Chronically low NMNAT2 expression causes sub-lethal SARM1 activation and altered response to nicotinamide riboside in axons

Nico=namide mononucleo=de adenylyltransferase 2 (NMNAT2) is an endogenous axon survival factor that maintains axon health by blocking ac=va=on of the downstream pro - degenera=ve protein SARM1 ( sterile alpha and TIR mo=f containing protein 1 ). While complete absence of NMNAT2 in mice results in extensive axon trunca=on and perinatal lethality, the removal of SARM1 completely rescues these phenotypes. Reduced levels of NMNAT2 can be compa=ble with life, however they compromise axon developme nt and survival. Mice born expressing sub-heterozygous levels of NMNAT2 remain overtly normal into old age but develop axonal defects in vivo and in vitro as well as behavioural phenotypes. Therefore, it is important to examine the effects of cons=tu=vely low NMNAT2 expression on SARM1 ac=va=on and disease suscep=bility. Here we demonstrate that chronically low NMNAT2 levels reduce prenatal viability in mice in a SARM1-dependent manner and lead to sub-lethal SARM1 ac=va =on in morphologically intact axons of superior cervical ganglion (SCG) primary cultures. T his is characterised by a deple=on in NAD(P) and compromised neurite outgrowth. We also show that chronically low NMNAT2 expression reverses the NAD- enhancing effect of nico=namide riboside (NR) in axons in a SARM1-dependent manner. These data indicate that low NMNAT2 levels can trigger sub -lethal SARM1 ac=va=on which is detectable at the molecular level and could predispose to human axonal disorders.

NMNAT2, SARM1, NAD, Programmed axon death .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 2 Introduc1on NMNAT2 (nico=namide mononucleo=de adenylyltransferase 2) is an essen=al axon survival molecule, the loss of which triggers axon death in vitro and in vivo [1–3]. Being one of three NMNAT isoforms that catalyse the final step in NAD biosynthesis, NMNAT2 is the predominant enzyme in axons . Mice lacking NMNAT2 die at birth with a severe axonal phenotype, characterised by widespread axon trunca=on in the peripheral nervous system (PNS) and central nervous system (CNS) [2,3]. Axon loss resul=ng from NMNAT2 deple=on requires the prodegenera=ve protein and toll-like receptor adaptor SARM1 (sterile alpha and TIR mo=f containing protein 1). Absence of SARM1 delays degenera=on caused by NMNAT2 deple=on in vitro, and remarkably, completely rescues the axonal outgrowth and perinatal lethality in mice lacking NMNAT2, which remain healthy for up to two years and retain normal innerva=on of distal leg muscles [4,36]. SARM1 has a cri=cal NAD(P) glycohydrolase (NAD(P)ase) ac=vity. This is regulated by NMN and NAD, which are the substrate and product of NMNATs respec=vely. NMN ac=vates SARM1 NAD(P)ase by binding to an allosteric site in the autoinhibitory ARM domain [5,6], while NAD competes for binding to the same site and opposes SARM1 ac=va=on [6–8]. Loss of labile NMNAT2, the major axonal NMNAT isoform, leads to a rise in axonal NMN and a decline in NAD, resul=ng in the ac=va=on of SARM1 NAD(P)ase and axon degenera=on. The interplay between pro-degenera=ve SARM1 and its pro-survival upstream regulator NMNAT2 is cri=cal for axon degenera=on following injury and in several models of neurodegenera=on [9,10]. Accumula=ng evidence supports roles for NMNAT2 loss in human disease. B iallelic loss-of- func=on (LOF) muta=ons (R232Q and Q135Pfs*44) in the NMNAT2 gene have been reported in two s=llborn fetuses with fetal akinesia deforma=on sequence (FADS) and mul=ple congenital abnormali=es [11], resembling the mouse phenotype where complete absence of NMNAT2 leads to perinatal death. H omozygous par=al LOF muta=ons in NMNAT2 (T94M) were reported in two siblings with childhood onset polyneuropathy and accompanying erythromelalgia that is exacerbated by infec=on [12]. More recently, two NMNAT2 missense variants (V98M and R232Q , which confer par=al and complete LOF respec=vely), were iden=fied in two brothers with a progressive neuropathy syndrome , who also have erythromelalgia that worsens with infec=on [13]. Reduced NMNAT2 mRNA levels have also been reported in Parkinson’s, Alzheimer’s and Hun=ngton’s disease pa=ents [14,15]. These observa=ons highlight the need to further characterise the mechanisms of NMNAT2 - mediated pathogenesis in humans. Although the interplay between NMNAT2 and SARM1 following complete loss of NMNAT2 is well-established, the effect of chronically low NMNAT2 expression on SARM1 ac=va=on is less clear. Crucially, there seems to be widespread variability in NMNAT2 mRNA levels among individuals in the human popula=on [15], which could underlie differen=al suscep=bility to various neurodegenera=ve stresses, and the par=al LOF coding variants described above may have similar effects. Thus, a beger understanding of how low NMNAT2 expression influences SARM1 ac=va=on, and in turn axon health, is needed. The remarkable rescue achieved by removing SARM1 in mice lacking NMNAT2 [4,36], raises the intriguing ques=on of whether targe=ng SARM1 could have such striking outcomes in humans with par=al LOF muta=ons in NMNAT2 or with low expression level [16]. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 3 In addi=on, NMN and other NAD precursors, such as nico=namide (NAM) and nico=namide riboside (NR), are widely used as a strategy to boost NAD levels with the purpose of promo=ng longevity and healthy aging [17,18]. Ini=al reports in small cohorts suggest these are safe [19,20] but it is important to ask whether there can be excep=ons. In par=cular, NMN is the endogenous ac=vator of SARM1 and while it does not have adverse effects when NMNAT ac=vity is intact to convert it immediately to NAD and prevent its accumula=on, this could differ when NMNAT ac=vity is insufficient. Thus, it is crucial to inves=gate the impact of these molecules on SARM1 ac=va=on and axon integrity under condi=ons of compromised NMNAT ac=vity, such as in the presence of low NMNAT2 expression in axons [16]. The present study sought to inves=gate whether cons=tu=vely low levels of NMNAT2 can ac=vate SARM1 in morphologically intact axons. We have previously demonstrated that compound heterozygous mice with one silenced and one par=ally silenced Nmnat2 allele, which express sub-heterozygous levels of NMNAT2 are overtly normal but present with in vivo and in vitro axonal defects and behavioural abnormali=es [21]. These include an early reduc=on of myelinated sensory axons with accompanying temperature insensi=vity, and a later loss of motor axons with a decline in motor performance. In culture, superior cervical ganglia (SCG) derived from NMNAT2 compound heterozygous mice have impaired neurite outgrowth and are more sensi=ve to the chemotherapy drug vincris=ne [21] and the mitochondrial toxin CCCP [22]. However, the underlying mechanism was not previously studied. We now demonstrate that sub -heterozygous NMNAT2 expression in mice reduces the number of live births in a SARM1 -dependent manner. Furthermore, chronic and par=al SARM1 ac=va=on underlies the NAD deple=on and neurite outgrowth defect in primary SCG cultures with sub-heterozygous NMNAT2 expression. Most surprisingly, the administra=on of the NAD precursor NR fails to increase NAD as it does in wild -type or heterozygous cultures, and instead causes a SARM1-dependent deple=on of NAD in axons where NMNAT2 levels are low. These data indicate that low NMNAT2 expression leads to sub-lethal SARM1 ac=va=on in at least some neuron types, increasing suscep=bility to otherwise innocuous s=muli, with poten=al to prime for axon degenera=on disorders in humans. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 4

