Combinatorial chemistry identifies additional compounds that selectively inhibit DLK-dependent retrograde signaling while minimally affecting other axonal roles of DLK

Abstract

Dual leucine-zipper kinase (DLK) is implicated in at least two distinct processes that drive neurodegeneration: retrograde (axon-to-soma) signaling to activate pro-degenerative transcription programs, and axon-intrinsic action to drive Wallerian and Wallerian-like axon degeneration. Inhibiting DLK-dependent signaling is thus an attractive neuroprotective strategy, but compounds that inhibit all cellular pools of DLK cause unintended side effects. We recently successfully deployed a complementary approach to identify compounds that selectively block DLK retrograde (axon-to-soma) signaling and subsequent neurodegeneration by inhibiting acute, axonal palmitoylation of DLK. Here, we explored chemical space for one of our two most effective compounds and identified multiple analogs that are equally neuroprotective in a model of transcription-dependent neurodegeneration that requires DLK-dependent retrograde signaling. In contrast, our original hits and these additional analogs had minimal effect in two models of DLK-dependent, but transcription-independent, distal axon degeneration. Moreover, our original hits did not phenocopy the stabilization of axon survival factor proteins that is a well-described effect of conventional DLK kinase domain inhibitors. These findings reveal additional potential neuroprotective compounds and further support the notion that the pool of DLK that conveys axonal retrograde signals can be selectively targeted therapeutically.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-24517-3.

Introduction

Neurodegeneration is an irreversible process characterized by damage to, and loss of, neuronal structure and function¹. Neurodegeneration underlies multiple significant neurological disorders, ranging from acute conditions like traumatic nervous system injuries, to more chronic conditions such as amyotrophic lateral sclerosis (ALS), Alzheimer’s disease (AD), Parkinson’s disease (PD), glaucoma and chemotherapy-induced peripheral neuropathy (CIPN)^2–4. Unfortunately, in most cases, the progression of these diseases remains uncontrolled and hence they are incurable. Understanding the mechanisms underlying neurodegeneration is critical to develop therapeutic strategies that may halt or even reverse a diverse range of neuropathological conditions.

One protein that has emerged as a potential therapeutic target to slow or prevent neurodegeneration is the neuronally-enriched dual leucine-zipper kinase (DLK, gene name MAP3K12)². DLK knockout, knockdown or pharmacological inhibition is protective in preclinical models of many of the neurodegenerative conditions mentioned above^2,3. At the cellular/molecular level, DLK acts as an evolutionarily conserved sensor of axonal damage and stress and mediates at least two distinct downstream processes. In the proximal segment of stressed and damaged axons, DLK conveys retrograde signals back to neuronal nuclei, initiating transcription programs that often result in the degeneration of the neuronal cell body and proximal axons^5–8. The importance of this role is exemplified by the striking neuroprotection afforded by DLK loss when embryonic dorsal root ganglion (DRG) neurons are deprived of axonal neurotrophic support, and when axons of retinal ganglion cells (RGCs) are subjected to a crush injury (the latter serving as a partial model of the RGC axon damage that occurs in glaucoma)^5,7–11. In addition, DLK acts in damaged distal axons to facilitate an axon self-destruction program that involves Sterile alpha and Toll/interleukin-1 receptor motif-containing 1 (SARM1)^12–15. This axon self-destruction is classically referred to as Wallerian degeneration (WD) when induced by acute axonal injury and it is now clear that mechanistically and phenotypically similar ‘Wallerian-like’ degeneration is central to distal axon loss in chronic neuropathies^16–20. However, in contrast to retrograde signaling-induced degeneration, this role of DLK in driving Wallerian/Wallerian-like degeneration of distal axons is transcription-independent¹⁴. DLK’s role in distal axon degeneration is more prominent in DRG neurons compared to RGCs, and likely involves contributions from DLK’s close paralog, leucine-zipper kinase (LZK)^11–13,15.

These compelling preclinical data, obtained by numerous different groups using an array of experimental systems, all pointed to inhibition of DLK as a promising neuroprotective strategy. Based on these and other findings, a DLK kinase domain inhibitor, GDC-0134, progressed to a Phase I clinical trial²¹. However, patients who received high doses of GDC-0134 showed elevated plasma levels of neurofilament light chain (NFL; a widely-used biomarker for neurodegeneration²², and several also developed symptoms consistent with sensory neuropathy²¹. The effect of GDC-0134 on NFL was considered to be ‘on target’ because NFL was also elevated in the plasma of mice in which DLK was knocked out in adulthood²¹. These findings raised concerns about therapeutic strategies that directly target DLK’s kinase domain.

In seeking a more selective approach, we found that acute, axonal palmitoylation is required for retrograde pro-degenerative signaling by DLK²³. We therefore used a high content screening approach to identify small molecules that modulate DLK’s palmitoylation-dependent localization^23,24. This screen identified compounds (termed 8 and 13 in our original report²³ that are highly neuroprotective, likely due to their ability to selectively block this acute, axonal palmitoylation of DLK. However, we did not generate analogs of our initial hits to enhance their efficacy/potency and/or pharmacokinetic properties. We also did not assess the action of our initial hits and/or their analogs on DLK’s roles in transcription-independent degeneration of distal axons.

Here we report the neuroprotective properties of 22 analogs of 13. Several of these analogs blocked DLK-dependent retrograde signaling equally as effectively as 13 itself, although none was significantly superior to 13 in a neurodegeneration assay. Compared with 13, two analogs showed greatly improved pharmacokinetic stability, but these analogs were not effective inhibitors of DLK-dependent retrograde signaling. Importantly, though, 8, 13, and their most effective analogs only minimally affected transcription-independent Wallerian and Wallerian-like degeneration. In addition, although a conventional DLK kinase domain inhibitor stabilized axon survival factor proteins (a well-described result, taken to mean that a pool of DLK usually encounters and downregulates these proteins on anterograde vesicles), 8 and 13 did not. Together, these findings reveal additional neuroprotective compounds and further support the notion that the pool of DLK that conveys axonal pro-degenerative retrograde signals can be selectively targeted pharmacologically, without affecting other DLK-dependent processes.

