Abstract
Treatments accelerating axon regeneration in the nervous system are still clinically unavailable. However, parthenolide promotes adult sensory neurons’ axon growth in culture by inhibiting microtubule detyrosination. Here, we show that overexpression of vasohibins increases microtubule detyrosination in growth cones and compromises growth in culture and in vivo. Moreover, overexpression of these proteins increases the required parthenolide concentrations to promote axon regeneration. At the same time, the partial knockdown of endogenous vasohibins or their enhancer SVBP in neurons facilitates axon growth, verifying them as pharmacological targets for promoting axon growth. In vivo, repeated intravenous application of parthenolide or its prodrug di-methyl-amino-parthenolide (DMAPT) markedly facilitates the regeneration of sensory, motor, and sympathetic axons in injured murine and rat nerves, leading to acceleration of functional recovery. Moreover, orally applied DMAPT was similarly effective in promoting nerve regeneration. Thus, pharmacological inhibition of vasohibins facilitates axon regeneration in different species and nerves, making parthenolide and DMAPT the first promising drugs for curing nerve injury.
Significance Statement
Here, we demonstrate, using knockdown and pharmacological approaches in cell culture and in vivo, the role of vasohibins in limiting axon regeneration, specifically in the adult peripheral nervous system. Vasohibins increase axonal microtubule detyrosination in growth cones of adult neurons compared with postnatal ones. Reducing vasohibin activity in adult neurons markedly accelerates axon growth. In vivo, pharmacological inhibition by repeated intravenous application of parthenolide or oral treatment with its prodrug di-methyl-amino-parthenolide (DMAPT) markedly facilitates the regeneration of sensory, motor, and sympathetic axons in injured murine and rat nerves, leading to acceleration of functional recovery. Thus, pharmacological inhibition of vasohibins facilitates axon regeneration in different species and nerves, making parthenolide and DMAPT promising drugs for treating nerve injury.
Introduction
Axonal lesions in the peripheral nervous system (PNS) disconnect peripheral targets from the central nervous system (CNS), resulting in motor, sensory, and autonomic impairments. While PNS axons are, in principle, capable of regeneration, functional recovery often remains incomplete because axons can only regenerate at a maximal rate of 1–2 mm/d (Sunderland, 1947; Sulaiman and Gordon, 2013). Moreover, this growth rate can only be sustained for a few months, after which the intrinsic growth potential of neurons typically decreases, and Schwann cells lose their growth-supporting functions. This time frame limits regenerating axons to overcome maximal distances of ∼9–18 cm, which is insufficient for most peripheral nerve injuries (Sulaiman and Gordon, 2013). Axons that fail to reinnervate their appropriate targets may leave patients with permanent functional deficits and chronic pain caused by inappropriate innervation and misguidance. Therefore, an axon growth rate elevation can quantitatively and qualitatively improve motor and sensory recovery after peripheral nerve injury (Ma et al., 2011).
Despite extensive research for new treatment strategies, little progress has been made in developing clinically relevant nerve repair drugs. Although factors such as nerve growth factor (NGF) or brain-derived neurotrophic factor (BDNF) accelerate axon growth and neuronal survival in animal models (Grinsell and Keating, 2014; Faroni et al., 2015), their administration causes severe side effects in humans and therefore is not applicable (Diekmann and Fischer, 2016). Similarly, the immunosuppressor FK506 (Tacrolimus) enhances axonal regeneration upon autologous nerve transplantation (Grinsell and Keating, 2014). However, prolonged and systemic administration of this drug, as required for long-distance nerve injury treatment, inflicts high infection risks, bone fractures, and hypertension (Tung, 2015). Hence, treating peripheral nerve injuries still mainly relies on surgical intervention and depends on the severity of the damage (Grinsell and Keating, 2014; Diekmann and Fischer, 2016). While nerve contusions are usually not treated and, to some extent, heal spontaneously, fully severed nerves normally are readapted at both ends. Autologous nerve transplants often bridge nerve gaps, which require sacrificing healthy nerves. The sural or the antebrachial cutaneous nerve, typically utilized as transplants, leads to numbness of the respective outer foot or inner arm, respectively. Lately, synthetic nerve guides can be implanted into smaller lesion sites, but these enable insufficient nerve regeneration (Grinsell and Keating, 2014). Surgical interventions to readapt severed nerves cannot sufficiently solve slow regeneration and often result in incomplete functional recovery.
Increased glycogen synthase kinase 3 (GSK3) activity in genetically modified mice raises peripheral nerves’ axon growth rate. Furthermore, it accelerates functional recovery, mediated by maintaining microtubules in a dynamically unstable state (Gobrecht et al., 2014). This effect is likely conveyed by microtubule-associated protein 1B (MAP1B) phosphorylation, which reduces tubulin detyrosination through interaction with tubulin tyrosine ligase (Utreras et al., 2008). In addition, MAP1B has been shown to prevent microtubule detyrosination by direct binding to tyrosinated tubulin (Goold et al., 1999). Therefore, it was hypothesized that active GSK3 reduces tubulin detyrosination by phosphorylating MAP1B, thereby increasing microtubule dynamics and, consequently, axonal growth (Lucas et al., 1998; Goold et al., 1999; Owen and Gordon-Weeks, 2003; Gobrecht et al., 2014, 2016).
Since genetic alterations are not yet suitable for clinical application, we tested whether pharmacological approaches can similarly reduce microtubule detyrosination in peripheral axon tips. We identified the sesquiterpene lactone parthenolide as a molecule able to induce such effects on microtubules and promote the axon growth of adult sensory neurons in culture (Gobrecht et al., 2016). Interestingly, parthenolide derivative di-methyl-amino-parthenolide (DMAPT) improves axon growth of central neurons and also functional recovery after a complete spinal cord crush (Leibinger et al., 2023). However, the pharmacological target and mechanism underlying parthenolide's effect on microtubule detyrosination and axon growth remained elusive because the identity of the mediating carboxypeptidase(s) responsible for detyrosination was unknown. However, recently, vasohibin 1 and vasohibin 2 (VASH1, VASH2) in a complex with their enhancer small vasohibin binding protein (SVBP) were identified as the first two tubulin carboxypeptidases (Aillaud et al., 2017; Nieuwenhuis et al., 2017; Li et al., 2019). Parthenolide was shown to bind VASH1 and VASH2 with high affinity in situ (Aillaud et al., 2017; Li et al., 2019). However, whether parthenolide promoted axon regeneration by binding to VASH in vivo was unknown. This is particularly true because at high concentrations, parthenolide reportedly also forms adducts on both cysteine and histidine residues on tubulin itself, suggesting an indirect inhibition by reducing the polymerization-competent pool of tubulin (Hotta et al., 2021).
Using genetic and pharmacological approaches, the current study suggests that vasohibins are the carboxypeptidase involved in axon regeneration and identifies the molecular targets of parthenolide in this context. Moreover, it also demonstrates that systemic application of parthenolide/DMAPT leads to an acceleration of functional motor and sensory recovery in various nerves and species upon injury. Thus, inhibiting vasohibins pharmacologically is a promising and potentially clinically feasible approach for treating nerve injuries, and parthenolide and DMAPT are the first drug candidates.
Materials and Methods
Sensory neuron cultures and immunocytochemical staining
As described previously, sensory neurons were obtained from adult or postnatal (P3) mice's dissociated dorsal root ganglia (DRG; Gobrecht et al., 2014, 2016). In short, isolated DRG (ca. T3–L6) were incubated with 0.25% trypsin/EDTA (GE HealthCare) and 0.3% collagenase type IA (Sigma) in DMEM (Invitrogen) at 37°C and 5% CO2 for 45 min and mechanically dissociated. Cells were resuspended in DMEM containing 2% (v/v) B27 (Thermo Fisher Scientific) and penicillin/streptomycin (500 U/ml; Merck, Millipore) and cultured at 37°C and 5% CO2 on poly-d-lysine (PDL, 0.1 mg/ml, molecular weight <300 kDa; Sigma) and laminin (20 μg/ml; Sigma)-coated 8-well plates (Sarstedt). Cells were treated with vehicle (DMSO), 1–100 nM parthenolide. For baculoviral transduction, 10% (v/v) virus solution was added to the culture 5 h after plating. A GFP-encoding baculovirus served as a control. For replating experiments, 5 d after dissociation, sensory neurons were rinsed off from their original culture dish with fresh medium after removing the old. The cell-containing medium was then transferred to new PDL and laminin-coated culture dishes. After an additional 24 h, the cells were fixed with 4% PFA (Sigma). The replating experiments are particularly mentioned as such in the figure legends. For the remaining experiments, sensory neurons were cultured in their original dishes for 24–72 h and then fixed with 4% PFA. All fixed cell cultures underwent immunocytochemical staining with antibodies against βIII-tubulin (1:2,000; BioLegend, RRID:AB_2313773). Imaging was performed automatically with the Olympus VS210-S5 slide scanner. The total axon length was automatically quantified with the NeuriteTracer plugin for ImageJ, avoiding experimenter-induced quantification bias. The average axon length per neuron of three wells per condition and neuron counts per experimental group were normalized to control groups. Data represent the mean ± SEM of two to six independent experiments. Significances of intergroup differences were evaluated using two-way ANOVA followed by the Holm–Sidak post hoc test.
