domains remains to be determined.

Although SARM1 inhibitors hold therapeutic promise, efforts to study the purified enzyme have been complicated by the fact that SARM1 appears to lose activity during purification and only shows measurable NAD⁺ hydrolase activity when assayed at a final enzyme concentration of ≥ 20 µM (Horsefield et al., 2019; Loring et al., 2020a). By contrast, enzymatic activity is 800-fold higher in lysates or when activity is measured on–bead (Essuman et al., 2017; Horsefield et al., 2019; Loring et al., 2020a). These data indicate that the pure protein does not fully recapitulate all features of SARM1 catalysis. Herein, we report that macroviscogens induce a phase transition that increases in vitro SARM1 activity by >2000-fold. By exploiting this finding, we report the first detailed studies of SARM1 catalysis. Additionally, we demonstrate that a point mutation which reduces multimerization and precipitation in vitro, also disrupts this phase transition in cell culture. We further show that citrate induces axonal degeneration in C. elegans that is dependent on the C. elegans orthologue of SARM1 (i.e. TIR–1). Notably, citrate induces the formation of larger puncta indicating that TIR–1/SARM1 undergoes a phase transition that is essential for degeneration in vivo. In total, these findings will aid efforts to therapeutically target SARM1 to treat neurodegenerative diseases as they improve our understanding of SARM1 activation and its role as a critical switch in the degenerative process resulting in catastrophic axonal fragmentation.

We previously established that SARM1 loses activity during purification and is essentially inactive unless purified to high micromolar concentrations (Figure 2A; Loring et al., 2020a). We initially hypothesized that the lack of activity could be explained by the loss of an essential element during enzyme purification. However, the addition of purified SARM1 to cell lysates did not activate the enzyme (Figure 2B). We then hypothesized that pure SARM1 adopts an inactive conformation. To investigate this hypothesis, we evaluated the effect of enzyme concentration (5–35 µM) on the rates of catalysis. Notably, the concentration dependence has parabolic character, increasing with upward curvature, as opposed to linearly, suggesting the formation of a more active oligomer (Figure 2C). The fact that pure SARM1 regains activity upon concentration, is inconsistent with the loss of an essential cofactor during purification. To investigate the oligomerization status of the SARM1 TIR domain, we next performed analytical ultracentrifugation (AUC). SARM1 was largely monomeric (~90%), as indicated by an S-value of 1.9–2.0S (Figure 2D). A smaller peak at ~3.9–4.2S, which showed rightward tailing indicates that the remaining ~10% exists as a trimer or tetramer (~71–88 kDa) with a smaller fraction in higher ordered forms. Similar oligomerization levels were observed at the two different SARM1 concentrations tested (Figure 2D).

Figure 2 with 2 supplements see all

Download asset Open asset

Having shown that SARM1 activity increases nonlinearly with concentration, we next hypothesized that molecular crowding might activate the enzyme. To test this hypothesis, we measured the activity of SARM1 in the presence of several macroviscogens (i.e. PEG 8000, 3350, 1500, 400, and Dextran) and microviscogens (i.e. Sucrose and Glycerol). Microviscogens are often used as protein stabilizers and affect the ability of small molecules to diffuse in solution, whereas macroviscogens alter the free volume available for a molecule to move within a solution with little effect on diffusion (Blacklow et al., 1988; Gadda and Sobrado, 2018; Hagen, 2010). Therefore, microviscogens can reveal effects on diffusional processes, whereas macroviscogens affect molecular crowding (Gadda and Sobrado, 2018). Notably, we found that all the macroviscogens tested increased activity, whereas the microviscogens had negligible effects (Figure 2E). Specifically, PEGs 8000, 3350, 1500, and 400 increased activity by 1500, 2000, 1900, and 200–fold, respectively. By contrast, dextran (36–50 kDa) only increased activity by 14–fold (Figure 2E). To better visualize the effect of PEG 3350 on the rate of the reaction, representative progress curves are shown at 20 µM SARM1 with and without PEG (Figure 2F). The 20 µM reaction with PEG 3350 plateaus after ~ 350 s, whereas the fluorescence of the reaction without PEG increases at a much slower rate. In total, these data indicate that a subset of macroviscogens dramatically increase SARM1 activity by up to 2000–fold.

To determine how macroviscogens increase SARM1 catalysis, the activity of SARM1 (10 µM final) was next measured at 0, 10, 20, and 30% w/v of the macroviscogens and glycerol. Consistent with our hypothesis, we observed a dose–dependent increase in activity for all the macroviscogens tested (Figure 2G). Notably, activity increases according to macroviscogen size until a critical point is reached where PEG 1500 and PEG 3350 have similar effects (Figure 2G). Note that the effect on activity is not purely due to a change in viscosity, as the enhancement in activity shows a poor correlation (R²=0.46) with viscosity (Figure 2H). For example, when dextran and PEG 3350 are assayed at similar relative viscosities, the fold activation is dramatically different (Figure 2H–I). Additionally, when the viscosity of glycerol was compared to that of PEG 3350, we found that glycerol did not enhance activity to the same degree as PEG 3350, despite having a similar relative viscosity (Figure 2—figure supplement 1A–B).

As PEG 3350 showed the greatest ability to enhance activity, we proceeded to investigate its effect on SARM1 activity further. Initially, we evaluated the effect of enzyme concentration on activity in the presence and absence of PEG 3350 (Figure 2J). SARM1 concentration was varied from 300 nM to 20 µM with 25% w/v PEG 3350 and from 5 to 35 µM in the absence of PEG 3350. Our data show that the addition of 25% w/v PEG 3350 accelerates the rate of reaction 7700-fold at 5 µM, 1400-fold at 10 µM, and ~280-fold at 20 µM of SARM1 (Figure 2J). These data indicate that the effect of PEG 3350 is dependent on enzyme concentration. Furthermore, SARM1 activity appears to increase almost linearly with respect to concentration in the presence of PEG after a slight lag phase, but parabolically in its absence (Figure 2J). This trend suggests that in the absence of PEG, there is a gradual build up in activity as the concentration of SARM1 increases, until the concentration surpasses a critical threshold allowing for increased transient interactions and a consequent large increase in activity (Figure 2J). By contrast, the addition of PEG reduces the concentration of enzyme needed to reach this threshold. One potential explanation is that PEG induces molecular crowding by excluding water from the protein surfaces, thereby promoting multimerization and catalysis. Along these lines, we note that in the absence of PEG the Hill coefficient is 3.8, indicating a high degree of cooperativity, whereas in the presence of PEG (0–25%) the Hill coefficient decreases to 1.3, suggesting that PEG decreases the dependence on the cooperative formation of an active multimer (Figure 2—figure supplement 1C–D).