Results

Multiple analogs of compound 13 inhibit DLK-dependent c-Jun phosphorylation in trophic factor-deprived DRG neurons

Our initial high content screen identified 13 as a highly neuroprotective compound. However, 13 showed only limited potency (low micromolar IC₅₀ in a neurodegeneration assay²³ and was also predicted to be labile to hydrolysis in preliminary absorption, distribution, metabolism and excretion (ADME) assays²³. We therefore synthesized 22 analogs of 13 in an attempt to improve these properties. These analogs (named 13−2 – 13–23; structures shown in Fig. 1) were first assessed for their ability to inhibit DLK-dependent signaling in DRG neurons subjected to trophic factor deprivation (TD). We and others have used TD-induced phosphorylation of the transcription factor c-Jun, mediated by DLK’s downstream targets the c-Jun N-terminal kinases (JNKs), as a readout of DLK-dependent pro-degenerative signaling^5,9,23. In line with previous studies, the DLK kinase domain inhibitor (DLKi, GNE-3511) effectively blocked TD-induced c-Jun phosphorylation^8,25. Though less effective than GNE-3511 in this assay, 13 still blunted TD-induced c-Jun phosphorylation. (Fig. 2A, B). Notably, several analogs of 13 also significantly reduced TD-induced c-Jun phosphorylation (Fig. 2A, B), indicating their ability to attenuate DLK-dependent pro-degenerative signaling in neurons.

Fig. 1.

Structures of analogs of Hit 13 used in this study. Analogs with activity in a DLK-dependent neuronal signaling assay (see Fig. 2) are denoted with an orange label. 

Fig. 2.

Multiple analogs of Hit 13 reduce TD-induced c-Jun phosphorylation. (A) Western blots of lysates from cultured embryonic rats DRG neurons that had been treated with vehicle (DMSO) or the indicated analogs of 13 from Fig. 1, prior to TD. The secondary antibody used on the p-cJun blot also weakly recognizes residual anti-NGF IgG used during TD (asterisk). Subsets of compounds were assayed side-by-side, but in batches indicated by spaces between individual blots. (B) Quantified data from (A), of p-c-Jun: tubulin, normalized to TD (vehicle) condition. Orange bars denote compounds whose effect on p-c-Jun: tubulin differed significantly from the TD (vehicle) condition. Statistical significance versus TD (vehicle) condition was as follows: 13: p = 0.0316; 13 − 7: p = 0.0255; 13–13: p = 0.034; 13–14: p = 0.008; 13–18: p = 0.0031; 13–19: p = 0.0027, Kruskal Wallis tests with Dunn’s post hoc test for multiple comparisons. Error bars represent the standard error of the mean (SEM).

A subset of analogs of 13 also protect DRG cell bodies and axons against TD-induced neurodegeneration

We further investigated whether analogs of 13 that inhibit p-c-Jun signaling could also protect DRG neurons from degeneration after prolonged TD. DRG cultures subjected to TD for 45 h in the presence of vehicle (DMSO) displayed pronounced axonal blebbing/beading and cell body degeneration in phase contrast images (Fig. 3A). Consistent with our prior work, the parent compound 13 preserved both neuronal cell bodies and axons from the deleterious effects of prolonged TD (Fig. 3A, B). The analogs 13−7 and 13–13 also protected neuronal cell bodies and axons in this assay, although neither was more neuroprotective than 13 itself (Fig. 3A, B).

Fig. 3.

A subset of Hit13 derivatives reduce TD-induced neurodegeneration. (A) Phase contrast images of DRG neurons that were maintained in NGF (+ NGF) or subjected to trophic factor deprivation (TD) for 45 h in the presence of vehicle (DMSO) or the indicated hit compounds that significantly reduced TD-induced c-Jun phosphorylation in Fig. 2. Right-hand panels show magnified views of the boxed region in the corresponding left-hand panel. Scale bars, 20 μm (all panels). (B) Extent of neural degeneration, quantified from images from A. Orange bars denote compounds whose effect on degeneration differed significantly from TD (vehicle) condition. Statistical significance versus TD (vehicle) condition was as follows: 13: p = 0.0021; 13−7: p = 0.0017; 13–13: p = 0.0429. TD (vehicle) condition also differed significantly from + NGF (vehicle) condition: p < 0.0001. Kruskal Wallis tests with Dunn’s post hoc test for multiple comparisons. Error bars represent the standard error of the mean (SEM).

ADME properties of analogs of 13

Although no analog was more protective than 13 itself in our neurodegeneration assay (Fig. 3), these analogs might nonetheless have preferential pharmacokinetic properties that could be important for translational studies. We therefore assessed the drug-like properties of all analogs, focusing on their aqueous solubility and their stability in rat liver microsomes (Table 1). All analogs of 13 displayed low to moderate maximum kinetic aqueous solubility. With two exceptions (13−6, 13–22), all analogs also displayed low stability in mouse liver microsomes in the presence of NADPH, indicating lability to P450 oxidation, enzymatic hydrolysis or a combination of both (Table 1). However, despite their more encouraging pharmacokinetic properties, neither 13−6 nor 13–22 significantly inhibited TD-induced c-Jun phosphorylation (Fig. 2A, B).

Table 1.

In vitro ADME data for analogs of hit 13. Data for the parent compound (13) are taken from²³.

We also assessed the extent to which analogs of 13 inhibit three major metabolizing CYP450 enzymes (CYPs 3A4, 2D6 and 2C9). Most analogs displayed no appreciable inhibition of these enzymes. The exception to this was the isoxazole moiety-containing analog 13−12, which detectably inhibited CYP450 3A4 and 2D6 (Table 1).

Finally, we assessed the activity of the two most neuroprotective analogs, 13−7 and 13–13, in the MDCK-MDR1 assay, a measure of likely CNS penetration. 13−7 displayed good permeability through MDCK-MDR1 cells with no evidence of efflux (Table 2). CNS penetration for this compound is predicted to be high. 13–13 displayed good permeability through MDCK-MDR1 cells but the efflux ratio indicates that some efflux through p-glycoprotein may be occurring (Table 2). CNS penetration is predicted to be moderate for this compound. These findings suggest that, despite their low stability, the predicted CNS penetration and lack of CYP450 inhibition seen with 13 and its most neuroprotective analogs increase their potential to treat DLK-dependent degeneration in nervous system disorders.

Table 2.

MDCK-MDR1 assay results. The indicated compounds were assayed as described in Methods to determine their likely blood-brain barrier permeability.

