The metalloprotease ADM-4/ADAM17 promotes axonal repair.

Axonal fusion is an efficient means of repair following axonal transection, whereby the regenerating axon fuses with its own separated axonal fragment to restore neuronal function. Despite being described over 50 years ago, its molecular mechanisms remain poorly understood. Here, we demonstrate that the Caenorhabditis elegans metalloprotease ADM-4, an ortholog of human ADAM17, is essential for axonal fusion. We reveal that animals lacking ADM-4 cannot repair their axons by fusion, and that ADM-4 has a cell-autonomous function within injured neurons, localizing at the tip of regrowing axon and fusion sites. We demonstrate that ADM-4 overexpression enhances fusion to levels higher than wild type, and that the metalloprotease and phosphatidylserine-binding domains are essential for its function. Last, we show that ADM-4 interacts with and stabilizes the fusogen EFF-1 to allow membranes to merge. Our results uncover a key role for ADM-4 in axonal fusion, exposing a molecular target for axonal repair.

INTRODUCTION

The axon, the longest neuronal process, is essential for the transmission of electrochemical signals to other neurons or muscles. Damage to axons in the central or peripheral nervous system can result in lifelong disabilities. Functional recovery is achieved when the axon of the injured neuron regenerates and, either directly or indirectly, reinnervates its original target. A highly efficient method of functional axonal repair, known as axonal fusion, has been observed in many invertebrate species, including the nematode Caenorhabditis elegans (–). Axonal fusion occurs when the proximal axonal fragment, which is still attached to the cell body, regrows, reconnects, and fuses with its own severed distal axonal fragment, reestablishing the original axonal tract () and restoring neuronal function (, ).

Using the two C. elegans mechanosensory PLM (posterior lateral microtubule) neurons as a model system, we have previously shown that molecules of the apoptotic clearance machinery regulate axonal fusion (). During apoptotic cell corpse engulfment, apoptotic cells expose the lipid phosphatidylserine (PS) on the outer leaflet of their plasma membrane, which functions as an “eat me” signal for the engulfing phagocyte (). Analogously, we have shown that, following axonal injury, PS exposure on the outer membrane of the distal fragment functions as a “save me” signal for recognition by the regrowing proximal fragment (). This specific recognition of the distal fragment by the proximal fragment is mediated by PS-binding molecules, such as the PS receptor PSR-1 and the secreted transthyretin TTR-52 (). Last, physical merging between the two membranes of the separated axonal fragments is mediated by the nematode fusogen epithelial fusion failure-1 (EFF-1), which we, and others, have shown to relocalize to the axonal membrane following axotomy, restoring both membrane and cytoplasmic continuity (, ). Overexpression of EFF-1 within the injured neuron can compensate for the lack of PSR-1 and TTR-52, revealing a key and sufficient role for EFF-1 in the axonal fusion process ().

EFF-1 was initially identified in C. elegans through genetic screens for cell-cell fusion failure phenotypes and was found to be required for correct epidermis, vulva, and pharynx formation during development (, ). It is a type I single-pass transmembrane glycoprotein with structural and functional similarity to class II viral fusion proteins (). Expression of EFF-1 is able to induce cell fusion in many organisms and cell types, including C. elegans embryos (, ), cultured insect cells (), and mammalian cells (, ). Fusion mediated by EFF-1 has been shown to require the presence of EFF-1 on both opposing cell membranes (, –). A trans-trimerization model for EFF-1–driven membrane fusion has been proposed, whereby formation of a trans-trimer made of two monomeric EFF-1 molecules on one membrane and a single EFF-1 molecule on the other membrane leads to conformational changes in EFF-1 (). As a result, EFF-1 transmembrane segments anchored in the two opposing membranes are brought into contact, pulling the two membranes into one (). Previous studies have shown that monomeric EFF-1 is metastable () and that EFF-1 internalization from the cell membrane can prevent axonal fusion (), whereas trimeric EFF-1 is stable and irreversible (). However, the molecules that regulate the formation of functional EFF-1 trimers from EFF-1 monomers across opposing membranes to allow axonal fusion after injury remain unknown.

Once postulated to be fusogens themselves, the ADAM (a disintegrin and metalloprotease) protein family are type I single-pass transmembrane metalloproteases that contain three signature extracellular domains: the metalloprotease, disintegrin, and cysteine-rich domains. ADAM1, ADAM2, and ADAM3 were initially thought to be fusogens required for membrane fusion between egg and sperm in mice (). However, it was later found that these ADAM proteins might be indirectly involved in the fertilization process and that cell-cell fusion is not dependent on their fusogenic potential (). ADAM metalloproteases have also been shown to cleave regulators of axonal regeneration in both mammalian systems and C. elegans (, ).

Four ADAM proteins are present in the C. elegans genome, UNC-71, ADM-2, ADM-4, and SUP-17. Here, we show that ADM-4, an ortholog of human ADAM17/TACE (tumor necrosis factor–α–converting enzyme), is essential for EFF-1–mediated axonal fusion. We reveal that ADM-4 mediates axonal fusion cell-autonomously within the PLM neurons and that its overexpression promotes axonal fusion above wild type (WT) levels. We further demonstrate that ADM-4 metalloprotease activity and its binding to PS are required for its function within the axonal fusion context. Last, we show that ADM-4 binds EFF-1 and stabilizes this fusogen after injury. Thus, we propose a model in which ADM-4 promotes axonal fusion in the regrowing axon by stabilizing monomeric EFF-1, and in turn driving the fusogenic EFF-1 trans-trimer.

RESULTS

ADM-4 is required for axonal fusion

C. elegans mechanosensory PLM neurons undergo axonal fusion following laser-induced transection (Fig. 1, A and B) (, , ). The key regulator of axonal fusion is the fusogen EFF-1 (, ). To discover novel regulators of this process, we performed a candidate genetic screen focusing on molecules known to regulate cell-cell fusion in different contexts, such as sperm-egg fusion or myoblast fusion. We selected five classes of conserved transmembrane molecules involved in cell-cell fusion events that occur during development: tetraspanins (tsp-1, tsp-3, tsp-4, tsp-5, tsp-7, tsp-8, tsp-12, and tsp-17) (–), annexins (nex-1, nex-2, nex-3, and nex-4) (, ), scavenger receptors (scav-2, scav-3, and scav-5) (, ), transthyretins (ttr-1, ttr-30, and ttr-33) (, ), and ADAM proteins (unc-71, adm-2, adm-4, sup-17, and mig-17) (–) (table S1). The cell and axonal morphology of PLM neurons was examined in animals carrying mutations in individual genes before assessing axonal regrowth, reconnection, and fusion. From this screen, we identified adm-4, an ortholog of human ADAM17, which encodes an active metalloprotease. adm-4(ok265) mutant animals containing a deletion spanning the disintegrin and cysteine domains () presented with a severe reduction of axonal fusion compared to WT animals (Fig. 1, C and D). Neuronal morphology, as well as the rate and length of axonal regrowth, were not impaired in these mutant animals (fig. S1, A and B), indicating a specific defect in the axonal fusion process. Two lines of evidence further confirmed that the observed defect was specific to mutations in adm-4. First, two alleles that contain either a frameshift mutation in the disintegrin domain, adm-4(gk951512) (), or a CRISPR-Cas9–engineered deletion of the adm-4 locus, adm-4(vd116), also exhibited a severe reduction in axonal fusion (Fig. 1D). Second, the axonal fusion defect of adm-4(ok265) mutant animals was rescued in transgenic animals carrying the WT genomic adm-4 driven by its endogenous promoter (Padm-4::adm-4; Fig. 1D). Together, these results conclusively demonstrate that ADM-4 is a key molecule necessary for successful axonal fusion.

Fig. 1.

ADM-4 is required for axonal fusion.