Next, we evaluated the potency of several metals that were previously shown to inhibit SARM1 when assayed in lysates (Loring et al., 2020a). Notably, NiCl₂ and ZnCl₂ completely eliminated activity in the presence of PEG (Figure 2—figure supplement 2A). This result is consistent with our previous data (Loring et al., 2020a; Loring et al., 2020b). Expanding on these studies, we also evaluated several other metal ions including CaCl₂, CdCl₂, CoCl₂, CuCl₂, MgCl₂, and MnCl₂ (Figure 2—figure supplement 2A). Of note, CdCl₂, CoCl₂, and CuCl₂, decreased activity in the presence of the pure protein but not in lysates. This inconsistency could be due to elements in the lysate preferentially sequestering these metals and thereby preventing their ability to act on SARM1. Given the high potency of CdCl₂ and CuCl₂, we used these inhibitors to titrate the enzyme and determine the percentage of active enzyme (Figure 2—figure supplement 2B). Based on these data, all the enzyme appears to be active. Although speculative, the fact that thiophilic metals (i.e. CdCl₂, CoCl₂, CuCl₂, NiCl₂, and ZnCl₂) inhibit SARM1 suggests the presence of a catalytically important thiol. Along these lines, we previously showed that C629 and C635, which are present in the SARM1 specific loop, are critical for activity in the absence of PEG (Loring et al., 2020b).

We next determined an IC₅₀ value for ZnCl₂ as this compound was previously shown to potently inhibit the SARM1 TIR domain assayed in lysates. Consistent with our prior data, the IC₅₀ value obtained in the presence of PEG is 10 ± 1 µM, which is very similar to that in lysates (10 ± 3 µM) (Loring et al., 2020a), and with purified protein without PEG (10 ± 1 µM) (Loring et al., 2020b; Figure 2K, Table 1). To further confirm that the addition of PEG 3350 was not introducing any confounding variables, we repeated the mechanism of inhibition studies with ZnCl₂. As before, we found that ZnCl₂ inhibits SARM1 noncompetitively with a K_i value of 1.0 ± 0.1 µM. This value is similar to the value obtained when lysate derived SARM1 was assayed (K_i = 3.3 ± 0.1 µM Figure 2L; Loring et al., 2020b).

For a secondary confirmation that PEG does not impact the in vitro properties of SARM1, we tested a different inhibitor, berberine chloride that was previously identified from a high-throughput screen (Loring et al., 2020b). Berberine chloride inhibited SARM1 in the presence of PEG with an IC₅₀ value of 140 ± 30 µM. This value is quite similar to its potency in lysates (140 ± 20 µM) and when assayed with purified SARM1 without PEG 3350 (140 ± 20 µM) (Figure 2M and Table 1; Loring et al., 2020b). As before, we found that berberine chloride inhibits SARM1 noncompetitively in the presence of PEG 3350 with a K_i value of 120 ± 10 µM. This value and inhibition pattern are identical to that obtained when SARM1 was assayed in lysates (K_i = 120 ± 10 µM, Figure 2N; Loring et al., 2020b).

We next sought to investigate how PEG enhances the activity of SARM1. Since PEGs are typically used as precipitants, we hypothesized that PEG3350 was precipitating and activating SARM1. Consistent with this hypothesis, we found that the addition of PEG 3350 (25% w/v) results in the formation of an active precipitate, whereas the protein in buffer without PEG remains soluble (Figure 3A). These data indicate that the observed rate enhancement is driven by a phase transition. Next, we evaluated the effect of protein and PEG concentration on this phase transition. We found that both enzyme (5–20 µM) and PEG (0–25%) concentration enhanced precipitate formation (Figure 3B–C). As the concentration of SARM1 in the supernatant decreases, the precipitate concentration increases with a corresponding increase in enzyme activity. This effect is evident by the absence of a pellet for 0% PEG, a faint band at 10% PEG, and darker bands at 17.5% and 25% PEG at 5, 10, and 20 µM SARM1, and the inverse trend for the supernatant (Figure 3C). To determine whether other precipitants increase SARM1 activity, we screened several common precipitants and found that citrate both precipitated and acted as a second highly potent activator of SARM1 activity (Figure 3—figure supplement 1A).

Figure 3 with 2 supplements see all

Download asset Open asset

To explore the nature of the interactions involved in precipitate formation, we tested the effect of 1,6–hexanediol, an aliphatic alcohol that is often used to distinguish between hydrophobic interactions involved in liquid-to-liquid phase separations and liquid-to-solid phase transitions (Kroschwald et al., 2015; Patel et al., 2007; Peskett et al., 2018). We found that while 1,6–hexanediol inhibited the activity of pure protein alone, it did not substantially affect the activity of SARM1 in the presence of PEG or sodium citrate (Figure 3D–E). These data suggests that the activity of the pure protein is dominated by weak hydrophobic interactions or a liquid-to-liquid phase separation (Figure 3D). On the other hand, the fact that 1,6-hexanediol does not eliminate the activity of pure protein in the presence of PEG or sodium citrate, supports SARM1 undergoing a liquid-to-solid phase transition in these cases (Figure 3E). To confirm that precipitate formation is reversible, we performed the centrifugation experiment in the presence of PEG or sodium citrate and tested the activity of the fractions (pre-centrifugation, supernatant, precipitate resuspended in buffer, and precipitate resuspended with buffer plus additive) (Figure 3—figure supplement 1B). The lack of both a precipitate and activity in the precipitate resuspended in buffer confirm that precipitation is reversible. To see if select additives/inhibitors affect precipitation of SARM1, we quantified the precipitation percentage with PEG or citrate in the presence of Zn, NMN, Ca, and berberine and found that none of these additives affected precipitation (Figure 3—figure supplement 1C–D).

Having established that both PEG 3350 and sodium citrate enhance SARM1-mediated NAD⁺ hydrolysis via a phase transition, we next sought to investigate how this phase transition enhances activity. To that end, we performed negative stain EM in the absence and presence of sodium citrate (500 mM) to determine how the precipitant affects activity and oligomeric state (Figure 3F–G). Surprisingly, we were able to detect the SARM1 TIR domain even in the absence of additive, which suggests the TIR domain alone is forming a higher ordered oligomer. The median diameter of the particles is ~8 nm or ~80 Å, which closely aligns with the diameter predicted for an octameric ring of the TIR domain; recent crystal and cryoEM structures of the SAM domains and full-length protein, respectively, show that SARM1 forms an octameric ring (Bratkowski et al., 2020; Sporny et al., 2020; Sporny et al., 2019). In the presence of sodium citrate, the SARM1 TIR domain forms significantly larger ring–like structures, potentially by wrapping around on itself (Figure 3G, p<0.00001). Notably, the mean area is 172 nm² (SEM = 5.2, n=100) in the presence of citrate compared to 66 nm² (SEM = 1.9, n = 100) without additive (Figure 3H). These data suggest that precipitants like PEG 3350 and sodium citrate enhance SARM1 activity by inducing the formation of a higher ordered oligomer and facilitating inter–TIR domain contacts necessary for NAD⁺ hydrolysis.