Animals Animal work was approved by the University of Cambridge and performed in accordance with the Home Office Animal Scien=fic Procedures Act (ASPA), 1986 under project licence P98A03BF9. Animals were kept under standard specific pathogen free (SPF) condi=ons and fed ad libitum. Mice of both sexes were studied in experiments. Genera=on of mice carrying the Nmnat2gtE and Nmnat2gtBay gene trap alleles and crosses to generate Nmnat2gtBay/gtE compound heterozygous mice have been described previously [3,23]. Animals were transferred to a new facility before this work was ini=ated, which resulted in a bogleneck in numbers. Nmnat2gtBay/gtE compound heterozygous mice homozygous for the Sarm1 dele=on were generated by crossing Nmnat2gtBay/gtE mice with Sarm1 knockout (Sarm1- /-) mice. F1 mice from this cross heterozygous for either Nmnat2 gene trap allele and heterozygous for the Sarm1 dele=on i.e. Nmnat2+/gtBay;Sarm1-/+ and Nmnat2+/gtE;Sarm1-/+ were crossed again with Sarm1 null mice in order to introduce the Nmnat2 gene trap alleles on a homozygous Sarm1 null background. All mice used in this study originated from the same breeding colony and ligermates were used wherever possible. While it would be preferable to perform all experiments with equal numbers of Sarm1 null and Sarm1 wild-type neurons in parallel, this was usually not possible because the random assortment of genotypes within each liger meant it was neither cost effec=ve nor ethical to breed large enough numbers of mice to enable this. Genotyping Separate duplex polymerase chain reac=on (PCR) was performed to assess the presence of each of the two gene trap alleles, Nmnat2gtE and Nmnat2gtBay. A duplex PCR was performed to determine the Sarm1 genotype. Primers 5ʹ-ctcagtcaatcggaggactggcgc-3ʹ (forward for gene trap), 5ʹ -gctggcctaggtggtgagtgc-3ʹ (forward for wild -type) and 5ʹ-cacaaggccgtctcagacggc-3ʹ (common reverse for both) were used to amplify a 215 bp product from the Nmnat2gtE gene trap allele and a 389 bp product from the corresponding wild-type locus. Temperatures used were 94 °C for denatura=on and 60 °C for primer annealing. Primers 5ʹ-aggaagcagggagaggcag- 3ʹ (reverse for wild -type), 5ʹ-tgcaaggcgagaagggggtaacg-3ʹ (reverse for gene trap) and 5ʹ - gagccacagactagtgactgggg-3ʹ (common forward for both) were used to amplify a 206 bp product from the Nmnat2gtBay gene trap allele and a 310 bp product from the corresponding wild-type locus. Temperatures used were 94 °C for denatura=on and 65 °C for primer annealing. Primers 5ʹ-acgcctgggtcgactctacg-3ʹ and 5ʹ -ccgacctcggcgggtgatgc-3ʹ were used to amplify a >500 bp product (~508 bp) from the wild -type Sarm1 allele and primers 5ʹ - ggtagccggatcaagcgtatgc-3ʹ and 5ʹ -ctcatctccgggccgtcgacc-3ʹ were used to amplify a <500 bp product (~450 bp) from the neomycin resistance cassege retained in the knockout allele in place of deleted exons 3-6 [24]. Temperatures used were 94 °C for denatura=on and 60 °C for primer annealing. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 5 Primary neuronal explant cultures Superior cervical ganglia (SCGs) were dissected from P0 -P3 mouse pups and dorsal root ganglia (DRGs) were dissected from E13-14 mouse embryos. Explants were plated in 3.5 cm =ssue culture dishes pre-coated with poly-L-lysine (20 mg/ml for 1 hour; Sigma) and laminin (20 mg/ml for 1 –2 hours; Sigma). Explants were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) with high glucose, glutamine and sodium pyruvate (Gibc o), with 1% penicillin/streptomycin (Invitrogen), 50 ng/ml 2.5S NGF (Invitrogen), and 2% B -27 (Gibco). Aphidicolin 4 μM (Calbiochem) was used to restrict the prolifera=on and viability of small numbers of non-neuronal cells. Nico=namide riboside (NR) was prepared as a 100 mM stock from Tru Niagen capsules (Chromadex) and stored at 4 °C. The contents of the capsules were dissolved in PBS without Ca2+ and Mg2+ (Merck) and passed through a 0.22 μm filter. Nico=namide (NAM) (Sigma - Aldrich) was prepared in water and stored frozen as 100 mM stock aliquots. Cell culture media was supplemented with NR (2 mM) or NAM (1 mM) at the days in vitro (DIV) indicated in the figures. Acquisi1on of neurite images and quan1fica1on of neurite outgrowth Phase contrast images were acquired on a DMi8 upright fluorescence microscope (Leica microsystems) coupled to a monochrome digital camera (Hammamatsu C4742 -374 95). Neurite outgrowth from SCG and DRG explants was assessed from low magnifica=on (NPLAN 5x/0.12 objec=ve) images on the DIV indicated in the figures. Two measurements of radial outgrowth were recorded for each ganglion, by taking the maximal outgrowth from the edge of the ganglion to the point where the bulk of neurites terminated. The average length was calculated for each day. Confocal imaging of PAD6 in primary neuronal cultures Compound PC 6, synthesised and provided by AstraZeneca, was used to visualise SARM1 ac=va=on in primary neuron al cultures through its conversion to PAD6 [29]. PC6 was administered in cell culture media at 50 μM and images were acquired 30 minutes later using a Confocal -LSM780 (1.4a) microscope, 40x oil objec=ve, Ex/Em: 405/525 nm. Two representa=ve images were taken for each sample. Measurement of NAD and NADP levels from primary neuronal cultures NAD and NADP levels were measured using the commercially available kits NAD/NADH -Glo Assay and NADP/NADPH-Glo Assay (Promega G9071, Promega G9081), respec=vely. Neurons were collected in Eppendorf tubes by disrup=ng adhesion to the plate with a jet of me dium and washed twice in ice -cold PBS with Ca 2+ and Mg2+ (Merck), supplemented with complete, ethylenediaminetetraace=c acid (EDTA)-free protease inhibitor cocktail tablets (plus protease inhibitors) (Roche). For experiments where separate measurements we re made in ganglia versus neurites, a scalpel was used to cut around and isolate the ganglia and the two compartments were collected in separate tubes. Neurons were lysed in ice-cold Pierce IP lysis buffer (Sigma) plus protease inhibitors. Lysates were centrifuged for 10 min at 13k rpm in a .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 6 microfuge at 4 °C to pellet insoluble material. Supernatants were collected on ice and diluted to 0.15 μg/μl in ice -cold Pierce IP lysis buffer plus protease inhibitors awer protein concentra=ons had been determined using the Pierce BCA assay (Thermo Fisher Scien=fic). For NAD and NADP measurements , 25 μl of each extract was mixed with 12.5 μl 0.4 M HCl and heated to 60 °C for 15 min before being allowed to cool at room temperature (RT) for 10 minutes. Reac=ons wer e neutralised by adding 12.5 μl 0.5 M T ris base and 10 μl of each neutralised reac=on was mixed with 10 μl of the NAD -Glo or NADP -Glo reagent (prepared following manufacturer’s instruc=ons) on ice in wells of a 384 -well white polystyrene microplate (Corning). The plate was incubated for 40 min at RT before reading luminescence using GloMax Explorer (Promega) plate reader. Concentra=ons of NAD and NADP were determined from standard curves generated from dilu=on series of the relevant nucleo=des. Values are expressed as nmol/mg of protein. ImmunobloYng Neuronal cultures were collected in Eppendorf tubes by disrup=ng adhesion to the plate with a jet of medium and washed twice in ice-cold PBS with Ca2+ and Mg2+ plus protease inhibitors. Neurons were directly lysed into 15 μl 2x Laemmli buffer containing 10% 2-mercaptoethanol and samples were incubated at 100 °C for 5 min. The total amount (15 μl) for each sample was loaded on a 4 -20% SDS-PAGE (Bio-Rad). For comparing protein levels between SCG and DRG cultures, samples were first diluted to the same protei n concentra=on, which w as determined using the Pierce BCA assay. Briefly, following the washes in PBS, neurons were lysed in ice-cold Pierce IP lysis buffer plus protease inhibitors. Lysates were centrifuged for 10 min at 13k rpm in