(A) Image and scheme of a WT animal with mechanosensory neurons labeled with cytosolic green fluorescent protein (GFP) zdIs5(Pmec-4::GFP). Anterior is left and ventral is down in this and all following images. Out-of-focus neurons are shown in gray in the schematic. Scale bar, 25 μm. Image representative of 24 animals. (B) Image and scheme of successful axonal fusion in a WT animal, 24 hours after axotomy. The black filled arrowhead points to the cut site; gray arrowhead points to the site of fusion. Scale bar, 25 μm. Image representative of 24 animals. (C) Image and scheme of defective axonal fusion in an adm-4(ok265) mutant animal, 24 hours after axotomy. Filled arrowhead points to the cut site. Scale bar, 25 μm. Image representative of 26 animals. (D) Quantification of axonal fusion in the adm-4(gk951512), adm-4(vd116), and adm-4(ok265) mutant animals compared to WT, and animals carrying the transgene vdEx1753(Padm-4::adm-4gDNA) (+) compared to nontransgenic siblings (−) in adm-4(ok265) background. Error bars indicate the standard error of proportion. n values within each bar. P values derived from unpaired t test with Welch’s correction and compared to WT unless marked on graph.

ADM-4 functions cell-autonomously within the PLM neurons to promote axonal fusion

To determine where ADM-4 functions, we visualized its expression pattern with a transgene driving green fluorescent protein (GFP) expression using the endogenous adm-4 promoter (Padm-4::GFP) while simultaneously visualizing the PLM neurons with a cytosolic red fluorescent protein (RFP) (Pmec-17::TagRFP). We found that ADM-4 was expressed in PLM neurons as well as in other neurons and tissues, including the epidermis, intestine, vulva, gonads, and pharynx (fig. S2). To determine whether ADM-4 functions cell-autonomously in the PLM neurons during axonal fusion, or within the surrounding epidermal tissue in which the axon is embedded, we performed cell-specific rescue experiments. First, we expressed WT adm-4 cell-specifically in the mechanosensory neurons ALM (anterior lateral microtubule), AVM (anterior ventral microtubule), PLM, and PVM (posterior ventral microtubule) (Pmec-4::adm-4) and observed a full rescue of the axonal fusion defect of adm-4(ok265) mutant animals (Fig. 2A). In contrast, expression of WT adm-4 selectively in the epidermis (Pdpy-7::adm-4) could not rescue the axonal fusion defect of adm-4(ok265) mutant animals (Fig. 2B). These results suggest a cell-autonomous function of ADM-4 in the PLM neurons, consistent with the function of the fusogen EFF-1, which also functions cell-autonomously within these neurons (). We observed that the rate of axonal fusion in transgenic adm-4(ok265) mutant animals carrying multicopy arrays of WT adm-4 (Pmec-4::adm-4) was higher than that in WT animals (Figs. 1D and 2A), suggesting that overexpression of ADM-4 could enhance axonal fusion. We found that overexpression of this Pmec-4::adm-4 transgene in WT animals significantly increased the level of axonal fusion compared to that in nontransgenic WT siblings (Fig. 2C).

Fig. 2.

ADM-4 functions cell-autonomously to regulate axonal fusion.

(A) Cell-autonomous rescue and enhanced axonal fusion in adm-4(ok265) mutant animals with expression of WT adm-4 genomic DNA (vdEx1801) or cDNA (tagged with wrmScarlet-I) (vdEx1957/2020) in three independent transgenic strains. (B) Expression of transgene adm-4 genomic DNA under an epidermal-specific promotor (vdEx1793/94/95) fails to rescue axonal fusion in adm-4(ok265) mutant animals in three independent transgenic strains. (C) Enhanced axonal fusion in WT animals with expression of adm-4 cDNA (vdEx1957) expressed selectively in the mechanosensory neurons. For (A) to (C), error bars indicate the standard error of proportion. n values within each bar. P values derived from unpaired t test with Welch’s correction. Transgenic animals are indicated with (+) and are compared to nontransgenic siblings indicated with (−). Dotted line indicates the percentage of successful axonal fusion in WT animals.

To determine where ADM-4 localizes within the PLM neuron, we generated a functional C-terminal wrmScarlet-tagged ADM-4, expressed it selectively in these neurons (Pmec-4::adm-4::wrmScarlet-I), and analyzed its localization. Before axotomy, ADM-4 localized in clusters within the PLM cell body and along the axon (fig. S3, A, C, and D). We observed a change in the localization of ADM-4::wrmScarlet following axotomy; in particular, the number of ADM-4::wrmScarlet clusters decreased, and the clusters that were present became larger and enriched at the growing tip of the proximal axonal fragment as well as at the site of reconnection and fusion (fig. S3, B to F). We qualitatively observed that the localization of ADM-4::wrmScarlet following successful axonal fusion returned to that observed pre-axotomy (fig. S3F). Together, these results reveal that ADM-4 functions cell-autonomously within the PLM neurons and localizes to the regrowing proximal fragment to promote axonal fusion, and that its overexpression is sufficient to enhance axonal repair in WT animals.

ADM-4 sheddase activity is required for its function

ADAM17/TACE regulates a large number of substrates, including cytokines, membrane receptors, and cell surface molecules, through its metalloprotease domain, which is responsible for its sheddase activity (). To test whether its enzymatic function is required to promote axonal fusion, we generated two ADM-4 transgenes containing mutations in the metalloprotease domain and individually expressed them cell-specifically in the PLM neurons of adm-4(ok265) mutant animals. First, we generated a mutant adm-4 complementary DNA (cDNA) carrying a deletion that truncated the signal sequence, prodomain, and metalloprotease domain (Δ Catalytic domain; Fig. 3A). The expression of this truncated variant of ADM-4 [Pmec-4::adm-4(ΔCat)] was unable to rescue the axonal fusion defect of adm-4(ok265) mutant animals (Fig. 3B). Second, we generated a catalytically inactive ADM-4 variant. The glutamic acid residue (E) in position 406 has been shown to be essential for ADAM17 to exert its sheddase activity (), and a mutation from E to alanine (A) results in a catalytically inactive ADAM17. The corresponding glutamic acid residue is conserved in ADM-4 at position 371 (Fig. 3A). The expression of the predicted catalytically inactive variant ADM-4(E371A) [Pmec-4::adm-4(E371A)] in adm-4(ok265) mutant animals was unable to rescue the axonal fusion defect (Fig. 3C). To confirm that these mutant variants were correctly expressed and localized, we visualized these versions by tagging them with wrmScarlet [Pmec-4::adm-4(ΔCat)::wrmScarlet-I and Pmec-4::adm-4(E371A)::wrmScarlet-I]; we observed clusters along the axon and at the distal end of the regrowing axon similar to those in WT ADM-4 (fig. S4, A to D), albeit with slight differences in the size but not the number of these clusters after axotomy (fig. S4, E and F). Together, these results reveal that ADM-4 requires the metalloprotease domain and its sheddase activity to promote axonal fusion.

Fig. 3.

ADM-4 catalytic activity is necessary to promote axonal fusion.

(A) Scheme of WT and mutant ADM-4, with their respective protein domains. Signal sequence (SS), prodomain, metalloprotease domain (Metalloprotease), disintegrin domain (Disintegrin), cysteine-rich domain (Cysteine), membrane-proximal domain (MPD), conserved ADAM17 dynamic interaction sequence (CANDIS), transmembrane region (TM), and cytoplasmic tail. (B) Quantification of axonal fusion in adm-4(ok265) mutant animals carrying a transgene (either vdEx1983 or vdEx1984 or vdEx1985), in which adm-4 cDNA lacks the signal sequence, prodomain, and metalloprotease domain in PLM neurons; Pmec-4::adm-4(ΔCat) fails to rescue the axonal fusion defect of the mutants. (C) Quantification of axonal fusion in adm-4(ok265) mutant animals carrying the transgene (either vdEx2099 or vdEx2102 or vdEx2104), in which adm-4 cDNA is mutated in the catalytic site (E371A) and expressed in the PLM neurons; Pmec-4::adm-4(E317A) fails to rescue the axonal fusion defect of the mutants. For (B) and (C), transgenic animals are indicated with (+) and are compared to nontransgenic siblings indicated with (−). Error bars indicate the standard error of proportion. n values within each bar. P values derived from unpaired t test with Welch’s correction. Dotted line indicates the percentage of successful axonal fusion in WT animals.