We next sought to exploit the additive–induced rate enhancement to perform the first detailed analysis of SARM1 catalysis. First, we determined how pH affects activity and precipitate formation. We found that the catalytic efficiency (k_cat/K_m) peaks from pH 6–8, which is primarily due to an effect on k_cat, as the K_m value remains nearly constant across the range tested (Figure 3I). Furthermore, the effect of pH on activity appears to be independent of precipitation as the precipitate percentage varies by only 10% across the range tested. Notably, the pH profile for k_cat is bell-shaped with pK_a values of 4.9 ± 0.2 and 8.8 ± 0.2 for the ascending and descending limbs, respectively. The pK_a value for the ascending limb is consistent with the previous suggestion that a glutamate, likely E642, acts as a key catalytic residue (Essuman et al., 2017).

We next evaluated the effect of PEG on the steady state kinetic parameters in the presence of 10 µM SARM1. In the presence of 25% PEG, the K_m decreases 3.3-fold, whereas k_cat and k_cat/K_m increase by 2000- and 7000-fold, respectively (Figure 4A). The fact that the increase in k_cat/K_m is dominated by an increase in k_cat, suggests that PEG promotes multimerization and activity via precipitate formation. To further interrogate how PEG impacts the activity of SARM1, we performed detailed steady state kinetic experiments at various PEG concentrations from 0 to 25% (Figure 4B). We found that increasing concentrations of PEG 3350 cause the K_m value to decrease and level off at ~500 µM (Figure 4C) and the k_cat to increase almost linearly (Figure 4D). The catalytic efficiency also increases and plateaus at ~1700 M^–1s^–1 (Figure 4E, Supplementary file 1a). Interestingly, this value is nearly identical to that recorded in the lysate and on-bead (1500 M^–1s^–1) (Essuman et al., 2017; Loring et al., 2020a; Figure 4B–E, Table 2).

Figure 4

Download asset Open asset

We also performed steady-state kinetic analyses at different concentrations of SARM1 in the absence and presence of 25% w/v PEG 3350 (Figure 4F–J). As with increasing PEG concentration, an increase in SARM1 concentration led to a small decrease in K_m and an increase in k_cat (Figure 4F–J, Supplementary file 1b), such that the k_cat/K_m values are comparable to those obtained for the enzyme assayed in lysate (Loring et al., 2020a). Of particular note is the fact that PEG makes kinetic studies feasible at 500 nM of SARM1, which is near the concentration studied in lysates (300 nM) (Loring et al., 2020a). Specifically, at 500 nM of SARM1 with PEG, the K_m is 1600 µM, the k_cat is 0.24 s^–1, and the catalytic efficiency is 150 M^–1s^–1 (Figure 4F–J, Supplementary file 1b). Notably, the K_m value at 500 nM with PEG is similar to that at 20 µM SARM1 without PEG (1200 µM); however, the k_cat is ~100-fold greater in the presence of PEG (Figure 4F–J, Supplementary file 1b). These data suggest that by inducing a phase transition, PEG drastically enhances enzymatic turnover, but has little effect on substrate binding. Although the K_m value slightly decreases as SARM1 or PEG concentration increases, it remains much higher than the lysate value of 40 ± 10 µM (Loring et al., 2020a), indicating that PEG/SARM1 concentration does not completely recapitulate SARM1 activity in the lysate or on-bead (Figure 4, Table 2). The similarities in k_cat/K_m, however, suggest compensation (i.e. alterations in rate limiting steps) between these two parameters. Moreover, the fact that k_cat is directly correlated with enzyme and PEG concentration, suggests that turnover rate is largely dependent on the multimeric state and the formation of an ordered oligomer. While the k_cat/K_m values obtained for SARM1 (~2000 M-1s-1) are not particularly robust, they are in-line with other eukaryotic enzymes that we have studied (e.g. the protein arginine deiminases, protein arginine methyltransferases, and nicotinamide N-methyltransferase Knuckley et al., 2007; Knuckley et al., 2010; Loring and Thompson, 2018; Osborne et al., 2007), and the slow kinetics likely relate to the fact that degeneration occurs on the hour time scale (Mack et al., 2001; Osterloh et al., 2012).

Next, we evaluated the product specificity of SARM1 with PEG and sodium citrate using an orthogonal LC-MS-based assay. We found that the addition of PEG or citrate led to a time-dependent decrease in NAD⁺ levels that coincided with increased levels of nicotinamide and a mixture of ADPR and cADPR (Figure 5A–B, Figure 5—figure supplement 1A–B). Of note, the ADPR to cADPR ratios are consistent over time under each condition (Figure 5C), indicating that cADPR is a poor substrate relative to NAD⁺. Interestingly, the addition of PEG increases the production of ADPR relative to cADPR from a ratio of 10 ± 2 to a ratio of 16 ± 1. This ratio is further increased in the presence of citrate to 23 ± 3 (Figure 5D). We hypothesize that this could be due to the multimeric state of the enzyme constraining NAD⁺ in such a position that it favors hydrolysis by water as compared to reaction with the N1 of the adenosine, which is required for cyclization. We also evaluated whether SARM1 could cleave cADPR in the presence or absence of additive (Figure 5—figure supplement 2A–C). Consistent with our prior finding that SARM1 follows an ordered uni–bi kinetic mechanism instead of a sequential intermediate mechanism, SARM1 is unable to cleave cADPR in the absence of additive and in the presence of sodium citrate (Loring et al., 2020a; Figure 5—figure supplement 2A–C). Further support comes from a single turnover experiment where SARM1 concentration exceeded the NAD⁺ concentration and the ADPR/cADPR ratio remains nearly constant over the time range observed, indicating that SARM1 does not cleave cADPR under these conditions (Figure 5—figure supplement 3A–B). Interestingly, however, SARM1 slowly cleaves cADPR in the presence of PEG, resulting in the time-dependent production of ADPR (Figure 5—figure supplement 3C–D). Additionally, we found that cADPR inhibits SARM1 in the presence of PEG with an IC₅₀ of 100 ± 10 µM and a competitive inhibition pattern, supporting its role as a poor substrate under these conditions (Figure 5—figure supplement 3E–F).