Microtubule detyrosination in axon tips was evaluated using antibodies against βIII-tubulin (1:2,000; BioLegend, RRID:AB_2313773) and detyrosinated tubulin (1:500; Sigma, RRID:mAB_477583) as described before (Freund et al., 2019a,b). In brief, axon tips were defined as the last 15 μm of βIII-tubulin–positive neurite extensions. Detyrosinated tubulin staining intensity of the axon tip was measured using ImageJ, and the adjacent background intensity was subtracted. Intensities were normalized to the respective control group. Data represent mean ± SEM of three replicated wells with 30–60 tips per well from three independent experiments. Significances of intergroup differences were evaluated using two-way ANOVA followed by the Holm–Sidak post hoc test.
Baculoviral vectors and knockdown
Plasmids: HA-tags were attached to mouse VASH1 (Origene, MR222520) and VASH2 (Origene, MR203958) by PCR (vash1-HA fw: cgatcgccatgccaggggg; VASH1-HA rev: atccgggtgtacccatacgatgttccagattacgcttaactcgagaaaa; VASH2-HA fw: cgatcgccatgtggctgcacg; VASH2-HA rev: gatccggatctacccatacgatgttccagattacgcttaactcgagaaaa. FLAG-tagged SVBP was cloned from cDNA derived from murine heart tissue by PCR. FLAG-SVBP fw: gactacaaagacgatgacgacaagatggatccacctgcc; FLAG-SVBP rev: gcagccgcctggggagtgactcgag. For SVBP or VASH1/2, knockdown primers containing the shRNA sequence SVBP shRNA fw: gatccgatgagttctgtaagcagatgctcgagcatctgcttacagaactcatcttttta; SVBP shRNA rev: agcttaaaaagatgagttctgtaagcagatgctcgagcatctgcttacagaactcatcg; VASH1sh fw: gatccgctgtgatcctgggaatttacctcgaggtaaattcccaggatcacagctttttaa; VASH1sh rev: agctttaaaaagctgtgatcctgggaatttacctcgaggtaaattcccaggatcacagcg; VASH2sh fw: gatccgacttcgaagattcctataagctcgagcttataggaatct tcgaagtcttttta; VASH2sh rev: agcttaaaaagacttcgaagattcctataagctcgagcttataggaatcttcgaagtc) were cloned into U6 plasmids.
Baculovirus production was performed as described recently (Levin et al., 2016, 2019): According to the manufacturer's protocols, recombinant baculoviruses were produced using the ViraPower BacMam Expression System (Thermo Fisher Scientific). In brief, recombinant bacmid DNA was purified and transfected into adherent Sf9 cells (Thermo Fisher Scientific) using Cellfectin reagent to generate P1 recombinant baculovirus stock. Baculoviruses were amplified by inoculation of 50 ml Sf9 suspension cultures (106 cells/ml) in Sf-900 III SFM Medium (Thermo Fisher Scientific) supplemented with 12.5 U/ml penicillin/streptomycin (Biochrom) in 125 ml polycarbonate Erlenmeyer flasks with vent cap (Corning) with 1 ml virus stock solution and incubation at 27°C and 110 rpm for 4 d. Baculovirus preparations were pretested on AAV293 cells overnight by adding 1% (v/v) virus. Transduction efficiencies of ≥90% were regarded as appropriate for further use. Otherwise, the virus stock was subjected to further amplification cycles. For transduction of sensory neurons, 10% (v/v) baculovirus solution was added to the cells 2 h after plating. Knockdown efficiencies were verified by immunostaining against HA (1:500; Sigma-Aldrich, RRID:AB_2600700) for VASH1 and VASH2 or FLAG (1:500; Sigma, RRID:AB_262044) for SVBP after baculoviral transduction in sensory neurons. The staining intensity was calculated using the formula intensity = integrated density − area * mean gray value background. Transduction was verified by immunostaining against GFP (1:500; Novus; RRID:AB_10128178), and neurons were identified by βIII-tubulin staining (1:1,000; BioLegend; RRID:AB_2313773). Additionally, the knockdown was verified by quantitative real-time PCR (qRT-PCR) with cDNA from sensory neurons after baculoviral-induced VASH1/2 or SVBP knockdown after 5 d in culture. qRT-PCR primers are as follows: VASH1 fw: TGGCCAAGATCCACCCAGATG; VASH1 rev: TCGTCGGCTGGAAAGTAGGCAC; VASH2 fw: AGGGGGAGAGATGGTAGGCGC; VASH2 rev: AGCCAGTCTGGGATCGTCATGG; SVBP fw: AACCAGCCTTCAGAGTGGAGAAGG; SVBP rev: GCTCCGTCATGACTCTGTTGAGAGC.
Surgical procedures
All animal protocols adhered to animal care guidelines and were approved by the local authorities (LANUV Recklinghausen). Male and female adult (8–12 weeks) C57/BL6J mice and Lewis rats were maintained in cages with 1–5 animals on a 12 h light/dark cycle with ad libitum access to food and water. Sciatic nerve crush (SNC) was performed as described previously (Gobrecht et al., 2014, 2016). In brief, mice were anesthetized by intraperitoneal injections of ketamine (80–100 mg/kg, Pfizer) and xylazine (10–15 mg/kg, Bayer), rats with 2% isoflurane in oxygen, and a skin incision of ∼10 mm was made above the gluteal region. Then, the ischiocrural musculature was carefully spread with minimal tissue damage to expose the right sciatic nerve from the sciatic notch to the point of trifurcation. Next, the crush injury was performed for 10 s proximal to the tibial and peroneal divisions using graphite powder-dipped Dumont #5 forceps (Hermle) to mark the crush site. The skin was then closed using 6-0 sutures.
A skin incision of ∼10 mm from the right axillary region to the elbow was made for the murine median nerve crush (MNC), thus exposing the nerve from its origin at the brachial plexus to the elbow. Subsequently, the crush injury was performed for 10 s using graphite powder-dipped Dumont #5 forceps (Hermle) to mark the crush site. The skin was then closed using 6-0 sutures. After the surgery, animals received daily intravenous or oral doses of parthenolide or DMAPT in 100 µl vehicle.
Intrathecal AAV1 injection
For intrathecal AAV application, a 5 mm skin incision was made above the L4–L5 spines. The muscle tissue was carefully spread, and 2.5 µl of the viral solution was injected into the cerebrospinal fluid between the L4 and L5 spines using a Nanoject III injector (Drummond Scientific) with three consecutive 833 nl injections with a speed of 7 nl/s and 20 s intervals between injections. The skin was closed using 6-0 sutures. Transduction was performed 2 weeks before SNC.
Adeno-associated virus
AAV1-GFP was obtained from Addgene (#37825-AAV1) with a 7 × 1012 vg/ml titer. The other AAV1 viruses were produced in our laboratory using a CMV-driven pAAV-IRES_hrGFP expression vector for VASH1-HA. AAV plasmids carrying either cDNA for respective genes were cotransfected with pAAV2/1 (Addgene #112862) encoding the AAV genes rep and cap, and the helper plasmid (Stratagene) encoding E24, E4, and VA into AAV-293 cells (Stratagene) for recombinant AAV1 generation. Virus particles were purified as described previously (Leibinger et al., 2013). The titer of the AAVs ranged from 1.2 × 1014 to 2 × 1014 vg/ml. Mainly neurons are transduced upon intrathecal injection of AAV1, as this virus serotype is highly neurotropic.