PS binding is necessary for ADM-4 to promote axonal fusion

We have previously shown that, after injury, exposed PS on the membrane of the PLM axon is essential for recognition between the separated axonal fragments, leading to successful axonal fusion (, ). Recent work has revealed that PS binds to the membrane-proximal domain (MPD) of ADAM17 and that this interaction is required for ADAM17 to exert its sheddase function (). Therefore, we asked whether the ability to bind PS is conserved in ADM-4, and if this is required for ADM-4 to regulate axonal fusion. To test this notion, we first expressed and purified recombinant glutathione S-transferase (GST)–tagged MPD of ADM-4 (GST-ADM-4-MPD-3xFLAG) from E. coli BL21(DE3) cells (fig. S5A), and then performed a lipid-binding assay whereby recombinant ADM-4 protein was incubated on a membrane spotted with different species of lipids. ADM-4-MPD exhibited clear binding to PS and, to a lesser extent, to phosphatidic acid, phosphatidylinositol-4-phosphate, phosphatidylinositol-4,5-diphosphate, phosphatidylinositol-3,4,5,-triphosphate, and cardiolipin, but not to other phospholipids (Fig. 4A and fig. S5B). We next performed sequence alignment of ADAM17 and ADM-4 to predict the corresponding PS-binding domain on ADM-4. A cluster of positively charged amino acids [arginine (R) and lysine (K)] located in the MPD of ADAM17 has been shown to interact with the negatively charged PS (). Alignment between ADAM17 and ADM-4 revealed a cluster of six positively charged residues in ADM-4 corresponding to the residues important for PS binding in ADAM17, and a cluster of five positively charged residues upstream, also in the MPD region (Fig. 4B). To test whether these positively charged residues located in the ADM-4 MPD region are necessary for its ability to bind PS, we mutated all R and K residues in the MPD domain of ADM-4 to neutral glycine (G) amino acids (Fig. 4B) and performed lipid-binding assay with this mutated version of ADM-4 [ADM-4(ALLxPS)]. We found that ADM-4(ALLxPS) had lost the capacity to bind PS, indicating that some, or all, of these 11 positively charged residues in the MPD region are crucial for PS binding (fig. S5B). To address whether binding between ADM-4 and PS is required for ADM-4 to regulate axonal fusion, we generated transgenic strains that expressed ADM-4(ALLxPS) cell-specifically in PLM neurons of adm-4(ok265) mutant animals. As predicted, ADM-4(ALLxPS) was unable to rescue the axonal fusion defect of the adm-4(ok265) mutants (Fig. 4C). Overall, these results reveal that, following injury, ADM-4 binds PS via positively charged residues within the MPD-CANDIS (conserved ADAM17 dynamic interaction sequence) domain, suggesting that this interaction is required for ADM-4 to regulate axonal fusion.

Fig. 4.

PS binding is required for ADM-4 to promote axonal fusion.

(A) ADM-4 binds phosphatidylserine (PS) in vitro. Scheme of lipids spotted on membrane (left). Affinity-purified GST-ADM-4(PS domain)-3xFLAG binds most strongly to PS on a lipid strip (indicated by arrowhead) (right). Three independent experiments were performed. (B) Prediction of the ADM-4 PS domain based on sequence alignment to human ADAM17 PS domain. Potential PS-binding residues in WT ADAM17 or ADM-4 are marked in blue, whereas marked in red are ADAM17-5x (K625G/K626G/K628G/K643G/R644G) or ADM-4 (ALLxPS) PS-binding mutants with R/K to G mutations throughout the MPD region and the CANDIS region (K556G/K561G/K562G/R574G/R578G/R594G/R598G/K601G/R605G/R607G/K608G). (C) Expression of adm-4 cDNA mutated in its PS-binding domain, vdEx1978/1979/1980 [Pmec-4::adm-4(ALLxPS)], selectively in the mechanosensory neurons of adm-4(ok265) mutant animals fails to rescue the axonal fusion defect. Transgenic animals are indicated with (+) and are compared to nontransgenic siblings indicated with (−). Error bars indicate the standard error of proportion. n values with each bar. P values derived from unpaired t test with Welch’s correction. Dotted line indicates the percentage of successful axonal fusion in WT animals.

ADM-4 binds EFF-1 to regulate axonal fusion

The nematode-specific fusogen EFF-1 is the main effector of axonal fusion that mediates the physical merging of two separated axonal membranes. Overexpression of EFF-1 can compensate for the reduced fusion observed in animals with mutations in membrane recognition molecules PSR-1 and TTR-52, indicating that EFF-1 acts downstream of these proteins (). To determine whether EFF-1 has a similar effect with ADM-4, we overexpressed EFF-1 in the PLM neurons of adm-4(ok265) mutant animals. Overexpression of EFF-1 could not overcome the fusion defect of the adm-4(ok265) mutants (Fig. 5A), indicating that ADM-4 is essential for the fusogenic activity of EFF-1. Consistent with the notion that ADM-4 might function by regulating EFF-1, the increased axonal fusion caused by the overexpression of ADM-4 (Fig. 2C) was abolished in eff-1(ok1021) mutant animals (Fig. 5B). To examine whether ADM-4 interacts with EFF-1, we performed coimmunoprecipitation experiments. We expressed both full-length FLAG-tagged ADM-4 (ADM-4-3xFLAG) and GFP-tagged EFF-1A (EFF-1A-eGFP) in human embryonic kidney (HEK) 293T cells. ADM-4-3xFLAG proteins were then immunoprecipitated by incubating cell lysates with agarose beads conjugated with anti-FLAG antibodies. We found that ADM-4-3xFLAG coprecipitated with EFF-1A-eGFP (Fig. 5C), thereby revealing that ADM-4 binds EFF-1.

Fig. 5.

ADM-4 associates with EFF-1 to promote axonal fusion.

(A) Overexpression of EFF-1 with a transgene (either vdEx661, vdEx663, or vdEx851) in adm-4(ok265) mutant animals does not rescue the axonal fusion defect in three independent transgenic strains. (B) Overexpression of ADM-4 with a transgene (either vdEx1801, vdEx1957, or vdEx2020) in eff-1(ok1021) mutant animals does not rescue the axonal fusion defect in three independent transgenic strains. For (A) and (B), transgenic animals are indicated with (+) and are compared to nontransgenic siblings indicated with (−). Dotted line indicates the percentage of successful axonal fusion in WT animals. Error bars indicate the standard error of proportion. n values within each bar. P values derived from unpaired t test with Welch’s correction; P value in the first bar of (B) is from a comparison with WT animals. (C) Interaction between ADM-4 and EFF-1 in vitro. HEK293T cells were transfected with plasmids encoding ADM-4-3xFLAG and/or EFF-1A-eGFP. Complex formation was detected by immunoprecipitation (IP) with an anti-FLAG antibody, followed by immunoblotting (IB) with an anti-GFP antibody. Representative blot of five independent experiments. Uncropped images are shown in fig. S12.

PS has previously been shown to function in the same process as EFF-1 to regulate axonal fusion (). To determine whether the binding between ADM-4 and EFF-1 requires PS, we examined the binding between the PS-binding ADM-4 mutant ADM-4(ALLxPS)-3xFLAG and EFF-1A-eGFP in HEK293T cells. We observed a strong reduction in the binding between ADM-4(ALLxPS) and EFF-1 compared to ADM-4(WT) and EFF-1 (fig. S6, A and B), suggesting that the binding of PS to ADM-4 is required for an efficient interaction between ADM-4 and EFF-1.