Figure 5 with 4 supplements see all

Download asset Open asset

To further interrogate the substrate specificity of SARM1, we analyzed the capacity for SARM1 to cleave 16 NAD⁺ analogues (Figure 5—figure supplement 4). Notably, we found that adenosine substitutions are well tolerated as etheno–NAD (ENAD), nicotinamide guanine dinucleotide (NGD) and nicotinamide hypoxanthine dinucleotide (NHD) are all processed by SARM1. By contrast, SARM1 does not cleave NADH or NHDH, the reduced forms of NAD⁺ and NHD, in the absence or presence of PEG. This finding likely relates to nicotinamide being a better leaving group than its reduced form, and/or conformational preferences; nicotinamide is planar whereas the reduced form adopts a puckered conformation. Along these lines, SARM1 can still cleave nicotinamide adenine dinucleotide phosphate (NADP⁺) but cannot accommodate the flavin substitution for nicotinamide in flavin adenine dinucleotide (FAD). SARM1 can also cleave thionicotinamide adenine dinucleotide (sNAD), which has a thiol substitution for the carbonyl. Interestingly, low level cleavage was observed for nicotinamide mononucleotide (NMN), but not for nicotinic acid mononucleotide (NaMN) nor nicotinic acid adenine dinucleotide (NaAD), suggesting that SARM1 cannot accommodate a carboxylic acid substitution at this position. However, SARM1 can accommodate a methyl substitution at the same site as it will cleave 3-acetylpyridine adenine dinucleotide (3apAD). Adenosine triphosphate (ATP), adenosine diphosphate (ADP), guanosine triphosphate (GTP) and ara-F-NAD⁺ were also not cleaved by SARM1, nor did they inhibit SARM1 cleavage of NAD⁺. Ara-F-NAD⁺ has a fluorine substitution for a hydroxyl with opposite stereochemistry, which likely interferes with coordination to the active site as no cleavage or inhibition was evident.

Next, we generated mutant constructs to understand how specific residues contribute to catalysis and multimerization. Residues were selected based on their proposed roles in multimerization, catalysis, or location in the SARM1 specific loop (Figure 5E). All the mutants examined showed a marked reduction in catalytic efficiency, k_cat/K_m, ranging from 100- to 2000-fold (Figure 5F). Interestingly, the most extreme effect was observed for the E642A mutant, which has previously been suggested to be a key catalytic residue based on structural alignments with other glycohydrolases (Essuman et al., 2017; Figure 5F). Consistent with a role in multimerization the Y568, G601, H640, and H685 mutants showed reduced precipitation, suggesting that the loss in activity observed upon their mutation is due in part to a reduction in multimer formation (Figure 5G).

While k_cat values were reduced by ~100- to 400-fold for all mutants examined, a subset showed a >4–fold increase in K_m, including residues thought to be involved in multimerization, that is, Y568, G601, W638, H640, and E686, and possible catalytic residues, K628 and E642. Several of these mutants have been previously characterized in the context of the purified protein and degenerating axons (Horsefield et al., 2019; Summers et al., 2016). Consistent with our results, they also found that TIR domain mutants E642A, Y568A, H685A, and BB loop mutations (D594A and G601P) reduce NAD⁺ hydrolase activity (Horsefield et al., 2019). Moreover, several of these mutations, including G601P, D627K, K628D, and C629S, were sufficient to prevent axonal degeneration post–axotomy, indicating that they render SARM1 non-functional (Summers et al., 2016).

To investigate the multimeric state of SARM1 in cells, we transfected HEK293T cells with a GFP-tagged SAMTIR–E642A construct; this mutant construct should multimerize like wild-type SARM1 without exhibiting toxicity due to the rapid depletion of NAD⁺ (Figure 6A–B). We found that this GFP tagged SARM1 SAMTIR construct forms puncta in cells (Figure 6A–B). To demonstrate that these puncta are due to SARM1 oligomerization we transfected cells with a construct encoding GFP–tagged SAMTIR-E642A plus an additional G601P mutation. G601 has been previously implicated as an association interface for TIR domains during SARM1 activation, and we demonstrated that the BB loop mutation G601P, reduces both activity and precipitation in vitro (Figure 5F–G). Moreover, the G601P mutation has previously been shown to block axon degeneration in cells (Gerdts et al., 2013; Sporny et al., 2019). Notably, the G601P point mutation also reduces SARM1 puncta formation in HEK293T cells (Figure 6A–C). Based on spot counts, the IDEAS software spot count wizard identified buckets of distinct sub–populations (0–28; R1 and R2). Specifically, cells expressing the E642A mutant show a higher frequency of puncta (34.4% of cells) whereas the majority of cells expressing the E642A/G601P double mutant were classified into the R1 or baseline fluorescence group (96.3%), with a smaller percentage exhibiting puncta characterized as the R2 group (3.7%) (Figure 6A–C). Since the SAM domains are involved in octamerization, which is not impacted by the G601P mutation, these data indicate that puncta formation is largely driven by the TIR domain.

Figure 6 with 2 supplements see all

Download asset Open asset

Having determined that SARM1 oligomerizes in cells, we next sought to determine whether SARM1 oligomerizes in vivo and if so, investigate how oligomerization affects SARM1-mediated degeneration. To this end, we opted to study the SARM1 orthologue, TIR-1 in the GABA motor nervous system of C. elegans. C. elegans encodes several TIR-1 splice variants including TIR-1b and TIR-1d which lack the ARM domain (Figure 6—figure supplement 1A) but contain the tandem SAM domains essential for octamer formation. Since TIR-1b was used in our in vivo studies (see below), we expressed and purified an MBP-tagged ceSAMTIR expression construct as a mimic of TIR-1b and evaluated whether citrate activates this construct. As a control, we also evaluated the effect of citrate on the C. elegans TIR domain (ceTIR). Like the human TIR domain, the NAD⁺ hydrolase activity of both ceTIR and ceSAMTIR was greater with citrate, where ceTIR and ceSAMTIR activity were enhanced 22-fold and 1.4-fold respectively (Figure 6—figure supplement 1B). While citrate does not activate the ceSAMTIR construct to the same extent as the ceTIR domain alone, mutations that disrupt the TIR–ARM interface or deletion of the ARM domain activate SARM1 to a similar extent (i.e. 1.8- to 2.4-fold) (Jiang et al., 2020).

Having demonstrated that ceSAMTIR activity is enhanced by citrate, we sought to determine whether ceSAMTIR undergoes a phase transition. To do this, we incubated ceSAMTIR with increasing concentrations of citrate, centrifuged the samples, removed the soluble fraction, and resuspended the insoluble pelleted fraction. All fractions were analyzed by SDS-PAGE for the presence of ceSAMTIR. At citrate concentrations 250 mM and below, ceSAMTIR was predominantly located in the supernatant. By contrast, at citrate concentrations of 500 mM and above, the protein was increasingly found in the pellet, such that at 1000 mM citrate, ceSAMTIR was primarily located in the pellet. Moreover, enzymatic activity in the fractions correlated with the location of ceSAMTIR (Figure 6—figure supplement 1C). These data are consistent with the hypothesis that longer versions of SARM1/TIR-1 also undergo a phase transition that correlates with enzyme activation.

