CRISPR/Cas9-mediated SARM1 knockout and epitope-tagged mice reveal that SARM1 does not regulate nuclear transcription, but is expressed in macrophages

,

Sterile alpha and HEAT/Armadillo motif-containing protein (SARM1) is an ancient and highly conserved immune protein, with orthologues in lower organisms including Caenorhabditis elegans, and D. melanogaster as well as mammals (1). It is the most recently identified member of the toll-like receptor (TLR) adaptor family of proteins and comprises two sterile alpha motif (SAM) domains, armadillo repeats, and a toll/ interleukin-1 receptor (TIR) domain (2). The SAM domains mediate homotypic protein-protein interactions and thereby facilitate SARM1 oligomerization into an octamer in both its active and inactive states (3,4). The presence of a TIR domain predicts a role for SARM1 in innate immune signaling. However, in contrast to the other TLR adaptor proteins, SARM1 does not facilitate TLR signal transduction. In fact, human SARM1 has been shown to antagonize TLR signaling through both TIR domain-containing adaptor-inducing interferon-β (TRIF) (5) and myeloid differentiation factor 88 (MyD88) (6) through TIR-TIR interactions (7). As well as having an opposing role in signaling to other TLR adaptors, SARM1 is also distinguished from the family by the presence of NADase activity within its TIR domain (8). Interestingly, phylogenetic analysis revealed that SARM1 TIR domain clusters more closely with bacterial TIRs than other animal TIR domain-containing proteins, suggesting it emerged in animals from bacteria via lateral gene transfer (9).
It has become clear in recent years that the functions of SARM1 extend beyond innate immunity. SARM1 is highly expressed in neurons, and hence many studies focus on the roles for SARM1 in cell death and axon degeneration in the nervous system. SARM1 was shown to mediate the death of mouse hippocampal neurons following glucose and oxygen deprivation (10), and murine SARM1 has also been shown to drive programmed cell death in sensory neurons downstream of oxidative stress (11). In addition to regulating neuronal cell death, SARM1 has been identified as the central executor of axon degeneration in response to a range of triggers. Following axotomy, the distal portion of the transected axon disintegrates by an active process known as Wallerian degeneration. In mice lacking SARM1, severed axons are protected from this and exhibit remarkable long-term survival (12). Sarm1 −/− mice also show attenuated surrogate markers of axon degeneration and reduced functional deficits after traumatic brain injury (13). Additionally, in the absence of SARM1, mouse neurons are protected from the axon degeneration which is induced by the chemotherapeutic agents vincristine (14), paclitaxel (15), and cisplatin (16). This axon degeneration limits the usefulness of these chemotherapies, as it results in peripheral neuropathy. In these circumstances, absence of SARM1 is advantageous, as it ameliorates the pathological outcomes associated with axon degeneration. There are situations, however, when SARM1induced axon degeneration is beneficial. Recently, it was demonstrated that SARM1 is protective in a mouse model of ulcerative colitis (17). SARM1-mediated axon degeneration in the enteric nervous system resulted in reduced secretion of norepinephrine, a neurotransmitter that drives inflammatory IL-17 secretion from T H 17 cells and ILC3s causing local inflammation in the colon. In addition, in the context of rabies infection, SARM1 mediates axonal degeneration and so diminishes the spread of the virus (18). The role of SARM1 in axon degeneration can be accounted for by the intrinsic NADase activity in the TIR domain. This NADase activity becomes activated when the inhibitory lock between the TIR domain and auto-inhibitory ARM domain is disrupted (19), resulting in catastrophic NAD and ATP depletion within the axon, which is temporally coupled with axon degeneration (8).
Independent from its role in cell death and axon degeneration, SARM1 is also known to regulate transcription in neurons under certain conditions. Following traumatic injury of neurons, SARM1 is required for the induction of CCL2, CCL7, CCL12, and CSF-1 (20). Remarkably, the expression of these chemokines and cytokines could be induced by forced dimerization of the SARM1 TIR domain alone. In addition, altered transcription has been seen in the brain of SARM1deficient mice in a number of neurotropic viruses. Following challenge with West Nile Virus (WNV), Sarm1 −/− mice displayed reduced TNF production in the brainstem, which resulted in enhanced susceptibility to the infection (21). In response to vesicular stomatitis virus, Sarm1 −/− mice showed reduced expression of a range of cytokines in the brain, and in contrast to WNV, this was protective and Sarm1 −/− mice did not succumb to infection (22). We also previously described a role for SARM1 in regulating transcription in immune cells (23). We reported that following stimulation of pattern recognition receptors with a range of ligands, Sarm1 −/− bonemarrow-derived macrophages (BMDM) showed reduced Ccl5 expression compared with WT cells, for both TLR-dependent and -independent stimulation of cells, suggesting a direct transcriptional role for SARM1 for select genes.
Compared with mechanistic insights on how SARM1 modulates axonal degradation that has arisen from elucidation of the structure and enzyme activity of SARM1 protein, it is unclear how SARM1 may regulate transcription and how general or specific this regulation might be. To address this, we performed RNA sequencing on BMDM from TLR-stimulated WT and 129 ES cell-generated Sarm1 −/− mice backcrossed onto a C57BL/6 background (herein referred to as B6 congenic Sarm1 −/− mice). This revealed that a disproportionate fraction of differentially expressed genes (DEGs) flanked the Sarm1 locus. A high density of single-nucleotide polymorphisms (SNPs) and insertions and deletions (indels) associated with 129 mice were observed on either side of the Sarm1 locus, indicating that these are passenger genes neighboring the Sarm1 −/− mutation that were not removed during backcrossing to C57BL/6 mice. Three independent SARM1deficient mice generated by CRISPR/Cas9-mediated genome editing confirmed that transcriptional phenotypes, which we previously attributed to the absence of SARM1, were actually accounted for by the presence of passenger genes.
Further, there has been uncertainty as to where SARM1 is expressed, which has been an obstacle in gaining greater understanding of SARM1 function. The lack of a reliable commercially available antibody has precluded comprehensive investigation into which tissues and cells express SARM1. By generating a mouse expressing epitope-tagged SARM1 endogenously, we clarify here that SARM1 is indeed expressed in macrophages as well as the brain. We verify that macrophages from these mice exhibit normal cytokine induction following TLR stimulation and show that the epitope tag does not interfere with SARM1 function in axon degeneration in primary neurons.