We then asked whether ADM-4 is necessary for other EFF-1–dependent fusion events. EFF-1 is required for developmental cell-cell fusion in C. elegans and is critical in the formation of the epidermis where it fuses 139 cells into a syncytium (). To test whether ADM-4 also regulates EFF-1–mediated epidermal cell fusion, we examined the formation of epidermal syncytia in adm-4(ok265) mutant animals. The border between epidermal cells can be visualized with AJM-1::GFP, a fluorescent protein that localizes to adherens junctions between unfused cells (). In WT animals at the L4 stage, epidermal cells and seam cells fused to form syncytia surrounded by a membrane labeled with AJM-1::GFP (fig. S7A). In eff-1(ok1021) mutant animals, the lack of epidermal cell-cell fusion was observed, with AJM-1::GFP decorating individual unfused cell borders (fig. S7B). adm-4(ok265) mutant animals had normal epidermal syncytia, suggesting that ADM-4 does not function in this specific EFF-1–mediated developmental fusion event (fig. S7C). Together, our results suggest that ADM-4 functions together with EFF-1 in a complex to regulate EFF-1–mediated axonal fusion, but not EFF-1–mediated developmental fusion within the epithelia.

ADM-4 stabilizes monomeric EFF-1 after injury

We observed that the levels of EFF-1A increased markedly with coexpression of ADM-4(WT) in coimmunoprecipitation experiments in HEK293T cells (Fig. 5C). However, this increase was strongly reduced with the coexpression of ADM-4(ALLxPS) (fig. S6, A and C). To investigate this further, we expressed increasing levels of ADM-4 in HEK293T cells together with either free eGFP or EFF-1A-eGFP. We consistently saw an increase in the level of EFF-1A-eGFP but not free eGFP, which became elevated at high levels of ADM-4 expression (Fig. 6, A and B). This indicates a threshold at which ADM-4 can increase the levels of EFF-1, potentially by stabilizing EFF-1 posttranslationally. To determine whether ADM-4 affects the stability and rate of turnover of EFF-1, we performed a pulse-chase assay with the protein synthesis inhibitor cycloheximide. We observed that the rate of EFF-1 degradation was slower when we cotransfected EFF-1 with ADM-4 compared to EFF-1 alone (Fig. 6, C and D). These experiments demonstrate that ADM-4 stabilizes EFF-1 posttranslationally and that the binding of PS to ADM-4 might be required for the stabilization of EFF-1.

Fig. 6.

ADM-4 stabilizes EFF-1.

(A) Cotransfection of ADM-4-3xFLAG and free eGFP or EFF-1A-eGFP in vitro. HEK293T cells were transfected with plasmids encoding eGFP or EFF-1A-eGFP with increasing amounts of ADM-4-3xFLAG. Total proteins were collected and analyzed by Western blotting against GFP, FLAG, and α-tubulin. Uncropped images are shown in fig. S12. (B) Levels of GFP signals are shown as normalized intensity values of GFP/tubulin. Error bars indicate the SEM in three independent experiments. Sidak’s multiple comparisons test was used after two-way analysis of variance (ANOVA) to generate P values. (C) HEK293T cells expressing EFF-1A-eGFP alone or EFF-1A-eGFP and ADM-4-3xFLAG were subjected to cycloheximide (CHX) (100 μg/ml) for the indicated periods. Total proteins were collected and analyzed by Western blotting against GFP, FLAG, and α-tubulin. Uncropped images are shown in fig. S12. (D) Quantification of EFF-1-eGFP levels was normalized to tubulin. Normalized EFF-1-eGFP/tubulin levels were then normalized against respective untreated samples (0-hour cycloheximide treatment). Error bars indicate the SEM from four independent experiments. A mixed-effects model was used to compare HEK293T cells transfected with EFF-1A-eGFP or EFF-1A-eGFP and ADM-4-3xFLAG across different cycloheximide treatment time points to generate P value indicated on the graph. Dotted lines are the nonlinear fit of the dataset using a two-phase decay model.

Next, we investigated whether this stabilization of EFF-1 by ADM-4 occurs in C. elegans neurons during axonal fusion. We performed laser axotomy and analyzed EFF-1(WT)::GFP intensity within the PLM axon of eff-1(ok1021) as well as eff-1(ok1021); adm-4(vd116) mutant animals but found no difference in the intensity levels of EFF-1(WT)::GFP between genotypes before injury (fig. S8, A to C) or 8 hours after axotomy in either the proximal or distal axonal fragments (fig. S9, A to D). This indicates a possibly more subtle mechanism of regulation that cannot be detected due to the presence of all EFF-1 conformations, both monomeric and trimeric EFF-1. To further investigate how ADM-4 regulates EFF-1, we tested two hypotheses. First, we asked if the effect of ADM-4 on EFF-1 occurs only at the cell membrane and not throughout the whole axon. To test this hypothesis, we cultured C. elegans neurons from animals with a mechanosensory neuron marker [uIs115(Pmec-17::TagRFP)] and endogenous EFF-1 tagged with GFP using CRISPR-Cas9 [eff-1(cas618)]. We then visualized EFF-1::GFP in RFP-positive cells using total internal reflection fluorescence microscopy (TIRFM), which enables visualization of EFF-1::GFP at the cell membrane. Before axotomy, we observed that the endogenous EFF-1::GFP levels between WT and adm-4(vd116) neurons were unaltered (Fig. 7, A, B, and E). However, we detected an increase in EFF-1::GFP levels in WT neurons 1 hour after axotomy, as previously observed (Fig. 7, C and E) (). The increase in EFF-1::GFP level was not observed in adm-4(vd116) neurons (Fig. 7, D and E), indicating that ADM-4 is required for the increase of EFF-1 after injury and that this stabilization occurs specifically at the cell membrane. Second, we asked if we could detect the stabilization of monomeric EFF-1 in vivo. EFF-1 has been proposed to mediate fusion by a trans-trimerization model, whereby a trans-trimer, made of two EFF-1 molecules on one membrane and one on the other, leads to a conformational change and merging of the two membranes. In this model, the metastable monomeric EFF-1 is an essential intermediate step for trans-trimers to form and is therefore essential for fusion to occur. It is possible that ADM-4 stabilizes a small population of monomeric EFF-1 in PLM neurons, which we cannot detect in vivo in the presence of the trimeric form. To address this notion, we expressed an EFF-1 mutant that is unable to form trimers [Pmec-4::EFF-1A(G260A/D321E/D322E)::GFPnovo2] () and compared the GFP intensity in the PLM axon between eff-1(ok1021) and eff-1(ok1021); adm-4(vd116) mutant animals. We observed no difference before injury (fig. S10, A to C); however, 8 hours after axotomy, we observed a reduction in EFF-1A(G260A/D321E/D322E)::GFP in the regrowing and distal PLM axon of eff-1(ok1021); adm-4(vd116) mutants compared to eff-1(ok1021) animals (Fig. 7, F to I). Consistent with these data, we saw an increase in EFF-1A(G260A/D321E/D322E)-eGFP when we coexpressed EFF-1A(G260A/D321E/D322E)-eGFP and ADM-4 in HEK293T cells, suggesting that ADM-4 can stabilize monomeric EFF-1 (fig. S11). These results are consistent with a model in which ADM-4 functions to stabilize EFF-1 monomers on the membrane during axonal repair, thereby promoting trans-trimer formation.

Fig. 7.

ADM-4 stabilizes monomeric EFF-1.

(A) Representative image of a cultured mechanosensory neuron from WT animals and (B) adm-4(vd116) animals with endogenously labeled EFF-1::GFP. (C) Representative image of an axotomized cultured mechanosensory neuron from WT animals and (D) adm-4(vd116) animals with endogenously labeled EFF-1::GFP. Arrowheads indicate the cut site. For (A) to (D), images for Pmec-17::TagRFP were taken using epifluorescence microscopy and average intensity projection images for EFF-1::GFP were acquired using TIRFM. Scale bars, 10 μm. (E) Quantification of endogenous EFF-1::GFP in cultured mechanosensory neurons in adm-4(vd116) animals as compared to WT animals, pre- and post-axotomy. Error bars indicate the SEM. Each dot represents an individual animal, with the total n value also indicated. Tukey’s multiple comparisons test was used after two-way ANOVA to generate P values. a.u., arbitrary units. (F) Representative image of EFF-1A (G260A/D321E/D322E)::GFP in eff-1(ok1021) and (G) eff-1(ok1021); adm-4(vd116) animals. Images are maximum intensity projections of deconvolved spinning-disk confocal images of PLM axons carrying the transgene vdEx2455(Pmec-4::eff-1A(G260A/D321E/D322E)::GFPnovo2; Pmec-4::mCherry). Arrowhead points to the cut site. Scale bars, 10 μm. (H) Quantification of EFF-1A(G260A/D321E/D322E)::GFP in the proximal axon and (I) distal axon of eff-1(ok1021) and eff-1(ok1021); adm-4(vd116) animals. Mean EFF-1::GFP signals were normalized to mean mCherry signals for individual animals. Error bars indicate the SEM. Each dot represents an individual animal, with the total n value also indicated. P values from unpaired t test with Welch’s correction.

