Axonal Degeneration Is Blocked by Nicotinamide Mononucleotide Adenylyltransferase (Nmnat) Protein Transduction into Transected Axons

Abstract

Axonal degeneration is an early and important component of many neurological disorders. Overexpression of nicotinamide mononucleotide adenylyltransferase (Nmnat), a component of the slow Wallerian degeneration (Wld^s) protein, protects axons from a variety of insults. We found that transduction of Nmnat protein into severed axons via virus-like particles prevented axonal degeneration. The post-injury efficacy of Nmnat indicates that its protective effects occur locally within the axon and provides an opportunity to develop novel agents to treat axonal damage.

Introduction

Axon degeneration is an important component of many neurological disorders including peripheral neuropathies, neurodegenerative diseases, and traumatic injuries (1). Damaged axons fragment via an orchestrated self-destructive process; however, the molecular pathways regulating axonal degeneration remain to be elucidated (2). Axonal degeneration can be delayed by overexpression of nicotinamide mononucleotide adenylyltransferase 1 (Nmnat1), a nuclear protein involved in NAD synthesis that is the principle component of the Wld^s fusion protein (3, 4). The therapeutic implications of preventing axonal degeneration have fostered intense interest in understanding how Nmnat² overexpression prevents axonal degradation (2, 4–6).

EXPERIMENTAL PROCEDURES

DRG Culture and Axon Degeneration Assays

Dorsal root ganglia (DRG) drop cultures were performed as described previously (7). Axotomy was performed after 5–7 days in culture (days in vitro) by severing axons from the cell body cluster with a microscalpel. Axon degeneration was assessed by analyzing phase contrast images as described (7).

Lentivirus and Adenovirus Production

Lentiviruses were generated as described previously (4). DRG neurons were infected at 3 days in vitro with 10⁴-10⁵ colony-forming units (cfu) of lentivirus expressing Cherry-cytNmnat1 (7). Viral infection was monitored using fluorescence microscopy to visualize the Cherry reporter. Adenovirus expressing cytNmnat1 was produced and utilized as described previously (8).

Monitoring of Growth Cone (GC) Retraction and Axonal Swellings

Time-lapse microscopy was performed with a climate-controlled chamber (In Vivo Scientific) at 37 °C and 5% CO₂, and images were acquired every 8 min with a CoolSNAP HQ² CCD camera (Photometrics) mounted on a Nikon Eclipse Ti-U microscope. Either VLP-cytNmnat1 or VLP-cytNmnat1(H24A) (control) was added to neurons 5 min after axonal severing, and fields containing ∼6 GCs were traced for 12 h after the injury. Neurons with GC retraction (i.e. disappearance of lamellipodia or filopodia) and axonal swellings (i.e. structure within the axon) were detected morphologically from images taken directly after axotomy (0 h) or 3 h later by a blinded observer. We confirmed continued axonal protection by VLP-cytNmnat1 by monitoring the same fields 12 h after axotomy.

Production of Virus-like Particles

Virus-like particles (VLPs) were prepared by transfecting 293T cells with vesicular stomatitis virus G (VSV-G) and Nmnat protein expression plasmids (unless otherwise indicated) (1:4 ratio) and collecting the culture media 48–96 h after transfection. For most experiments, His₆-tagged Nmnat1, cytNmnat1, or cytNmnat1(H24A) was expressed using pcDNA3.1 instead of the lentivirus transfer vector to remove all viral elements from the system. VLP-containing media (30 μl) were added to DRG neurons grown in 24-well plates at the indicated times after axonal severing. To purify VLPs, culture medium of transfected 293T cells was centrifuged at 45,000 rpm for 90 min (TLA 100.3, Beckman). The supernatant was removed, the pelleted VLPs were suspended in an equivalent volume of PBS, and the VLPs were repelleted by centrifugation. The washed VLPs were suspended in one-tenth of the original volume and used for experiments. For antibody-blocking experiments, equal amounts of VLP-cytNmnat1 and anti-SV-2 or anti-VSV-G hybridoma supernatant were mixed and incubated at 25 °C for 30 min. The mixture (30 μl) was added to DRG cultures 5 min after axotomy. We confirmed that VSV-G antibody did not alter Nmnat enzymatic activity by incubating it with purified Nmnat protein and performing Nmnat enzymatic assays as described previously (8).