Compound	Recovery %		A to B Papp (× 10−6 cm/s)	B to A Papp (× 10−6 cm/s)	Efflux ratio (A-B/B-A)	Brain penetration classification
	A-B	B-A
(13–7)	52.3	89.7	10.3	11.2	1.09	High
(13–13)	33.4	74.9	10.2	56.4	5.54	Moderate

Neuroprotective hits only partially and transiently protect DRG distal axons from injury induced-Wallerian degeneration

In seeking to further evaluate their therapeutic potential, we assessed the ability of our original compounds (8 and 13), and 13’s two most neuroprotective analogs (13−7 and 13–13) to inhibit DLK-dependent Wallerian/Wallerian-like degeneration of distal axons^12,13. In contrast to the transcription-dependent mechanism that drives TD-induced degeneration²⁶, acute roles of DLK in WD must be transcription-independent (because severed distal axons are physically disconnected from the cell body) and thus do not involve axon-to-soma signaling^12,13. We therefore plated dissociated DRG neurons as ‘spot’ cultures, in which neuronal cell bodies are plated in the center of a well and extend axons outwards towards the periphery of the well¹³. Axons can then be selectively transected using a biopsy punch. Treatment of DRG spot cultures with two established WD inhibitors, DLKi-3511 and DSRM-3716 (the latter an inhibitor of the axo-degenerative NADase SARM1 and henceforth referred to here as SARM1i;²⁷ at the time of injury resulted in significant protection against degeneration 4 h later. At this timepoint, 8, 13 and 13−7 were slightly protective (Fig. 4A, B). However, although GNE-311 and SARM1i still protected injured axons at 6 h post-injury, 8, 13, 13−7 and 13–13 did not (Fig. 4A, C). These findings suggest that, despite their effective block of DLK-dependent retrograde signaling, 8, 13 and analogs of the latter only minimally affect roles of DLK in WD.

Fig. 4.

Neuroprotective hits only transiently protect against injury-induced Wallerian degeneration of distal axons. (A) Representative phase contrast images of distal axons of DRG neurons in ‘spot’ cultures, taken at the indicated times after axonal injury, which was performed immediately after treatment with the indicated compounds. Control axons (top panels) were left uninjured. (B) Quantified extent of degeneration of axons from A at 4 h post-injury. Statistical significance versus injured vehicle control was as follows: SARM1i: p = 0.0036; DLKi-3511: p = 0.0011; 8: p = 0.0432; 13: p = 0.0498; 13−7: p = 0.0301. Injured (vehicle) condition also differed significantly from uninjured (vehicle) condition: p < 0.0001. ANOVA, Dunnett’s post hoc test. (C) As (B), but for axons imaged 6 h post-injury. Scale bars, 20 μm (all panels). Statistical significance versus vehicle control was as follows: SARM1i: p < 0.0001; DLKi-3511: p < 0.0001. Injured (vehicle) condition also differed significantly from uninjured (vehicle) p < 0.0001. ANOVA, Dunnett’s post hoc test. Error bars represent the standard error of the mean (SEM).

Hit compounds, even when pre-incubated, do not protect DRG distal axons against vincristine-induced Wallerian-like degeneration

Lastly, we assessed the ability of 8, 13 and analogs of the latter to inhibit Wallerian-like degeneration induced by the chemotherapeutic drug vincristine. Vincristine-induced axon degeneration is reported to be DLK-dependent but is transcription-independent^12–14. This latter finding increases the likelihood that roles of DLK in vincristine-induced degeneration involve intrinsic DLK action in distal axons rather than retrograde signaling. Consistent with this model, neither 8, 13 nor analogs of the latter protected axons against vincristine-induced degeneration whereas SARM1i, added at the same time as vincristine, was significantly axo-protective. GNE-3511 added at the same time as vincristine did not protect axons from vincristine-induced degeneration (Fig. 5A, B). However, we realized that prior reports of DLK involvement in vincristine-induced degeneration involved long-term genetic KO, or pre-treatment with pharmacologic inhibitors^12,13. Consistent with these reports, adding GNE-3511 to cultures prior to vincristine treatment significantly protected axons against degeneration (Fig. 5C, D). However, neither 8, 13 nor analogs of the latter protected axons against vincristine-induced degeneration, even when preincubated (Fig. 5E, F). Surprisingly, SARM1i was less axo-protective when preincubated, perhaps reflecting some instability of this compound during the extended pre-treatment period (Fig. 5E, F).

Fig. 5.

Hit compounds do not protect DRG distal axons from vincristine-induced degeneration, even when pre-incubated. (A) Images of distal axons of DRG neuron spot cultures 18 h after treatment with 40 nM vincristine and concomitant addition of the indicated compounds. Scale bar: 20 μm. (B) Quantification of axonal degeneration from (A). SARM1i reduced vincristine-induced degeneration, compared to vehicle treatment (p = 0.0386). The vehicle-only (no vincristine) condition also differed significantly from the vincristine/vehicle-treated condition (p < 0.0001). ANOVA with Dunnett’s post hoc test; error bars represent SEM. (C) Distal axons of DRG neuron spot cultures imaged 18 h after treatment of cultures with vincristine or vehicle. DLKi-3511 was added simultaneously with vincristine (time of vin) or 24 h prior to vincristine (24 h pre-treatment). Scale bar: 20 μm. (D) Quantification of (C). DLKi-3511 pretreatment reduced vincristine-induced degeneration versus vincristine/vehicle control (p = 0.0002). The vehicle-only (no vincristine) condition also differed significantly from the vincristine/vehicle-treated condition (p < 0.0001). ANOVA with Dunnett’s post hoc test; error bars represent SEM. (E) Images of distal axons 18 h after treatment with 40 nM vincristine, from DRG spot cultures that had been treated with the indicated compounds 24 h prior to vincristine addition. Scale bar: 20 μm. (F) Quantification of E. DLKi-3511 pretreatment reduced vincristine-induced degeneration (p = 0.0123). The vehicle-only (no vincristine) condition also differed significantly from the vincristine/vehicle-treated condition (p = 0.0004). ANOVA with Dunnett’s post hoc test; error bars represent SEM.

Finally, we assessed whether the stabilization of the unstable axon survival factors nicotinamide mononucleotide adenylytransferase-2 (NMNAT2) and Stathmin-2 (STMN2) that is seen with conventional inhibition or genetic loss of DLK and its close paralog leucine-zipper kinase (LZK;¹³, is also seen with our original compounds. This stabilization is taken to mean that DLK/LZK usually downregulate NMNAT2/STMN2 on the anterograde vesicles upon which these survival factors are transported^13,28,29. We readily reproduced the stabilization of NMNAT2 and STMN2 caused by DLKi-3511, but did not observe similar stabilization by either 8 or 13 (Fig. 6A-C). Consistent with prior reports^13,30,31, we observed rapid, injury-induced degradation of both NMNAT2 and STMN2 in injured axons (Fig. 6A), which both supported the validity of antibodies used in these experiments and further confirmed the efficacy of the Wallerian degeneration mechanism in our cultures.

Fig. 6.