DISCUSSION

Following injury to the axon, intrinsic and extrinsic factors need to act in concert to achieve successful axonal fusion and functional repair. Here, we found a metalloprotease, ADM-4, that not only consolidates our current knowledge about axonal fusion but also furthers our understanding of how axonal fusion is regulated at the molecular level. Our data demonstrate that ADM-4 functions cell-autonomously in PLM neurons to mediate axonal fusion. This function is dependent on its sheddase activity, and we propose that ADM-4 is activated by PS binding and exerts its effect by acting upon the fusogen EFF-1 to regulate membrane fusion in response to injury. Overexpression of ADM-4 promotes axonal fusion, making it a potential therapeutic target to exploit this form of repair.

ADAM proteins are known for processing their substrates through their proteolytic activity. For example, ADAM17, which is the human ortholog of ADM-4, cleaves the axonal guidance receptor Neogenin, which is responsible for the inhibition of axonal regeneration in optic nerve injury (). In C. elegans, overexpression of ADAM10/sup-17 and ADAM17/adm-4 in GABAergic (γ-aminobutyric acid–releasing) motor neurons reduces axonal regeneration (). However, there are some ADAMs that lack the necessary Zn2+-binding sequences in their catalytic domain to be proteolytically active. These proteolytically inactive ADAMs participate in intercellular communication through their adhesive properties rather than through proteolytic cleavage. Our results are consistent with a role for ADM-4 as an active metalloprotease given that its proteolytic activity is required for axonal fusion to proceed.

One of the early events that occur upon axonal injury is the flipping of PS to the external leaflet of the plasma membrane (). Exposed PS on the outer membrane has been shown to interact with positively charged amino acids located in the MPD of ADAM17, bringing the protease into position to process its substrate (). Our results suggest that surface PS is required to activate ADM-4 in a similar way to ADAM17. We present evidence suggesting that positively charged amino acids located in the MPD of ADM-4 are required for the binding of PS in vitro and for axonal fusion in vivo. These data support a model in which surface PS, triggered by axotomy, binds the MPD of ADM-4 to activate this protein for further substrate processing.

The proteolytically active ADAM17 is known to process more than 80 different substrates (), with no consensus motifs among them. To identify the substrate on which ADM-4 is acting to regulate axonal fusion, we performed coimmunoprecipitation of ADM-4 and EFF-1, the key effector of axonal fusion. We found that ADM-4 forms a complex with EFF-1 to regulate axonal fusion. Known posttranslational regulators of EFF-1 include molecules involved in endocytosis, guanosine triphosphatases (GTPases) RAB-5 and DYN-1 (), actin regulator VAB-10 (), and Notch (). These molecules have been shown to control the presence of EFF-1 to fusion sites in larval hypodermal cells. In the context of axonal repair, we have previously shown that RAB-5 regulates fusion by controlling the levels of EFF-1 in C. elegans (). There was a notable increase in the amount of EFF-1 with cotransfection of ADM-4 in HEK293T cells. Furthermore, cotransfection of ADM-4 with EFF-1 stabilizes the levels of EFF-1. We speculate that ADM-4 may stabilize monomeric EFF-1, thereby facilitating EFF-1 trans-trimer formation across opposing membranes, to promote axonal fusion.

Our results, together with other proposed models of cell-cell fusion, lead us to propose a model for how ADM-4 regulates EFF-1 to promote axonal fusion (Fig. 8). Data from this study suggest that exposed PS, which flips to the outer membrane after injury, activates ADM-4; this, in turn, either directly or indirectly stabilizes the metastable monomeric EFF-1, allowing the formation of functional EFF-1 trans-trimers that drive axonal fusion (Fig. 8A). Previous studies have shown that recombinant EFF-1 can form trimers in cis and trans, but fusion is only driven by EFF-1 trimers that are formed in trans (). As the metalloprotease activity of ADM-4 is required for axonal fusion, we speculate that ADM-4 may also, either directly or indirectly, cleave aberrant nonfusogenic EFF-1 cis-trimers (Fig. 8B). The PLM neuron is surrounded by a specialized extracellular matrix (ECM) (), and ADAM proteins are known to regulate the ECM (–). Therefore, it is also possible that ADM-4 functions to remodel the ECM at the site of fusion to allow a favorable environment for EFF-1 to form trans-trimers and promote fusion.

Fig. 8.

Proposed model depicting the function of ADM-4 on EFF-1 to promote axonal fusion.

In each panel, the membranes of the separated axonal fragments are represented at either side, with the intercellular space in between. Leaflets of the plasma membrane are differentiated in dark and light colors. EFF-1 is drawn with purple domain II, light purple domain I, blue domain II, and gray transmembrane domain. ADM-4 is drawn with pink metalloprotease, yellow disintegrin domain, green cysteine-rich domain, cyan PS-binding domain, and dark gray transmembrane domain. (A) Stabilization of EFF-1 monomers by ADM-4 directly (cell 2) or indirectly (cell 1) to allow the formation of EFF-1 trans-trimers that allows axonal fusion to proceed (). (B) ADM-4 may also directly or indirectly minimize the levels of nonfunctional cis-trimeric EFF-1 through its proteolytic activity, allowing the formation of EFF-1 trimers in trans that promote axonal fusion.

Here, we report a previously unknown function of an ADAM protein as a key regulator of axonal fusion, and linked PS induction following injury with the final fusogenic effector EFF-1. Axonal fusion has not yet been observed as a spontaneous mechanism of axonal repair in vertebrate neurons. However, treatment with polyethylene glycol to promote axonal fusion has shown promising results, suggesting that axons in rodents and humans can undergo fusion (–). Hence, it is possible that axonal fusion occurs in a subset of neurons that is yet to be found or that the ability for axonal fusion to spontaneously occur is suppressed by a tight regulation of the endogenous fusogens in vertebrate nervous systems. However, it remains unknown whether the expression of endogenous fusogens in vertebrate neurons, or treatment with exogenous fusogens, is sufficient to confer fusion competency and promote axonal repair by fusion. Our findings identify a molecular element and mechanism that regulate axonal fusion through modulation of the fusogen EFF-1, which in turn could potentially be a target to promote axonal repair in mammalian systems.

MATERIALS AND METHODS

Strains

Standard techniques were used for C. elegans maintenance, crosses, and other genetic manipulations (). All experiments were performed on hermaphrodites grown at room temperature (22°C) unless otherwise stated, and all strains were grown on Escherichia coli OP50 bacteria. The WT Bristol N2 strain and the following mutations were used: zdIs5(Pmec-4::GFP), adm-4(ok265) X; adm-4(gk951512) X; adm-4(vd116); eff-1(ok1021) I; eff-1(cas618(eff-1::GFP)) I was a gift from G. Ou; the morphology marker uIs115(Pmec-17::TagRFP) was a gift from M. Chalfie.