DNA

Lentivirus transfer plasmids encoding His₆-tagged Nmnat1, cytNmnat1, cytNmnat1(H24A), and Cherry-cytNmnat1 were described previously (7, 9). To generate expression constructs lacking all viral elements, Nmnat1, cytNmnat1, and cytNmnat1(H24A) were cloned into pcDNA3.1.

Antibodies

Hybridoma supernatant containing antibodies directed against synaptic vesicle glycoprotein 2 (SV2) developed by K. Buckley was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD, National Institutes of Health, and maintained by The University of Iowa, Department of Biology, Iowa City, IA. VSV-G (clone 8G5F11) hybridoma supernatant was obtained from M. Whitt (10). Anti-His₆ antibody (clone AD1.1.10) was purchased from R&D Systems.

RESULTS AND DISCUSSION

In pursuing the mechanism of Nmnat-mediated axonal protection, we have extensively utilized an in vitro system that uses lentiviruses to alter gene expression in DRG sensory neurons. Using this system, we previously demonstrated that Nmnat enzymatic activity was required for axonal protection and that protection was enhanced when Nmnat was localized to the cytoplasm/axon (i.e. cytNmnat1 mutant) (4, 8). Further, studies of transgenic mice expressing Nmnat proteins in neurons demonstrated that these proteins also promote axonal protection in vivo (8, 11). Curiously, we discovered during these studies that lentivirus expressing cytNmnat1 provided robust protection even when it was added after the axons were severed from the neuronal cell body (Fig. 1a). This caused us to question whether de novo gene expression directed by the viral genome was required for the observed protection.

FIGURE 1.

Post-injury addition of lentivirus expressing cytNmnat1 prevents axonal degeneration. Lentivirus expressing Nmnat1 or cytNmnat1 or adenovirus expressing cytNmnat1 was added 5 min after axotomy, and axonal degeneration was analyzed over time. a, quantification of axonal degeneration showed strong axonal protection only in neurons exposed to lentivirus expressing cytNmnat1. b and c, conditioned media containing viral particles collected from 293T cells expressing various combinations of the indicated lentivirus production plasmids (pVSV-G, Δ8.9, FCIV expression vector (see under “Results”)) were applied to DRG neurons after axonal severing. Representative images at 72 h after axotomy are displayed. Quantification of axonal degeneration at 0, 24, 48, and 72 h after axotomy is shown. Transfection of 293T cells with pVSV-G and FCIV-cytNmnat1 was sufficient to produce a “factor” in the cell media that protects axons when delivered after injury. Control DRGs received culture media from non-infected 293T cells. Scale bar, 100 μm. * indicates significant difference from control axons at the corresponding time after axotomy (n = 16, p < 0.0001 Student's t test; error bars indicate standard deviation).

During the lentivirus life cycle, proteins from the host cell cytoplasm are incorporated into the budding viral particle (12). In turn, these sequestered proteins are released into the infected target cells (13). We hypothesized that the acute axonal protective phenomenon we observed occurs by transfer of cytNmnat1 protein sequestered within the viral particle into the axon during the infection process, which would imply that Nmnat-mediated protection occurs directly within the axon via a local mechanism. To pursue this idea, we tested whether lentivirus expressing Nmnat1 (a nuclear protein that would not be trapped in the budding viral particle) or adenovirus expressing cytNmnat1 (which generates viral particles via an entirely different process that should preclude inclusion of host cell proteins (14)) could also protect axons if added after axotomy. Unlike cytNmnat1 lentivirus, these viral preparations did not promote axonal protection if delivered after axon severing (Fig. 1a), although, as in previous studies, axons of DRG neurons infected with these viruses prior to injury were protected (supplemental Fig. 1). These results are consistent with the observation that cytoplasmic but not nuclear host cell proteins are incorporated into the lentivirus particle. Moreover, the failure of cytNmnat1 adenovirus to promote post-injury protection further supports the idea that lentivirus particles directly transfer cytNmnat1 to the severed axon.