A conventional DLK kinase domain inhibitor elevates axonal survival factor levels in DRG neurons but neuroprotective hits do not. (A) Left panels: Western blots of lysates from cultured DRG neurons that had been treated with the indicated compounds for 4 h prior to lysis. Right panels: Western blots of axon-only lysates of DRG spot cultures that had been subjected to injury or left uninjured. Left and right panel images are taken from the same blot but are separated to indicate removal of intervening spacer lanes. The co-migrating, injury-sensitive bands seen in the right panels support the specificity of the NMNAT2 and STMN2 antibodies used. (B) Quantification of (A). DLKi-3511, but not 8 or 13, significantly increased NMNAT2 compared to the vehicle-treated control (p = 0.0339), ANOVA with Dunnett’s post hoc test; error bars represent SEM. (C) DLKi-3511, but not 8 or 13, significantly increased STMN2 compared to the vehicle-treated control (p = 0.0112), ANOVA with Dunnett’s post hoc test; error bars represent SEM.

Discussion

Targeting the DLK pathway has proven effective in a variety of preclinical models, including traumatic brain injury, traumatic optic/glaucoma neuropathy, excitotoxic neurodegeneration, AD, PD, and CIPN^{3,9,10,32–35}. However, translating these preclinical results into clinical applications has thus far been problematic, exemplified by the symptoms of sensory neuropathy and elevated plasma neurofilament light chain (NFL) levels reported for patients receiving high doses of a DLK kinase domain inhibitor²¹. We recently reported that these two phenotypes might be related because a DLK kinase domain inhibitor disrupts the axonal cytoskeleton of DRG neurons²³, and we described two compounds, 8 and 13, that prevent DLK-dependent retrograde signaling without inducing this cytoskeletal disruption²³. In this current report, we expanded these studies to identify additional neuroprotective compounds and found that our original hits 8 and 13, and the two most protective analogs of the latter, have only minor, transient effects on another axonal role of DLK, its ability to control WD/Wallerian-like degeneration (Figs. 4 and 5). Moreover, in contrast to the effect of a conventional DLK kinase domain inhibitor (GNE-3511), neither 8 nor 13 lead to stabilization of anterogradely-supplied axon survival factor proteins (Fig. 6). These findings further support the notion that it is possible to identify compounds that selectively inhibit DLK-dependent retrograde signaling while sparing other DLK functions in distal axons.

Although neither 8 nor 13 provides prolonged protection against WD, there is a slight protective effect at 4 h post-injury (Fig. 4A, B). What might be the reason for this protection? Both 8 and 13 were identified in a screen that assesses DLK’s palmitoylation-dependent localization, and both compounds prevent stimulus-dependent DLK palmitoylation in axons²³. The modest protection seen at 4 h could therefore be due to transient palmitoylation (of likely a minor pool) of DLK that also contributes to distal WD. By 6 h (a time at which 8 and 13 are no longer protective) this palmitoylation-dependent mechanism may no longer be essential, although the continued protection by DLKi-3511 at 6 h post-injury suggests that DLK activity is still important for WD at this later time point. However, we cannot rule out that the modest effect at 4 h is due to off-target activity of our compounds, and we further note that these two possible explanations are not mutually exclusive.

Although our novel compounds only minimally protected against injury-induced WD and did not protect axons against vincristine-induced degeneration, our findings support prior reports that GNE-3511 can protect axons against these insults¹³. For vincristine-induced degeneration, we found that GNE-3511 was only protective when added prior to vincristine treatment. Despite questions regarding deleterious side effects of current DLK kinase domain inhibitors related to GNE-3511²¹, this mode of action is still compatible with treatment of CIPN, in which a patient can be treated with a potential axo-protective compound prior to taking the chemotherapeutic agent itself. The requirement for preincubation is consistent with prior studies that prolonged block of the DLK pathway increases levels of the axo-protective survival factors NMNAT2 and STMN2^13,36. One interpretation of this finding is that a sub-pool of DLK regulates, and perhaps physically encounters, the anterograde transport vesicles via which NMNAT2 and STMN2 are trafficked into axons. Consistent with the model that our compounds preferentially inhibit DLK-dependent retrograde signaling, though, neither 8 nor 13 phenocopied the increase in NMNAT2 and STMN2 protein levels seen with GNE-3511 (Fig. 6A-C).

In prior work, we showed that virally-infected DLK-GFP co-traffics with other DLK/JNK pathway proteins on axonal transport vesicles, and that a subset of these vesicles are transported retrogradely^8,37. Moreover, our initial compounds 8 and 13 block stimulus-dependent recruitment of DLK to axonal puncta that likely represent this vesicle population²³, see also³⁸. It would therefore be interesting to see if derivatives of 8 and 13 also prevent DLK vesicular recruitment and/or trafficking. However, because we initially tested a large number of analogs (22) in this study, we decided to focus on functional assays that could readily be run in parallel (western blotting of c-Jun phosphorylation; phase contrast imaging of neurodegeneration at lower magnification (Figs. 2 and 3)). In the future, it would be informative to perform live imaging studies in axons treated with the most promising of these analogs.

While our study represents a promising step forward, several challenges remain if compounds based on 8 and/or 13 are to be developed therapeutically to block DLK-dependent pro-degenerative signaling. In particular, it will be important to identify compounds with enhanced potency and improved pharmacokinetic properties. Our efforts to optimize 13 improved such properties but only at the expense of neuroprotective ability (Table 1; Fig. 2A, B). However, our synthetic chemistry efforts focused on analogs that retained the adamantyl group of 13, which may be an oxidative metabolic weak spot. Compounds that substitute the adamantyl group for a structurally similar but more stable moiety may have greater therapeutic promise. Moreover, we did not attempt to synthesize analogs of our other initial hit 8; such compounds might also show increased stability and/or potency compared with the parent molecule.

Together with our prior work, this study raises the possibility that targeting specific, functionally critical post-translational modifications is an attractive orthogonal approach to direct enzymatic inhibition when seeking to inhibit DLK or other pro-degenerative signaling pathways.

Experimental procedures

All procedures followed the National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) at Temple University. All methods were performed in accordance with the relevant guidelines and regulations of the IACUC and Institutional Biosafety Committee at Temple University. All methods are reported in accordance with ARRIVE guidelines (https://arriveguidelines.org).

Chemicals

DLK inhibitor GNE-3511 and SARM1 inhibitor DSRM-3716 were from Cayman Chemicals and R&D systems, and were used at final concentrations of 500 nM and 10 µM, respectively, in DMSO. Fresh stocks of 8 and 13, and analogs of 13 were synthesized at Temple University Moulder Institute for Drug Discovery (synthesis details available on request) and purity and integrity were confirmed by HPLC. All other reagents were from ThermoFisher Scientific and were of the highest reagent grade.