Results
Sarm1 −/− macrophages show reduced expression of Ccl5, but not Tnf or Il6, compared with WT cells We previously demonstrated that the expression of Ccl5 is diminished in BMDM from Sarm1 −/− mice compared with WT mice following stimulation with a diverse range of both TLR and non-TLR ligands (23). Here, we found that Ccl5 transcription was significantly reduced in Sarm1 −/− pBMDM relative to WT following TLR7 stimulation with CL075 and TLR4 stimulation with MPLA over the course of 24 h (Fig. 1, A and C). Similarly, following TLR4 stimulation with lipopolysaccharide (LPS), Ccl5 expression was reduced at the mRNA level, and while it fell short of statistical significance, it is a consistent trend (Fig. 1B). The diminished transcription led to substantially reduced secretion of CCL5 protein by Sarm1 −/− BMDM relative to WT following stimulation with LPS, MPLA, or CL075 ( Fig. 1, J-L). This reduced CCL5 expression was not reflective of a global impairment in induction of inflammatory genes in BMDM from Sarm1 −/− mice since TNF (Fig. 1, D-F and M-O) and IL-6 ( Fig. 1, G-I and P-R) mRNA and protein expression were not consistently lower in these cells relative to WT counterparts. This does not support an inhibitory role for murine SARM1 in TRIF signaling, contrary to human SARM1 (5).
We also previously showed that activation of transcription factors commonly associated with TLR-stimulated genes in macrophages, i.e., NFκB and interferon regulators factors (IRFs), was normal in Sarm1 −/− BMDM. We therefore sought to determine the cellular localization of SARM1 to understand how it may regulate transcription. While it has been reported that SARM1 can enter the nucleus to stabilize lamins (24), most reports show that SARM1 exists at the mitochondria (25,26) and in the cytosol (27). To assess if SARM1 could enter the nucleus to regulate transcription in BMDM, we used cellular fractionation of immortalized B6 congenic Sarm1 −/− BMDM stably overexpressing flag-tagged SARM1 to isolate a pure nuclear fraction and mixed cytoplasmic fraction. Flag-tagged SARM1 could not be detected in the nuclear fraction either before or after LPS treatment (Fig. S1) showing that in BMDM, SARM1 does not reside at the nucleus, nor does it translocate there following LPS stimulation to regulate transcription.
Passenger genes flank the Sarm1 locus and confound interpretation of differential gene expression Knowing that TLR-stimulated CCL5 is diminished in B6 congenic Sarm1 −/− BMDM, we wondered what other genes, if any, may be dependent on SARM1 for optimal expression and the mechanism by which they could be SARM1-regulated. We therefore employed RNA sequencing to provide a broad, unbiased overview of transcription in BMDM from B6 congenic Sarm1 −/− mice. RNA from unstimulated BMDM was analyzed for differential gene expression to determine if any differences exist basally in the transcriptional landscape of BMDM from WT and B6 congenic Sarm1 −/− mice. Similarly, RNA from CL075-stimulated BMDM was analyzed for differential gene expression to identify potential SARM1-dependent TLR inducible genes. Differentially expressed genes (DEGs) from both comparisons were compiled into a single gene list, which was interrogated for patterns, which may shed light on the mechanism by which SARM1 influenced their expression. No clear patterns emerged following transcription factor binding site analysis or pathway analysis, suggesting that these identified DEGs were not differentially expressed due to regulation of a known regulatory protein or transcription factor by SARM1. However, when a heatmap was generated with DEGs clustered by chromosomal location, it became clear that chromosome 11, on which the Sarm1 and Ccl5 loci reside, was disproportionately represented ( Fig. 2A).
This prompted a re-examination of the B6 congenic Sarm1 −/− mouse, which was generated using targeted gene Figure 2. Transcriptome analysis of BMDMs from Sarm1 −/− mice compared with WT BMDMs reveals differentially expressed genes and passenger genes that cluster near the Sarm1 locus. A, heatmap displaying genes, which are differentially expressed between WT and Sarm1 −/− pBMDM, either basally or following CL075 stimulation for 3 h, grouped according to chromosome. B, graph showing density of SNPs and indels per 10 kb section of chromosome 11 in Sarm1 −/− BMDM relative to the C57BL/6 reference sequence. The Sarm1 locus is indicated by a green arrow. C, WT and Sarm1 −/− pBMDMs were stimulated with 5 μg/ml CL075 (C-E) or 100 ng/ml LPS (F-H) for 3 h or medium as a control. Acap1 (C and F), Ccl9 (D and G), and Ccl6 (E and H) mRNA were measured by qRT-PCR, normalized to the housekeeping gene β-actin, and are presented relative to the untreated WT control. Graphs show mean ± SEM from six mice per genotype, performed in triplicate. **p < 0.01, ***p < 0.001, multiple Mann-Whitney tests with Holm-Sídák multiple comparisons test. See also Figs. S2 and S3. SARM1 CRISPR-Cas9 mice clarify SARM1 function and expression disruption. In this mouse, exons 3 through 6 of the Sarm1 locus were disrupted by replacement with a targeting vector containing a neomycin-resistance cassette in embryonic stem cells derived from an unspecified 129/SvJ mouse, known as the donor (10). These cells were implanted to a C57BL/6 blastocyst, the recipient, and following germline transmission and at least 15 generations of backcrossing, a SARM1-deficient mouse on the C57BL/6 background was achieved. However, it is now well established that mice generated in this way retain donor-derived genetic material, termed passenger genes or passenger mutations, in the region flanking the targeted gene (28). The probability of a gene being of donor origin depends on the proximity to the targeted locus and the number of generations of backcrosses to C57BL/6 mice. To investigate the presence of these passenger genes in our mice, we compared the RNA sequence from B6 congenic Sarm1 −/− BMDM to the C57BL/6 mm10 reference sequence. Using a sliding window approach, we examined the density of SNPs and insertions/deletions (indels) per 10 Kb along chromosome 11 in the RNA from BMDM from B6 congenic Sarm1 −/− mice, relative to the C57BL/6 reference sequence, and found a high density of SNPs and indels in the region flanking the Sarm1 locus (Fig. 2B). This pattern was not seen on other chromosomes (Fig. S2).
Among the differentially expressed genes in this region were Ccl6, Ccl9, and Acap1, which are 4.32, 4.31, and 2.955 cM, respectively, from the Sarm1 locus, as determined by the JAX Mouse Map Converter (based on (29,30)). Assuming 16 to 20 generations of backcrossing to C57BL/6, the probability of a passenger gene being retained within 5 cM from the targeted gene is 46.33 to 37.74% (31). Thus, we examined the sequence of RNA from B6 congenic Sarm1 −/− BMDM for 129-associated variants at these loci. When the RNA sequence from B6 congenic Sarm1 −/− BMDM was examined using the Integrative Genomics Viewer (IGV) from the Broad Institute, deviations from the reference sequence were visible in the Acap1 and Ccl9 loci (Fig. S3, A and B). By manual comparison, these deviations were found to correspond to 129-associated SNPs as listed by the Wellcome Sanger Institute's Mouse Genomes Project (32). No SNPs or indels were visible in the Ccl6 locus in B6 congenic Sarm1 −/− BMDM (Fig. S3C), which corresponds with the lack of exonic 129-associated variations listed in the Mouse Genomes Project. DEG analysis showed that Ccl6 and Ccl9 expression were lower in B6 congenic Sarm1 −/− BMDM compared with their WT counterparts, and that Acap1 was more highly expressed ( Fig. 2A). We validated these differences by quantitative RT-PCR: following either CL075 or LPS stimulation, Sarm1 −/− BMDM showed significantly less Ccl6 and Ccl9 expression, and significantly greater Acap1 expression, than their WT counterparts (Fig. 2, C-H). Importantly, these differences also existed in unstimulated cells, which could have implications for experimental models using these mice. Overall, RNA sequencing revealed that a number genes are differentially expressed in B6 congenic Sarm1 −/− BMDM compared with WT cells, but it is unclear whether this is due to the absence of SARM1 or the presence of passenger genes which flank the Sarm1 locus in these cells.