To establish which of the components used to produce lentivirus were required to produce viral particles capable of promoting acute axonal protection, we produced viral particles from 293T cells transfected with various combinations of lentivirus production plasmids. These plasmids included pVSV-G that expresses the VSV-G protein required for viral cell attachment and membrane fusion and is alone sufficient to produce VLPs (15); plasmid Δ8.9 that expresses viral proteins (gag, pol, and rev) necessary for virus packaging and integration (16); and transfer vector FCIV-cytNmnat1, which contains the viral LTRs and other elements necessary for robust viral transgene (e.g. cytNmnat1) expression (17). We found that only pVSV-G and FCIV-cytNmnat1 were required to promote robust acute axonal protection (Fig. 1, b and c). We next substituted pcDNA3.1-cytNmnat1 for FCIV-cytNmnat1 to eliminate all lentiviral elements from the system and found that VLPs produced in this manner also promoted acute axonal protection (supplemental Fig. 2). Importantly, VLPs from cells co-transfected with pVSV-G and pcDNA3.1-cytNmnat1(H24A), which expresses an enzymatically inactive cytNmnat1, did not mediate acute axonal protection. These striking results indicate that transduction of cytNmnat1 into the axon via VLPs is responsible for the observed axonal protection and strongly suggests that Nmnat acts locally within the injured axon to prevent its fragmentation.

To confirm that cytNmnat1 is present in VSV-G derived VLPs, we performed Western blots on purified VLPs using an antibody to the His₆ epitope tag linked to all of the Nmnat constructs. Both cytNmnat1 and cytNmnat1(H24A) were detected in the VLP fraction (pellet). Conversely, only trace amounts of native Nmnat1 (nuclear protein) were present in VLP preparations (Fig. 2c). These results are consistent with the incorporation of cytoplasmic but not nuclear proteins into the VLPs that form at the cell surface. Furthermore, these results correlate with the acute axonal protection assays showing that only VLPs (and not supernatant fractions) produced from 293T cells expressing enzymatically active cytNmnat1 provided axonal protection when delivered post-injury (Figs. 1a and 2, a and b).

FIGURE 2.

Post-injury axonal protection is mediated via VLP-directed cytNmnat1 transduction to severed axons. Culture media containing VLPs collected from 293T cells expressing Nmnat1, cytNmnat1, or cytNmnat1(H24A) together with VSV-G was centrifuged. The resultant supernatant (sup) and pellet (ppt) fractions were applied to DRG neurons 5 min after axotomy. a, representative images of axons of neurons treated with the pellet fractions at 72 h after injury are shown. Scale bar, 100 μm. b, quantification of axonal degeneration in experiment described above. * indicates significant difference from VLP-Nmnat1-treated axons at the corresponding time after axotomy (n = 16, p < 0.0001 Student's t test; error bars indicate standard deviation). Note that only the fraction (pellet) containing VLP-cytNmnat1 prevented axon degeneration. c, Western blot of VLPs produced by cells expressing VSV-G and Nmnat1, cytNmnat1, or cytNmnat1(H24A) was probed with His₆ antibody. Note the low level of the nuclear protein Nmnat1 in the VLP fraction. d, VLP-cytNmnat1 was added to DRG neurons after axotomy in the presence of antibodies directed against VSV-G or SV2 (control). Quantification of axonal degeneration demonstrates that blocking VSV-G function prevents cytNmnat1 axonal protection. * indicates significant difference from VLP-cytNmnat1-treated axons at the corresponding time after axotomy (n = 16, p < 0.0001 Student's t test; error bars indicate standard deviation). e, DRG neurons were infected with lentivirus expressing Cherry-cytNmnat1 in the presence or absence of anti-VSV-G. Fluorescence microscopy showed that Cherry-cytNmnat1 was not expressed in the presence of anti-VSV-G, confirming the ability of this antibody to block lentivirus attachment and infection.

We next determined whether cytNmnat1-mediated protection required VLP fusion to the axon; that is, was the protein delivered to the axon via transduction? VLPs produced from cells transfected with pVSV-G and pcDNA3.1-cytNmnat1 were applied to neurons after axon severing in the presence or absence of an antibody that blocks VSV-G-mediated cell attachment and membrane fusion (10). In the presence of the VSV-G-blocking antibody, axonal degeneration was extensive, indicating that VLP binding and/or fusion to the axonal membrane are critical for acute axonal protection (Fig. 2d), whereas robust protection persisted after the addition of anti-SV2 (control). The blocking efficacy of the VSV-G antibody was also demonstrated via lentivirus infection as monitored by Cherry-cytNmnat1 expression using fluorescence microscopy (Fig. 2e). These results suggest that VSV-G-mediated fusion of VLPs to the axonal membrane allows entry of sequestered cytNmnat1 into the severed axons where it acts locally to delay axonal degeneration.