The transgenes used were the following: vdEx1753(Padm-4::adm-4(20ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(20ng/μL)); vdEx1801(Pmec-4::adm-4gDNA(5ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(35ng/μL)); vdEx1957(Pmec-4::adm-4cDNA::wrmScarlet-I(10ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(30ng/μL)); vdEx2020(Pmec-4::adm-4cDNA::wrmScarlet-I(20ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(20ng/μL)); vdEx1793/4/5(Pdpy-7::adm-4gDNA(1ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(39ng/μL)); vdEx2099/2102/04(Pmec-4::adm-4cDNA(E371A)(20ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(20ng/μL)); vdEx1983/84/85(Pmec-4::adm-4cDNA(ΔCat)(10ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(30ng/μL)); vdEx1978/79/80(Pmec-4::adm-4cDNA(ALLxPS)(10ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(30ng/μL)); vdEx661(Pmec-4::eff-1gDNA::GFP(5ng/μL); Podr-1::DsRed(60ng/μL); Pmec-4::mCherry(20ng/μL); pSM vector(20ng/μL)); vdEx663(Pmec-4::eff-1gDNA::GFP(5ng/μL); Podr-1::DsRed(60ng/μL); Pmec-4::mCherry(20ng/μL); pSM(20ng/μL)); vdEx851(Pmec-4::eff-1gDNA::GFP(5ng/μL); Podr-1::DsRed(60ng/μL)); jcIs1 (ajm-1::GFP + unc-29(+) + rol-6(su1006)), vdEx2455(Pmec-4::eff-1A(G260A/D321E/D322E)::GFPnovo2(5ng/μL); Pmec-4::mCherry(20ng/μL); Podr-1::GFP(60ng/μL); pSM vector(35ng/μL)), vdEx2479(Pmec-4::eff-1A::GFPnovo2(5ng/μL); Pmec-4::mCherry(20ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(35ng/μL)), sEx10699(rCes ZK154.7::GFP + pCeh361); vdEx2592(Pmec-4::adm-4(ΔCat)::wrmScarlet-I(20ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(20ng/μL)); vdEx2590(Pmec-4::adm-4(E371A)::wrmScarlet-I(20ng/μL); Podr-1::DsRed(60ng/μL); pSM vector(20ng/μL)).

Molecular biology and CRISPR-Cas9 gene editing

Molecular biology was performed using standard techniques (). Enzymes and reagents for polymerase chain reaction (PCR) genotyping or cloning were sourced from New England Biolabs. Genomic DNA templates were obtained from whole-animal lysates (). Plasmids were transformed into E. coli (strain DH5α) and purified using the QIAprep Spin Miniprep Kit (QIAGEN). DNA sequencing of each construct was performed by the Australian Genome Research Facility (Brisbane). Gene-edited adm-4 knockout strains (adm-4(vd116)) were generated using CRISPR-Cas9 gene editing with Alt-R CRISPR-Cas9 system [Integrated DNA Technologies (IDT)], and progeny containing adm-4 deletion were screened with the dpy-10 Co-CRISPR approach (, ). Cas9 target sites were chosen using the CHOP-CHOP design tool (https://chopchop.cbu.uib.no/) (). To generate lines, Cas9 (62 μM), IDT tracRNA (trans-activating CRISPR RNA) (10 μM), crRNAs (CRISPR RNAs) (crHXY5 and crHXY6, 10 μM each), crRNA targeting the dpy-10 locus, and single-stranded donor template dpy-10(cn64) () were injected into the gonads of young adult animals. Injected animals were transferred to new OP50 plates and allowed to generate progeny for 3 days at room temperature. The progeny were then individually transferred onto fresh OP50 plates and genotyped for the adm-4 deletion.

Oligonucleotides used in this study

Plasmids used in this study

The vector that expresses adm-4 under its endogenous promoter (pXH8) was generated as follows: a 5-kb region upstream of adm-4 was amplified with primers HXY160 and HXY161 engineered to contain Not I and Fse I, respectively. This PCR amplicon was then digested and inserted into a plasmid containing adm-4 amplified from N2 genomic DNA using HXY162 and HXY163, flanked with Asc I and Xma I restriction sites.

The vectors that express adm-4 in mechanosensory neurons under the control of the mec-4 promoter (pXH13 and pXH22) were generated as follows: a pSM::Pmec-4::unc-54 3′UTR plasmid was first obtained by amplifying a 1020–base pair (bp) promoter region upstream of mec-4 and cloned into a pSM vector with the restriction sites Not I and Fse I. adm-4 was amplified either from genomic DNA (pXH13) or from a cDNA clone (λZAPII phage suspension yk187d12) (pXH22) using HXY162 and HXY163 primers, flanked with Asc I and Xma I restriction sites, and cloned into pSM::Pmec-4::unc-54 3′UTR plasmid.

Vectors that express adm-4::wrmScarlet-I (pXH19) under the control of the mec-4 promoter were generated as follows: wrmScarlet-I was amplified from pSEM90 using HXY215 and HXY216 primers. Gibson cloning () was performed to fuse wrmScarlet-I to the 3′ end of adm-4 cDNA with a linearized pXH22.

To express adm-4 in the hypodermis under the control of the dpy-7 promoter (pXH14), a Pdpy-7::unc-54 3′UTR plasmid was first generated by amplifying a 1261-bp region upstream of dpy-7 using HXY126 and HXY127 primers and cloned into a pSM vector with the restriction sites Not I and Fse I. Genomic adm-4 DNA from pXH13 was then cloned into this plasmid using Asc I and Xma I restriction sites.

To test the functionality of the ADM-4 metalloprotease domain in axonal fusion, two ADM-4 mutants, pXH29 and pXH30, were generated. In the pXH29 plasmid, the catalytic-dead ADM-4 was expressed in the mechanosensory neurons using the mec-4 promoter; pXH22 was linearized with HXY242 and HXY243 primers. Gibson cloning was used to mutate the target nucleotide with HXY244 and HXY245 primers. In pXH30, mutant ADM-4 that lacks the prodomain and the metalloprotease domain was expressed in the mechanosensory neurons using the mec-4 promoter. A truncated form of ADM-4 was amplified from pXH22 using HXY222 and HXY163 primers and cloned into a pSM::Pmec-4::unc-54 3′UTR plasmid using Asc I and Xma I restriction sites. To visualize the two catalytic-dead ADM-4 mutants, wrmScarlet-I was amplified from pSEM90 using HXY466 and HXY467 primers that contain homologous sequences to pXH29 and pXH30. Gibson cloning was then used to tag wrmScarlet-I to the C-terminal end of mutated adm-4(E371A) and adm-4(ΔCat), using pXH29 and pXH30 as the backbone vector to generate pXH81 and pXH83, respectively.

The plasmid pXH31 was built to test whether PS binding is required for ADM-4 to regulate axonal fusion. ADM-4 with mutated PS-binding sites [ADM-4(ALLxPS)] was expressed in mechanosensory neurons using the mec-4 promoter; a fragment of ADM-4 containing mutated PS-binding sites was generated by IDT gene blocks. pXH22 was linearized with HXY223 and HXY224 primers, and the fragment was cloned in using Gibson cloning. Plasmid pXH51 was generated to express ADM-4(ALLxPS) in HEK293T cells, and the mutated PS-binding domain was excised from pXH31 with BstX II and Bgl II and switched with the corresponding region in pXH50, which contains ADM-4(WT) driven under a mammalian promoter.

The membrane lipid-binding assay was done by expressing pXH60 and pXH67. pXH60 was created by amplifying the PS domain with HXY494 and HXY461 from pXH22 and cloned into a pGEX4-T-2::3xFLAG plasmid using SaI I and Hind III restriction sites. pXH67 was created by amplifying the PS domain from pXH31 using HXY510 and HXY461 and cloned into pXH60 using SaI I and Hind III restriction sites.

pXH77 and pXH78 were created using Gibson cloning. HXY569 and HXY570 were used to amplify both EFF-1A(WT) and EFF-1(G260A/D321E/D322E) with overlapping sequences to Pmec-4 and GFPnovo2.

pXH50, pXH64, and pXH68 were used as expression vectors to perform immunoprecipitation in HEK293T cells. pXH50 was created by amplifying adm-4::3xFLAG with HXY395 and HXY396 from a pSM::Pmec-4::adm-4cDNA::3xFLAG plasmid and cloned into pMAX vector with Hind III and Not I. Gibson cloning was used to create pXH64. Gibson cloning was used to create pXH68 with HXY530.