Generation and validation of three SARM1-deficient mice using CRISPR/Cas9
To study the effect of SARM1 deficiency on transcription in BMDM without the confounding factor of passenger genes, new Sarm1 −/− mice were generated by CRISPR/Cas9-mediated genome engineering. Figure 3A shows the CRISPR/Cas9 targeting strategy used, which generated three lines with indels of 2 bp, 5 bp, and 34 bp resulting in premature stop codons. Each of the three novel lines contained deletions within the first exon of the Sarm1 locus; Sarm1 em1.1Tftc (2 bp deletion), Sar-m1 em1.2Tftc (34 bp deletion), and Sarm1 em1.3Tftc (5 bp deletion) (Fig. 3B). These deletions each interrupt the recognition site of the restriction enzyme BsaWI, which facilitated the use of this enzyme in genotyping (Fig. 3B). Heterozygous breeding pairs were established to produce WT and SARM1-deficient littermate mice (Fig. 3C). As SARM1 is highly expressed in neurons, we verified that SARM1 protein expression was lost in the brain of each line by Western blot (Fig. 3D). SARM1 is difficult to reliably and specifically detect using commercially available antibodies, so we generated a polyclonal antibody by immunizing chicken with human SARM1 TIR domain recombinant protein. Indeed, SARM1 could be detected in the brains of WT mice, but not of their Sarm1 em1.1Tftc , Sarm1 em1.2Tftc , and Sar-m1 em1.3Tftc littermates, nor in Sarm1 −/− brains (Fig. 3D).
The role of SARM1 as a driver of axon degeneration has been extensively studied in the context of traumatic brain injury, axotomy, and mitochondrial dysfunction, but mainly in B6 congenic Sarm1 −/− mice. The chemotherapeutic drug vincristine is widely reported to induce SARM1-dependent axon degeneration (14,33,34). We exploited this to assess and confirm the loss of SARM1-dependent axonal degeneration in the Sarm1 em1.1Tftc , Sarm1 em1.2Tftc , and Sarm1 em1.3Tftc mice. Using live-cell imaging microscopy, we examined vincristine-induced neurite degeneration in postnatal primary neurons cultured for 7 days in vitro (DIV7) from WT, B6 congenic Sarm1 −/− mice, and each of the three CRISPR knockout lines (Fig. 3E). Neurons were treated with vincristine and scanned every 6 h for 3 days, starting 1 h posttreatment. WT neurons exhibited altered morphology and clear neurite loss in the 24 h following vincristine treatment (Fig. 3F, Movies S1 and S2). In line with the literature, Sarm1 −/− neurons were protected from axon degeneration following insult with the microtubule poison vincristine. Neurons from Sarm1 em1.3Tftc mice were also robustly protected from axon degeneration (Fig. 3F). In contrast to prenatal dorsal root ganglion neurons, which are often used in such assays, fragments from degenerated primary neuron neurites disappear and could not be counted to quantitate. Therefore, to quantitatively appraise the axon degeneration, we measured neurite length, which is commonly used to determine the effects of a particular substrate on neural outgrowth integrity. We measured neurite length at 1 h, 24 h, and 72 h after vincristine treatment. Vincristine treatment resulted in dramatic and significant loss SARM1 CRISPR-Cas9 mice clarify SARM1 function and expression Figure 3. Generation of SARM1 knockout mice by CRISPR/Cas9 and validation of neuronal degeneration phenotype. A, CRISPR/Cas9 targeting strategy for SARM1 knockout. The mouse Sarm1 gene comprises nine exons (numbered boxes). A gRNA target sequence was identified downstream of the ATG start site on the antisense strand (blue). CRISPR/Cas9 cutting at this site generated indels of 2 bp, 5 bp, and 34 bp resulting in premature stop codons. B, multiple sequence alignment showing disrupted Sarm1 locus compared with WT for Sarm1 em1.1Tftc (2 bp deletion), Sarm1 em1.2Tftc (34 bp deletion) and Sarm1 em1.3Tftc (5 bp deletion). The start codon (black bar), the guide sequence (red bar), and the BsaWI, restriction enzyme site, used for genotyping are indicated. C, genotyping results for CRISPR littermates (WT, hets, homs). PCR amplicon from mouse genomic DNA was incubated with (+) or without (−) BsaWI enzyme. The amplicon containing WT sequence is recognized by the enzyme and shows a cleaved band, whereas the disrupted sequence does not. D, immunoblot analysis of brain lysate for SARM1 and vinculin (loading control). WT1 is a littermate of Sarm1 em1.1Tftc , same as WT2 and Sarm1 em1.2Tftc , of neurite length in WT neurons after 24 h, which was maintained after 72 h ( Fig. 3G; unnormalized and raw data for this is shown in Fig. S4, A and B). As expected from previous studies, Sarm1 −/− neurons were protected from loss of neurite length following vincristine treatment. Importantly, neurons from each of the three new CRISPR knockout lines retained their full neurite length, even 72 h after vincristine treatment. Thus this assay demonstrated that neurons from Sar-m1 em1.1Tftc , Sarm1 em1.2Tftc , and Sarm1 em1.3Tftc mice faithfully recapitulate the axoprotective phenotype of Sarm1 −/− neurons.
In the absence of passenger genes, SARM1-deficient BMDM displays normal transcription Having verified that SARM1 protein is not expressed in Sarm1 em1.1Tftc , Sarm1 em1.2Tftc , and Sarm1 em1.3Tftc and having shown that the CRISPR knockout lines are phenotypically similar to Sarm1 −/− mice with regard to axon degeneration, we then sought to interrogate the transcriptional phenotype in BMDM without the confounding factor of passenger genes. In particular, since Ccl5 resides on chromosome 11, less than 5 cM from the Sarm1 locus, passenger genes could explain the differential Ccl5 expression seen in Sarm1 −/− BMDM (Fig. 1). Indeed, contrary to what was seen in Sarm1 −/− BMDM, Ccl5 transcription and subsequent secretion were not significantly different in BMDM from Sarm1 em1.3Tftc mice compared with their WT littermates following stimulation with LPS over a 24 h period (Fig. 4, A and C). A similar phenotype was observed in BMDM from both Sarm1 em1.1Tftc and Sar-m1 em1.2Tftc mice (Fig. S5). The reduced Ccl5 expression previously observed in B6 congenic Sarm1 −/− BMDM was therefore not as a result of SARM1 deficiency, but of passenger genes. We then measured expression of three genes, which were identified by RNA sequencing as differentially expressed in Sarm1 −/− and validated by RT-PCR: Acap1, Ccl9, and Ccl6. Compellingly, all three genes were expressed equally in BMDM from Sarm1 em1.3Tftc mice and their WT littermates, both basally and following LPS stimulation over a 24 h time course (Fig. 4, E-G). Therefore, the previously observed differential expression resulted from the presence of passenger genes and not the absence of SARM1.
Transcription is normal in the brainstem of Sarm1 em1.1Tftc , Sarm1 em1.2Tftc , and Sarm1 em1.3Tftc mice Although SARM1 deficiency does not alter transcription in BMDM in the absence of passenger genes, we wondered if differences in transcription would persist in the brain, where SARM1 is reported to be more highly expressed. Other transcriptional phenotypes previously ascribed to SARM1 deficiency in the brain have also been shown to be attributable to passenger genes in Sarm1 −/− mice. Recently, the García-Sastre lab generated two independent SARM1-deficient mice using CRISPR technology and demonstrated that their previously published phenotype of altered cytokine and chemokine production in the brain of Sarm1 −/− mice following vesicular stomatitis virus infection (22) was not recapitulated where passenger genes are absent (35). In that study, they found using RNA sequencing that a number of genes, including Tfrc, Rps29, Rpl38, Ndufb3, and Atp5k were homeostatically differentially expressed in the brainstem of SARM1-deficient mice generated by CRISPR relative to WT. Having verified that Sarm1 em1.1Tftc , Sarm1 em1.2Tftc , and Sarm1 em1.3Tftc mice do not express SARM1 protein in the brain, we therefore examined expression of Ccl5, Tfrc, Rps29, Rpl38, Ndufb3, and Atp5k in the brainstem in each of these independent lines. Using RT-PCR, we found no significant differences in expression of any of these genes in the brainstem of Sarm1 em1.1Tftc mice compared with their WT littermates (Fig. 5, A-F), nor in the brainstems of Sarm1 em1.2Tftc and Sarm1 em1.3Tftc mice compared with their WT littermates (Fig. S6). Thus data from our three independent SARM1-deficient mice do not support a role for SARM1 in transcription in the absence of passenger genes, neither in BMDM nor in the brainstem.