Quantification of regenerating axons in the sciatic nerve
Sciatic nerves were isolated 3 d after SNC, postfixed in 4% PFA overnight, dehydrated in 30% sucrose again overnight, and embedded in Tissue-Tek (Sakura). Longitudinal sections (14 μm thick) were cut on a cryostat (Leica), and thaw-mounted onto coated glass slides (Superfrost Plus, Fisher). Cryosections were immunohistochemically stained with antibodies against the regeneration-associated sensory axon marker SCG10 (1:1,000; Novus Biologicals, RRID:AB_10011569; Shin et al., 2012), the motor axon marker CHAT (1:100; Sigma, RRID:AB_90661) or the sympathetic axon marker tyrosine hydroxylase (1:500; Novus Biologicals, RRID:AB_10077691). In addition, labeled axons were quantified beyond the graphite-labeled injury site. Therefore, the antibody-stained axons at 1.5, 2, 2.5, 3, 3.5, 4, 4.5, and 5 mm past the lesion site were counted and the thickness of the section at this distance was determined. From these two values, the number of axons per millimeter was calculated. Experimental groups comprised four to five animals, and five different sections were analyzed per animal. Statistical significances of intergroup differences were evaluated using two- or three-way ANOVA followed by the Holm–Sidak post hoc test.
Immunohistochemical DRG staining
L4 DRGs were removed 2–6 weeks after intrathecal AAV1 transduction and fixed in 4% PFA overnight at 4°C. Then, DRGs were dehydrated in 30% sucrose at 4°C again overnight and embedded in Tissue-Tek (Sakura). Sections (14 μm) were cut on a cryostat (Leica) and thaw-mounted onto coated glass slides (Superfrost Plus, Fisher). Cryosections were immunohistochemically stained with antibodies against βIII-tubulin (1:2,000; BioLegend, RRID:AB_2313773) or GFP (1:500, Novus, RRID:AB_10128178) overnight.
Static sciatic index
After SNC, motor recovery was determined by calculating the static sciatic index (SSI), as previously described. At the same time of day (11:00 A.M. to 2:00 P.M.) at various time points after SNC (mice: 0, 1, 4, 7, 9, 14, 21, and 28 d; rats: 0, 1, 7, 9, 11, 14, 16, 18, 21, 24, 28, 30, 32, 35, 37, 42, 45, 49, 53, and 63 d), mice or rats were lifted from the ground to photograph the left and right hind feet, respectively. The treatment was blinded to the experimenter during the experiment. Toe spreading on the contralateral (C, left) and the ipsilateral (I, right) sides relative to the SNC was assessed by measuring paw length (PL) and the distance between the first and fifth toe (FF). The SSI was then calculated using the previously described formula: SSI = 101.3 ((IFF − CFF) / CFF) − 54.03 ((IPL − CPL) / CPL) − 9.5 (Baptista et al., 2007). Data represent mean ± SEM per experimental group. Statistical significances of intergroup differences were evaluated using two-way ANOVA followed by the Holm–Sidak post hoc test.
Grip strength
After the MNC, the right forepaw's functional motor recovery was assessed by determining the grip strength return. Therefore, the grip strength was tested at various time points after MNC (0, 1, 4, 7, 9, 11, and 14 d) at the same time of day (11:00 A.M. to 2:00 P.M.).
For the test, the mouse was held by the tail close enough to the bar of the computerized grip strength meter (Model 47230, Ugo Basile) to grasp it. When the mouse held onto the bar, it was carefully pulled away until losing grip. The metal bar was connected to a force transducer that automatically recorded each measurement's peak force in grams. At each time point, five measurements were recorded. The mean of the three intermediate values was determined as the grip strength of the five measurements. Statistical significances of intergroup differences were evaluated using two-way ANOVA followed by the Holm–Sidak post hoc test.
Von Frey test
The von Frey test was determined as described recently (Gladman et al., 2012; Gobrecht et al., 2014, 2016):
Sensory recovery after SNC and MNC was determined at various time points (mice, SNC: 0, 1, 4, 7, 9, 14, 21, 28 d; mice, MNC: 0, 1, 4, 7, 9, 11, and 14 d; rats, SNC: 0, 1, 7, 9, 11, 14, 16, 18, 21, 24, 28, 30, 32, 35, 37, 42, 45, 49, 53, and 56 d) after SNC with the von Frey filament test. Tests were performed at the same time of day by the same experimenter, unaware of the treatment. Mice or rats were placed on an elevated metal grid (pore size: 2 mm) and allowed to acclimate for 15 min before testing. Starting with the smallest, differently sized, innocuous von Frey filaments (Muromachi Kikai) were consecutively poked six times into the ipsilateral hindpaw to elicit a paw withdrawal. The filament size was recorded if the paw was withdrawn at least two of these six times. Statistical significances of intergroup differences were evaluated using two-way ANOVA followed by the Holm–Sidak post hoc test.
Analysis of muscle reinnervation after SNC
Mice were killed before crush or 10 d after SNC. The musculus extensor hallucis longus was dissected, fixed in 4% PFA for 1 h, and permeabilized in 2% Triton-X-100/PBS overnight. Axons were labeled with an antibody against heavy and medium-chain neurofilament (1:2,000; Cell Signaling) and Alexa488-conjugated secondary antibody. Re-established synapses in neuromuscular endplates were visualized by incubation with Alexa594-conjugated α-BTX (1:1,000; Invitrogen) for 1 h at room temperature. Representative pictures of each experimental group were taken on an SP8 confocal microscope (Leica).
Analysis of footpad reinnervation after SNC
The sciatic nerve of mice was crushed. Mice were injected intravenously with 2 µg/kg parthenolide (Sigma) or vehicle (DMSO) daily for 10 d before tissue isolation. The footpads innervated by the sciatic nerve were isolated. The tissue was harvested from uninjured animals and animals 1 and 10 d after SNC. The tissue was fixed in 4% PFA at 4°C overnight and dehydrated in 30% sucrose at 4°C again overnight. Twenty micrometers of cryosections were permeabilized in methanol and blocked (2% BSA, 5% donkey serum). Axons were identified by immunostaining against βIII-tubulin (1:1,000, Covance) overnight, and antibodies were detected with anti-rabbit antibodies conjugated with Alexa-594. In addition, nuclei were stained with DAPI (Merck).
Axons in the stratum spinosum of 25 sections per mouse were counted. Four mice per condition were analyzed. The experimental conditions were blinded for the experimenter. Statistical significances of intergroup differences were evaluated using two-way ANOVA followed by the Holm–Sidak post hoc test.
RNA isolation and qRT-PCR
L3, L4, and L5 DRGs of animals with or without a 3 d SNC were isolated. Total RNA was isolated using the RNeasy Mini kit (QIAGEN) according to the manufacturer's instructions. RNA was reverse transcribed using the SuperScript II reverse transcriptase kit (Invitrogen). Glyceraldehyde 3-phosphate dehydrogenase (gapdh) was amplified using QuantiTect primers (QIAGEN), and Vash1 and Vash2 were amplified with self-designed primers. DNA was visualized with a SYBR Green PCR Master Mix (Applied Biosystems) on an Applied Biosystems 7500 real-time PCR system (Thermo Fisher Scientific) using 45 amplification cycles. PCR specificity was verified with the dissociation curve analysis feature. Vash1 and Vash2 mRNA expression levels were quantified and compared using the ΔΔCt method. All reactions were duplicated, and three independent samples were analyzed per experimental group.
Results
Neuronal overexpression of VASH1/2 induces microtubule detyrosination and compromises axon growth of primary neurons
As VASH1 and VASH2 were recently identified as the first carboxypeptidases interacting with parthenolide (Aillaud et al., 2017; Li et al., 2019; Hotta et al., 2021), we addressed whether VASH1 and VASH2 are also molecular targets relevant for axon growth. To this end, we tested, whether SNC affects VASH1 or VASH2 expression. As the expression was not affected by injury (Fig. 1H,I), we virally overexpressed both enzymes or their common enhancer, SVBP, in adult sensory neurons and measured the effects on microtubule detyrosination in axonal tips and, functionally, on axonal growth. Baculoviral protein expression (GFP) was detected in neurons 48 h after transduction and resulted in transduction rates >75% (Fig. 1A,B). Furthermore, overexpression of either VASH1-HA or VASH2-HA significantly increased microtubule detyrosination in axon tips after 48 h in culture compared with GFP-transduced controls (Fig. 1C–E), indicating their activity as tubulin carboxypeptidases in primary neurons. We also overexpressed SVBP in this model. Although SVBP alone does not affect microtubule detyrosination, it interacts with VASH1 and VASH2 to strongly increase their activity (Aillaud et al., 2017). Accordingly, SVBP overexpression alone had no measurable effect on detyrosination (Fig. 1C–E). Consistent with the dependence of axon extension on the detyrosination state in axonal growth cones, the overexpression of each VASH alone, but not SVBP, reduced the average length of axons grown in culture significantly (Fig. 1F,G).