a microfuge at 4 °C to pellet insoluble material. Supernatants were collected on ice and diluted to the same concentra=on (based on the sample with the lowest concentra=on) in ice-cold Pierce IP lysis buffer. Samples were diluted 1 in 2 with 2 x Laemmli buffer and were incubated at 100 °C for 5 min. For detec=ng NAMPT, SARM1 and GAPDH, 1/6 of the total amount was loaded on a 4-20% SDS-PAGE (Bio-Rad), while the remaining sample was used for detec=ng NMNAT2. Samples were transferred to Immobilon-FL PVDF membrane (Millipore) using the BioRad Mini-PROTEAN III wet transfer system. Blots were blocked in Tris buffered saline (TBS) (20 mM Tris p.H. 8.3, 150 mM NaCl) with 0.05% Tween 20 (Merck) (TBST) and 5% skimmed milk powder, for 1 hour at RT. Blots were incubated overnight at 4 ° C with primary an=bodies in TBST containing 5% milk. Awer three × 10 min washes in TBST, blots were incubated for 1 hour at RT with appropriate HRP-conjugated secondary an=bodies (Bio- Rad) diluted 1 in 3,000 in TBST with 5% milk. Awer three × 10 min wa shes in TBST blots were incubated with Pierce ECL Western Bloyng Substrate or SuperSignal West Dura Extended Dura=on Substrate (both Thermo Fisher Scien=fic) and imaged using an Alliance chemiluminescence imaging system (UVITEC Cambridge). Fiji sowware was used to determine the rela=ve intensi=es of specific bands from captured digital images. The following primary an=bodies were used: mouse an=-SARM1 monoclonal an=body (1 in 5000, [25]), mouse an= -NAMPT monoclonal an=body (1 in 2000, Cayman Chemical 10813), mouse an= -NMNAT2 monoclonal an=body (1 in 250, Merck WH0023057M1), mouse an= - GAPDH monoclonal an=body (1 in 2000, Abcam ab8245). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 7 Sta1s1cal Analysis Sta=s=cal analysis was conducted using Prism Sowware (GraphPad Sowware Inc, La Jolla, CA, USA). The n numbers and specific sta=s=cal tests used for each experiment are described in the figure legends. A p value < 0.05 was considered significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 8

Sub-heterozygous NMNAT2 expression reduces viability in a SARM1-dependent manner In an agempt to inves=gate the effects of sub -heterozygous levels of NMNAT2 expression, mice heterozygous for two dis=nct Nmnat2 gene trap alleles were crossed. Although a gene trap cassege is located in the first intron of the Nmnat2 gene in each case, the degrees of gene silencing differ, with the Nmnat2gtE allele being completely silenced and the Nmnat2gtBay allele being only par=ally silenced [3,21]. We have previously reported that mice of all four genotypes generated from these crosses are born quite close to the expected ra=os [21], although Nmnat2gtBay/gtE mice, which express sub-heterozygous levels of NMNAT2, were slightly under-represented rela=ve to the expected frequencies without this effect reaching sta=s=cal significance. However, w e now report a significant loss in p renatal viability of Nmnat2gtBay/gtE mice, as well as a similar trend to lower than expected numbers at embryonic stage E13-E14 (which may only fail to reach sta=s=cal significance because of the smaller sample size) (fig. 1a, b; supplementary fig. 1). Remarkably, knocking out Sarm1 rescues this prenatal loss, restoring the Nmnat2gtBay/gtE genotype ra=o to the expected level (fig. 1c). Thus, sub-heterozygous NMNAT2 expression can lead to a modest but significant SARM1-dependent loss of viability that is likely to manifest at an embryonic stage. Poten=al explana=ons for why one of these studies crosses the threshold for significance and the other did not are discussed below. Fig. 1 Absence of SARM1 rescues prenatal lethality in Nmnat2gtBay/gtE mice. (a) Genotype frequencies of embryos from crosses between Nmnat2+/gtE and Nmnat2+/gtBay mice on a Sarm1+/+ background. The observed embryo frequencies are not significantly different from expected frequencies: 𝜒2 = 2.919, d.f. = 3, p = 0.4042. ( b) Genotype frequencies of viable offspring from crosses between Nmnat2+/gtE and Nmnat2+/gtBay mice, on a Embryos x Genotype Observed Expected Nmnat2+/gtBay;Sarm1+/+ x Nmnat2+/gtE;Sarm1+/+ Nmnat2+/+;Sarm1+/+ 61 55.75 Nmnat2+/gtBay;Sarm1+/+ 57 55.75 Nmnat2+/gtE;Sarm1+/+ 60 55.75 Nmnat2gtBay/gtE;Sarm1+/+ 45 55.75 Viable Offspring x Genotype Observed Expected Nmnat2+/gtBay;Sarm1+/+ x Nmnat2+/gtE;Sarm1+/+ Nmnat2+/+;Sarm1+/+ 182 167.5 Nmnat2+/gtBay;Sarm1+/+ 161 167.5 Nmnat2+/gtE;Sarm1+/+ 191 167.5 Nmnat2gtBay/gtE;Sarm1+/+ 136 167.5 Viable Offspring x Genotype Observed Expected Nmnat2+/gtBay;Sarm1-/- x Nmnat2+/gtE;Sarm1-/- Nmnat2+/+;Sarm1-/- 131 128.75 Nmnat2+/gtBay;Sarm1-/- 135 128.75 Nmnat2+/gtE;Sarm1-/- 114 128.75 Nmnat2gtBay/gtE;Sarm1-/- 135 128.75 Nmnat2+/+ Nmnat2+/gtBay Nmnat2+/gtE Nmnat2gtBay/gtE Total = 223 20.18% 27.35% 26.91% 25.56% Sarm1+/+ Nmnat2+/+ Nmnat2+/gtBay Nmnat2+/gtE Nmnat2gtBay/gtE 20.30% 27.16% 28.51% 24.03% Sarm1+/+ Total = 670 Live births Live births * E13-E14a b c Nmnat2+/+ Nmnat2+/gtBay Nmnat2+/gtE Nmnat2gtBay/gtE 26.21% 25.44% 22.14% 26.21% Sarm1-/- Total = 515 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 9 Sarm1+/+ background. The observed birth frequencies are significantly different from expected frequencies: 𝜒2 = 10.728, d.f. = 3, p = 0.0133. (c) Genotype frequencies of viable offspring from crosses between Nmnat2+/gtE and Nmnat2+/gtBay mice, on a Sarm1-/- background. The observed birth frequencies are not significantly different from expected frequencies: 𝜒2 = 2.336, d.f. = 3, p = 0.5057. Viable offspring numbers include animals between P0-P3 and post-weaning. Sub-lethal SARM1 ac@va@on underlies the NAD(P) decrease and neurite outgrowth defect in SCG neurons from Nmnat2gtBay/gtE mice Injury and other insults that ac=vate SARM1 induce its enzyma=c ac=vity resul=ng in the consump=on of NAD prior to degenera=on [26]. As well as NAD-consuming ac=vity, SARM1 is also an NADPase, cleaving the phosphorylated form of NAD, NADP [27,28]. In an agempt to test for molecular markers of SARM1 ac=va=on in non -degenera=ng axons, NAD and NADP levels were measured in SCG whole explant cultures from wild -type (Nmnat2+/+), Nmnat2+/gtE and Nmnat2gtBay/gtE mice. While NAD and NADP levels in SCG neurons from Nmnat2+/+ and Nmnat2+/gtE mice were indis=nguishable, neurons from Nmnat2 compound heterozygotes had significantly lower levels of both metabolites (fig. 2a). NAD levels were ~50% lower in Nmnat2gtBay/gtE neurons, while a ~20% reduc=on in NA DP levels was also observed. Thus, halving of NMNAT2 levels does not lower NAD or NADP , whereas sub-heterozygous NMNAT2 expression significantly reduces the levels of both metabolites in SCG primary cultures. NMNAT2 is an NAD-synthesising enzyme. It is therefore possible that the deple=on of NAD(P) observed in primary cultures of mice with sub-heterozygous NMNAT2 expression reflects only a lack of NAD synthesis (due to limited NMNAT2) rather than increased consump=on of NAD (due to ac=vated SARM1). In order to establish whether the observed NAD and NADP deple=on are SARM1-dependent, Nmnat2gtBay/gtE mice were crossed to the Sarm1-/- mice to introduce the Sarm1 dele=on to the Nmnat2 compound heterozygote mice. Remarkably, absence of SARM1 completely rescues the NAD and NADP deple=on in SCG neurons from low- NMNAT2 expressing mice (fig. 2b), providing evidence in support of increased SARM1 ac=vity in morphologically intact, non-degenera=ng axons. As an independent, more direct indica= on of SARM1 ac=va=on in Nmnat2gtBay/gtE SCG neurites, PC6, a recently developed marker of SARM1 ac=va=on was used. PC6 is a pyridine base that gets converted by SARM1-dependent base exchange to PAD6, a molecule with increased fluorescence emission at 525nm [29]. Incuba=on of primary SCG neurons with PC6 gave a markedly higher signal in primary SCG neuron cultures from Nmnat2gtBay/gtE mice compared to wild -type controls (fig. 2c), sugges=ng that SARM1 is chronically ac=vated in these axons. No significant difference in SARM1 protein levels was observed between Nmnat2+/+ and Nmnat2gtBay/gtE SCG neurons (fig. 2d, e) , ruling out the possibility that the difference in the PAD6 signal is due to variability in SARM1 levels between the two genotypes. We previously demonstrated that complete absence of NMNAT2 severely compromises neurite outgrowth in primary neuronal cultures, whereas a 50% reduc=on in NMNAT2 levels has no detectable phenotype [3]. Interes=ngly, sub-heterozygous levels of NMNAT2 are consistent with an intermediate phenotype, as SCG neurites from Nmnat2gtBay/gtE mice have a reduced outgrowth rate compared to both wild -types and Nmnat2+/gtE single heterozygotes, especially once neurites extend beyond several millimetres [21]. Here we show that a bsence of SARM1 restore s outgrowth of Nmnat2gtBay/gtE neurites to control levels (fig. 2f-i), .