The C. elegans GABA motor nervous system is composed of neurons that have cell bodies along the ventral nerve cord and extend axon commissures circumferentially around the body of the worm to the dorsal nerve cord. We visualized the GABA motor nervous system and TIR-1 using a transgenic strain that expresses cytosolic GFP and mScarlet::TIR-1b specifically in GABA motor neurons (Figure 6D; Julian and Byrne, 2020). Consistent with our cell culture studies, TIR-1 forms puncta in axons under basal conditions (Figure 6D). Given our demonstration that citrate activates SARM1 in vitro, we sought to determine whether citrate could alter TIR-1 expression and puncta formation in vivo. For these studies, L4 stage worms were exposed to 8 mM citrate and imaged 2–3 hr and 24 hr later. After 2–3 hr, no difference was observed in the mean number of puncta between untreated control worms and worms that were exposed to citrate (Figure 6E). However, there was a significant increase in mean puncta area seen in citrate treated worms (mean=84.83, SEM=4.03, n=902) compared to untreated control animals (mean=58.42, SEM=3.45, n=739) (Figure 6F–G). Note that the increase in puncta size was not a secondary consequence of gross axon morphology as the presence and size of mScarlet puncta does not coincide with the expression pattern of cytosolic GFP (Figure 6H). Together these data support that citrate treatment not only enhances the activity of SARM1 by inducing oligomerization in vitro but also facilitates multimerization in vivo as evidenced by the formation of larger puncta in C. elegans axons.

Given that citrate induces the formation of larger TIR-1 puncta in C. elegans axons, we determined whether this translates into downstream axonal degeneration. To this end, axons were analyzed at 2–3 hr and again at 24 hr post-citrate treatment. Axon commissures that were broken or fragmented, truncated, or missing were counted as damaged (Figure 6I). No significant change in the percentage of damaged axons was observed in the citrate treated animals 2–3 hr post-treatment (Figure 6I). However, when the same animals were analyzed 24 hr post-treatment, there was a significant increase in the percentage of damaged axons in citrate treated animals compared to untreated controls (Figure 6I). Note the fact that axons are significantly fragmented or degraded at 24 hr precludes quantification of puncta number and area at this time point. To confirm that citrate-induced damage was dependent on TIR-1 function, we exposed tir–1(qd4) mutant animals, which have a null mutation in tir–1, to citrate for 24 hr and quantified axon damage (Figure 6J). We found that 98% of axons remain intact while only 2% were damaged, indicating that citrate exposure induces TIR-1-dependent axonal damage (Figure 6K). In total, these findings indicate that citrate induces TIR-1 multimerization and axonal degeneration in C. elegans.

Conclusions

Our data demonstrate that SARM1 activity is largely dependent on concentration and crowding alone and that as SARM1 multimerizes it becomes more active. Moreover, we show that SARM1 activity is enhanced by a phase transition in vitro. These data are consistent with cellular and in vivo data showing that SARM1 forms puncta and that puncta size increases in response to neuronal injury. These findings are notable because SARM1 was previously thought to function as a dimer. The formation of a higher ordered multimer is, however, consistent with recent structural studies showing that the SAM domains in isolation and in the context of the full-length enzyme form octamers (Bratkowski et al., 2020; Gerdts et al., 2013; Horsefield et al., 2019; Jiang et al., 2020; Sporny et al., 2020; Summers et al., 2016). However, the resolution of these structures was not sufficient to determine the active conformation of the TIR domains (Bratkowski et al., 2020), and it is unclear whether the TIR domains form dimers, tetramers, octamers, or larger multimers via two or more octameric rings interacting with one another (Bratkowski et al., 2020; Sporny et al., 2020). Our data suggests that TIR domain multimerization is required to upregulate SARM1 activity.

Significant cellular and biochemical data supports the notion that the ARM domain interacts with and inhibits the TIR domain. However, mutations that disrupt this interaction cause subtle increases (~2-fold) in SARM1 activity (Bratkowski et al., 2020; Jiang et al., 2020; Sporny et al., 2020). By contrast, our data demonstrates that SARM1 multimerization increases catalytic efficiency and turnover rate by >2000-fold. Pre-organization of the TIR domains within the context of the octameric structure likely increases the effective concentration of the TIR domain such that multimerization can occur more readily. This conclusion is supported by our finding that our MBP-tagged ceSAMTIR expression construct has a higher basal level of activity that can be further increased by the addition of citrate. Moreover, the data suggest that the ARM domain prevents TIR domain multimerization and limits full activation of the enzyme.

Our observation that SARM1 becomes progressively more active as it undergoes a phase transition and forms higher ordered multimers, is analogous to other systems including purine biosynthesis where higher ordered oligomers, known as purinosomes, form under certain conditions to enhance purine synthesis (An et al., 2010; Chan et al., 2015). Moreover, SARM1 multimerization is reminiscent of the structures that are formed by inflammasomes and apoptosomes during an immune response or apoptosis. Interestingly, other TIR domain–containing proteins (e.g. MAL) form insoluble fibrils (Ve et al., 2017).

Supporting the notion that SARM1 multimerizes, we also show that SARM1 forms puncta in cells and that mutations in the BB loop interface (i.e. G601P) is sufficient to disrupt puncta formation. Notably, punctate expression of SARM1 in neurons (both soma and dendrites) has been observed in numerous studies (Chen et al., 2011; Osterloh et al., 2012), further confirming that SARM1 can form higher ordered structures. In the context of a degenerating axon, TIR domain multimerization, could translate into a feed-forward mechanism to induce degeneration. In fact, we demonstrate that increasing TIR–1/SARM1 multimerization in C. elegans via citrate treatment is sufficient to induce axonal degeneration. Furthermore, SARM1 protein levels have been documented to increase significantly in response to injury prior to Wallerian degeneration (Massoll et al., 2013), which is consistent with our data demonstrating that increasing SARM1 concentration enhances activity and inducing multimerization in vivo causes axonal degeneration. These data are also consistent with a prior report showing that upregulation of SARM1 is sufficient to induce axonal degeneration (Gerdts et al., 2013).