Parthenolide-mediated VASH1 and VASH2 inhibition promotes axon growth. A, Representative photos of adult murine sensory neurons after 2 d in culture, baculovirally transduced with GFP, VASH1 (V1), VASH2 (V2), or SVBP. Axons were stained for GFP (green) and βIII-tubulin (tubulin, red). Scale bar, 250 μm. B, Quantification of the transduction rate of sensory cultures described in A. Transduced neurons were recognized by colocalization of βIII-tubulin and GFP expression, which was coexpressed by the baculovirus. Data represent the mean ± SEM of three independent experiments. A dot represents each of the three independent experiments. C, Representative photos of axon tips of adult murine sensory neurons after 2 d in culture, baculovirally transduced with GFP, VASH1 (V1), VASH2 (V2), or SVBP. Axons were stained for detyrosinated tubulin (detyr tub, red) and βIII-tubulin (tub, green). Scale bar, 5 μm. D, Quantification of axon tips positive for detyrosinated tubulin in cultures described in C additionally exposed to 0, 1, 10, or 100 nM parthenolide. E, Quantification of detyrosinated tubulin intensity in axon tips described in D. F, Representative photos of adult murine sensory neurons after 2 d in culture, baculovirally transduced with GFP, VASH1 (V1), VASH2 (V2), or SVBP and treated with 0, 1, 10, or 100 nM parthenolide. Neurons were stained for βIII-tubulin (tub). Scale bar, 250 μm. G, Quantification of axon growth of cultures as depicted in F. Data were normalized to vehicle-treated GFP controls with an average axon length of 543 μm/neuron. Data represent the mean ± SEM of at least three independent experiments. Each of the three independent experiments is represented by a dot. p-values were determined using ANOVA on the ranks followed by Dunns’ post hoc test. The lines connect the bars between for which a significant p-value was determined. H, Relative VASH1 mRNA levels in L3, L4, and L5 DRG from three animals before the injury and 3 d after SNC determined with real-time PCR. I, Relative VASH2 mRNA levels in L3, L4, and L5 DRGs from three animals before the injury and 3 d after SNC determined with real-time PCR. B, D, E, G, Data represent the mean ± SEM of at least three independent experiments. Each of the three independent experiments is represented by a dot. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. The lines connect the bars between for which a significant p-value was determined.
We then tested whether these VASH1/2 or SVBP overexpressing neurons were still responsive to parthenolide. Neurons transduced with a GFP control virus showed the most robust parthenolide-induced axon growth at 1 nM. In comparison, 10 nM showed no effect, and 100 nM even reduced axon growth due to a more pronounced and prevailing microtubule destabilization at high concentrations, thereby confirming the previously described bell-shaped concentration–response curve (Fig. 1F,G; Gobrecht et al., 2016). The bell-shaped curve also indicates that a defined equilibrium range of tyrosinated and detyrosinated microtubules is required for optimal growth, corresponding to a 20% reduced detyrosinated tubulin staining intensity in our assay to (Fig. 1E). Also, parthenolide was effective in VASH1 or VASH2 overexpressing neuronal cultures, with basically all axon tips being positive for detyrosinated microtubules before treatment. However, to reach a reduction of 20% in detyrosinated tubulin staining intensity of the axon tips, at least 10 times higher concentrations (10 nM) were required. Consistently, in VASH overexpressing neurons, the effective concentration of parthenolide to promote axon growth shifted from 1 to 10 nM (Fig. 1G). Moreover, even concentrations ≥100 nM still slightly promoted axon growth in these cells (Fig. 1G), showing that higher parthenolide concentrations can counteract elevated microtubule detyrosination by VASH1 or VASH2 overexpression (Fig. 1C–E). On the other hand, as GFP controls, SVBP overexpressing neurons showed improved axon growth in the presence of 1 nM parthenolide (Fig. 1G), supporting the notion that parthenolide directly targets VASH1 and VASH2.
Endogenous VASH1/2 and SVBP expression affects microtubule detyrosination and axonal growth in adult neurons
We then examined the relevance of endogenous VASH1, VASH2, and SVBP in microtubule detyrosination in axonal tips. We created baculoviruses for a shRNA-mediated knockdown in sensory neurons. We used recombinant coexpressed HA- or FLAG-tagged proteins to confirm the specificity and efficiency of the shRNA constructs in sensory neurons (Fig. 2A–F). Moreover, we confirmed the efficient knockdown of endogenous VASH1, VASH2, or SVBP through qRT-PCR (Fig. 2G–J), as specific antibodies for VASH1/2 or SVBP are unavailable. Interestingly, the data show a slight but nonstatistically significant upregulation of VASH2 upon VASH1 knockdown (Fig. 2J).
VASH1, VASH2, and SVBP can be efficiently knocked down with specific shRNA. A–C, Representative photos of immunostained murine sensory neurons after baculoviral transduction with GFP, VASH1-HA, VASH-2-HA, or FLAG-SVBP and 2 d in culture. Cultures were cotransduced with either VASH1 shRNA (V1sh), VASH2 shRNA (V2sh), or SVBP shRNA (sh). Neurons were stained for HA or FLAG (red), GFP (green), and tub (cyan). Scale bar, 20 μm. D–F, Quantification of HA- or FLAG-immunostaining intensity in sensory neurons, as described in A–C. Data represent the mean ± SEM of three independent experiments (n = 3). Each of the three independent experiments is represented by a dot. p-values were determined using one-way ANOVA followed by the Holm–Sidak post hoc test. The lines connect the bars between which a significant p-value was determined. G–I, Verification of baculoviral-induced knockdown of (G) VASH1, (H) VASH2, and (I) SVBP in qRT-PCR using cDNA from cultivated sensory neurons. For this experiment, sensory neurons isolated from six animals were pooled. All samples were measured five times. Data represent the mean ± SEM. p-values were determined using Student's t-test. J, Verification of baculoviral-induced knockdown of VASH1 and its effect on vash1/2 expression in qRT-PCR using cDNA from cultivated sensory neurons. Data represent the mean ± SEM of four technical replicates. The expression was normalized to the GAPDH expression and the respective VASH1 or VASH2 expression of the GFP control. p-values were determined using Student's t-test.
In the first set of experiments, we transduced sensory neurons with the shRNAs 2 h after cell dissociation and quantified axon growth after 3 d. As the knockdown-mediated decrease in mRNA expression occurred not immediately but rather took several hours, the reduction of microtubule detyrosination reached effective levels to promote axon growth for some time between 2 and 3 d in culture (Fig. 3A–C). Afterward, microtubule detyrosination fell below effective levels (Fig. 3A–C). Consistently, the 3 d knockdown of endogenous VASH1, VASH2, or SVBP significantly increased axon growth compared with respective controls (Fig. 3D,E), demonstrating that each protein contributes to microtubule detyrosination and is relevant for axon growth in adult neurons. We then investigated whether a longer-lasting knockdown approach of these proteins was also effective in accelerating axon growth despite microtubule detyrosination levels having fallen below respective levels (Fig. 3A–C). Therefore, we cultured sensory neurons for 5 d after baculoviral transfection using the same shRNA constructs. Since differences in axon growth cannot be determined after this time because axons cover the entire floor of the culture dish, the cells were then replated to allow new axon growth. At that time, axotomized sensory neurons are in a preconditioned state and, consequently, show an increased speed of growth. Therefore, instead of after 48 h, axon growth and axon tip detyrosination were determined 24 h after replating. Consistent with ineffective detyrosination levels after this time, the knockdown of any of these proteins alone had decreased microtubule detyrosination in regrown axons to about. 65% of detyr+ axon tips (similar to ineffective levels of 10 nM parthenolide; Fig. 3B,F–H,M–O). Accordingly, these proteins’ prolonged and efficient knockdown did not increase axon growth (Fig. 3I). Moreover, adding 1 nM parthenolide to these cultures also showed no effect (Fig. 3I), while adding 10 or 100 nM reduced microtubule detyrosination and axon growth after VASH1 knockdown stronger than respective GFP controls (Fig. 3J–L). Also, the double knockdown of VASH1 and VASH2 reduced microtubule detyrosination stronger than each knockdown alone and significantly reduced neurite growth compared with GFP-treated controls (Fig. 3M–O). Thus, these data verify the importance of optimal detyrosination levels in axonal growth cones and demonstrate that a basic knockdown strategy is challenging for improving axon regeneration as it can reduce microtubule detyrosination beyond effective levels.