Generation and validation of mouse expressing epitope-tagged SARM1
Our data from these three new CRISPR/Cas9-generated Sarm1 −/− mice confirms a role for SARM1 in neuronal axonal degeneration but does not reveal any phenotypic transcriptional differences between Sarm1 −/− and WT littermate BMDMs. Although we and others have reported roles for SARM1 in macrophages, the lack of a commercially available antibody, which can reliably and specifically detect mouse SARM1 in tissues outside of the brain, has hindered studies into the functions for SARM1 outside of the nervous system, including in macrophages. To overcome this, we used CRISPR technology to successfully generate a mouse expressing an epitope-tagged SARM1 endogenously, with a triple flag tag and double strep tag on the C-terminal end (Fig. 6A), herein referred to as Sarm1 Flag mice (see Experimental procedures for details on the targeting and repair strategy used). These Sarm1 Flag mice showed similar Sarm1 mRNA expression to their WT counterparts across a number of tissues, being most highly expressed in the brain, and expressed at a lower level in the liver, spleen, and kidneys (Fig. 6B).
We used the Sarm1 Flag mice to interrogate SARM1 protein expression in the brain, the brainstem, and BMDM. Western blotting showed that SARM1 is as highly expressed in the brainstem as it is in the cerebrum (Fig. 6C). SARM1 was undetectable by anti-flag immunoblot in BMDM from Sarm1 Flag mice (Fig. 6C), although it was detectable at the mRNA level  Fig. S4, Movies S1 and S2. SARM1 CRISPR-Cas9 mice clarify SARM1 function and expression (Fig. 6D). To overcome experimental constraints associated with limited primary BMDM numbers, we established immortalized BMDMs from Sarm1 Flag mice and littermate WT mice. SARM1 was clearly detected in these immortalized Sarm1 Flag BMDM using flag immunoprecipitation followed by anti-flag immunoblot, albeit at lower levels than in the brain of Sarm1 Flag mice (Fig. 6E). Hence, we confirmed that SARM1 protein expression is not limited to brain.
To facilitate future studies using the Sarm1 Flag mice, we performed experiments to confirm that SARM1-flag protein retained functionality in these mice. Firstly we assessed cytokine induction in pBMDM from Sarm1 Flag mice following LPS stimulation and found that Sarm1 Flag BMDM showed normal Ccl5 and Tnf transcription and secretion (Fig. 7, A-D). Importantly, we used the vincristine-induced neurite degeneration assay, which previously confirmed that this process is SARM1-dependent (Fig. 3G) to assess whether SARM1-flag protein in the Sarm1 Flag mice retained the ability to mediate axonal degeneration. We found that neurons from Sarm1 Flag mice showed similar loss of neurite outgrowth to WT neurons after 24 and 72 h, indicating that epitope-tagged SARM1 protein retains its ability to be activated and mediate axon degeneration (Fig. 7E, unnormalized and raw data for this is shown in Fig. S7). Thus, using these Sarm1 Flag mice, we clarify that SARM1 protein expression is not limited to the brain and show that the epitope tags allow for detection of SARM1 without interfering with SARM1 expression or neuronal functional. These mice therefore are a useful tool for further studies into SARM1 expression and function.