Microscopy for C. elegans

Animals were immobilized in 0.05% tetramisole hydrochloride on 3.5% agar pads. Axotomies were performed as previously described (), using the MicroPoint Laser System Basic Unit attached to a Zeiss Axio Imager A1 microscope [Objective EC Plan-Neofluar 100×/1.30 numerical aperture (NA) Oil M27]. Axotomies were completed by delivering pulses of 0.2 ± 0.03 μW of 435-nm power. Axons were severed in L4-stage animals, approximately 50 μm anterior to the PLM cell body, visualized using a Zeiss Axio Imager Z1 microscope equipped with a Photometrics camera (Cool Snap HQ2). Images acquired in MetaMorph software were further processed in FIJI ().

Quantification of axonal reconnection and fusion

Animals were assayed for regeneration 24 hours after axotomy. Axonal reconnection was deemed to occur when the proximal and distal axons appeared to be touching within the same focal plane when observed at high magnification, as previously described (). Successful axonal fusion was scored based on the prevention of distal fragment degeneration after reconnection. The rate of fusion was expressed as a percentage of the total number of successfully reconnected axons; this eliminates the possibility of incorrectly scoring axon guidance or regrowth defects as failed axonal fusion ().

Quantification of axonal regrowth

Animals were assayed for regeneration 24 hours after axotomy. The rate of regrowth of the proximal fragment was expressed as a percentage of the total axons that were severed but had not reconnected. Length of regrowth was measured from the cut site to the end of the longest proximal regrowing axon using FIJI (). Axons that were scored as reconnected or fused were excluded from the quantification of regrowth.

Visualization of ADM-4 and EFF-1

To study the localization of ADM-4::wrmScarlet and EFF-1::GFP in the PLM neurons, acquisitions were performed on animals mounted on 3.5% agarose pads in M9 buffer, in 0.05% tetramisole. Individual animals were imaged using a spinning-disk confocal system (Marianas; 3i Inc.) consisting of Axio Observer Z1 (Carl Zeiss) equipped with a CSU-W1 spinning-disk head (Yokogawa Corporation of America), an ORCA-Flash4.0 v2 sCMOS camera (Hamamatsu Photonics), and a 100×/1.46 NA oil-immersion Plan Apochromat objective with sampling intervals x,y = 63 nm and z = 130 nm. Image acquisition was performed using SlideBook 6.0 (3i Inc.). Green fluorescence was visualized with a 488-nm laser, and red fluorescence was visualized with a 561-nm laser, using a Quad 405/488/561/640 dichroic mirror, and 525/30-nm and 617/40-nm single-band emission filters. All spinning-disk confocal microscopy images were deconvolved with Huygens Professional, v.18.04 (Scientific Volume Imaging) using the CMLE algorithm, with a signal-to-noise ratio (SNR) of 20 and 40 iterations. Mean GFP fluorescence was normalized with mean RFP fluorescence to reduce variability from the expression of extrachromosomal arrays between animals. Quantification of fluorescence intensities was done with FIJI (). To quantify the localization of ADM-4, images were thresholded using minimum values in FIJI, and clusters within 15 μm from the tip of the regrowing axon were counted and their size was measured using “analyze particles” in FIJI.

Microinjections

Microinjections were performed according to standard methods (), using an inverted Zeiss Axio Observer microscope equipped with differential interference contrast, a Narishige needle holder, and an Eppendorf FemtoJet pump.

Lipid-binding assay

GST-ADM-4(PS domain)-3xFLAG and GST-ADM-4(ALLxPS domain)-3xFLAG were affinity-purified from E. coli BL21(DE3) cells and used for the in vitro lipid-binding assay. Ten micrograms of purified protein was resolved on a precast 4 to 15% gradient gel (Bio-Rad) to determine its purity. Membrane strips spotted with various phospholipids (P-6002; Echelon Biosciences) were first incubated with blocking buffer {TBS-T [tris-HCl (pH 7.4), 150 mM NaCl, 0.1% Tween 20]} + 3% bovine serum albumin (BSA; fatty acid free, A7030, Sigma-Aldrich) for 1 hour at room temperature. Fifteen micrograms of purified PS domain of both WT and mutant GST-ADM-4(PS domain)3xFLAG proteins was then added to the blocking buffer and incubated for 1 hour at room temperature. The membranes were then washed with TBS-T three times for 5 min each, after which they were subjected to immunoblotting using an anti-GST antibody (ProteinTech, 66001-2-lg; 1:2000).

Coimmunoprecipitation assay

HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum at 37°C in a tissue culture incubator supplemented with 5% CO2. Transfection of HEK293T cells with pMAX::adm-4::3xFLAG (pXH50)/pMAX::adm-4(ALLxPS)::3xFLAG (pXH51) and/or pMAX::eff-1A::eGFP (pXH64) was performed with X-tremeGENE 9 DNA Transfection Reagent (Sigma-Aldrich). Transfected cells were then lysed 48 hours later with ice-cold radioimmunoprecipitation assay (RIPA) buffer [1% Triton X-100, 0.5% sodium deoxycholate, 100 mM NaCl, 0.1% SDS, 2 mM EDTA, 2 mM EGTA, 50 mM NaF, 10 mM sodium pyrophosphate in tris-buffered saline (TBS)], supplemented with cOmplete EDTA-free protease inhibitor mixture (Sigma-Aldrich). Cell lysates were centrifuged at 14,000 rpm for 30 min at 4°C and precleared with Protein G Sepharose beads (GE Healthcare) for 1 hour at 4°C. Precleared lysates were then incubated with anti-FLAG M2 agarose beads (Sigma-Aldrich) or anti-GFP agarose beads (ChromoTek GFP-Trap) for 3 to 4 hours at 4°C. The beads were subsequently washed four times for 15 min with ice-cold RIPA buffer and eluted in 2× SDS sample buffer containing β-mercaptoethanol and dithiothreitol. The immunoprecipitated proteins were then resolved by SDS–polyacrylamide gel electrophoresis (PAGE) and visualized by Western blot analysis.

Proteins were separated by SDS-PAGE on 7.5 or 10% or 4 to 15% precast (Bio-Rad) acrylamide gels and transferred to polyvinylidene difluoride (PVDF) membranes under a constant voltage of 100 V for 2 hours. After transfer was completed, membranes were blocked in 5% BSA (in TBS containing 0.1% Tween 20), washed, and incubated with the primary antibodies anti-eGFP {Invitrogen (A-6455), 1:2800 or Roche [from mouse immunoglobulin G1κ (clones 7.1 and 13.1) #11814460001], 1:1000} or anti-FLAG [Cell Signaling Technology (#14793), 1:500 or Sigma (F1804), 1:1000] overnight at 4°C. They were then washed and incubated with horseradish peroxidase–conjugated secondary antibodies (1:10,000) for 1 hour at room temperature. Following extensive washing with TBS-T, blots were analyzed using the enhanced chemiluminescence method. Images were acquired digitally on the Odyssey Fc imaging system (LI-COR) and analyzed using Image Studio Lite (LI-COR).

C. elegans primary cultures

C. elegans primary cultures were prepared as described in previous studies (, ). Gravid hermaphrodites were treated with a bleach solution containing 20% bleach and 0.5 N NaOH to release eggs. Eggs were washed with egg buffer [118 mM NaCl, 48 mM KCl, 2 mM CaCl2, 2 mM MgCl2, and 25 mM Hepes (pH 7.3), 340 mOsm] and pelleted through centrifugation at 900g for 3 min. Isolated eggs were then resuspended in 500 μl of chitinase (2 mg/ml) (Sigma-Aldrich, catalog no. C6137) to digest the eggshell and incubated on a rotator at room temperature for around 40 min. Embryonic cells were dissociated by resuspending chitinase-treated eggs in Leibovitz’s L-15 medium (Gibco, catalog no. 21083027) containing 10% fetal bovine serum (Life Technologies), penicillin (50 U/ml), streptomycin (50 μg/ml), and osmolarity of 345 mOsm and by pipetting up and down with a 1-ml pipettor until approximately 50% of the embryos were dissociated into single cells. Cell clumps, undigested eggs, and hatched larvae were then removed from dissociated embryonic cells by filtration through a 5-μm Durapore filter with a hydrophilic PVDF membrane (Millipore). Clean single dissociated cells were plated on 35-mm glass bottom dishes with 14-mm microwells and 1.5 cover glass (Mat-Tek P35G-1.5-14-CGRD) that were coated with peanut lectin (0.5 mg/ml) (Sigma-Aldrich, L0081) in L-15 growth medium. Dishes containing cells were placed in a plastic container lined with moist tissue papers at room temperature on a laboratory bench for 3 days.