A 3 d VASH knockdown improves axon growth and suppresses the effects of parthenolide. A, Representative pictures of axon tips of adult murine sensory neurons after 3 d in culture. Cells were baculovirally transduced with GFP, VASH1sh (V1sh), VASH2sh (V2sh), or SVBPsh, and axons were stained for detyrosinated tubulin (detyr tub, red), and βIII-tubulin (tub, green). Three days after the VASH knockdown, detyrosination levels significantly fall below this level. Scale bar, 10 μm. B, Quantification of detyrosinated tubulin in axon tips of cultures described in A with additional time points at 1 and 2 d in culture. The dashed line represents the axon growth-promoting microtubule detyrosination level determined in Figure 1C–G with parthenolide. C, Quantification of detyrosinated tubulin intensity of axon tips described in B. D, Representative photos of 3 d cultured murine sensory neurons virally transduced with either GFP or shRNAs against VASH1 (V1sh), VASH2 (V2sh), or SVBP (SVBPsh). Grown axons were stained for tubulin (tub). Scale bar, 250 μm. E, Quantification of axon growth of sensory neuron cultures, as described in D. Data represent the mean ± SEM of four technical replicates. Each of the replicates is represented by a dot. The results were confirmed in an independent experiment. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. The lines connect the bars between which a significant p-value was determined. F, Representative photos of axon tips of murine sensory neurons virally transduced with either GFP or shRNA against VASH1 (V1sh), VASH2 (V2sh), or SVBP (SVBPsh). Neurons were replated after 5 d and treated with 1 nM parthenolide. Grown axons were stained for detyrosinated tubulin (detyr tub, red) and tubulin (tub, green) after an additional 24 h in culture. Scale bar, 5 μm. G, Quantification of detyr positive axon tips of replated sensory neuron cultures described in F. H, Quantification of detyrosinated tubulin intensity of axon tips described in G. I, Quantification of axon growth of replated sensory neuron cultures, as described in F. J, Quantification of detyrosinated tubulin in axon tips of sensory neuron cultures treated with 1, 10, or 100 nM parthenolide (Par) or a vehicle (0). Neurons were transduced with a VASH1sh (V1sh) expressing baculovirus or GFP as a control for a total of 6 d. K, Quantification of detyrosinated tubulin intensity of axon tips described in J. L, Quantification of axon growth of replated sensory neuron cultures, as described in K. M, Quantification of detyr positive axon tips of replated sensory neuron cultures, as described in F. Neurons had received either a VASH1, a VASH2, or a VASH1 and VASH2 knockdown or GFP as control. N, Quantification of detyrosinated tubulin intensity of axon tips described in M. O, Quantification of axon growth of replated sensory neuron cultures, as described in M. B, C, G–O, Data represent the mean ± SEM of separate experiments. A dot represents each of the experiments. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. The lines connect the bars between which a significant p-value was determined. The p-values above the bars represent the difference between the gfp control group and the bar on the respective day.
To confirm these findings in vivo, we developed a method to efficiently transduce sensory and motor neurons in mice. To this end, adult mice received a single intrathecal AAV1-GFP injection (Fig. 4A), which transduced, on average, 92.5% of sensory neurons of L3–L5 DRG (Fig. 4B,C). Moreover, AAV1-tdTomato injections, which resulted in a similar transduction rate, enabled the detection of red-fluorescent intradermal axon tips in the footpad (Fig. 4F). Additionally, 82.4% of neuromuscular junctions (NMJs) in the musculus extensor hallucis longus were innervated by transduced axons 2 weeks after injection (Fig. 4D,E), indicating efficient transduction of motor and sensory neurons.
VASH1 overexpression reduces functional sciatic nerve regeneration. A, Schematic drawing depicting intrathecally injected AAV1 transducing motor neurons (left) and sensory neurons (right), whose axons project through the sciatic nerve to muscles or the footpad skin, respectively. B, L4 DRG section immunostained for GFP (green) and βIII-tubulin (tub, red). Scale bar, 200 µm. C, Quantification of transduced DRG neurons after intrathecal AAV1-GFP injection, showing transduction rates of >90%. The data represent mean ± SEM (bars) and single values (dots) of individual animals (n = 9 animals). D, Immunostaining of the musculus extensor hallucis longus shows GFP-positive motor axons forming NMJs, identified by postsynaptic bungarotoxin-staining (BTX, red) 2 weeks after intrathecal AAV1 injection. Scale bar, 50 µm. E, Quantification of synapses innervated with GFP-positive axons 14 d after intrathecal AAV1-GFP injection, showing transduced innervation rates of 80%. The data represent mean ± SEM (bars) and single values (dots) of individual animals. F, Section of the footpads showing intradermal fibers from transduced sensory neurons (red) 2 weeks after intrathecal AAV1-tdTomato injection. The section was counterstained with DAPI. Scale bar, 50 µm. G, Photos of right hindpaws of mice before (0) or 1, 7, and 14 d (dpi) after SNC. Two weeks before SNC, a VASH1- or GFP-expressing AAV1 was injected intrathecally to transduce motor and sensory neurons projecting into the sciatic nerve. H, Quantifying motor recovery in adult mice as depicted in G, showing the SSI over 28 d after SNC. Each group contained six mice (n = 6). I, Quantification of sensory recovery in the same mice described in G and H determined with the von Frey test. J, Quantifying motor recovery in adult mice, using the SSI over 28 d after SNC. Two weeks before SNC, a VASH1 shRNA or GFP-expressing AAV1 was injected intrathecally to transduce motor and sensory neurons projecting into the sciatic nerve. K, Quantification of sensory recovery in the same mice described in J determined with the von Frey test. H–K, Data represent the mean ± SEM. Each hollow dot represents an individual animal for the respective time point. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. p-values are displayed for the time points with a significant difference for the two treatment groups.
We then used this approach for neuronal expression of either GFP (control) or VASH1 and performed a SNC 2 weeks afterward. Like in cultured sensory neurons, VASH1-overexpressing mice showed a significant delay in motor (assessed by the SSI) and sensory recovery after SNC (determined by the von Frey test; Fig. 4G–I). Additionally, we used this approach to knock down VASH1 expression by expressing a VASH1sh construct. Compared with GFP-expressing controls, VASH1 depletion did not affect sensory or motor sciatic nerve recovery after crush (Fig. 4J,K), resembling the in vitro findings shown in Figure 3I.
Microtubule detyrosination and axon growth-promoting effect of parthenolide is age dependent
As injured axons of young animals usually show a higher growth rate than adults (Pestronk et al., 1980; Niwa et al., 2002; Painter et al., 2014), we investigated whether parthenolide can also accelerate axon growth further in postnatal neurons. To this end, we cultured sensory neurons from postnatal (P3) and adult (P70) mice. Postnatal neurons expectedly showed significantly longer axons than adult neurons after 48 h in culture (Fig. 5A,B). However, parthenolide had no significant effect on axon growth in P3 cultures, while in P70 cultures, 1 nM of parthenolide increased axon growth to similar values as determined in postnatal cultures (Fig. 5A,B). We, therefore, analyzed the degree of microtubule detyrosination in axon tips. Strikingly, axonal growth cones from postnatal neurons showed markedly lower microtubule detyrosination than adult cultures (P70; Fig. 5C,D), suggesting higher microtubule dynamics in growth cones of postnatal neurons. We then tested to what extent parthenolide treatment affected the detyrosination in these axons. While at 1 nM, parthenolide significantly reduced microtubule detyrosination in neurons from P70 mice to similar levels as in naive postnatal neurons, it failed to decrease microtubule detyrosination in postnatal neurons further (Fig. 5C,D). Thus, the higher detyrosination levels in adult neuron axonal growth cones compromise axonal growth. Moreover, a shift of microtubule detyrosination into the range of physiological levels of postnatal neurons facilitates axon growth of adult neurons.
Postnatal neurons show reduced microtubule detyrosination in axon tips. A, βIII-tubulin (tub)-stained neurons were treated with either 0, 0.5, 1, or 5 nM par. Scale bar, 100 μm. B, Quantification of axon growth of neurons as depicted in A. Data from treated neurons were normalized to vehicle controls with an average axon length of 277 μm/neuron. Data show the mean ± SEM of three independent experiments (n = 3). Each of the three independent experiments is represented by a dot. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. The lines connect the bars between which a significant p-value was determined. C, Representative photos of axon tips of sensory neurons from postnatal (p3) and adult (p70) mice after 2 d in culture and treatment with either vehicle (−) or 1 nM parthenolide (par). Axons were stained for detyrosinated tubulin (detyr tub, red) and tub. Scale bar, 5 μm. D, Quantification of detyr tub in axon tips of cultured sensory neurons, as described in C. Data represent the mean ± SEM of three independent experiments (n = 3). Each of the three independent experiments is represented by a dot. p-values were determined using one-way ANOVA followed by the Holm–Sidak post hoc test. The lines connect the bars between which a significant p-value was determined.