Discussion
We initially set out to understand the mechanism by which SARM1 positively regulates TLR-dependent Ccl5 transcription in murine macrophages. Given that we previously demonstrated that transcription factor translocation to the nucleus was normal in Sarm1 −/− BMDM, but transcription factor and RNA polymerase II recruitment to the Ccl5 promoter was diminished relative to WT (23) we had considered the possibility that SARM1 could reside at or enter the nucleus of BMDM following stimulation. There, we speculated that it may physically interact with transcription factors, or catalyze the local depletion of NAD, reducing its availability to enzymes, which modulate chromatin accessibility and thereby regulate transcription. However we found that SARM1 does not reside at the nucleus or translocate there following stimulation, so we therefore undertook a full analysis of the TLRmodulated transcriptome in WT and Sarm1 −/− BMDM in A-G, WT and Sarm1 em1.3Tftc pBMDMs were stimulated with 100 ng/ml LPS for the indicated times or medium as a control. Expression of Ccl5 (A), Tnf (B), Acap1 (E), Ccl9 (F), and Ccl6 (G) mRNA in were assayed by quantitative RT-PCR, normalized to the housekeeping gene β-actin, and are presented relative to the untreated WT control. Supernatants were assayed for CCL5 (C) and TNF (D) protein by ELISA. Graphs show mean ± SEM from three mice per genotype, performed in triplicate. There were no significant differences determined by multiple Mann-Whitney tests with Holm-Sídák multiple comparisons test. SARM1 CRISPR-Cas9 mice clarify SARM1 function and expression order to obtain insights into the role of SARM1 in regulating nuclear transcription. Through analysis of RNA sequencing data, it became apparent that proximity to the Sarm1 locus was a common factor among a number of DEGs. Interrogation of the RNA sequence from Sarm1 −/− BMDM revealed a high density of SNPs and indels relative to the C57BL/6 reference sequence on chromosome 11, flanking the Sarm1 locus. Further analysis revealed that a number of the DEGs had sequence deviations, which corresponded to 129-associated variants, indicating that these were passenger genes and were introduced to the genome during the generation of the B6 congenic Sarm1 −/− mouse. Some DEGs did not display any sequence deviations, such as Ccl6; however, their altered expression may be a result of SNPs or indels within promoters or enhancers or disruption of the 3D chromosomal landscape. These differences would not be detectable by examination of the transcribed RNA alone. A number of genes were differentially expressed basally between BMDMs from WT and B6 congenic Sarm1 −/− mice, and this could have implications for the interpretation of experiments involving B6 congenic Sarm1 −/− mice or tissues or cells derived from these mice.
It has been well established that generation of C57BL/6 congenic mice following target gene disruption in 129 ES cells results in residual donor genetic material in the recipient mouse in the form of passenger genes, and that the probability of a certain locus retaining this donor material is a function of its proximity to the target gene and the number of generations of backcrosses to the recipient mouse strain. In this case, a number of genes we observed to be differentially expressed in Sarm1 −/− BMDM by RNA sequencing lie near the Sarm1 locus. Ccl6, Ccl9, Acap1, and Ccl5 are all within 5 cM of the Sarm1 locus, where the probability of genes being of 129 origin is 46% (31), assuming 16 backcrosses to the C57BL/6 genome. The generation of three independent SARM1deficient mice using CRISPR/Cas9 allowed us to delineate whether it was the absence of SARM1 or presence of passenger genes that accounted for differential expression. Our data from these mice did not support a role for SARM1 in transcription in BMDM. This is in agreement with the García-Sastre lab's report that Ccl3, Ccl4, and Ccl5 transcriptions are defective in B6 congenic Sarm1 −/− mice, but normal in CRISPR knockout mice (35).
Also in line with findings from the García-Sastre lab, we noted that an alternative Xaf1 transcript with an extended 3 0 UTR was expressed in B6 congenic Sarm1 −/− BMDM (data not shown). This has implications for a previous report that used these mice to implicate SARM1 in regulation of XAF1 and hence neuronal apoptosis (36). In contrast to findings from the García-Sastre study (35), we did not observe any alterations in basal expression of brainstem Tfrc, Rps29, Rpl38, Ndufb3, or Atp5k between WT and CRISPR-generated SARM1-deficient mice, across three independent lines. These differences between the two studies remain to be resolved. Our findings add to a growing body of literature implicating passenger genes as a possible cause of phenotypes previously attributed to the targeted gene in congenic mice (37)(38)(39)(40).
While transcriptional phenotypes previously reported in Sarm1 −/− mice did not persist in our CRISPR-generated SARM1-deficient mice, the axoprotective phenotype did. We found that primary neurons from each of the three CRISPRgenerated SARM1-deficient mice were robustly protected from vincristine-induced axon degeneration, which is the first demonstration of SARM1-dependent protection from SARM1 CRISPR-Cas9 mice clarify SARM1 function and expression chemotherapeutic agent-induced axonal loss outside of congenic Sarm1 −/− mice. As such, this demonstrates that these mice are a valuable tool, which may be useful for further studies pertaining to murine SARM1 function in neurons, particularly where altered transcription due to passenger genes may confound interpretation results. Having such reliable animal models to probe SARM1 function in vivo is especially important and urgent currently since new data on SARM1 structure and function is rapidly emerging (41)(42)(43)(44), leading to a heightened focus on SARM1 as a potential therapeutic target for neurodegenerative disease, traumatic brain injury, and peripheral neuropathies (45,46) Comprehensive investigation into which cells and tissues mouse SARM1 are expressed in, particularly in regard to macrophages, has been hindered by difficulty in detecting SARM1 using commercially available antibodies and has been also noted by others (35). To overcome this, we generated to our knowledge the first epitope-tagged SARM1 expressing mouse using CRISPR/Cas9-mediated genome engineering. In this Sarm1 Flag mouse, we found robust SARM1 protein Figure 6. Generation of epitope-tagged SARM1 mice, Sarm1 Flag . A, CRISPR/Cas9 targeting strategy to generate SARM1 FLAG-Strep (Sarm1 Flag ) mice. The mouse Sarm1 gene comprises nine exons (numbered boxes). A gRNA target sequence was identified in exon 9 (blue) proceeding the stop codon. A repair template was used to insert a triple FLAG, double Streptag II (purple and orange boxes, respectively) at the C-terminus via homology directed repair. A cc > tt base change was engineered into the repair construct, underlined, to ablate the PAM sequence of the original gRNA target site. B, expression of Sarm1 mRNA in the brain, liver, kidney, spleen assayed by quantitative RT-PCR, normalized to the housekeeping gene β-actin, and are presented relative to the WT spleen. Data are mean ± SEM (n = 4-8). Each datapoint represents one mouse. A two-tailed Mann-Whitney test was used to calculate the p values. C, immunoblot comparing expression of flag-tagged SARM1 in the brain, brainstem, and BMDM from Sarm1 Flag mice, with WT brain lysate acting as a negative control. In total, 30 μg of protein was loaded per sample as determined by BCA assay. β-actin was used as a loading control. Blot is representative of four independent experiments. D, expression of Sarm1 mRNA in Sarm1 Flag BMDM as measured by qRT-PCR, normalized to the housekeeping gene β-actin, and presented relative to the WT control. Data are mean ± SEM from three mice per genotype, performed in triplicate. E, SARM1 is expressed weakly in BMDM compared to brain. Flag-tagged SARM1 was detected by immunoprecipitation in Sarm1 −/− iBMDM stably overexpressing SARM1 or Sarm1 Flag iBMDM (iMac), and Sarm1 Flag brain. β-actin was used as a loading control for input. Representative of three independent experiments. ns, not significant. expression in the brainstem, where no transcriptional phenotype was observed, and in the brain. In BMDM where SARM1 expression has been contested, we saw that SARM1 was expressed, though to a lesser degree than in the brain. It was readily detectable by Western blot following immunoprecipitation in immortalized BMDM from Sarm1 Flag mice. However, using the same approach we were unable to detect tagged SARM1 in primary BMDM, so it may be that immortalization increases the expression of SARM1. We verified that these mice resemble WT in both Sarm1 expression and function. In the Sarm1 Flag mice, the mRNA levels of Sarm1 were normal in the brain, liver, spleen, and kidney. Further, we found that BMDM from Sarm1 Flag and WT mice showed similar cytokine induction following stimulation. Importantly, by characterizing axon degeneration in Sarm1 Flag primary neurons, we could demonstrate that epitope-tagged SARM1 retains its ability to function as an executioner of axon degeneration. These mice are therefore a useful resource for further studies relating to SARM1 in the brain, in macrophages, and in other tissues and will allow biochemical studies to proceed involving endogenous SARM1.
As well as the previously reported roles of SARM1 in neuronal degeneration and transcriptional regulation of chemokines, a third reported role for SARM1 by us was in regulating cell survival and cytokine release after inflammasome activation (25). Studies are currently ongoing in order to fully understand the role of SARM1 in inflammasome regulation, both using these mice and human cell models.
Overall, this work has revealed that SARM1 is expressed in mouse macrophages but does not have a broad role in transcriptional regulation in these cells and has provided important new animal models to further explore SARM1 function.