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 10 demonstra=ng that the defect is SARM1-dependent and is not solely driven by the lower levels of NMNAT2 expression. Fig. 2 Absence of SARM1 restores NAD(P) levels and rescues the neurite outgrowth defect in SCG neurons from Nmnat2gtBay/gtE mice. (a) NAD and NADP levels in SCG explants of the indicated genotypes on a Sarm1+/+

(mean ± SEM; n = 11-17 pups per genotype; ****p < 0.0001, ***p < 0.001, **p < 0.01, *p 0.05, one-way ANOVA with Tukey’s mul`ple comparisons test). (b) NAD and NADP levels Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 2 4 6 8 NAD (nmol/mg protein) ns ✱✱✱✱ ✱✱✱ Sarm1+/+ Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0.0 0.1 0.2 0.3 0.4 NADP (nmol/mg protein) ns ✱ ✱✱ Sarm1+/+ a b 20 µm Nmnat2+/+Nmnat2gtBay/gtE SARM1 75 37 GAPDH Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/ gtE kDa Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0.0 0.5 1.0 1.5 SARM1 (Relative Intensity) ns ns ns c d e Nmnat2+/gtE Nmnat2+/+ Nmnat2gtBay/gtE 1 mm Sarm1+/+ 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Days in culture Neurite outgrowth (mm) Nmnat2+/+ Nmnat2+/gtE Nmnat2gtBay/gtE ** **** Sarm1+/+ Sarm1-/- Nmnat2+/gtE Nmnat2+/+ Nmnat2gtBay/gtE 1 mm 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Days in culture Neurite outgrowth (mm) Nmnat2+/+ Nmnat2+/gtE Nmnat2gtBay/gtE Sarm1-/- f g h i Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 2 4 6 8 10 NAD (nmol/mg protein) ns ns ns Sarm1-/- Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0.0 0.1 0.2 0.3 0.4 0.5 NADP (nmol/mg protein) ns ns ns Sarm1-/- .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 11 in SCG explants of the indicated genotypes, all on a Sarm1-/- background (mean ± SEM; n = 11 -13 pups per genotype; ns (not significant) = p > 0.05, one-way ANOVA with Tukey’s mul`ple comparisons test). For a -b, cultures were collected at DIV7. (c) Representa`ve images (from n=3) of Nmnat2+/+ and Nmnat2gtBay/gtE SCG neurons (on a Sarm1+/+ background) 30 min acer incuba`on with PC6 (50 μM). (d) Representa`ve immunoblot of SCG neurite extracts of the indicated genotypes (Sarm1+/+) probed for SARM1 and GAPDH (loading control). Cultures were collected at DIV7. ( e) Quan`fica`on of normalised SARM1 levels (to GAPDH) in SCG neurite extracts for the indicated genotypes (Sarm1+/+) (mean ± SEM; n = 3; ns (not significant) = p > 0.05, one -way ANOVA with Tukey’s mul`ple comparisons test). (f) Representa`ve images of neurite outgrowth at DIV7 in SCG explant cultures of the indicated genotypes on a Sarm1+/+ background. (g) Quan`fica`on of neurite outgrowth in SCG explant cultures of the indicated genotypes on a Sarm1+/+ background, between DIV0 and DIV7 (mean ± SEM; n = 8 -9 pups per genotype; ****p < 0.0001 and **p < 0.01, two -way repeated measures ANOVA with Tukey’s mul`ple comparisons test for between genotype effects at each `me point. Significance is shown for the Nmnat2+/+ vs Nmnat2gtBay/gtE comparison). (h) Representa`ve images of neurite outgrowth at DIV7 in SCG explant cultures of the indicated genotypes on a Sarm1-/- background. (i) Quan`fica`on of neurite outgrowth in SCG explant cultures of the indicated genotypes on a Sarm1-/- background, between DIV0 and DIV7 (mean ± SEM; n = 7 pups per genotype; ns (not significant) = p > 0.05, two-way repeated measures ANOVA with Tukey’s mul`ple comparisons test for between genotype effects at each `me point). Sub-heterozygous NMNAT2 expression does not result in NAD(P) deple@on or neurite outgrowth defect in DRG neurons Having obtained evidence in favour of sub-lethal SARM1 ac=va=on in SCG neurons, we next addressed whether low NMNAT2 expression has a similar effect in other neuron types. DRG neurons from E13-E14 embryos of the same Nmnat2 genotypes (i.e. Nmnat2+/+, Nmnat2+/gtE, Nmnat2gtBay/gtE) were used. Surprisingly, in contrast to our findings in SCG cultures, no NAD or NADP deple=on was observed in DRG neurons from Nmnat2gtBay/gtE mice (fig. 3a). The fact that nucleo=de measurements were made in whole explant cultures, encompassing both the cell body and neurite compartments, prompted us to ask whether the presence of NMNAT1, the nuclear NAD-synthesising enzyme, could mask a poten=al NAD loss occurring solely in the axons where the effect of NMNAT2 is more promine nt. For this reason, metabolite measurements were made separately in cell bodies versus neurites. However, no significant difference in NAD or NADP was observed among the three genotypes in separate cell body or neurite frac=ons (fig. 3b, 3c), although a downwards trend with lower NMNAT2 expression was observed in the neurites (fig. 3c). In addi=on, in agreement with our previous report [21], we found no evidence for any outgrowth defect in Nmnat2gtBay/gtE DRG cultures (fig. 3d, e). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 12 Fig. 3 No NAD(P) deple`on or neurite outgrowth defect in DRG neurons from Nmnat2gtBay/gtE mice. (a) NAD and NADP levels in DRG whole explant cultures of the indicated genotypes on a Sarm1+/+ background (mean ± SEM; n = 8-9 embryos per genotype; ns (not significant) = p > 0.05, one-way ANOVA with Tukey’s mul`ple comparisons test). (b) NAD and NADP levels in DRG ganglia of the indicated genotypes on a Sarm1+/+ background (mean ± SEM; n = 6 -7 embryos per genotype; ns (not significant) = p > 0.05, one -way ANOVA wit h Tukey’s mul`ple comparisons test). (c) NAD and NADP levels in DRG neurites of the indicated genotypes on a Sarm1+/+ background (mean ± SEM; n = 6-7 embryos per genotype; ns (not significant) = p > 0.05, one-way ANOVA with Tukey’s mul`ple a b Whole cultures Ganglia Neuritesc Nmnat2+/gtE Nmnat2+/+ Nmnat2gtBay/gtE 1 mm Sarm1+/+ 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Days in culture Neurite outgrowth (mm) Nmnat2+/+ Nmnat2+/gtE Nmnat2gtBay/gtE Sarm1+/+ d e Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0.0 0.2 0.4 0.6 NADP (nmol/mg protein) ns ns ns Sarm1+/+ Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 2 4 6 8 NAD (nmol/mg protein) ns ns ns Sarm1+/+ Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 1 2 3 4 NAD (nmol/mg protein) ns ns ns Sarm1+/+ Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0.0 0.1 0.2 0.3 0.4 NADP (nmol/mg protein) ns ns ns Sarm1+/+ Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 2 4 6 NAD (nmol/mg protein) ns ns ns Sarm1+/+ Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0.0 0.2 0.4 0.6 NADP (nmol/mg protein) ns ns ns Sarm1+/+ .