Antibodies

The following primary antibodies, raised in the indicated species, were used: anti-NGF (sheep, CedarLane, #CLMCNET-031); phospho–c-Jun (Ser⁶³) (rabbit, Cell Signaling Technology, #91952); alpha-tubulin (mouse, Cell Signaling Technology, #3873), STMN2 (rabbit, Protein Tech, #10586-1-AP). An antibody against a previously reported epitope on NMNAT2 (amino acids 232–253 from mouse NMNAT2;³⁹⁾ was raised in rabbits and affinity purified by Yenzym Antibodies.

Western blotting

Western blotting was performed using standard methods, with horseradish peroxidase-conjugated secondary antibodies. Signals were visualized by enhanced chemiluminescence. Full, uncropped images of western blots presented in the main Figures are shown in Supplementary Figs. 1 and 2.

Chemical synthesis

General

The common intermediate N-1-adamantyl-4-amino-benzenesulfonamide was purchased from Combiblocks (San Diego, CA, cat. # QY-8266). Structure and purity were confirmed by 1H-NMR and LC/MS. Other reagents were purchased from commercial suppliers and used without further purification. Unless stated otherwise, reactions were under a nitrogen atmosphere. ¹H-NMR data were collected on a Bruker 400 mHz Avance III spectrometer at ambient temperature in the identified solvent. Peak positions are given in parts per million downfield from tetramethylsilane as the internal standard. LC-MS analysis was performed on an Agilent Technologies 1200 series LC system coupled to a 6300 quadrapole MS. Normal phase silica gel chromatography was performed using a Teledyne ISCO Combiflash Rf system with UV detection at 220 nM and 254 nM.

Synthesis of compounds 13−2 – 13–23

Synthesis of compounds 13−2 – 13−12

Amides 13−2 – 13−12 were prepared by acylating N-1-adamantyl-benzenesulfonamide with appropriate acid chlorides using a variation of a previously described procedure⁴⁰ The synthesis of compound 13 − 2 serves as an example of the general procedure.

Synthesis of N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-4-fluorobenzamide (13−2)

A solution of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol) in anhydrous tetrahydrofuran (10 mL) was treated with 4-fluorobenzoyl chloride (21.2 µL, 179 µmol) followed by dry triethylamine (35 µl, 244.5 µmol). The resulting mixture was stirred at room temperature for 18 h, during which time a white precipitate formed. The mixture was concentrated on a rotary evaporator and then partitioned between dichloromethane and 5% aqueous sodium bicarbonate. The aqueous layer was washed twice more with dichloromethane and the combined organic layers were dried over anhydrous sodium sulfate, filtered and concentrated on a rotary evaporator. Purification of the residue by chromatography on silica gel using a gradient of methanol in dichloromethane provided 67 mg (86% yield) of the desired product as a white solid. ¹H-NMR (CDCl₃) δ 9.34 (s, 1H), 8.32 (m, 2 H), 7.99 (d, 2 H, J = 8.2 Hz), 7.45 (d, 2 H, 8,2 Hz), 7.35 (s, 1H), 7,33 (m, 2 H), 2,19 (m, 2 H), 1.96 (m, 4 H), 1.85 (m, 3 H), 1.74 (m, 6 H); MS (ESI+): m/z 429 (M + H)⁺.

The following compounds were prepared using the method described for compound 13−2:

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-2-chlorobenzamide (13−3). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 2-chlorobenzoyl chloride (22.7 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 76% yield and isolated as a light yellow solid (55 mg). ¹H-NMR (CDCl₃) δ 9.82 (s, 1H), 7.94 (d, 2 H, J = 8.7 Hz), 7.52 (d, 2 H, J = 8.7 Hz), 7.44 (m, 4 H), 5.06 (s, 1H), 2.26 (m, 2 H), 1.98 (m, 2 H), 1.88 (m, 3 H), 1.78 (m, 6 H); MS (ESI+): m/z 445/447 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-3-chlorobenzamide (13−4). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3-chlorobenzoyl chloride (23 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 78% yield and isolated as a yellow solid (56 mg). ¹H-NMR (CDCl₃) δ 9.75 (s, 1H), 7.81 (d, 2 H, J = 8.5 Hz), 7.77 (m, 1H), 7.59 (m, 2 H), 7.41 (d, 2 H, J = 8.5 Hz), 5.10 (s, 1H), 2.20 (m, 2 H), 1.90 (m, 3 H), 1.75 (m, 6 H); MS (ESI+): m/z 445/447 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-3,5-dichlorobenzamide (13−5). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3,5-dichlorobenzoyl chloride (34 mg, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 78% yield and isolated as a yellow solid (56 mg). ¹H-NMR (CDCl₃) δ 9.68 (s, 1H), 7.83 (d, 2 H, J = 8.3 Hz), 7.73 (m, 1H), 7.67 (m, 1H), 7.45 (d, 2 H, J = 8.3 Hz), 5.13 (s, 1H), 2.19 (m, 2 H), 1.88 (m, 3 H), 1.74 (m, 6 H); MS (ESI+): m/z 480/482 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-4-methoxybenzamide (13−6). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol),4-methoxy-benzoyl chloride (24 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 84% yield and isolated as a white solid (60 mg). ¹H-NMR (CDCl₃) δ 9.70 (s, 1H), 7.72 (d, 2 H, J = 8.0 Hz), 7.89 (d, 2 H, J = 8.2 Hz), 7.45 (d, 2 H, J = 8.0 Hz), 7.15 (d, 2 H, J = 8.2 Hz), 5.08 (s, 1H), 3.79 (s, 3 H), 2.17 (m, 2 H), 1.85 (m, 3 H), 1.71 (m, 6 H); MS (ESI+): m/z 441 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-4-(trifluoromethyl)benzamide (13−7). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 4-(trifluoromethyl)benzoyl chloride (24.2 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 71% yield and isolated as a beige solid (55 mg). ¹H-NMR (CDCl₃) δ 9.38 (s, 1H), 7.88 (d, 2 H, J = 8.5 Hz), 7.75 (d, 2 H, J = 8.3 Hz), 7.67 (d, 2 H, J = 8.3 hz), 7.23 (d, 2 H, J = 8.5 Hz), 5.011 (s, 1H), 2.20 (m, 2 H), 1.87 (m, 3 H), 1.75 (m, 6 H); MS (ESI+): m/z 479 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-3,4-fluorobenzamide (13−8). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3,5-dichloro-benzoyl chloride (22.5 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 70% yield and isolated as a white solid (51 mg). ¹H-NMR (CDCl₃) δ 9.70 (s, 1H), 7.85 (d, 2 H, J = 8.2 Hz), 7.88 (m, 1H), 7.63 (m, 1H), 7.43 (d, 2 H, J = 8.2 Hz), 7.35 (m, 1H), 5.03 (s, 1H), 2.22 (m, 2 H), 1.90 (m, 3 H), 1.73 (m, 6 H); MS (ESI+): m/z 447 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-3,5-difluorobenzamide (13−9). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3,5-difluorobenzoyl chloride (22.4 µL 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 72% yield and isolated as an off-white solid (52 mg). ¹H-NMR (CDCl₃) δ 9.47 (s, 1H), 7.90 (d, 2 H, J = 8.1 Hz), 7.49 (d, 2 H, J = 8.1 Hz), 7.40 (m, 2 H), 6.95 (m, 1H), 2.22 (m, 2 H), 1.87 (m, 3 H), 1.79 (m, 6 H); MS (ESI+): m/z 447 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)-2-fluorobenzamide (13−10). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 2-fluorobenzoyl chloride (20.2 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 70% yield and isolated as a beige solid (51 mg). ¹H-NMR (CDCl₃) δ 9.78 (s, 1H), 8.10 (m, 1H), 7.88 (d, 2 H, J = 8.8 Hz), 7.55 (m, 1H), 7.50 (d, 1H, J = 8.8 Hz), 7.25 (m, 1H), 7.10 (m, 1H), 5.09 (s, 1H), 2.23 (m, 2 H), 1.96 (m, 2 H), 1.88 (m, 3 H), 1.75 (m, 6 H); MS (ESI+): m/z 429 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)cyclohexanecarboxamide (13−11). Prepare from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), cyclohexane carboxylic acid chloride (24.3 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 87% yield and isolated as a yellow solid (59 mg). ¹H-NMR (CDCl₃) δ 9.33 (s, 1H), 7.66 (d, 2 H, J = 8.1 Hz), 7.44 (d, 2 H, J = 8.1 Hz), 5.03 (s, 1H), 2.49 (m, 1H), 2.47 (m, 1H), 2.20 (m, 2 H), 1.88 (m, 4 H),1.84–1.76 (broad m, 3 H), 1.76–1.66 (broad m, 8 H), 1.56–1.46 (broad m, 8 H); MS (ESI+): m/z 417 (M + H)⁺.