Intravenous application of parthenolide or DMAPT promotes nerve regeneration
As the knockdown of VASH1/2 and SVBP appeared less favorable to reduce microtubule detyrosination to sciatic nerve regeneration-inducing levels, we next tested whether systemically applied parthenolide or one of its derivatives, DMAPT, is sufficient to facilitate axon growth in vivo. Therefore, we applied vehicle (PBS) or different doses of either drug into adult mice's tail veins daily, starting directly after SNC. Three days later, axon regeneration of superior cervical ganglion-10-positive (SCG10+) sensory neurons, choline-acetyltransferase-positive (CHAT+) motoneurons, and tyrosine hydroxylase-positive (TH+) sympathetic neurons were evaluated in immunostained longitudinal sciatic nerve sections (Fig. 6A–F). While low doses of parthenolide (0.2 µg/kg) and DMAPT (2 µg/kg) showed only a slightly improved axon regeneration, results were more substantial at 2 µg/kg for parthenolide and 20 µg/kg for DMAPT, respectively (Fig. 6A,B). At higher doses of 20 µg/kg, the beneficial effect of parthenolide was decreased again (Fig. 6A,B), confirming the bell-shaped concentration–response curve previously reported in cell culture (Gobrecht et al., 2016). Intravenous treatment with either parthenolide (2 µg/kg) or DMAPT (20 µg/kg) increased all neuronal subpopulations’ axonal lengths compared with respective vehicle-treated control groups (Fig. 6A–F). TH+-axons were the longest among these different fiber types. They reached up to 5 mm beyond the lesion site at this early time, indicating higher sympathetic neurons’ axonal regeneration rates compared with sensory and motor neurons.
Systemic parthenolide or DMAPT applications accelerate the regeneration of different neuronal types in various lesion models. A, Longitudinal sciatic nerve sections from adult mice stained with the sensory axon marker SCG10 3 d after crush injury. Animals had received daily injections of either vehicle (−), parthenolide (par), or DMAPT at indicated doses. Asterisks indicate the injury site. B, Quantification of regenerating axons at indicated distances beyond the injury site in the sciatic nerve of mice depicted in A. C, Longitudinal sections of the animals described in A stained for CHAT-positive motor axons with asterisks indicating the injury site. D, Quantification of regenerating axons at indicated distances beyond the injury site in the sciatic nerve of mice depicted in C. E, Longitudinal sections of the same animals described in A and C, stained for TH-positive sympathetic axons with asterisks indicating the injury site. F, Quantification of regenerating axons at indicated distances beyond the injury site of mice depicted in E. B, D, F, Data represent the mean ± SEM. Each dot represents four averaged nerve sections from an individual animal, and nerves from four animals were analyzed (n = 4). p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistically significant p-values are compared with the control group (−) of the respective distance from the crush site. Scale bar for A, C, E, 500 μm. G–I, Longitudinal sciatic nerve sections from adult mice stained with either sensory (SCG10, G), motor (CHAT, H), or sympathetic (TH, I) axon markers, 7 d after nerve transection and anastomosis. Mice had been intravenously treated daily with vehicle (−) or parthenolide (par, 2 µg/kg). Asterisks indicate the injury site. Scale bar, 1 mm. J–L, Quantification of regenerating sensory (J), motor (K), and sympathetic (L) axons at indicated distances beyond the injury site in the sciatic nerves as depicted in G–I, respectively. At least five sections per animal were analyzed from four animals per group. Data in J–L represent the mean ± SEM. Each dot represents four averaged nerve sections from an individual animal, nerves from four animals were analyzed (n = 4). p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistically significant p-values are compared with the control group (−) of the respective distance from the crush site.
As a systemic parthenolide application enabled axon growth promotion after a crush injury, we tested whether intravenously applied parthenolide enhances axon regeneration after complete nerve transection (neurotmesis) and anastomosis, where axonal regeneration is generally slower and less efficient than after nerve crush (de Ruiter et al., 2008; Ma et al., 2011; Diekmann and Fischer, 2016). To this end, we cut the sciatic nerve and performed surgical anastomosis. While controls, daily treated with vehicle, showed only a few short axons in the nerve's distal part 7 d after injury, intravenously injected parthenolide (2 µg/kg/d) significantly increased the length of the sensory, motor, and sympathetic axons (Fig. 6G–L). Thus, intravenous administration of parthenolide or DMAPT promotes the regeneration of sensory, motor, and sympathetic axons upon nerve injury.
Systemically applied parthenolide or DMAPT accelerate sensory and motor recovery in sciatic and median nerves
Next, we addressed whether intravenous application of parthenolide or DMAPT accelerates anatomic and clinically relevant functional recovery after nerve injury. To this end, we assessed motor and sensory function using the SSI and von Frey test, respectively, over 4 weeks after SNC (Gobrecht et al., 2014). Strikingly, daily repeated doses of parthenolide (2 µg/kg) or DMAPT (20 µg/kg) significantly improved the SSI score compared with vehicle-treated controls regarding both the onset of recovery and time to reach full recovery (Fig. 7A,B). In contrast to vehicle-treated controls, parthenolide- and DMAPT-treated mice showed the first measurable improvements in motor function 4 d after injury (Fig. 7A,B). In contrast, treatment effects concerning the touch response were first detectable 7 d after injury (Fig. 7C). Consistent with the accelerated motor recovery by parthenolide, re-established NMJs were found in the musculus extensor hallucis longus (EHL) of parthenolide-treated mice 4 d after SNC (Fig. 7D,E). In contrast, very few re-established NMJs were found in controls after this early time point. We also evaluated skin reinnervation in the footpads (Fig. 7F–I). In cross sections of the footpads, the axon numbers in the stratum spinosum (SS) dramatically dropped from an average of 28 per slice in uninjured mice to 1 when assessed 1 d after SNC (Fig. 7H,I), indicating extensive Wallerian degeneration. When evaluated 10 d after injury, parthenolide treatment significantly increased the number of axons in the SS layer compared with the vehicle-treated control (Fig. 7H,I). Thus, improved skin reinnervation was also correlated with parthenolide-mediated sensory recovery.
Intravenous application of parthenolide or DMAPT accelerates target reinnervation and functional recovery. A, Photos of right hind feet of mice before (0) or 1, 14, and 28 d (dpi) after SNC. Animals had received daily intravenous doses of either vehicle (−), 2 µg/kg of parthenolide (par), or 20 µg/kg DMAPT. B, Quantification of motor recovery in adult mice as depicted in A, showing the SSI over 28 d after SNC and daily intravenous injection with vehicle (−), parthenolide (par), or DMAPT. C, Quantification of sensory recovery in the same mice described in A and B was determined with the von Frey test. B, C, Each group contained six mice (n = 6), represented by the hollow dots. Data represent the mean ± SEM. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistical differences were calculated within each time point to compare the control (−) and each treatment group, respectively. The significant p-values are displayed in the color of the line of the treatment group. D, Representative pictures of musculus extensor hallucis longus (EHL) reinnervation 4 d (4 dpi) after SNC + parthenolide (par) or vehicle (−) application. Muscular whole-mounts were stained for neurofilament (NF, green) and α-bungarotoxin (BTX, red). Arrows indicate reinnervated synapses. Scale bar, 50 µm. E, Quantification of innervated synapses, as representatively depicted in D. Data represent mean ± SEM (bars) and single values (dots) of individual animals (n = 4). F, Areas of isolated footpads innervated by the sciatic (encircled) or saphenous nerves. G, Schematic drawing of a footpad cross section showing the stratum granulosum (SG), stratum spinosum (SS), and stratum basale (SB). H, Quantification of the footpad reinnervation as depicted in G. Of each mouse, 25 slices per footpad were evaluated. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. p-values are presented in comparison with the control group of the respective time point. I, Footpad sections stained for βIII-tubulin (red) and DAPI (cyan). Scale bar, 10 µm. E, H, Data represent the mean ± SEM, each dot an individual animal. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistical differences were calculated within each time point to compare the control (−) and each treatment group, respectively. The significant p-values are displayed in the color of the line of the treatment group.
We then tested whether orally applied parthenolide or DMAPT is also effective. Daily applied oral DMAPT doses of 20 µg/kg fully mimicked the effects of intravenously injected DMAPT of similar dosages, while orally administered parthenolide showed no effect (Fig. 8A–C). However, even higher oral doses of parthenolide (20 or 200 µg/kg) failed to improve nerve regeneration (Fig. 8D,E), reflecting poor oral bioavailability of parthenolide (Guzman et al., 2007).