Based on these studies, we propose a model whereby, increasing SARM1 protein levels induces TIR domain multimerization that enhances SARM1 activity via a phase transition and that this process is mimicked by PEG 3350 and sodium citrate (Figure 7). We hypothesize that the enhanced activity associated with this phase transition is due to the formation of a higher ordered oligomer as we see larger multimers by negative stain, formation of puncta in HEK293T cells, and induction of larger puncta with citrate treatment in C. elegans. We expect that the formation of larger puncta facilitates increased inter-TIR domain contacts that allow for enhanced NAD⁺ hydrolysis and subsequent axonal degeneration. Similar phase transitions have been observed in multiple signaling pathways. For example, the liquid phase condensation of cyclic GMP–AMP synthase (cGAS) activates innate immune signaling (Du and Chen, 2018) and liquid-to-solid phase transitions have been reported for FUS and huntingtin, during the formation of aggregates associated with ALS and Huntington’s disease (Patel et al., 2015; Peskett et al., 2018). Given that citrate is a mitochondrial metabolite, it is tempting to speculate that citrate, either from stressed mitochondria or from inhibition of the cytosolic enzyme ATP-citrate lyase, might promote a similar phase transition in vivo. However, the concentrations of citrate in plasma and cerebrospinal fluid are on the order of 0.1–0.4 mM (Costello and Franklin, 2013; Westergaard et al., 2017) versus the mM levels of citrate required to activate SARM1 in vitro. Perhaps in the context of a living cell a lower threshold of citrate could promote this phase transition. Consistent with the latter possibility is our demonstration that lower amounts of citrate promote puncta formation and degeneration in vivo. Further research is needed to fully address this possibility.

Figure 7

Download asset Open asset

In summary, we show that SARM1 catalytic activity is enhanced by TIR domain multimerization and a phase transition reminiscent of that documented with other neurodegenerative disease-causing proteins. Therefore, inhibiting multimerization or this phase transition could be a target for future therapeutic development for SARM1-associated diseases. In the context of a degenerating axon, we rationalize this phase transition as SARM1 exhibiting gradual NAD⁺ hydrolase activity upon initial activation (which is demonstrated by in vitro ARM domain mutations relieving inter-TIR ARM domain contacts) (Bratkowski et al., 2020; Jiang et al., 2020) until a critical threshold is reached whereby SARM1 levels increase allowing SARM1 to multimerize and form larger puncta. We anticipate that higher ordered multimerization translates into increased activity and rapid depletion of NAD⁺, which further relieves autoinhibition by NAD⁺ (Jiang et al., 2020) and causes the axon to undergo catastrophic fragmentation and granular disintegration. This could shed light on a built–in protective mechanism to control the toxic effects of SARM1; thus, ensuring the neuron has ample time to circumvent the degenerative path, but when committed to it, degeneration is executed in a positive feed-forward mechanism resulting in abrupt fragmentation and granular disintegration during the latter stages of Wallerian degeneration.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. An S
   2. Kyoung M
   3. Allen JJ
   4. Shokat KM
   5. Benkovic SJ

   (2010) Dynamic Regulation of a Metabolic Multi-enzyme Complex by Protein Kinase CK2*

   Journal of Biological Chemistry 285:11093–11099.

   https://doi.org/10.1074/jbc.M110.101139

   • Google Scholar
2. 1. Blacklow SC
   2. Raines RT
   3. Lim WA
   4. Zamore PD
   5. Knowles JR

   (1988) Triosephosphate isomerase catalysis is diffusion controlled appendix: analysis of triose phosphate equilibria in aqueous solution by 31P NMR

   Biochemistry 27:1158–1167.

   https://doi.org/10.1021/bi00404a013

   • PubMed
   • Google Scholar
3. 1. Bratkowski M
   2. Xie T
   3. Thayer DA
   4. Lad S
   5. Mathur P
   6. Yang Y-S
   7. Danko G
   8. Burdett TC
   9. Danao J
   10. Cantor A
   11. Kozak JA
   12. Brown SP
   13. Bai X
   14. Sambashivan S

   (2020) Structural and Mechanistic Regulation of the Pro-degenerative NAD Hydrolase SARM1

   Cell Reports 32:107999.

   https://doi.org/10.1016/j.celrep.2020.107999

   • Google Scholar
4. 1. Carlsson E
   2. Ding JL
   3. Byrne B

   (2016) SARM modulates MyD88-mediated TLR activation through BB-loop dependent TIR-TIR interactions

   Biochimica et Biophysica Acta (BBA) - Molecular Cell Research 1863:244–253.

   https://doi.org/10.1016/j.bbamcr.2015.11.021

   • Google Scholar
5. 1. Chan CY
   2. Zhao H
   3. Pugh RJ
   4. Pedley AM
   5. French J
   6. Jones SA
   7. Zhuang X
   8. Jinnah H
   9. Huang TJ
   10. Benkovic SJ

   (2015) Purinosome formation as a function of the cell cycle

   PNAS 112:1368–1373.

   https://doi.org/10.1073/pnas.1423009112

   • Google Scholar
6. 1. Chen C-Y
   2. Lin C-W
   3. Chang C-Y
   4. Jiang S-T
   5. Hsueh Y-P

   (2011) Sarm1, a negative regulator of innate immunity, interacts with syndecan-2 and regulates neuronal morphology

   Journal of Cell Biology 193:769–784.

   https://doi.org/10.1083/jcb.201008050

   • Google Scholar
7. 1. Chuang C-F
   2. Bargmann CI

   (2005) A Toll-interleukin 1 repeat protein at the synapse specifies asymmetric odorant receptor expression via ASK1 MAPKKK signaling

   Genes & Development 19:270–281.

   https://doi.org/10.1101/gad.1276505

   • Google Scholar
8. 1. Costello LC
   2. Franklin RB

   (2013) A review of the important central role of altered citrate metabolism during the process of stem cell differentiation

   Journal of Regenerative Medicine and Tissue Engineering 2:1.

   https://doi.org/10.7243/2050-1218-2-1

   • PubMed
   • Google Scholar
9. 1. Du M
   2. Chen ZJ

   (2018) DNA-induced liquid phase condensation of cGAS activates innate immune signaling

   Science 361:704–709.

   https://doi.org/10.1126/science.aat1022

   • Google Scholar
10. 1. Essuman K
    2. Summers DW
    3. Sasaki Y
    4. Mao X
    5. DiAntonio A
    6. Milbrandt J

    (2017) The SARM1 Toll/Interleukin-1 Receptor Domain Possesses Intrinsic NAD + Cleavage Activity that Promotes Pathological Axonal Degeneration

    Neuron 93:e1335–e1343.

    https://doi.org/10.1016/j.neuron.2017.02.022

    • Google Scholar
11. 1. Gadda G
    2. Sobrado P

    (2018) Kinetic Solvent Viscosity Effects as Probes for Studying the Mechanisms of Enzyme Action

    Biochemistry 57:3445–3453.

    https://doi.org/10.1021/acs.biochem.8b00232

    • Google Scholar
12. 1. Geisler S
    2. Doan RA
    3. Strickland A
    4. Huang X
    5. Milbrandt J
    6. DiAntonio A

    (2016) Prevention of vincristine-induced peripheral neuropathy by genetic deletion of SARM1 in mice

    Brain 139:3092–3108.

    https://doi.org/10.1093/brain/aww251

    • Google Scholar
13. 1. Gerdts J
    2. Summers DW
    3. Sasaki Y
    4. DiAntonio A
    5. Milbrandt J

