The effect of recombinant botulinum neurotoxin A on neuropathic pain in the spared nerve injury mouse model

Abstract

Neuropathic pain following traumatic nerve injury is a disabling chronic pain disorder characterized by sensory abnormalities such as mechanical allodynia. Botulinum neurotoxin type A (BoNT/A) has shown analgesic properties in a range of clinical pain conditions and in animal models. Here, we investigated analgesic efficacy of recombinant BoNT/A1 (rBoNT/A1; IPN10260) in the spared nerve injury (SNI) mouse model of neuropathic pain. Potential tissue site and mechanisms of action were explored by analyzing a series of pain biomarkers in the ipsilateral dorsal root ganglion (DRG) and the spinal cord. C57Bl6 mice received either SNI- or a sham surgery 14 days before being treated with either rBoNT/A1 or vehicle. Mechanical sensitivity was evaluated in von Frey tests performed at baseline and throughout the experiment. DRGs and spinal cords were collected for quantitative microscopy of immunohistochemically labelled pain-related targets. rBoNT/A1-injection resulted in significant and prolonged (up to 14 days) increases in mechanical threshold compared to vehicle in SNI-operated mice. Volume of type B DRG neurons and number density of Iba1-positive cells in DRG were significantly increased in the SNI-operated animals in comparison to the sham-operated controls, however no significant effect of rBoNT/A1 could be demonstrated. Among spinal cord biomarkers, no effects were observed. These results demonstrate that rBoNT/A1 reduces mechanical allodynia following peripheral nerve injury, but the mechanisms remain elusive. Investigating these biomarkers in a challenged system (diabetes, chemotherapy, etc.) might extend the window of activation, possibly better exposing analgesic mechanisms of rBoNT/A1.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00702-025-02978-0.

Introduction

Neuropathic pain is caused by a lesion or disease affecting the somatosensory system including peripheral sensory fibers (Aβ, Aδ and C fibers) and central nervous system (CNS), and affects 7–10% of the general population (Colloca et al. 2017; Karlsson et al. 2015). Traumatic nerve injury, is a common cause of neuropathic pain (Dualé et al. 2014; Sansone et al. 2015). Due to the aging global population, improved survival following chemotherapy and increased diabetes mellitus incidence, the incidence of neuropathic pain is likely to increase. However, available treatment options are limited and provide only modest pain relief (Finnerup et al. 2015), often accompanied by dose-limiting adverse effects. Hence, novel analgesic drugs without systemic side effects are urgently needed (Attal et al. 2016).

Botulinum neurotoxin type A (BoNT/A), produced by Clostridium botulinum, cleaves SNAP-25 (synaptosomal associated protein 25 kDa) essential for neurotransmitter release (Wang et al. 2017). BoNT/A has a growing number of clinical indications, including limb spasticity, dystonia, and migraine (Steward et al. 2021). BoNT/A has been shown to have analgesic effects in a range of conditions in human patients, including peripheral neuropathic pain (Attal et al. 2016). The dorsal root ganglia (DRG) and spinal cord are proposed as potential sites of analgesic efficacy of BoNT/A following its retrograde axonal transport (Antonucci et al. 2008; Marinelli et al. 2012; Matak and Lacković 2014; Périer et al. 2021). However, the mechanisms mediating the analgesic properties of BoNT/A remain largely unexplored.

Injury of a peripheral nerve results in a cascade of inflammatory mediators resulting in an inflammatory response throughout the inflicted nerve, the corresponding DRG and innervated spinal cord segment. Immune and glial cells such as macrophages and satellite glia cells (SGC) in the DRG, and microglia and astrocytes in the spinal cord are involved in protection of neural tissue in naïve condition. However, these cells may initiate release of cytokines following nerve injury, enhancing the inflammatory response and contributing to development and maintenance of neuropathic pain. Peripheral nerve fibers conducting pain and temperature (C-fibers) and proprio- and mechanosensory signals (Aδ-fibers), originate from type B and type A neurons, respectively. Cell volume regulation reflects pathophysiological changes, consequently, peripheral nerve damage affects cell volume of type A and B DRG neurons. The major inhibitory neurotransmitter GABA plays an important role in regulating nociceptive input, and downregulation of GABA_A receptor (GABA_AR)-subunits has been observed following peripheral nerve injury (Wang et al. 2021). Disinhibition in the spinal dorsal horn is widely accepted as part of the neuropathic pain pathophysiology. The potassium chloride cotransporter 2 (KCC2) in the spinal cord regulates the GABA_AR–subunit composition. KCC2 expression is decreased following nerve damage, thereby negatively affecting spinal inhibitory neurotransmission (Coull et al. 2003).