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 13 comparisons test). For a-c cultures were collected at DIV7. (d) Representa`ve images of neurite outgrowth at DIV7 in DRG explant cultures of the indicated genotypes on a Sarm1+/+ background. (e). Quan`fica`on of neurite outgrowth in DRG explant cultures of the indicated genotypes on a Sarm1+/+ background, between DIV0 and DIV7 (mean ± SEM; n = 3 -4 embryos per genotype; ns (not significant) = p > 0.05, two-way repeated measures ANOVA with Tukey’s mul`ple comparisons test for between genotype effects at each `me point). Our observa=ons indicate that different neuron types display varying suscep=bility to NMNAT2 deple=on. In an agempt to iden=fy possible mechanisms rendering SCG neurons more suscep=ble, the expression levels of proteins involved in NAD synthesis (NAMPT, NMNAT2) and NAD consump=on (SARM1) were compared between wild-type cultures of the two neuron types (fig. 4a). Whereas levels of SARM1 appeared to be similar between the two neuron types (fig. 4b, f), NAMPT (nico=namide phosphoribosyltransferase), the rate-limi=ng enzyme in the NAD biosynthe=c pathway from Nico=namide (NAM), w as significantly lower in DRG neurons compared to SCG neurons (fig. 4b, d) . Furthermore, NMNAT2, the enzyme conver=ng NMN to NAD, showed a non-significant trend towards lower expression in SCG s than DRGs (fig. 4b, e). As a result, the NMNAT2:NAMPT ra=o, which is likely to influence levels of the endogenous SARM1 ac=vator NMN, was significantly lower in SCG cultures with an effect size of more than twofold (fig. 4c). The higher ra=o in DRG neurons could explain why NMNAT2 levels can be reduced further than in SCG neurons before becoming rate limi=ng, thereby lowering the likelihood of NMN accumula=on and consequent SARM1 ac=va=on when NMNAT2 levels are chronically low. This could explain, at least partly, the absence of any NAD(P) deple=on and neurite outgrowth defect in DRG neurons from Nmnat2gtBay/gtE mice. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 14 Fig. 4 Higher NMNAT2 to NAMPT ra`o in DRG vs SCG neurons. (a) Pathway of NAD synthesis from precursor NAM and NAD consump`on by SARM1. (b). Representa`ve immunoblot of wild-type SCG and DRG extracts probed for NAMPT, NMNAT2, SARM1 and GAPDH (loading control). Cultures were collected at DIV7. (c) NMNAT2:NAMPT ra`o (mean ± SEM; n = 3; *p < 0.05 paired t -test). (d-f) Quan`fica`on of NAMPT, NMNAT2 and SARM1 levels normalised to GAPDH (mean ± SEM; n = 3; **p 0.05, paired t-test). NR causes a SARM1-dependent NAD deple@on in SCG neurites from Nmnat2gtBay/gtE mice The data provided thus far support the hypothesis that SARM1 can be chronically ac=vated without causing degenera=on , at least in shorter axons . We next sought to inves=gate whether the balance could be further shiwed in favour of SARM1 ac=va=on by eleva=ng levels of its endogenous ac=vator, NMN, when there is insufficient NMNAT2 to convert all of it rapidly to NAD. First, wild-type primary DRG cultures were supplemented with the NAD precursors NR and NAM in order to iden=fy condi=ons that lead to NAD accumula=on, thereby reflec=ng a prior increase in NMN. NAM is converted to NMN by the enzyme NAMPT, while NR is converted to NMN by nico=namide riboside kinase 1/2 (NRK1 and NRK2). While NR as well as the combina=on of NR and NAM resulted in a marked increase in NAD(P) levels awer a 24 -hour treatment, NAM applica=on alone did not significantly increase the levels of either metabolite (supplementary fig. 2). This observa=on could be due to NAMPT already opera=ng at satura=on, as it is the rate -limi=ng enzyme for NAD biosynthesis in mammals and NAM is SARM1 37 NAMPT NMNAT2 GAPDH DRGs SCGs 50 75 37 kDa DRGs SCGs 0.0 0.4 0.8 1.2 1.6NMNAT2:NAMPT ✱ DRGs SCGs 0.0 0.3 0.6 0.9 NAMPT (relative intensity) ✱✱ DRGs SCGs 0.0 0.3 0.6 0.9 NMNAT2 (relative intensity) ns DRGs SCGs 0.0 0.4 0.8 1.2 SARM1 (relative intensity) ns a b c d e f .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 15 already present in the culture medium, so any further increase in NAM does not increase the rate of NAD produc=on. For this reason, NR alone was used in subsequent experiments (fig. 5a). Supplementa=on with NR also significantly increased NAD levels in wild-type and Nmnat2+/gtE whole SCG cultures but had no effect in Nmnat2gtBay/gtE whole SCG cultures. NR and PBS - treated Nmnat2gtBay/gtE cultures were indis=nguishable (fig. 5b). The lack of NAD increase in Nmnat2gtBay/gtE SCG cultures in response to NR could be agributed to low NMNAT2 expression being unable to u=lise the excess of NMN for NAD synthesis. However, absence of SARM1 restored the ability of Nmnat2gtBay/gtE neurons to increase NAD levels following NR treatment (fig. 5c). This suggests that the lack of NAD accumula=on in the first instance is not due to lower NAD synthesising capacity, as this is also lower in the Sarm1-/- cultures, but rather to the lower capacity to remove NMN, whose accumula=on ac=vates SARM1 NADase. Unfortunately, it was not feasible to measure NMN levels in these primary cultures to confirm an increase. The limited number of pups carrying the genotypes of interest, coupled with the reduced viability of Nmnat2gtBay/gtE animals, hindered the acquisi=on of sufficient material necessary for metabolite analysis. The observa=on that NR administra=on had no net effect on NAD levels in Nmnat2gtBay/gtE cultures led to the hypothesis that the response to NR could differ between neurites and cell bodies, as cell bodies also have NMNAT1 to convert NMN to NAD (fig. 5d). For this reason, separate metabolite measurements were made in cell body and neurite compartments following NR administra=on in wild -type and Nmnat2gtBay/gtE SCG cultures. Interes=ngly, NR administra=on caused NAD to decline significantly in neurites with sub-heterozygous NMNAT2 expression ( fig. 5e). I n contrast, the cell bodies of Nmnat2gtBay/gtE SCG cultures showed increased NAD when supplemented with NR, consistent with conversion of NMN to NAD by NMNAT1, the nuclear NMNAT isoform (fig. 5e). Given that NMNAT1 levels are unaffected in Nmnat2gtBay/gtE cultures, this isoform is able to u=lise the NM N derived from NR, resul=ng in the increased levels of NAD seen in the cell bodies (fig. 5d). Hence, the observa=on that NAD levels in whole SCG Nmnat2gtBay/gtE cultures do not change with NR supplementa=on is likely to reflect the net effect of both increasing NAD in cell bodies and decreasing it in neurites. In Nmnat2gtBay/gtE neurites lacking SARM1, NAD levels were increased following NR administra=on (fig. 5f). A similar trend was observed with NADP , although the changes were less marked (supplementary fig. 3). Collec=vely, these observa=ons strongly suggest that under condi=ons of inadequate NMNAT2, the accumulated NMN resul=ng from NR administra=on further ac=vates SARM1, leading to a deple=on of NAD levels in Nmnat2gtBay/gtE neurites. Interes=ngly, despite NR leading to further ac=va=on of SARM1 in neurites with low NMNAT2, no morphological changes were seen in these cultures. There were no signs of frank axon degenera=on (data not show n) and somewhat counter -intui=vely the neurite outgrowth defect was actually slightly improved, albeit not significantly (fig. 5g, h). We propose that this could be agributed to the =ming of NR administra=on, with transient increases in NAD levels occurring shortly awer NR administra=on giving neurites an ini=al growth spurt before SARM1 ac=va=on and NAD deple=on take place. Alterna=vely, the increase in NAD levels occurring in Nmnat2gtBay/gtE cell bodies following NR administra=on could account for the improvement in the outgrowth phenotype. Interes=ngly, increased levels of NAD awer NR supplementa=on .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 16 had no effect on neurite outgrowth in Nmnat2+/+ and Nmnat2+/gtE cultures on a Sarm1+/+ background, or in any Nmnat2 genotype lacking SARM1 (fig 5h, i ). This indicates that enhanced NAD alone is not sufficient to boost neurite outgrowth. Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 5 10 15 NAD (nmol/mg protein) NAD PBS NR NR NR PBS NR PBS PBS PBS PBS PBS NR NR NR Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 5 10 15 NAD (nmol/mg protein) ✱✱ ✱✱ ✱✱ Sarm1-/- Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 5 10 15 NAD (nmol/mg protein) NAD PBS NR NR NR Sarm1+/+ PBS NR PBS PBS PBS PBS PBS NR NR NR Nmnat2 +/+ Nmnat2 +/gtE Nmnat2 gtBay/gtE 0 5 10 15 NAD (nmol/mg protein) ✱ ✱ ns Sarm1+/+ Whole cultures Whole cultures Nmnat2 +/+ Nmnat2 gtBay/gtE 0 2 4 6 8 NAD (nmol/mg protein) ✱ ✱ Sarm1+/+ Nmnat2 +/+ Nmnat2 gtBay/gtE 0 5 10 15 NAD (nmol/mg protein) ✱ ✱ Sarm1+/+ Nmnat2 +/+ Nmnat2 gtBay/gtE 0 2 4 6 8 NAD (nmol/mg protein) ✱ ✱ PBS PBSNR NR Sarm1-/- Nmnat2 +/+ Nmnat2 gtBay/gtE 0 2 4 6 8 NAD (nmol/mg protein) ✱✱ ✱✱ Sarm1-/- Nmnat2 +/+ Nmnat2 gtBay/gtE 0 5 10 15 20 NAD (nmol/mg protein) ✱ ✱ Nmnat2 +/+ Nmnat2 gtBay/gtE 0.0 0.1 0.2 0.3 0.4 NADP (nmol/mg protein) ✱✱ ✱✱ PBS PBSNR NR Ganglia Neurites Ganglia Neurites Nmnat2gtBay/gtE + PBS Nmnat2+/+ + PBS Nmnat2gtBay/gtE + NR Sarm1+/+ 1 mm 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Days in culture Neurite outgrowth (mm) Nmnat2+/+ + PBS Nmnat2+/+ + NR Nmnat2+/gtE + PBS Nmnat2+/gtE + NR Nmnat2gtBay/gtE + PBS Nmnat2gtBay/gtE + NR Sarm1+/+ ns Nmnat2+/+ + PBS Nmnat2gtBay/gtE + PBS Nmnat2gtBay/gtE + NR WT PBs 3 Ch nr bottom 1 mm Sarm1-/- 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Days in culture Neurite outgrowth (mm) Nmnat2+/+ + PBS Nmnat2+/+ + NR Nmnat2+/gtE + PBS Nmnat2+/gtE + NR Nmnat2gtBay/gtE + PBS Nmnat2gtBay/gtE + NR Sarm1-/- a b c e f g h i j d .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 17 Fig. 5 NR causes SARM1-dependent NAD deple`on in SCG neurites from Nmnat2gtBay/gtE mice. (a) Timeline of NR (2 mM) or PBS administra`on and collec`on of SCG cultures. (b) NAD levels in whole SCG explants of the indicated genotypes on a Sarm1+/+ background (mean ± SEM; n = 5; *p 0.05, mul`ple paired t-tests for PBS vs NR with Holm-Šídák correc`on method). (c) NAD levels in whole SCG explants of the indicated genotypes on a Sarm1-/- background (mean ± SEM; n = 4; **p < 0.01, mul`ple paired t-tests for PBS vs NR with Holm-Šídák correc`on method). (d) Schema`c of NMNAT1 and NMNAT2 localisa`on in neurons and effects of NR administra`on. (e) NAD levels in SCG ganglia and neurites of the indicated genotypes on a Sarm1+/+ background (mean ± SEM; n = 4; *p < 0.05, mul`ple paired t -tests for PBS vs NR with Holm -Šídák correc`on method). (f) NAD levels in SCG ganglia and neurites of the indicated genotypes on a Sarm1-/-

(mean ± SEM; n = 3; **p < 0.01 and *p < 0.05, mul`ple paired t-tests for PBS vs NR with Holm-Šídák correc`on method). (g) Representa`ve images of neurite outgrowth at DIV7 in SCG explant cultures of the indicated genotypes on a Sarm1+/+ background acer administra`on of NR (2 mM) or PBS control. ( h) Quan`fica`on of neurite outgrowth in SCG explant cultures of the indicated genotypes on a Sarm1+/+ background, between DIV0 and DIV7 (mean ± SEM; n = 5; ns (not significant) = p > 0.05, two -way repeated measures ANOVA with Tukey’s mul`ple comparisons test for between genotype effects at each `me point). (i) Representa`ve images of neurite outgrowth at DIV7 in SCG explant cultures of the indicated genotypes on a Sarm1-/- background acer administra`on of NR (2 mM) or PBS control. (j) Quan`fica`on of neurite outgrowth in SCG explant cultures of the indicated genotypes on a Sarm1-/- background, between DIV0 and DIV7 (mean ± SEM; n = 3; ns (not significant) = p > 0.05, two-way repeated measures ANOVA with Tukey’s mul`ple comparisons for between genotype effects at each `me point). Despite DRG cultures from Nmnat2gtBay/gtE mice having no phenotype in terms of baseline NAD levels, NR administra=on had similar effects to those seen in the more suscep=ble SCG cultures, causing NAD to decline in neurites with sub -heterozygous NMNAT2 expression (supplementary fig. 4). This suggests that in DRG neurons 30% of the C57BL /6 NMNAT2 expression level is insufficient to ac=vate SARM1 and cause NAD(P) deple=on under basal condi=ons. However, boos=ng NMN levels with the precursor NR, is able to =p the balance further in favour of SARM1 ac=va=on and lower NAD, specifically in the neurites, where NMNAT2 is limi=ng. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 18

The findings presented here indicate that chronically low NMNAT2 expression causes sub - lethal SARM1 ac=va=on, which can be enhanced by the NMN precursor NR. SARM1 ac=va=on is thus not a binary, all or nothing response but appears to lie on a spectrum, where par=al ac=va=on does not translate to frank axon degenera=on but is s=ll detectable at the molecular level. While SARM1 ac=va=on seems to occur when NMNAT2 levels fall below a heterozygous threshold of 50% of normal C57BL/6 mouse expression, it is unknown how this level compares to the human spectrum of NMNAT2 expression. It is also possible that a less pronounced decrease from mean NMNAT2 levels or ac=vity has comparable outcomes in humans, especially considering that human axons are longer, and exposed to mul=ple neurodegenera=ve stresses over a substan=ally longer lifespan. Our work supports the hypothesis that low NMNAT2 levels compromise prenatal survival in a SARM1-dependent manner, demonstra=ng that targe=ng SARM1 can be beneficial not only in condi=ons of complete [4] but also of par=al NMNAT2 loss. This observa=on reinforces the trend of reduced viability we previously reported [21], poten=ally strengthened by gene=c selec=on and/or environmental differences following a move of our mice to a new animal facility between the two studies. These findings support the idea that decreased NMNAT2 expression could be more problema=c in some people than others, given the widespread variability in genotype and environment within the human popula=on. Based on the Genome Aggrega=on Database (gnomAD), the probability of LOF intolerance (pLI) for human NMNAT2 is 0.9 8 sugges=ng intolerance of, and selec=ve pressure against hemizygosity. N evertheless, all six of the parents of the biallelic cases so far reported are neurologically healthy [11–13] sugges=ng that other gene=c and/or environmental factors may modify the outcome to explain their healthy survival despite this selec=ve pressure. This appears to parallel the extreme variability between outcomes in Nmnat2gtBay/gtE mice (albeit at sub-heterozygous level), where some pups are non-viable while the ones born alive remain overtly normal throughout life. Interes=ngly, this is the case despite our mice having greater gene=c and environmental homogeneity than that in the human popula=on. Our study has also shown that the NAD precursor supplement NR lowers NAD instead of increasing it in neurites expressing sub -heterozygous levels of NMNAT2. While NR supplementa=on is likely to be harmless and poten=ally beneficial in the majority of the popula=on, (with no toxicity being reported so far in human studies [30]), our findings suggest that NR and possibly other NAD precursors could be problema=c in a subset. As well as the known human cases with muta=ons in NMNAT2, there are other condi=ons where the levels, ac=vity, or transport of the protein could be compromised. The w idespread variability of human Nmnat2 expression [15] raises the possibility that individuals at the lower end of the spectrum could be at increased risk of SARM1 ac=va=on and poten=al neurotoxicity as a result of NR supplementa=on. In addi=on, ageing has been shown to decrease axonal transport of NMNAT2, at least in mice [31], while inhibi=on of protein synthesis [1] and mitochondrial dysfunc=on [22,32] decrease NMNAT2 levels in axons. Thus, the elderly and individuals undergoing treatments that are likely to inhibit axonal transport, including some chemotherapy treatments [33], or people .