Mice
Mice with a targeted gene deletion of exons 3 to 6 of Sarm1, Sarm1 tm1Aidi herein referred to as B6 congenic Sarm1 −/− , were and Tnf (B) mRNA were assayed by qRT-PCR, normalized to the housekeeping gene βactin, and are presented relative to the untreated WT control. Supernatants were assayed for CCL5 (C) and TNF (D) protein by ELISA. Data are mean ± SEM from 3 to 4 mice per genotype, performed in triplicate. There were no significant differences determined by multiple Mann-Whitney tests with Holm-Sídák multiple comparisons test. E, graph of relative neurite length over time for different mice and treatments. Neurite lengths are normalized to its mean of own 1 h length as 100%. All data are mean ± SEM (n = 2-4 as shown in Fig. S7B). Overall three WT mice and four Sarm1 Flag mice were used. Data were tested with a two-way ANOVA showing significant main effects of group F (3, 920) = 82.79, p < 0.0001; time F (2, 920) = 54.11, p < 0.0001 and interaction F (6, 920) = 22.26, p < 0.0001; Tukey's multiple comparisons test, ****p < 0.0001 vincristine versus vehicle (DMSO) in WT and Sarm1 Flag at 24 h and 72 h. See also Fig. S6. SARM1 CRISPR-Cas9 mice clarify SARM1 function and expression generated on a C57BL/6J background from 129 ES cells and were previously described (10). CRISPR knockout mice and epitope-tagged mice were generated as described below. Mice were housed in individually ventilated cages (IVC, Tecniplast) in a specific pathogen-free facility on a 12 h light/dark cycle with access to food and water ad libitum at the Trinity Biomedical Sciences Institute (TBSI), Trinity College Dublin. B6 congenic Sarm1 −/− were compared with wild-type (WT) C57BL/6J mice bred and housed at TBSI. CRISPR-generated strains were compared with wild-type littermates from heterozygous breeding. Both male and female mice were used for experiments. Procedures were conducted under licenses from the Health Products Regulatory Authority (Ref: AE19136/ P083) and with the approval of TCD Animal Research Ethics Committee.