N-(4-(N-Adamantan-1-yl)sulfamoyl)phenyl)isoxazole-5-carboxamide (13−12). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), isoxazole-5-carbonyl chloride (173 µL, 179 µmol) and triethylamine (35 µL, 244.5 µmol) in anhydrous tetrahydrofuran (10 mL) in 65% yield and isolated as a yellow oil (43 mg). ¹H-NMR (CDCl₃) δ 9.98 (s, 1H),8.30 (d, 1H, J = 4.7 Hz), 7.65 (d, 2 H, J = 8.3 Hz), 7.40 (d, 2 H, J = 8.3 Hz), 6.05 (d, 1H, N = 4.7 Hz), 5.13 (s, 1H), 2.20 (m, 2 H), 1.96 (m, 2 H), 1.84 (m, 3 H, 1.74 (m, 6 H); MS (ESI+): m/z 402 (M + H)⁺.

Synthesis of compounds 13–13 – 13–19

Ureas 13–13 – 13–19 were prepared by acylating N-1-adamantyl-benzenesulfonamide with appropriate isocyanates. The synthesis of compound 13–13 serves as an example of the general procedure.

Synthesis of N-Adamantan-1-yl)-4-(3-(4-fluorophenyl)ureido)benzenesulfonamide (13–13)

A solution of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol) in anhydrous tetrahydrofuran (10 mL) was treated with 4-fluorophenylisocyanate (22 µL, 195 µmol) and dry triethylamine (35 µL, 244.5 mmol). The resulting mixture was stirred at room temperature for 48 h. The mixture was concentrated to dryness on a rotary evaporator and the residue was purified by chromatography on silica gel using a gradient of methanol in dichloromethane. The desired product (39 mg, 41% yield) was isolated as a white solid. ¹H-NMR (CDCl₃) δ 8.80 (bs, 2 H), 7.70 (m, 4 H), 7.49 (d, 2 H, J = 8.6 Hz), 7.25 (t, 2 H, J = 8.4 Hz), 5.21 (s, 1H), 2.18 (m, 2 H), 1.88 (m, 3 H), 1.75 (m, 6 H); MS (ESI+): m/z 444 (M + H)⁺.

The following compounds were prepared using the method described for compound 13–13:

N-Adamantan-1-yl)-4-(3-(4-chlorophenyl)ureido)benzenesulfonamide (13–14). Prepared from of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 4-chlorophenylisocyanate (25 µL, 195 µmol) and triethylamine (35 µL, 244.5 mmol).in anhydrous tetrahydrofuran (10 mL). The desired product (32 mg, 43% yield) was isolated as an light yellow solid. ¹H-NMR (CDCl₃) δ 8.74 (bs, 2 H), 7.72 (d, 2 H, J = 8.4 Hz), 7.64 (d, 2 H, J = 8.1 Hz), 7.50 (d, 2 H, J = 8.1 Hz), 7.38 (d, 2 H, J = 8.4), 5.10 (s, 1H), 2.21 (m, 2 H), 1.84 (m, 3 H), 1.72 (m, 6 H); MS (ESI+): m/z 460/462 (M + H)⁺.

N-Adamantan-1-yl)-4-(3-(4-methoxyphenyl)ureido)benzenesulfonamide (13–15). Prepared from of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 4-methoxyphenylisocyanate (25 µL, 195 µmol) and triethylamine (35 µL, 244.5 mmol).in anhydrous tetrahydrofuran (10 mL). The desired product (26 mg, 35% yield) was isolated as an off-white solid. ¹H-NMR (CDCl₃) δ 8.70 (s, 1H), 8.40 (s, 1H), 7.72 (d, 2 H, J = 8.4 Hz), 7.70 (d, 2 H, J = 8.1 Hz), 7.46 (d, 2 H, J = 8.1 Hz), 7.25 (d, 2 H, J = 8.3), 6.91 (d, 2 H, J = 8.3 Hz), 5.16 (s, 1H), 3.85 (s, 3 H), 2.20 (m, 2 H), 1.82 (m, 3 H), 1.70 (m, 6 H); MS (ESI+): m/z 456 (M + H)⁺.

N-Adamantan-1-yl)-4-(3-(3-fluorophenyl)ureido)benzenesulfonamide (13–16). Prepared from of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3-fluorophenylisocyanate (22 µL, 195 µmol) and triethylamine (35 µL, 244.5 mmol).in anhydrous tetrahydrofuran (10 mL). The desired product (32 mg, 44% yield) was isolated as an off-white solid. ¹H-NMR (CDCl₃) δ 9.10 (s, 1H), 8.4 (s, 1H), 7.82 (m, 1H), 7.72 (d, 2 H, J = 8.2 Hz), 7,50 (d, 2 H, J = 8.2 Hz), 7.48–7.38 (broad m, 4 H), 7.00 (m, 1H), 5.16 (s, 1H), 2.16 (m, 2 H), 1.90 (m, 3 H), 1.72 (m, 6 H); MS (ESI+): m/z 444 (M + H)⁺.