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 19 with other axonal transport deficiencies, such as muta=ons in genes encoding motor proteins or tubulin cofactors [34], could be among the risk groups. Despite the SARM1-dependent NAD loss in Nmnat2gtBay/gtE neurites following NR administra=on, no effects on neurite morphology were observed. However, these findings raise important ques=ons as to whether prolonged SARM1 ac=va=on and NAD deple=on within axons, resul=ng from NR supplementa=on, could cause axon degenera=on in humans, where axons are much longer, are exposed to many more stressful s=muli, and need to be retained for many decades. In support of this, we previously showed that neurites with sub -heterozygous NMNAT2 expression are more suscep=ble to other neurodegenera=ve stresses [21,22]. Furthermore, the data presented here highlight the need to examine the long -term implica=ons of sub-lethal SARM1 ac=va=on. SARM1 ac=va=on in the absence of axon death has been reported previously in acute models, evident through accumula=on of the SARM1 marker cADPR following administra=on of NMN analogue CZ -48 [5,29], low doses of mitochondrial toxins [35] and treatment with NR in neurons overexpressing the enzyme NRK1 [6] and our data now indicate that par=al SARM1 ac=va=on can also occur chronically. Intact NMNAT ac=vity is able to compensate for increased SARM1 -dependent NAD consump=on, however prolonged SARM1 ac=va=on and accumula=on of associated products, such as cADPR and NAADP (nico=nic acid adenine dinucleo=de phosphate) , could affect cellular physiology leading to func=onal defects despite the lack of morphologically visible neurotoxicity. When NMNAT ac=vity is compromised, as we describe here, prolonged SARM1 ac=va=on and NAD deple=on are expected to have more severe consequences. For instance, chronic SARM1 ac=va=on likely results, or at least contributes, to the behavioural phenotypes and deficits in peripheral axon numbers previously reported in the Nmnat2gtBay/gtE mice [21]. In support, a SARM1-dependent increase in cADPR levels has been reported in scia=c nerves of 2-month-old mice harbouring the human NMNAT2 LOF muta=ons (Nmnat2V98M/R232Q), with SARM1 being required for the neuropathy phenotypes in these mice [13]. Thus, both reduced levels and ac=vity of NMNAT2 can chronically ac=vate SARM1 and compromise neuronal health and survival. It is likely that this will be more pronounced in longer-lived human axons and can poten=ally worsen with age. Finally, assessing the levels of relevant metabolites in accessible =ssues such as blood or CSF could serve as the basis of screening tools for condi=ons involving SARM1 ac=va=on in humans. The present study has provided evidence in support of chronic, sub-lethal SARM1 ac=va=on in non-degenera=ng axons. What sub-lethal SARM1 ac=va=on means in structural terms is nonetheless unknown. One possibility is that fewer SARM1 octamers exist in an ac=ve conforma=on compared to a fully ac=vated state that leads to degenera=on. Alterna=vely, the break of the ARM -TIR lock, which mediates the conforma=onal change required for SARM1 ac=va=on might itself be par=al. Another possibility would be transient ac=va=on when NMN binds followed by deac=va=on when it dissociates, or is replaced by NAD. Structural studies will thus be instrumental in answering this ques=on. Moreover, the absence of degenera=on could be agributed to compensatory mechanisms arising in response to cons=tu=vely low NMNAT2 expression. In support, we previously showed that acute deple=on of a single Nmnat2 allele causes axon degenera=on in primary cultures [3], whereas cons=tu=vely lower levels of NMNAT2 are compa=ble with axon survival. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 20 In summary, we have shown that cons=tu=vely low NMNAT2 levels reduce viability in mice and lead to sub-lethal SARM1 ac=va=on in morphologically intact axons, characterised by NAD(P) deple=on and the development of shorter neurites. The effect of chronic NMNAT2 deple=on is not uniform across different neuron types and varia=ons in the NMNAT2 to NAMPT ra=o might account for the differen=al suscep=bility. Finally, we argue that supplementa=on with NR and other NAD precursors may need addi=onal safe ty studies in condi=ons of reduced NMNAT func=on. Importantly, compromised or reduced NMNAT2 ac=vity could have a more profound effect in human axons considering their greater length and longer lifespan. Although LOF muta=ons in the NMNAT2 gene are rare, the widespread variability of NMNAT2 mRNA expression reported in humans, together with the mul=tude of pathological and physiological situa=ons that can compromise NMNAT2 transport or synthesis, could mean that the effects of NMNAT2 on SARM1 ac=va=o n might be more widespread than previously an=cipated. Finally, the findings of this study raise important ques=ons as to what other neurodegenera=ve stresses can par=ally ac=vate SARM1, to what extent they contribute to sporadic neurodegenera=ve diseases and importantly, if early detec=on and interven=on is possible in humans. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 21

[1] Gilley J, Coleman MP . Endogenous Nmnat2 Is an Essen=al Survival Factor for Maintenance of Healthy Axons. PLoS Biol 2010;8:e1000300. hgps://doi.org/10.1371/journal.pbio.1000300. [2] Hicks AN, Lorenzey D, Gilley J, Lu B, Andersson K-E, Miligan C, et al. Nico=namide Mononucleo=de Adenylyltransferase 2 (Nmnat2) Regulates Axon Integrity in the Mouse Embryo. PLoS ONE 2012;7:e47869. hgps://doi.org/10.1371/journal.pone.0047869. [3] Gilley J, Adalbert R, Yu G, Coleman MP . Rescue of Peripheral and CNS Axon Defects in Mice Lacking NMNAT2. Journal of Neuroscience 2013;33:13410–24. hgps://doi.org/10.1523/JNEUROSCI.1534-13.2013. [4] Gilley J, Orsomando G, Nascimento-Ferreira I, Coleman MP . Absence of SARM1 Rescues Development and Survival of NMNAT2-Deficient Axons. Cell Reports 2015;10:1974–81. hgps://doi.org/10.1016/j.celrep.2015.02.060. [5] Zhao ZY , Xie XJ, Li WH, Liu J, Chen Z, Zhang B, et al. A Cell-Permeant Mime=c of NMN Ac=vates SARM1 to Produce Cyclic ADP-Ribose and Induce Non-apopto=c Cell Death. iScience 2019;15:452–66. hgps://doi.org/10.1016/j.isci.2019.05.001. [6] Figley MD, Gu W, Nanson JD, Shi Y , Sasaki Y , Cunnea K, et al. SARM1 is a metabolic sensor ac=vated by an increased NMN/NAD+ ra=o to trigger axon degenera=on. Neuron 2021;109:1118-1136.e11. hgps://doi.org/10.1016/j.neuron.2021.02.009. [7] Jiang Y , Liu T, Lee C-H, Chang Q, Yang J, Zhang Z. The NAD+-mediated self-inhibi=on mechanism of pro-neurodegenera=ve SARM1. Nature 2020;588:658–63. hgps://doi.org/10.1038/s41586-020-2862-z. [8] Sporny M, Guez-Haddad J, Khazma T, Yaron A, Dessau M, Shkolnisky Y , et al. Structural basis for SARM1 inhibi=on and ac=va=on under energe=c stress. eLife 2020;9:e62021. hgps://doi.org/10.7554/eLife.62021. [9] Confor= L, Gilley J, Coleman MP . Wallerian degenera=on: an emerging axon death pathway linking injury and disease. Nat Rev Neurosci 2014;15:394–409. hgps://doi.org/10.1038/nrn3680. [10] Coleman MP , Höke A. Programmed axon degenera=on: from mouse to mechanism to medicine. Nat Rev Neurosci 2020;21:183–96. hgps://doi.org/10.1038/s41583-020- 0269-3. [11] Lukacs M, Gilley J, Zhu Y , Orsomando G, Angeley C, Liu J, et al. Severe biallelic loss-of- func=on muta=ons in nico=namide mononucleo=de adenylyltransferase 2 (NMNAT2) in two fetuses with fetal akinesia deforma=on sequence. Experimental Neurology 2019;320:112961. hgps://doi.org/10.1016/j.expneurol.2019.112961. [12] Huppke P , Wegener E, Gilley J, Angeley C, Kurth I, Drenth JPH, et al. Homozygous NMNAT2 muta=on in sisters with polyneuropathy and erythromelalgia. Experimental Neurology 2019;320:112958. hgps://doi.org/10.1016/j.expneurol.2019.112958. [13] Dingwall CB, Strickland A, Yum SW, Yim AKY , Zhu J, Wang PL, et al. Macrophage deple=on blocks congenital SARM1-dependent neuropathy. Journal of Clinical Inves=ga=on 2022;132:e159800. hgps://doi.org/10.1172/JCI159800. [14] Ali YO, Li-Kroeger D, Bellen HJ, Zhai RG, Lu H-C. NMNATs, evolu=onarily conserved neuronal maintenance factors. Trends in Neurosciences 2013;36:632–40. hgps://doi.org/10.1016/j.