Having shown that VSV-G-dependent VLP delivery of cytNmnat1 to severed axons prevents their degeneration, we used them to investigate the mechanism of Nmnat-mediated axonal protection. Neurons were treated with the protein synthesis inhibitor cycloheximide (CHX) prior to VLP-cytNmnat1 addition and axon injury. We found that CHX blocked protein synthesis but had no effect on VLP-cytNmnat1-mediated axonal protection (Fig. 3a and supplemental Fig. 4), indicating that protection does not require local protein synthesis within the axon. Further, these results demonstrate that gene expression from nucleic acid components potentially delivered to the axon by the VLPs is not responsible for this protective phenomenon.

FIGURE 3.

Nmnat inhibits early critical steps in the axonal degeneration process. a, VLP-cytNmnat1 was applied to DRG neurons after axonal severing in the presence or absence of CHX, and axonal degeneration was quantified. Axonal degeneration was efficiently inhibited by VLP-cytNmnat1 even in the absence of protein synthesis. DMSO, dimethyl sulfoxide. b, VLP-cytNmnat1 was added to DRG neurons at the indicated times after axotomy. Representative images of axons at 72 h after axotomy are shown. Scale bar, 100 μm. c, quantification of axonal degeneration showed that the addition of VLP-cytNmnat1 within 2 h of injury provided complete axonal protection. d, GCs (dotted circles) of DRG neurons were examined immediately after injury (0 h) or 3 h later. Severe GC retraction after axonal injury occurred even in the presence of VLP-cytNmnat1 (compare with VLP-cytNmnat1(H24A) control). Note the appearance of phase-dark swellings (blebs) within injured axons treated with VLP-cytNmnat1(H24A) at 3 h. These swellings are not observed in neurons treated with VLP-cytNmnat1. Scale bar, 20 μm. e, neurons exhibiting GC retraction or axonal swellings were counted using time-lapse images (VLP-cytNmnat1(H24A), n = 100; VLP-cytNmnat1, n = 81). VLP-cytNmnat1 blocked axonal swelling but not GC retraction after axotomy. * indicates significant difference from 0 h after axotomy in each condition (n = 16, p < 0.0001 Student's t test; error bars indicate standard deviation) for a and c. ** indicates significant difference from VLP-cytNmnat1(H24A) for e.

Next, we used cytNmnat1-VLPs to determine the critical window when injury-induced axonal degeneration could be effectively halted (i.e. the time span in which key molecular events required for axonal breakdown are carried out). VLP-cytNmnat1 was added at various times after axotomy, and axonal degeneration was assessed up to 72 h after injury. When VLP-cytNmnat1 was added within 2 h of injury, there was robust axonal protection. Addition at 4 h after injury provided significant, albeit reduced, protection, whereas addition at 6 h had only minimal protective effects on the axons (Fig. 3, b and c, and supplemental Fig. 3). This latency period (2–4 h) is similar to that identified using JNK inhibitors (18) and strongly suggests that critical steps in the dismantling of injured axons must occur within this short, post-injury time period.

To examine events within this time frame that might be influenced by VLP-cytNmnat1 and are important for the axon degradation process, we performed time-lapse microscopic analysis of degenerating axons. We focused on terminal axon segments as they are remodeled extensively during axonal guidance when axon termini undergo repeated cycles of growth cone retraction and axon breakdown. We found that growth cones rapidly retract within 1–3 h after axon severing. Growth cone retraction is then followed by the appearance of phase-dark swellings within the terminal axon segments themselves (supplemental video 1). We examined these events after the addition of either VLP-cytNmnat1 or VLP-cytNmnat1(H24A) (inactive control) 5 min after injury and found that Nmnat enzymatic activity does not influence the onset or progression of growth cone retraction (Fig. 3, d and e, and supplemental videos 1 and 2). In contrast, the development of axonal swellings was completely blocked by the addition of VLP-cytNmnat1 (Fig. 3e). These results suggest that the later axonal disruption, signified by the appearance of swellings shortly after injury, can be prevented by Nmnat but that the initial reaction of growth cones to axonal damage appears to lie upstream of (or parallel to) Nmnat-sensitive events.

Using protein transduction via lentivirus particle delivery, we have shown that Nmnat-mediated protection acts locally within the axon and that critical, Nmnat-sensitive steps in axonal breakdown occur within the first few hours after injury. These studies suggest that effective treatments for axonal degeneration could be delivered after injury (e.g. ischemic events) to minimize damage. The identification of Nmnat-regulated components responsible for activating the degeneration process in this early phase is a crucial next step in the development of therapeutic agents aimed at ameliorating the effects of axonal damage.