N-Adamantan-1-yl)-4-(3-(2-fluorophenyl)ureido)benzenesulfonamide (13–17). Prepared from of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 2-fluorophenylisocyanate (22 µL, 195 µmol) and triethylamine (35 µL, 244.5 mmol).in anhydrous tetrahydrofuran (10 mL). The desired product (29 mg, 40% yield) was isolated as a yellow solid. ¹H-NMR (CDCl₃) δ 8.92 (s, 1H), 8.70 (s, 1H), 7.90 (m, 1H), 7.72 (d, 2 H, J = 8.2 Hz), 7.51 (d, 2 H, J = 8.2 Hz), 7.20–7.05 (broad m, 2 H), 7.02 m, 1H), 5.20 (s, 1H), 2.18 (m, 2 H), 1.88 (m, 3 H), 1.75 (m, 6 H). MS (ESI+): m/z 444 (M + H)⁺.

N-Adamantan-1-yl)-4-(3-(3-methoxyphenyl)ureido)benzenesulfonamide (13–18). Prepared from of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3-methoxyphenylisocyanate (29 µL, 195 µmol) and triethylamine (35 µL, 244.5 mmol).in anhydrous tetrahydrofuran (10 mL). The desired product (27 mg, 37% yield) was isolated as a white solid. ¹H-NMR (CDCl₃) δ 8.72 (s, 1H), 8.50 (s, 1H), 7.70 (d, 2 H, J = 8.3 Hz), 7.49 (d, 2 H, J = 6.3 Hz), 7.30 (m, 1H), 7.32–7.24 (broad m, 2 H), 6.78 (m, 1H), 5.22 (s, 1H), 3.75 (s, 3 H), 2.16 (m, 2 H), 1.86 (m, 3 h), 1.75 (m, 6 H); MS (ESI+): m/z 456 (M + H)⁺.

N-Adamantan-1-yl)-4-(3-(3-chlorophenyl)ureido)benzenesulfonamide (13–19). Prepared from N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 3-chlorophenylisocyanate (24 µL, 195 µmol) and triethylamine (35 µL, 244.5 mmol).in anhydrous tetrahydrofuran (10 mL). The desired product (33 mg, 44% yield) was isolated as an light yellow solid. ¹H-NMR (CDCl₃) δ 8.60 (s, 1H), 8.39 (s, 1H), 7.88 (m, 1H), 7.70 (d, 2 H, J = 8.4 Hz), 7.50–7.40 (broad m, 4 H), 7.18 (m, 1H), 5.26 (s, 1H), 2.21 (m, 2 H), 1.86 (m, 3 H), 1.74 (m, 6 H); MS (ESI+): m/z 460/462 (M + H)⁺.

Synthesis of compounds 13–20 – 13–23

Carbamates 13–20–13–23 were prepared by acylating N-1-adamantyl-4-amino-benzenesulfonamide with appropriate chloroformates using a variation of a previously described procedure⁴¹. The synthesis of compound 13–20 serves as an example of the general procedure.

Synthesis of 4-fluorophenyl (4-(N-(adamantan-1-yl)sulfamoyl)phenyl)carbamate (13–20)

To a solution of N-1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol) in anhydrous tetrahydrofuran (10 mL) was added 4-fluorophenyl chloroformate (22.5 µL, 172 µmol), followed by 5 drops of anhydrous pyridine. The resulting mixture was stirred for 18 h at room temperature. The resulting reaction mixture was concentrated on a rotary evaporator and the resulting residue was partitioned between dichloromethane and 0.1 N aqueous HCl. The combined organic layers were dried over anhydrous sodium sulfate, filtered and concentrated on a rotary evaporator. The resulting residue was purified by chromatography on silica gel using a gradient of methanol in dichloromethane to provide the desired product (27 mg, 37% yield) as an off-white solid. ¹H-NMR (CDCl₃) δ 9.12 (s, 1H), 7,62 (d, 2 H, J = 8.4 Hz), 7,50 (d, 2 H, J = 8.4 Hz), 7.20 (m, 3 H), 5.25 (s, 1H), 2.14 (m, 2 H), 1.86 (m, 3 H), 1.74 (m, 6 H); MS (ESI+): m/z 445 (M + H)⁺.

The following compounds were prepared using the method described for compound 13–20:

4-Methoxyphenyl (4-(N-(adamantan-1-yl)sulfamoyl)phenyl)carbamate (13–21). Prepared from 1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 4-methoxyphenyl chloroformate (25.5 µL, 172 µmol) and pyridine (5 drops) in anhydrous tetrahydrofuran (10 mL). The desired product (23 mg, 31% yield) was isolated as an off-white solid. ¹H-NMR (CDCl₃) δ 9.04 (s, 1H),7.64 (d, 2 H, J = 8.2 Hz), 7.53 (d, 2 H, J = 8.2 Hz), 7.22 (d, 2 H, J = 8.4 Hz), 7.01 (d, 2 H, J = 8.4 Hz), 5.24 (s, 1H), 3.80 (s, 3 H), 2.23 (m, 3 H), 1.89 (m, 3 H), 1.72 (m, 6 H); MS (ESI+): m/z 457 (M + H)⁺.

2-Fluoroethyl-((N-adamantan-1-yl)sulfamoyl)phenyl)carbamate (13–22). Prepared from 1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), 2-fluoroethyl chloroformate (16.5 µL, 172 µmol) and pyridine (5 drops) in anhydrous tetrahydrofuran (10 mL). The desired product (26 mg, 39% yield) was isolated as a yellow oil. ¹H-NMR (CDCl₃) δ 9.00 (s, 1H), 7.61 (d, 2 H, J = 8.5 Hz), 7.50 (d, 2 H, J = 8.5 Hz), 5.17 (s, 1H), 3.28 (t, 1H, J = 6.8 Hz), 3.15–3.05 (broad m, 3 H), 2.15 (m, 2 H), 1.86 (m, 3 H), 1.74 (m, 6 H); MS (ESI+): m/z 397 (M + H)⁺.