    (2013) Sarm1-Mediated Axon Degeneration Requires Both SAM and TIR Interactions

    Journal of Neuroscience 33:13569–13580.

    https://doi.org/10.1523/JNEUROSCI.1197-13.2013

    • Google Scholar
14. 1. Gerdts J
    2. Brace EJ
    3. Sasaki Y
    4. DiAntonio A
    5. Milbrandt J

    (2015) SARM1 activation triggers axon degeneration locally via NAD+ destruction

    Science 348:453–457.

    https://doi.org/10.1126/science.1258366

    • Google Scholar
15. 1. Hagen SJ

    (2010) Solvent viscosity and friction in protein folding dynamics

    Current Protein & Peptide Science 11:385–395.

    https://doi.org/10.2174/138920310791330596

    • PubMed
    • Google Scholar
16. 1. Henninger N
    2. Bouley J
    3. Sikoglu EM
    4. An J
    5. Moore CM
    6. King JA
    7. Bowser R
    8. Freeman MR
    9. Brown RH

    (2016) Attenuated traumatic axonal injury and improved functional outcome after traumatic brain injury in mice lacking Sarm1

    Brain 139:1094–1105.

    https://doi.org/10.1093/brain/aww001

    • Google Scholar
17. 1. Horsefield S
    2. Burdett H
    3. Zhang X
    4. Manik MK
    5. Shi Y
    6. Chen J
    7. Qi T
    8. Gilley J
    9. Lai J-S
    10. Rank MX
    11. Casey LW
    12. Gu W
    13. Ericsson DJ
    14. Foley G
    15. Hughes RO
    16. Bosanac T
    17. von Itzstein M
    18. Rathjen JP
    19. Nanson JD
    20. Boden M
    21. Dry IB
    22. Williams SJ
    23. Staskawicz BJ
    24. Coleman MP
    25. Ve T
    26. Dodds PN
    27. Kobe B

    (2019) NAD ⁺ cleavage activity by animal and plant TIR domains in cell death pathways

    Science 365:793–799.

    https://doi.org/10.1126/science.aax1911

    • Google Scholar
18. 1. Jiang Y
    2. Liu T
    3. Lee C-H
    4. Chang Q
    5. Yang J
    6. Zhang Z

    (2020) The NAD+-mediated self-inhibition mechanism of pro-neurodegenerative SARM1

    Nature 588:658–663.

    https://doi.org/10.1038/s41586-020-2862-z

    • Google Scholar
19. Preprint

    1. Julian V
    2. Byrne AB

    (2020) TIR-1/SARM1 inhibits axon regeneration

    bioRxiv.

    https://doi.org/10.1101/2020.06.23.165852

    • Google Scholar
20. 1. Knuckley B
    2. Bhatia M
    3. Thompson PR

    (2007) Protein Arginine Deiminase 4: Evidence for a Reverse Protonation Mechanism

    Biochemistry 46:6578–6587.

    https://doi.org/10.1021/bi700095s

    • Google Scholar
21. 1. Knuckley B
    2. Causey CP
    3. Jones JE
    4. Bhatia M
    5. Dreyton CJ
    6. Osborne TC
    7. Takahara H
    8. Thompson PR

    (2010) Substrate specificity and kinetic studies of pads 1, 3, and 4 identify potent and selective inhibitors of protein arginine deiminase 3

    Biochemistry 49:4852–4863.

    https://doi.org/10.1021/bi100363t

    • Google Scholar
22. 1. Kroschwald S
    2. Maharana S
    3. Mateju D
    4. Malinovska L
    5. Nüske E
    6. Poser I
    7. Richter D
    8. Alberti S

    (2015) Promiscuous interactions and protein disaggregases determine the material state of stress-inducible RNP granules

    eLife 4:e06807.

    https://doi.org/10.7554/eLife.06807

    • Google Scholar
23. 1. Loring HS
    2. Icso JD
    3. Nemmara VV
    4. Thompson PR

    (2020a) Initial kinetic characterization of sterile alpha and toll/interleukin receptor motif-containing protein 1

    Biochemistry 59:933–942.

    https://doi.org/10.1021/acs.biochem.9b01078

    • Google Scholar
24. 1. Loring HS
    2. Parelkar SS
    3. Mondal S
    4. Thompson PR

    (2020b) Identification of the first noncompetitive SARM1 inhibitors

    Bioorganic & Medicinal Chemistry 28:115644.

    https://doi.org/10.1016/j.bmc.2020.115644

    • Google Scholar
25. 1. Loring HS
    2. Thompson PR

    (2018) Kinetic mechanism of nicotinamide N -methyltransferase

    Biochemistry 57:5524–5532.

    https://doi.org/10.1021/acs.biochem.8b00775

    • Google Scholar
26. 1. Loring HS
    2. Thompson PR

    (2020) Emergence of SARM1 as a potential therapeutic target for wallerian-type diseases

    Cell Chemical Biology 27:1–13.

    https://doi.org/10.1016/j.chembiol.2019.11.002

    • Google Scholar
27. 1. Mack TGA
    2. Reiner M
    3. Beirowski B
    4. Mi W
    5. Emanuelli M
    6. Wagner D
    7. Thomson D
    8. Gillingwater T
    9. Court F
    10. Conforti L
    11. Fernando FS
    12. Tarlton A
    13. Andressen C
    14. Addicks K
    15. Magni G
    16. Ribchester RR
    17. Perry VH
    18. Coleman MP

    (2001) Wallerian degeneration of injured axons and synapses is delayed by a Ube4b/Nmnat chimeric gene

    Nature Neuroscience 4:1199–1206.

    https://doi.org/10.1038/nn770

    • Google Scholar
28. 1. Massoll C
    2. Mando W
    3. Chintala SK

    (2013) Excitotoxicity upregulates sarm1 protein expression and promotes wallerian-like degeneration of retinal ganglion cells and their axons

    Investigative Opthalmology & Visual Science 54:2771–2780.

    https://doi.org/10.1167/iovs.12-10973

    • Google Scholar
29. 1. Osborne TC
    2. Obianyo O
    3. Zhang X
    4. Cheng X
    5. Thompson PR

    (2007) Protein arginine methyltransferase 1: positively charged residues in substrate peptides distal to the site of methylation are important for substrate binding and catalysis

    Biochemistry 46:13370–13381.

    https://doi.org/10.1021/bi701558t

    • Google Scholar
30. 1. Osterloh JM
    2. Yang J
    3. Rooney TM
    4. Fox AN
    5. Adalbert R
    6. Powell EH
    7. Sheehan AE
    8. Avery MA
    9. Hackett R
    10. Logan MA
    11. MacDonald JM
    12. Ziegenfuss JS
    13. Milde S
    14. Hou Y-J
    15. Nathan C
    16. Ding A
    17. Brown RH
    18. Conforti L
    19. Coleman M
    20. Tessier-Lavigne M
    21. Zuchner S
    22. Freeman MR