Oral application of DMAPT but not parthenolide accelerates functional sciatic nerve recovery. A, Representative photos of right hindpaws of mice before (0) or 1, 14, and 28 dpi. Mice received daily doses of a vehicle orally, 2 µg/kg par or 20 µg/kg DMAPT. B, Quantifying motor recovery in adult mice as depicted in A by the SSI (n = 6). C, Quantification of sensory recovery in the mice described in A determined by the von Frey test (n = 6). D, Longitudinal sciatic nerve sections from adult mice stained with the sensory axon marker SCG10 3 d after crush injury. Animals had received daily injections of either vehicle (−) or parthenolide (par) at indicated doses. Asterisks indicate the injury site. E, Quantification of regenerating axons at indicated distances beyond the injury site in the sciatic nerve of mice depicted in D. At least five sections per animal were analyzed from five animals per group. F, Quantification of the grip strength of mice receiving either daily intravenous doses of vehicle (veh) or 2 µg/kg of parthenolide (par) over 14 d at the indicated time points before and after MNC. G, Quantifying sensory recovery in adult mice described in F, determined with the von Frey test at the indicated time points before and after SNC. B, C, E–G, Data represent the mean ± SEM, each dot an individual animal. p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistical differences were calculated within each time point to compare the control and parthenolide as well as DMAPT. The significant p-values are displayed for the respective time point.
Finally, we investigated whether the daily repeated intravenous applications of parthenolide (2 µg/kg) also promote other peripheral nerves’ regeneration. To this end, we crushed the median nerve of mice and evaluated regeneration by the grip strength of the front paw and the von Frey test. As for the injured sciatic nerve, parthenolide treatment significantly accelerated the motor and sensory of the front paw (Fig. 8F,G).
Parthenolide accelerates functional nerve recovery in rats
After finding that parthenolide and DMAPT accelerates functional recovery in mice upon nerve injury, we investigated whether these drugs are effective in other species, where axons must overcome longer regeneration distances to reach functional recovery. Therefore, we performed SNC in adult rats and treated them daily by intravenous applications of either vehicle or different parthenolide doses (0.2, 2, and 20 µg/kg/d), starting directly after SNC (Fig. 9A–C). While daily doses of 0.2 and 20 µg/kg/d showed significant but only moderate effects, efficacy was markedly higher at 2 µg/kg/d (similar to mice). The first effects could be seen in the SSI at this dosage 11 d after injury (Fig. 9B). In contrast, vehicle-treated rats showed the first signs of recovery after 16 d. Moreover, parthenolide-treated rats reached preinjury SSI scores after 35 d, while the control group required another 14 d to reach these levels (Fig. 9A,B). Sensory recovery was likewise accelerated upon intravenous parthenolide treatment (Fig. 9C). Also, daily doses of 2 µg/kg showed the most potent effects: Improvements in the touch response were first detectable in this group after 14 d, while only seen in controls 21 d after injury. Furthermore, the function was restored to presurgery levels by parthenolide after 35 d, whereas the vehicle group took ∼49 d (Fig. 9C). In contrast to mice, SNC led to hypersensitivity in all rat groups, manifested earlier in parthenolide-treated animals. However, 56 d after SNC, vehicle-treated controls and groups treated with 0.2 or 20 µg/kg reached the same level of hypersensitivity (Fig. 9C). Thus, optimal doses for intravenous parthenolide application in mice and rats were similar.
Systemic application of parthenolide promotes functional recovery in rats. A, Representative photos of right hind feet of rats before (0) or 1, 14, and 28 d after SNC (dpi). Adult rats received daily intravenous doses of either vehicle (0) or 2 µg/kg of parthenolide (par). B, Quantifying motor recovery in adult rats as depicted in A by the SSI after SNC and daily intravenous injections of various doses of 0.2 µg/kg, 2 µg/kg, or 20 µg/kg parthenolide over 49 d. C, Quantification of sensory recovery in the same rats described in B determined by the von Frey test. B, C, Data represent the mean ± SEM, each hollow dot an individual animal (n = 6). p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistical differences were calculated within each time point to compare the control (0 µg/kg) and each treatment group, respectively. The significant p-values are displayed in the color of the line of the treatment group. D, Representative photos of the right hindpaw of rats before (0) or 1, 14, and 21 dpi and different treatment regimes. Animals received daily treatment (intravenous injections) of either vehicle (veh) or par from days 0–5 or 6–28 dpi as indicated. Another group received parthenolide treatment from 0 to 28 dpi. E, Quantification of motor recovery in animals as depicted in D by the SSI over 42 dpi. F, Quantification of sensory recovery in the same rats described in D and E determined by the von Frey test. E, F, Data represent the mean ± SEM, each hollow dot an individual animal (n = 6). p-values were determined using two-way ANOVA followed by the Holm–Sidak post hoc test. Statistical differences were calculated within each time point to compare the control (veh veh) and each treatment group, respectively. The significant p-values are displayed in the color of the line of the treatment group.
Delayed parthenolide treatment still improves functional recovery in rats
From a clinical point of view, knowing whether a delayed start of parthenolide treatment or just an initial treatment for some days is sufficient to accelerate functional recovery is relevant. As the onset of functional recovery occurs later in rats than in mice, we used adult rats. We examined four groups where 2 µg/kg of parthenolide was intravenously administered daily using different treatment regimes. Animals of group 1 received the vehicle for 28 d; rats of group 2 received the vehicle for the first 5 d and then parthenolide for the remaining 23 d; animals of group 3 received parthenolide for the first 5 d and then the vehicle for the remaining 23 d; rats of group 4 received parthenolide daily for the entire 28 d. As expected, continuous treatment (group 4) showed the fastest motor and sensory recovery, and group 1 (vehicle only) was the slowest (Fig. 9D–F). Moreover, animals of group 2 with a postponed treatment showed a delayed onset of functional improvement (Fig. 9D–F) but only a slightly weaker effect at later time points than in group 4. Finally, animals of group 3, treated for the initial 5 d only, showed similar initial improvement as group 4 but were less effective than group 2 at later time points (Fig. 9D–F). Thus, although continuous parthenolide treatment over the entire period showed the best results, a delayed start of the treatment was still effective.
Discussion
Although peripheral nerves have, in principle, the ability to regenerate injured axons, functional recovery often fails to manifest due to a limited axonal growth rate. Thus, the clinical demand for drugs accelerating axon growth speed is high (Diekmann and Fischer, 2015, 2016). We have previously demonstrated that peripheral nerve regeneration in adult mice is accelerated in knockin mice with constitutively active GSK3 leading to inhibition of microtubule detyrosination (Gobrecht et al., 2014, 2016). Effects on microtubule detyrosination and axon growth in culture can be mimicked by the sesquiterpene lactone parthenolide (Gobrecht et al., 2016). Here, we further established the relevant role of optimal levels of microtubule detyrosination to accelerate axon growth upon injury and demonstrate that while a substantial reduction of detyrosination in growth cones has no effect or even compromises axon growth, a relatively low decline enhances axon extension (Figs. 1C–G, 5). However, the mechanism underlying the parthenolide effect on microtubule detyrosination and axon growth remained elusive. The enzymes VASH1 and VASH2 were recently identified as the first carboxypeptidases for microtubules (Aillaud et al., 2017; Nieuwenhuis et al., 2017; Li et al., 2019; Wang et al., 2019). Although parthenolide inhibits VASH1/2 activity (Aillaud et al., 2017; Li et al., 2019), a functional role in promoting axon growth remains elusive.
Here, we demonstrate that overexpression of VASH1 or VASH2, but not its enhancer SVBP, increases microtubule detyrosination and compromises mature neurons’ axon regeneration. Moreover, the knockdown of endogenous VASH1, VASH2, or SVBP reduced detyrosination levels in axonal tips. When sensory neurons were transduced immediately after plating and evaluated after 3 d, knockdown of each protein resulted in a marked enhanced axon regeneration, demonstrating the relevance of these enzymes on microtubule detyrosination in mature neurons and axonal growth. This was due to effective rates of detyrosination being achieved during the gradual decrease in protein expression within this period. This resulted in increased outgrowth compared with controls. However, replating experiments did not show any effect on axonal growth because, by the time of replating (6 d after transduction), the rates of detyrosination had already reached subeffective levels. These findings are, therefore, consistent with the bell-shaped concentration–response curve obtained with parthenolide in pharmacological studies. Since functional sciatic nerve recovery takes place over ∼4 weeks, a VASH1 knockdown in vivo likely lowers axonal detyrosination levels below the threshold optimal for axon growth. This resembles the abovementioned findings in vitro, where a 3 d VASH1 knockdown improved axon growth, while a 6 d knockdown kept growth at GFP levels.