=ns.2013.07.002. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 22 [15] Ali YO, Allen HM, Yu L, Li-Kroeger D, Bakhshizadehmahmoudi D, Hatcher A, et al. NMNAT2:HSP90 Complex Mediates Proteostasis in Proteinopathies. PLoS Biol 2016;14:e1002472. hgps://doi.org/10.1371/journal.pbio.1002472. [16] Loreto A, Antoniou C, Merlini E, Gilley J, Coleman MP . NMN: The NAD precursor at the intersec=on between axon degenera=on and an=-ageing therapies. Neuroscience Research 2023:S0168010223000044. hgps://doi.org/10.1016/j.neures.2023.01.004. [17] Yoshino J, Baur JA, Imai S. NAD+ Intermediates: The Biology and Therapeu=c Poten=al of NMN and NR. Cell Metabolism 2018;27:513–28. hgps://doi.org/10.1016/j.cmet.2017.11.002. [18] Reiten OK, Wilvang MA, Mitchell SJ, Hu Z, Fang EF. Preclinical and clinical evidence of NAD+ precursors in health, disease, and ageing. Mechanisms of Ageing and Development 2021;199:111567. hgps://doi.org/10.1016/j.mad.2021.111567. [19] Hui F, Tang J, Williams PA, McGuinness MB, Hadoux X, Casson RJ, et al. Improvement in inner re=nal func=on in glaucoma with nico=namide (vitamin B3 ) supplementa=on: A crossover randomized clinical trial. Clinical Exper Ophthalmology 2020;48:903–14. hgps://doi.org/10.1111/ceo.13818. [20] Brakedal B, Dölle C, Riemer F, Ma Y , Nido GS, Skeie GO, et al. The NADPARK study: A randomized phase I trial of nico=namide riboside supplementa=on in Parkinson’s disease. Cell Metabolism 2022;34:396-407.e6. hgps://doi.org/10.1016/j.cmet.2022.02.001. [21] Gilley J, Mayer PR, Yu G, Coleman MP . Low levels of NMNAT2 compromise axon development and survival. Human Molecular Gene=cs 2019;28:448–58. hgps://doi.org/10.1093/hmg/ddy356. [22] Loreto A, Hill CS, Hewig VL, Orsomando G, Angeley C, Gilley J, et al. Mitochondrial impairment ac=vates the Wallerian pathway through deple=on of NMNAT2 leading to SARM1-dependent axon degenera=on. Neurobiology of Disease 2020;134:104678. hgps://doi.org/10.1016/j.nbd.2019.104678. [23] Mayer PR, Huang N, Dewey CM, Dries DR, Zhang H, Yu G. Expression, Localiza=on, and Biochemical Characteriza=on of Nico=namide Mononucleo=de Adenylyltransferase 2. Journal of Biological Chemistry 2010;285:40387–96. hgps://doi.org/10.1074/jbc.M110.178913. [24] Kim Y , Zhou P , Qian L, Chuang J-Z, Lee J, Li C, et al. MyD88-5 links mitochondria, microtubules, and JNK3 in neurons and regulates neuronal survival. J Exp Med 2007;204:2063–74. hgps://doi.org/10.1084/jem.20070868. [25] Chen C-Y, L i n C-W, Chang C-Y , Jiang S-T, Hsueh Y-P . Sarm1, a nega=ve regulator of innate immunity, interacts with syndecan-2 and regulates neuronal morphology. J Cell Biol 2011;193:769–84. hgps://doi.org/10.1083/jcb.201008050. [26] Essuman K, Summers DW, Sasaki Y , Mao X, DiAntonio A, Milbrandt J. The SARM1 Toll/Interleukin-1 Receptor Domain Possesses Intrinsic NAD + Cleavage Ac=vity that Promotes Pathological Axonal Degenera=on. Neuron 2017;93:1334-1343.e5. hgps://doi.org/10.1016/j.neuron.2017.02.022. [27] Essuman K, Summers DW, Sasaki Y , Mao X, Yim AKY , DiAntonio A, et al. TIR Domain Proteins Are an Ancient Family of NAD+-Consuming Enzymes. Current Biology 2018;28:421-430.e4. hgps://doi.org/10.1016/j.cub.2017.12.024. [28] Angeley C, Amici A, Gilley J, Loreto A, Trapanogo AG, Antoniou C, et al. SARM1 is a mul=-func=onal NAD(P)ase with prominent base exchange ac=vity, all regulated .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 23 bymul=ple physiologically relevant NAD metabolites. iScience 2022;25:103812. hgps://doi.org/10.1016/j.isci.2022.103812. [29] Li WH, Huang K, Cai Y , Wang QW, Zhu WJ, Hou YN, et al. Permeant fluorescent probes visualize the ac=va=on of SARM1 and uncover an an=-neurodegenera=ve drug candidate. eLife 2021;10:e67381. hgps://doi.org/10.7554/eLife.67381. [30] Damgaard MV, Treebak JT. What is really known about the effects of nico=namide riboside supplementa=on in humans. Sci Adv 2023;9:eadi4862. hgps://doi.org/10.1126/sciadv.adi4862. [31] Milde S, Adalbert R, Elaman MH, Coleman MP . Axonal transport declines with age in two dis=nct phases separated by a period of rela=ve stability. Neurobiology of Aging 2015;36:971–81. hgps://doi.org/10.1016/j.neurobiolaging.2014.09.018. [32] Summers DW, Frey E, Walker LJ, Milbrandt J, DiAntonio A. DLK Ac=va=on Synergizes with Mitochondrial Dysfunc=on to Downregulate Axon Survival Factors and Promote SARM1-Dependent Axon Degenera=on. Mol Neurobiol 2020;57:1146–58. hgps://doi.org/10.1007/s12035-019-01796-2. [33] LaPointe NE, Morfini G, Brady ST, Feinstein SC, Wilson L, Jordan MA. Effects of eribulin, vincris=ne, paclitaxel and ixabepilone on fast axonal transport and kinesin-1 driven microtubule gliding: Implica=ons for chemotherapy-induced peripheral neuropathy. NeuroToxicology 2013;37:231–9. hgps://doi.org/10.1016/j.neuro.2013.05.008. [34] Sferra A, Baillat G, Rizza T, Barresi S, Flex E, Tasca G, et al. TBCE Muta=ons Cause Early- Onset Progressive Encephalopathy with Distal Spinal Muscular Atrophy. The American Journal of Human Gene=cs 2016;99:974–83. hgps://doi.org/10.1016/j.ajhg.2016.08.006. [35] Sasaki Y , Engber TM, Hughes RO, Figley MD, Wu T, Bosanac T, et al. cADPR is a gene dosage-sensi=ve biomarker of SARM1 ac=vity in healthy, compromised, and degenera=ng axons. Experimental Neurology 2020;329:113252. hgps://doi.org/10.1016/j.expneurol.2020.113252. [36] Gilley J, Ribchester RR, Coleman MP . SARM1 dele=on, but not WldS, confers lifelong rescue in a mouse model of severe axonopathy. Cell Rep 2017; 21: 10-16. hgps://doi.org/10.1016/j.celrep.2017.09.027.

We thank Astra Zeneca for synthesising and providing PC6 and Yi-Ping Hsueh for providing the SARM1 monoclonal an=body. Author contribution Christina Antoniou: conceptualisation, data acquisition, data analysis, data interpretation, study design, writing—original draft, writing—review and editing. Andrea Loreto: data acquisition, data interpretation, study design, writing—review and editing . Jonathan Gilley: data interpretation, study design, writing—review and editing. Elisa Merlini: data acquisition, writing—review and editing . G iuseppe Orsomando : data interpretation, writing—review and editing . M ichael P Coleman: conceptualisation, data interpretation, study design, supervision, writing—original draft, writing—review and editing . All authors read and approved the final manuscript. Funding C.A. is funded by the MRC DTP Studentship and Gates Foundation; A.L. is funded by the Wellcome Trust [Grant number 210904/Z/18/Z]; E.M. is funded by the Cambridge Trust; J.G. is funded by ALS Finding a Cure and the ALS Association 959996; G.O. is funded by the .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint 24 Italian Grants RSA 20 20-2022 from UNIVPM. M.P.C. is funded by the John and Lucille van Geest Foundation. Data Availability The datasets generated and analysed during the current study are available from the corresponding author upon reasonable request. Declarations Competing Interests MPC consults for Nura Bio and Drishti Discoveries and the Coleman group is part funded by AstraZeneca for academic research projects but none of these activities relate to the study reported here. Ethics Approval Animal work was approved by the University of Cambridge and performed in accordance with the Home Office Animal Scientific Procedures Act (ASPA), 1986 under project licence P98A03BF9. Consent to Participate Not applicable. Consent for Publication Not applicable. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 14, 2024. ; https://doi.org/10.1101/2024.03.14.584798doi: bioRxiv preprint