    (2012) dSarm/Sarm1 is required for activation of an injury-induced axon death pathway

    Science 337:481–484.

    https://doi.org/10.1126/science.1223899

    • Google Scholar
31. 1. Patel SS
    2. Belmont BJ
    3. Sante JM
    4. Rexach MF

    (2007) Natively unfolded nucleoporins gate protein diffusion across the nuclear pore complex

    Cell 129:83–96.

    https://doi.org/10.1016/j.cell.2007.01.044

    • Google Scholar
32. 1. Patel A
    2. Lee HO
    3. Jawerth L
    4. Maharana S
    5. Jahnel M
    6. Hein MY
    7. Stoynov S
    8. Mahamid J
    9. Saha S
    10. Franzmann TM
    11. Pozniakovski A
    12. Poser I
    13. Maghelli N
    14. Royer LA
    15. Weigert M
    16. Myers EW
    17. Grill S
    18. Drechsel D
    19. Hyman AA
    20. Alberti S

    (2015) A liquid-to-solid phase transition of the ALS protein FUS accelerated by disease mutation

    Cell 162:1066–1077.

    https://doi.org/10.1016/j.cell.2015.07.047

    • Google Scholar
33. 1. Peskett TR
    2. Rau F
    3. O’Driscoll J
    4. Patani R
    5. Lowe AR
    6. Saibil HR

    (2018) A liquid to solid phase transition underlying pathological huntingtin exon1 aggregation

    Molecular Cell 70:e586–e601.

    https://doi.org/10.1016/j.molcel.2018.04.007

    • Google Scholar
34. 1. Schindelin J
    2. Arganda-Carreras I
    3. Frise E
    4. Kaynig V
    5. Longair M
    6. Pietzsch T
    7. Preibisch S
    8. Rueden C
    9. Saalfeld S
    10. Schmid B
    11. Tinevez JY
    12. White DJ
    13. Hartenstein V
    14. Eliceiri K
    15. Tomancak P
    16. Cardona A

    (2012) Fiji: an open-source platform for biological-image analysis

    Nature Methods 9:676–682.

    https://doi.org/10.1038/nmeth.2019

    • PubMed
    • Google Scholar
35. 1. Sporny M
    2. Guez-Haddad J
    3. Lebendiker M
    4. Ulisse V
    5. Volf A
    6. Mim C
    7. Isupov MN
    8. Opatowsky Y

    (2019) Structural evidence for an octameric ring arrangement of SARM1

    Journal of Molecular Biology 431:3591–3605.

    https://doi.org/10.1016/j.jmb.2019.06.030

    • Google Scholar
36. 1. Sporny M
    2. Guez-Haddad J
    3. Khazma T
    4. Yaron A
    5. Dessau M
    6. Shkolnisky Y
    7. Mim C
    8. Isupov MN
    9. Zalk R
    10. Hons M
    11. Opatowsky Y

    (2020) Structural basis for SARM1 inhibition and activation under energetic stress

    eLife 9:e62021.

    https://doi.org/10.7554/eLife.62021

    • Google Scholar
37. Book

    1. Sulston J
    2. Hodgkin J

    (1988)

     Methods in the Nematode C. elegans

    New York: Cold Spring Harbor Laboratory.

    • Google Scholar
38. 1. Summers DW
    2. Gibson DA
    3. DiAntonio A
    4. Milbrandt J

    (2016) SARM1-specific motifs in the TIR domain enable NAD ⁺ loss and regulate injury-induced SARM1 activation

    PNAS 113:E6271–E6280.

    https://doi.org/10.1073/pnas.1601506113

    • Google Scholar
39. 1. Ve T
    2. Vajjhala PR
    3. Hedger A
    4. Croll T
    5. DiMaio F
    6. Horsefield S
    7. Yu X
    8. Lavrencic P
    9. Hassan Z
    10. Morgan GP
    11. Mansell A
    12. Mobli M
    13. O'Carroll A
    14. Chauvin B
    15. Gambin Y
    16. Sierecki E
    17. Landsberg MJ
    18. Stacey KJ
    19. Egelman EH
    20. Kobe B

    (2017) Structural basis of TIR-domain-assembly formation in MAL- and MyD88-dependent TLR4 signaling

    Nature Structural & Molecular Biology 24:743–751.

    https://doi.org/10.1038/nsmb.3444

    • Google Scholar
40. 1. Wang JT
    2. Medress ZA
    3. Vargas ME
    4. Barres BA

    (2015) Local axonal protection by WldS as revealed by conditional regulation of protein stability

    PNAS 112:10093–10100.

    https://doi.org/10.1073/pnas.1508337112

    • Google Scholar
41. 1. Westergaard N
    2. Waagepetersen HS
    3. Belhage B
    4. Schousboe A

    (2017) Citrate, a Ubiquitous key metabolite with regulatory function in the CNS

    Neurochemical Research 42:1583–1588.

    https://doi.org/10.1007/s11064-016-2159-7

    • Google Scholar

Author details

1. Heather S Loring

   1. Department of Biochemistry and Molecular Pharmacology, UMass Medical School, Worcester, United States
   2. Program in Chemical Biology, UMass Medical School, Worcester, United States

   Contribution

   Conceptualization, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

2. Victoria L Czech

   Department of Neurobiology, UMass Medical School, Worcester, United States

   Contribution

   Formal analysis, Investigation, Methodology, Writing - original draft

   Competing interests

   No competing interests declared

3. Janneke D Icso

   1. Department of Biochemistry and Molecular Pharmacology, UMass Medical School, Worcester, United States
   2. Program in Chemical Biology, UMass Medical School, Worcester, United States

   Contribution

   Investigation, Writing - original draft

   Competing interests

   No competing interests declared

4. Lauren O'Connor

   Department of Neurobiology, UMass Medical School, Worcester, United States

   Contribution

   Formal analysis, Investigation, Methodology

   Competing interests

   No competing interests declared

5. Sangram S Parelkar

   1. Department of Biochemistry and Molecular Pharmacology, UMass Medical School, Worcester, United States
   2. Program in Chemical Biology, UMass Medical School, Worcester, United States

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

6. Alexandra B Byrne

   Department of Neurobiology, UMass Medical School, Worcester, United States

   Contribution

   Formal analysis, Investigation, Methodology, Writing - original draft

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7449-9188

7. Paul R Thompson

   1. Department of Biochemistry and Molecular Pharmacology, UMass Medical School, Worcester, United States
   2. Program in Chemical Biology, UMass Medical School, Worcester, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   Paul.Thompson@umassmed.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1621-3372

• 2,593

  views

• 481

  downloads

• 19

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.