Generation of CRISPR knockout and epitope-tagged mice Microinjection
Female C57BL/6J mice were superovulated with 5IU PMSG and 5IU hCG i.p., mated to C57BL/6J males, and the recovered zygotes microinjected with preassembled Cas9 ribonucleoparticles (RNP) according to standard protocols (47) before being surgically transferred into the oviducts of CD1 pseudo pregnant females. Sarm1 knockouts were made by targeting exon 1, 96 nucleotides downstream of the ATG start codon, using gRNA against the sequence GCCACCACTCC-GATCCGGTCCGG (Fig. 3A). Of the 137 zygotes microinjected, 11 offspring were recovered, screened by PCR, and selected mutants characterized by Sanger sequencing (MWG eurofins). Knockout mutants used in this study had a 2 bp (Sarm1 em1.1Tftc ), 34 bp (Sarm1 em1.2Tftc ), and 5 bp (Sarm1 em1.3Tftc ) deletions, respectively. Epitope tagging of SARM1 involved inserting a triple FLAG, double Streptag II sequence at the C-terminus of the gene. The gRNA recognition sequence in exon 9, CACAGTAGGGCACGG-GAACTGG, was targeted and a repair construct used to remove the stop codon and introduce the tag-sequences (Fig. 6A). Two base changes (cc > tt) were introduced at the original gRNA target site to remove the PAM sequence and prohibit recutting by Cas9. Pronuclear microinjection yielded 13 offspring, with one mutant male (Sarm1 em2(FLAG-Strep)Tftc ) identified by sequencing. Sarm1 em2(FLAG-Strep)Tftc mice are herein referred to as Sarm1 Flag mice.

Cell culture of primary neurons
Primary mouse neurons were dissected from P1 to P3 neonate brains and cultured as described (49). The cortex was dissected, trypsinized, and resuspended into the Neurobasal A medium (NBAM, GIBCO) containing B27 supplement (GIBCO), 10 μM Cytosine β-D-arabinofuranoside (Ara-C, Sigma), and 0.5 mM GlutaMax supplement (GIBCO). Cells were plated in wells coated overnight with 0.1 mg/ml Poly-D-lysine (Sigma). At DIV2 (Days in vitro), the medium was replaced with NBAM containing B27 and one-half of the media was replaced at DIV5. Cells were used for the assay at DIV7.

ELISA
Quantification of secreted CCL5, TNF, and IL-6 from cell supernatants was measured by ELISA (R&D, DuoSet ELISA kits) following the manufacturer's instructions.

RNA analysis by quantitative RT-PCR
For mice tissues, the brain, brainstem, liver, spleen, and kidney were dissected and kept in RNAlater (Invitrogen) until isolation. Total RNA was extracted from the tissues in TRIzol reagent (Ambion) according to the manufacturer's instructions. RNA was quantified using a Nanodrop 2000 (Thermo Fisher). Extracted RNA was treated with DNase I (Promega), and cDNA was generated as described below. For cells in culture, cells were seeded at 5 × 10 5 cells/ml in 24-well plates, allowed to adhere overnight, and stimulated as indicated the next day. Total RNA was extracted using the High Pure RNA Isolation Kit (Roche) and reversed transcribed with random hexamers (IDT) using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega) according to the manufacturers' instructions. The resulting cDNA was analyzed by quantitative RT-PCR using the PowerUp SYBR Green Master Mix (Applied Biosystems) and gene-specific primer pairs (sequences are listed in Table S1). Relative mRNA expression was calculated using the comparative C T method, normalizing the gene of interest to the housekeeping gene β-actin, and comparing it to an untreated WT sample.

RNA sequencing
Primary BMDMs were generated from femurs and tibiae of four 8-week-old female WT and four Sarm1 −/− mice. Cells were stimulated for 3 h with 5 μg/ml CL075 or left unstimulated. Total RNA was isolated prepared from samples as described above and stored at −80 C. Bioanalysis was performed using Agilent RNA 6000 Nano Kit and Agilent 2100 bioanalyzer according to manufacturer's instructions. Library preparation and sequencing were performed by Macrogen Inc. RNA libraries were prepared using TruSeq Stranded Total RNA with Ribo-zero (Illumina). Sequencing was carried out on the Illumina NovaSeq 6000 platform, in a 100 bp paired-ends read format, with a sequencing coverage of 40 million reads.
Sequencing data was received from Macrogen as Fastq files. Two different forms of alignment were carried out. Hisat2 (50) was used to map reads directly to the GRCm38.p6 assembly of the C57BL/6 mouse genome, whereas kallisto (51) was used to map reads against the repeat-masked transcriptome, which were subsequently transposed onto the genome.