Neopentyl-((4-(N-adamantan-1-yl)sulfamoyl)phenyl)carbamate (13–23). Prepared from 1-adamantyl-4-amino-benzenesulfonamide (50 mg, 163 µmol), neopentyl chloroformate (26 µL, 172 µmol) and pyridine (5 drops) in anhydrous tetrahydrofuran (10 mL). The desired product (21 mg, 30% yield) was isolated as a yellow oil. ¹H-NMR (CDCl₃) δ 9.10 (s, 1H), 7.59 (d, 2 H, J = 8.3 Hz), 7.47 (d, 2 H, J = 8.3 Hz), 5.16 (s, 1H), 3.85 Ism 2 H), 2.19 (m, 2 H), 1.93 (m, 3 H), 1.77 (m, 6 H), 1.35 (s, 9 H); MS (ESI+): m/z 421 (M + H)⁺.

Conventional and ‘spot’ cultures of DRG neurons

Primary dorsal root ganglion (DRG) neurons were isolated from embryonic day 16 (E16) rat embryos. Conventional ‘mass’ cultures were plated on tissue culture plastic pre-coated with poly-lysine and laminin, as previously described³⁷. ‘Spot’ cultures were plated on tissue culture plastic pre-coated with poly-lysine and laminin similar to¹³.

Trophic factor deprivation (TD) assay in conventional DRG cultures

Conventional DRG mass cultures were treated at 5 days in vitro (DIV 5) with either DMSO or 10 µM of hit compounds for 1 h prior to trophic factor deprivation (TD). For all TD experiments, NGF-containing medium was replaced with fresh Neurobasal medium lacking NGF but containing B27 supplement plus sheep anti-NGF antibody (25 µg/ml), in the continued presence of drug or vehicle. For biochemical experiments, cells were lysed 2.5 h post-TD in SDS-PAGE loading buffer and then processed for subsequent SDS-PAGE and immunoblotting analysis. For assays of neurodegeneration, cultures were imaged live 45 h post-TD using an Olympus CKX41 inverted microscope with 20X, 0.4 NA objective. The extent of degeneration was quantified manually on a 10-point scale, similar to prior studies^7,15, by an experimenter blinded to the treatment condition. TD-induced responses in conventional cultures require the retrograde motor protein dynein⁷ so this assay likely directly assesses retrograde axonal signaling, a process known to require DLK and its palmitoylation^7,37.

Injury in ‘spot’ cultures

A 3.5 mm biopsy punch (RoyalTek) was used to form a circular incision around the spot containing the DRG cell bodies, separating the cell bodies from the distal axons and thus mimicking mechanical axotomy. To assess axonal degeneration, images of the distal axon regions were captured using an Olympus CKX41 inverted microscope equipped with a 20X, 0.4 NA objective. Images of axon degeneration were auto-thresholded (Yen) in NIH ImageJ/Fiji. The ‘analyze particles’ algorithm was then used to detect axonal particles of size 20-infinity pixels and with circularity 0.3-1.0. The number of particles was then normalized to the fraction of the total axonal area.

Assay of vincristine-induced degeneration in spot cultures

On DIV7, DRG spot cultures were exposed to either vehicle (DMSO) or vincristine sulfate (40 nM in DMSO). Vincristine-treated cultures received, in parallel, one of the following co-treatments: SARM1 inhibitor (10 µM), DLKi-3511 (500 nM), or the indicated hit compound (10 µM). For pre-treatment experiments, DLKi-3511 (500 nM) or indicated hit compounds (10 µM), were applied on DIV6, 24 h prior to vincristine sulfate addition. Image capture and quantification of axonal degeneration were performed as described for injury-induced degeneration.

Assay of axon survival factor stabilization

On DIV9, DRG spot cultures were incubated with DMSO vehicle, DLKi-3511 (500 nM), or the indicated hit compound (10 µM). Cultures were lysed 4 h later and lysates were western blotted to detect NMNAT2 and STMN2. Additional cultures were subjected to injury or left uninjured, and axonal fractions were harvested 4 h later. Western blots of these latter samples confirmed rapid, injury-induced loss of protein bands recognized by our NMNAT2 and STMN2 antibodies, at the predicted molecular weight for each protein. These findings are consistent with prior reports^13,30,31 and suggest that our NMNAT2 and STMN2 antibodies specifically recognize their cognate targets.

Permeability assay in MDCK-MDR1 cells

Bidirectional MDCK-MR1 assay, a predictor of permeability and CNS penetration, was performed using standard procedures⁴², commercially available pre-plated cells (Pharmaron, Exton, PA) and 1 µM substrate concentrations to minimize transporter saturation. The MDCK-MR1 cell line was also used to monitor p-glycoprotein efflux liability.

Determination of maximum aqueous solubility

Compounds were assessed for their solubility at pH 7.4 using the Millipore MultiScreenTM Solubility filter system (Millipore, Billerica, MA). Analysis was performed by LC/MS/MS on a Waters Xevo TQ instrument (Waters, Milford, MA). MS/MS analyses use positive or negative electrospray or APCI ionization. Assay acceptance criteria were 20% for all standards and 25% for the LLOQ. Results are reported as the maximum concentration of test compound obtained.

Liver microsomal stability assay

Compounds were assessed for their stability in CD-1 mouse liver microsomes by incubating them at 37 °C in the presence or absence of an NADPH regenerating system according to standard procedures⁴³. Analyses were performed by LC/MS/MS. Results are reported as half-lives (t_1/2) for studies including NADPH and percentage remaining after 60 min for studies excluding NADPH.

Inhibition of CYP450 metabolizing enzymes

Compounds were assessed for their ability to inhibit the three major human cytochrome P450 enzymes, 3A4, 2D6 and 2C9. Expressed enzymes (Corning Gentest, Tewksbury, MA) were used to minimize non-specific binding and membrane partitioning issues⁴⁴. The 3A4 assay used midazolam as the substrate and was analyzed using LC/MS/MS. The 2D6 and 2C9 assays used fluorescent substrates and were analyzed on an Envision plate reader (ThermoFisher, Waltham, MA).

Experimental replicates and statistical analysis

For all experiments using cultured neurons, ‘n’ reflects the indicated number of cultures, each from different dissections, indicated as individual data points in each Figure. In some cases, replicate determinations from a single dissection were performed side-by-side and averaged to give a single biological ‘n’.

Statistical analysis was performed using GraphPad Prism. ANOVA was used for multi-condition experiments, with Dunnett’s post hoc test for comparisons against an indicated control condition. For non-normally distributed continuous variables, the Mann-Whitney U test was used for two-group comparisons, and the Kruskal-Wallis test with Dunn’s test for multiple comparisons for more than two groups.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: GMT; Experimentation/Investigation: XZ, HJ, EM, JG; Data acquisition: XZ, HJ, EM, JG; Data curation: XZ, HJ, EM, JG; Methodology: XZ, HJ, EM, JG; Writing: XZ, WEC, GMT; Funding acquisition: GMT; Resource: WEC, GMT; Supervision: WEC, GMT.

Data availability

The datasets generated during and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.