Laser axotomy and TIRFM of C. elegans primary cultures

Fluorescence in C. elegans primary cells was visualized using a Plan Apochromat 63×/1.4 NA oil-immersion objective on a confocal/two-photon laser-scanning microscope (LSM 710; Carl Zeiss Pty Ltd) built around an Axio Observer Z1 body (Carl Zeiss). The system was equipped with two internal gallium arsenide phosphide (GaAsP) photomultiplier tubes (PMTs), three normal PMTs for epi-(descanned) detection, and two external GaAsP PMTs for non-descanned detection, and controlled by Zeiss Zen Black software. Axotomies were performed on 3-day-old RFP-positive cells that had axons that were more than 50 μm long and that were severed approximately 15 μm from the distal end of the axon with the Spectra-Physics Mai Tai eHP Ti:sapphire 800-nm laser at 40% laser power. Cells were then fixed approximately 1 hour later with 4% paraformaldehyde and 4% sucrose in phosphate-buffered saline solution for 15 min at room temperature in the dark. Endogenously tagged EFF-1::GFP was then imaged with a standard Zeiss ELYRA PS.1 SIM/PALM/STORM system fitted with a Zeiss Alpha Plan-Apochromat 100×/1.46 NA oil-immersion objective, a Zeiss FC12 Definite Focus module, and an iXon3 897 EMCCD 512 × 512 pixel camera (Andor, Oxford Instruments). Cultured neurons were imaged by TIRFM at an angle of 66.92°. Excitation with a 561-nm laser and a 570- to 620-nm single-band emission filter was first used to detect RFP-positive cells imaged in epifluorescence mode. Excitation with a 488-nm laser and a 495- to 550-nm band-pass filter was then used to perform TIRFM and to visualize GFP. A minimum of 5000 frames at 30-ms exposure time with 5% laser power were recorded for each neuron. Zeiss Zen Black 2012 software was used to set the imaging parameters and capture the recordings. FIJI () was used for subsequent measurement of GFP fluorescence, and the background was subtracted for each dataset.

taagcagcggccgcatactcacgaccgcttg	HXY160-Fwd
tgcttaggccggccatgatgcggtcttaaaaata	HXY161-Rev
taagcaggcgcgccatgaagatacaggacagatcattac	HXY162-Fwd
tgcttacccgggctaattgacgtccgctttga	HXY163-Rev
ttcagcaagggagaggc	HXY215-Fwd
ttacttgtagagctcgtcca	HXY216-Rev
taagcagcggccgcaagcttcgttgtgagatgag	HXY126-Fwd
tggcatggccggccggatccttatctggaacaaa	HXY127-Rev
catgtgagactacaatgtca	HXY242-Fwd
gtacggccacgcttggggcg	HXY243-Rev
ccaaggaaattgacattgtagtctcacatgcgtacggccacgcttggggcgccactcatg	HXY244-Fwd
catgagtggcgccccaagcgtggccgtacgcatgtgagactacaatgtcaatttccttgg	HXY245-Rev
taagcaggcgcgccatgggaaaatgggaaagctg	HXY222-Fwd
ttcacagaacggtaagcaaacaccatttag	HXY223-Fwd
gtaaacgaagttgtcgataatgtccgcaac	HXY224-Rev
taagcagtcgaccttctgtgaaaaaatgagtattggaa	HXY494-Fwd
tgcttaaagcttgatgtgcgttttgatgaatt	HXY461-Rev
taagcagtcgaccttctgtgaaggaatgagtat	HXY510-Fwd
taagcaaagcttatgaagatacaggacagatcattac	HXY395-Fwd
tgcttagcggccgcttatctcttgtcatcgtcatcct	HXY396-Rev
aaaagaagggaaaggagcaaaagctcatttctgaggaagatctctaggagctcgatgagt	HXY530-Fwd
aatgctattttttttatcgctatcaagttatagaggccggccatggaaccgccgtttgag	HXY569-Fwd
gaattgggacaactccagtgaaaagttcttctcctttactaatgtactggctactgctat	HXY570-Rev
tcctgtatcttcatatgatg	crHXY5
tcgttcagtatttgtcaaag	crHXY6
tcgttcagtatttgtcaaagcggacgtcaatgtcagcaagggagaggcag	HXY466-Fwd
cgaaacatacctttgggtcctttggccaatttacttgtagagctcgtcca	HXY467-Rev

Plasmid name	Plasmid	Purpose
pXH8	pSM::Padm-4::adm-4::unc-54 3′UTR	To rescue fusion defect with vdEx1735 expressing adm-4 under endogenous promoter
pXH13and pXH22	pSM::Pmec-4::adm-4::unc-54 3′UTR	To rescue fusion defect cell-autonomously in PLM neurons with vdEx1801
pXH19	pSM::Pmec-4::adm-4::wrmScarlet-I::unc-54 3′UTR	To visualize ADM-4 and to rescue fusion defect cell-autonomously in PLM neurons with vdEx1957 and vdEx2020
pSEM90	Ptwk-18::wrmScarlet-I	Used to create ADM-4::wrmScarlet-I (pXH19), received from the Boulin laboratory
pXH14	pSM::Pdpy-7::adm-4::unc-54 3′UTR	To test rescue of fusion defect with expression in the hypodermis with vdEx1801
pXH29	pSM::Pmec-4::adm-4(E371A)::unc-54 3′UTR	To express catalytic-dead ADM-4 with vdEx2099, vdEx2102, and vdEx2104
pXH30	pSM::Pmec-4::adm-4(ΔCat)::unc-54 3′UTR	To express the truncated form of ADM-4 lacking the prodomain and metalloprotease domain with vdEx1983
pXH31	pSM::Pmec-4::adm-4(ALLxPS)::unc-54 3′UTR	To express the PS-binding ADM-4 mutant with vdEx1978
pXH51	pMAX::adm-4(ALLxPS)::3xFLAG	To express the PS-binding ADM-4 mutant tagged with FLAG in HEK293T cells
pXH60	pGEX-4T-2::adm-4(PS)::3xFLAG	To express the PS domain of ADM-4 tagged with GST and FLAG in BL21(DE3) bacterial cells
pXH67	pGEX-4T-2::adm-4(ALLxPS)::3xFLAG	To express the mutated PS domain of ADM-4 in BL21(DE3) tagged with GST and FLAG in bacterial cells
pXH50	pMAX::adm-4::3xFLAG	To express ADM-4::3XFLAG in HEK293T cells
pXH64	pMAX::eff-1A::eGFP	To express EFF-1A::eGFP in HEK293T cells
pXH77	pSM::Pmec-4::eff-1A(WT)::GFPnovo2	To express and visualize EFF-1::GFP with vdEx2479
pXH78	pSM::Pmec-4::eff-1A(G260A/D221E/D222E)::GFPnovo2	To express and visualize EFF-1::GFP with vdEx2455
pXH81	pSM::Pmec-4::adm-4cDNA(E371A)::wrmScarlet-I::unc-54 3′UTR	To visualize catalytic-dead ADM-4(E371A)::wrmScarlet in PLM neurons with vdEx2590
pXH83	pSM::Pmec-4::adm-4cDNA(ΔCat)::wrmScarlet-I::unc-54 3′UTR	To visualize catalytic-dead ADM-4(ΔCat)::wrmScarlet in PLM neurons with vdEx2592