Differential gene expression analysis
Following mapping using kallisto, transcripts that were too short to have been captured at the library preparation step (<300 bp in length) were filtered out of the analysis. Following this, transcripts that were very lowly expressed, having a transcript per million (TPM) count of less than one, were filtered out to remove background noise. DESeq2 (52) (version 1.22.2) was used for differential gene expression analysis of RNA sequencing data. As Sarm1 −/− replicates exhibited a lot of variability, these samples were split into two pairs with similar expression profiles (SKO1/SKO2 and SKO3/SKO5), then compared with WT. Unstimulated Sarm1 −/− and WT pBMDM were compared, and genes were deemed differentially expressed if the DESeq2 base mean was greater than 100, the fold change exceeded 1.6 fold, and the adjusted p-value was less than 0.05. Similarly, CL075-stimulated Sarm1 −/− and WT pBMDM were compared, and genes were deemed differentially expressed if the DESeq2 base mean was greater than 100, and the fold change exceeded 1.6 fold. The p-value cutoff for two replicates (SKO3 and SKO5) was relaxed to p ≤ 0.1 to include Ccl5, which was the positive control. A heatmap was then generated using the ComplexHeatmap (53) package in R, showing genes that were differentially expressed either before or after stimulation in both SKO1/2 and SKO3/5 relative to WT, grouped according to chromosomal location.

Analysis of density of SNPs and indels relative to the reference genome
The samtools mpileup (54) tool was used to identify genetic variants between a representative unstimulated Sarm1 −/− sample sequence (mapped using Hisat2) and the mm10 mouse reference genome. Genetics variants (both SNPs and indels) were filtered to only include: (1) Those with high-quality score (QUAL > 100) and (2) Those with high sequencing depth (DP > 100). The mouse reference genome was divided into sliding windows of 10 kb using bedtools (55) makewindows. Genetic variants were counted within each of these windows using bedtools intersect. The count data was visualized in R with the IdeoViz package.
Visualizing gene coverage, SNPs, and indels using integrative genomics viewer (IGV) The integrative genomics viewer (56) (IGV) from the Broad Institute was used to visualize reads, which were mapped to the genome using Hisat2. BAM files of these reads were loaded to IGV so that sequence variation from the reference sequence (denoted by a colored bar) and gene coverage could be examined at a given locus.

Cell fractionation
Cells were seeded at 5 × 10 5 cells/ml in 10 cm culture dishes and allowed to adhere overnight. The next day, cells were stimulated as indicated. Following stimulation, supernatants were removed and discarded, and cells were washed twice with ice-cold phosphate buffered saline (PBS) on ice. Cells were scraped in 5 ml ice-cold PBS, then pelleted by centrifugation at 1000g for 5 min at 4 C. Cells were resuspended in 200 μl cytosolic lysis buffer (10 mM Tris-HCl pH 7.4, 10 mM NaCl, 3 mM MgCl 2 , 10 mM EDTA, 1% Triton X-100) supplemented with 1 mM sodium orthovanadate, 10 μl/ml aprotinin (Sigma), and 1 mM PMSF (protease inhibitors). The cell suspension was incubated for 30 min on ice to lyse the plasma membrane, then centrifuged for 30 min at 2000 rpm at 4 C. The supernatant, containing an impure cytoplasmic fraction, was removed and stored at −80 C. The pellet, containing a pure nuclear fraction, was washed twice with 1 ml of unsupplemented cytoplasmic lysis buffer, then resuspended in 200 μl nuclear lysis buffer (30 mM HEPES pH 7.9, 36% glycerol, 600 mM NaCl, 2 mM MgCl 2 , 30 mM EDTA). During a 30 min incubation on ice, samples were vigorously vortexed three times. The samples were then centrifuged at 13,200 rpm. Supernatants containing nuclear proteins were removed and stored at −80 C.
Mice tissues, brain, and brainstems were dissected and homogenized in ice-cold lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, and cOmplete mini Protease Inhibitor Cocktail (Roche)). Following incubation on ice for 30 min, samples were centrifuged for 20 min at 15,000 rpm at 4 C. The supernatants (lysates) were transferred to a fresh tube, mixed with Laemmli sample buffer, and boiled for immunoblotting.
SDS-PAGE was carried out, and proteins were transferred to a nitrocellulose membrane using the semidry transfer method. Membranes were blocked using a 5% (w/v) solution of skimmed milk powder in PBS-Tween (0.1% Tween, v/v), then incubated with antibodies as indicated. Membranes were read on the Li-Cor Odyssey, and images were adjusted using Image Studio Lite ver. 4.

Immunoprecipitation
WT and Sarm1 Flag iBMDMs were seeded at 5 × 10 5 cells/ml in 15 cm dishes and allowed to adhere overnight. The following day, supernatants were removed and cells were washed three times with ice-cold PBS. Cells were then lysed using IP lysis buffer (50 mM Tris-HCl, pH 7.4, with 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100), supplemented with 1 mM sodium orthovanadate, 10 μl/ml aprotinin, and 1 mM PMSF. Immunoprecipitation was carried out using anti-FLAG M2 Magnetic Beads (Sigma, M8823) according to the manufacturer's instructions. Immunoprecipitated proteins were eluted from the beads by incubation with glycine-HCl (0.1 M, pH 3), then neutralized with 1 M Tris-HCl. IP lysates and input controls were subject to SDS-PAGE as described above.
Axon degeneration and measuring length of the neurite Neurons (DIV7) were treated with 1 nM vincristine sulfate salt (Sigma) or vehicle (DMSO), and then were set into Incucyte Live-Cell Analysis Systems (Sartorius). The first scan was made at 1 h after the treatment; the following scans were made every 6 h after the treatment. The neurite outgrowth was quantified using the semiautomatic tracing tool NeuronJ plugin (57) from the ImageJ package Fiji (58).

Data availability
All data generated or analyzed during this study are included within this article (and its supplementary information files). RNA sequencing data is available as GEO data set GSE182091 at https://www.ncbi.nlm.nih.gov/geo/query/acc. cgi?acc=GSE182091.
Supporting information-This article contains supporting information.