Interestingly, SVBP overexpression did not affect axon tip detyrosination or growth, while overexpression of VASH1/2 reduced axon growth. This is most likely because SVBP alone does not affect microtubule detyrosination but acts by enhancing VASH activity. SVBP already activates most VASH molecules in cultured sensory neurons under normal conditions. Thus, an SVBP overexpression does not affect axon growth or detyrosination. However, if SVBP is knocked down, endogenous levels gradually fall below the amount of VASH molecules and stop enhancing their activity. This causes a reduction of VASH activity to the extent that microtubule detyrosination becomes optimal for axon growth. This occurred in our experiments after a 2 to 3 d knockdown. A 2 d VASH1, VASH2, or SVPB knockdown lowered detyrosination to growth-promoting levels, while a 3 d knockdown dropped these levels to an amount unfavorable for axon growth. A further knockdown, as reached in our 6 d cultures, led to suboptimal detyrosination and did not promote axon growth (Fig. 3I,L). Additionally, the results indicate that a modest shift in the ratio of tyrosinated/detyrosinated microtubules within the physiological range of postnatal neurons is sufficient to promote axon growth, as seen with the effect of parthenolide. Accordingly, parthenolide failed to significantly accelerate the axon growth of postnatal neurons further (Fig. 5). These results show that a gene therapeutic approach by VASH knockdown can only be achieved by a dosed knock-out. Here we realized this by limiting the knockdown time to 3 d. This, of course, is challenging in vivo. On the other hand, parthenolide still promoted axon growth of VASH1, VASH2, and SVBP overexpressing neurons. However, at least 10 times higher drug concentrations were required to overcompensate for the higher VASH1/2 activity. In contrast, effective concentrations for adult neurons (1 nM par) did not affect axon growth in VASH1/2 overexpressing cells (Fig. 1F–G). Based on the data presented in the current study and the fact that parthenolide binds to VASH1 and VASH2 (Li et al., 2019), we conclude that these proteins likely are the inhibited molecular targets mediating axon growth promotion. Although we cannot exclude that parthenolide also indirectly inhibits microtubule detyrosination by reducing the polymerization-competent pool of tubulin, as reported recently in HeLA cells (Hotta et al., 2021), this possibility seems unlikely in neurons since the formation of adducts on cysteine and histidine residues on tubulins was reported at concentrations >1,000 times higher (Hotta et al., 2021) than the concentrations used to promote axon regeneration in the current study. Furthermore, the low expression of VASH1/2 in neurons, the covalent binding of parthenolide to these proteins, and the required relatively moderate shift of microtubule detyrosination are likely the reason for its high potency at the optimal low dosages to accelerate axon regeneration in vivo. Thus, our data support a model in which VASH1 and VASH2 are bound to an excess amount of SVBP and thus detyrosinate α-tubulin. Partial VASH inhibition by optimal parthenolide concentrations (1 nM) or a short nonactivation through a SVBP knockdown leads to a growth-favoring level of microtubule detyrosination. However, higher parthenolide concentrations (100 nM) or a longer SVBP knockdown reduce microtubule detyrosination levels to an extent that even reduces axon regeneration. Also, the knockdown of either VASH for 6 d reduces microtubule detyrosination in the same manner as 10 nM parthenolide, thus not improving axon growth. Consequently, adding parthenolide to these neurons even decreased axon growth below the levels of VASH-expressing cells (Fig. 3J–L). A knockdown of both VASHs for 6 d decreases microtubule detyrosination to similar levels as 100 nM parthenolide and consequently reduces axon growth (Fig. 2J–L).
While this study shows a reduced microtubule detyrosination and consequently that increased dynamics are beneficial for axon growth in the PNS, in previous studies, microtubule stabilization via taxol improved axon regeneration in the CNS (Hellal et al., 2011; Sengottuvel et al., 2011). These seemingly opposing results in PNS and CNS can be explained by the fact that taxol treatment leads to disinhibition toward inhibitory CNS myelin and CSPGs, which is not required in the growth-supportive environment of the injured PNS. How parthenolide affects regeneration in the injured CNS (alone or in combination with disinhibitory treatments) is currently under investigation (Leibinger et al., 2023).
In addition, the current study's findings suggest that the reduction in microtubule detyrosination in postnatal sensory neurons compared with adult neurons may play a significant role in the faster nerve regeneration observed in young animals. So far, it was believed that axon extrinsic factors, such as diminished Schwann cell repair responses (Le et al., 2001; Painter et al., 2014), were responsible for the reduced regeneration in adult nerves. However, further research is needed to investigate whether aging is correlated with increased VASH1/2 or SVBP expression and to understand how and why the degree of detyrosination is physiologically regulated in different neuronal populations.
Currently, no therapeutics are clinically available to accelerate functional regeneration upon nerve injury. Therefore, our findings that intravenous injections or oral application of DMAPT accelerate recovery in different nerve injury paradigms and species are a promising step forward. In addition, the possibility for repeated applications over several weeks by these application routes is essential because, as shown in the current study, a continuous treatment was functionally more effective than a treatment over just a few days (Fig. 9D–F). Furthermore, even a delayed onset of ongoing parthenolide treatment was sufficient to accelerate functional recovery. As regeneration distances usually are more extended in humans than in rodents, this aspect would be particularly relevant for potential clinical use in patients. Although it is currently unknown whether both drugs can also promote nerve regeneration in humans, their high potency and low effective dosages of 2 µg/kg (parthenolide) or 20 µg/kg (DMAPT) in rats and mice make severe side effects unlikely, since DMAPT reportedly indicates no signs of toxicity or hematological parameters in mice when intravenously applied at 100 mg/kg (5,000 times higher than used in our study) over 10 consecutive days (Guzman et al., 2007). Furthermore, since DMAPT accelerated functional sciatic nerve recovery at 5,000-fold lower doses (20 µg/kg), possible severe side effects at these dosages are most likely even more reduced. Similarly, administration of parthenolide at 25 mg/kg thrice weekly over 4 weeks or daily over 17 d showed no signs of toxicity (Kishida et al., 2007; Zhang et al., 2009).
To improve compliance, oral drug administration is often preferable over intravenous treatment. Consistent with parthenolide's low oral bioavailability (Guzman et al., 2007), even much higher oral dosages of this drug remained ineffective. However, DMAPT, which reportedly has an oral bioavailability of ∼80% (Guzman et al., 2007), was effective at similar oral and intravenous dosages, making it particularly attractive as a medication. The higher required dosages of DMAPT compared with parthenolide are likely due to the covalent binding of parthenolide to a cysteine in the catalytic center of the VASH1/2 through C13. Thus, the catalytically active thiol group of C169 reacts with the α-methylene of parthenolide through a Michael addition reaction (Li et al., 2019; Freund et al., 2020). As the C13 of DMAPT is reversibly bound to an amine, the Michael addition must first dissociate back into parthenolide before becoming biologically active (Matsuda et al., 2003; Hwang et al., 2006; Neelakantan et al., 2009). Consistent with this, higher doses of the prodrug DMAPT were needed to accelerate axon growth in vivo. Although parthenolide has been shown to bind to VASH1/2 covalently, it cannot be excluded that parthenolide might also bind to other cysteine-based proteases. While this might affect axon growth, as well, our data (Figs. 1, 3) demonstrate that the interaction with neuronal VASH1 and VASH2 is essential for the axon growth-promoting effects.
In conclusion, our findings indicate that the systemic application of parthenolide or DMAPT significantly accelerates the functional recovery of injured nerves in various species. This opens up the potential for developing the first widely applicable medication to increase regeneration distances within a critical and limited time window for the reinnervation of peripheral targets. This can lead to faster recovery of motor function and sensation and, ultimately, a better outcome following traumatic nerve injuries. Additionally, these drugs may also be useful in treating nerve diseases that are caused by or associated with axonal damage, such as diabetic neuropathy (Landowski et al., 2016; Chandrasekaran et al., 2019), axonal Guillain–Barre syndrome, or chemotherapy-induced nerve damage. Research exploring these possibilities is ongoing.
Data Availability
Most data generated or analyzed during this study are included in this published article (and its supplementary information files). The remaining data are available from the corresponding author upon reasonable request.
Footnotes
The Federal Ministry of Education and Research supported this work (03VP05200). We thank Anastasia Andreadaki, Nadine Hube, Swenja Henne, Cora Fried, Christopher Brennsohn, Kessy Brzozowski, and Celina Böse for technical support.
D.F. and P.G. hold parts of patent number 11298337 titled “Parthenolide and its derivative for use in treating axonal damage.”
- Correspondence should be addressed to Dietmar Fischer at dietmar.fischer{at}uni-koeln.de.