Here, we investigate potential analgesic mechanisms of a recombinant BoNT/A1 (rBoNT/A1) in the DRG and the spinal cord by studying a series of the pain-related biomarkers in SNI- or sham-operated animals.

Methods

Animals

Male C57BL/6JRj mice (Janvier Labs, Le Genest-Saint-Isle, France), aged 8–10 weeks, weighing 20–26 g at the start of the experiment were used. The mice were acclimatized for one week before the start of the experiment. Animals were group-housed (4 per cage) in type III cages (Tecniplast 1290D Euro standard, 425 × 276 × 153 mm) and maintained on a 12:12 h light–dark cycle (lights on from 07.00 to 19.00 h), 23 ± 2 °C temperature and 55 ± 5% humidity with food and water ad libitum. Bedding material (Tapvei 2HV and LBS soft paper wool) were changed weekly. An overview of the study design is shown in Fig. 1a, and the experimental timeline in Fig. 1b. All experiments were approved by the Danish Animal Experiment Inspectorate (permission No: 2017-15-0201-01192) and performed in full compliance with the ARRIVE guidelines, European Communities Council Directive 2010/63EU.

Fig. 1.

(a) The study design. rBoNT/Al = recombinant botulinum neurotoxin A. Vehicle = Gelatine phosphate buffer (GPB) (b) The timeline of the experiment. Day -1 von Frey = Baseline measurements. Surgery = SNI-surgery. Injection = Intraplantar injection of either 3.2 pg rBoNF/Al or the vehicle (GPB)

Botulinum neurotoxin

Recombinant BoNT/A1 (rBoNT/A1; IPN10260) was provided by Ipsen Bioinnovation (Milton Park, UK). rBoNT/A1, synthesized using the native amino acid sequence expressed in Escherichia coli and characterized in vitro and ex vivo, was found to be biochemically and functionally comparable to that of native BoNT/A (nBoNT/A1) (Périer et al. 2021). Furthermore, rBoNT/A1 was found similar to nBoNT/A1 in in vivo potency, onset, and duration of action in the mouse and rat digit abduction score assays following a single intramuscular administration (Périer et al. 2021). rBoNT/A1 showed efficacy in the complete Freund’s adjuvant (CFA) model of inflammatory pain in mice (Oehler et al. 2022). In the study by Oehler et al., rBoNT/A1-injected animals showed reduction in CFA-induced allodynia in comparison with vehicle-injected controls, whereas Ca²⁺ fluctuations in DRG neurons were similar in two groups. Following, suggesting that analgesic activity of rBoNT/A1 in inflammatory pain is likely mediated by central rather than peripheral mechanisms (Oehler et al. 2022).

Study design

Initially, a pilot study was performed to establish the dose of rBoNT/A1, which was free of systemic effects. As in a pilot intraplantar injection (IPL) of 4.0 pg/mouse rBoNT/A1 diluted in 20 µl gelatin phosphate buffer (GPB) resulted in a mild reduction in body weight, (data not shown), for the main study the dose was reduced to 3.2 pg/mouse and administered to 32 mice. Sixteen mice were subjected to unilateral SNI in left hind leg (ipsilateral), while the remaining 16 were unilaterally sham-operated. In each group, eight animals received IPL injection of 3.2 pg/mouse rBoNT/A1 and eight animals received IPL vehicle in the left hind paw (Fig. 1a). The contralateral side (right) was left intact. Injections were performed 14 days post-surgery allowing a stable pain threshold to manifest. Tissue samples were collected 14 days post injection (i.e., day 28 post-surgery), as our aim was to assess the sustained neurobiological effects. A schematic representation of the experimental design and the timeline is shown in Fig. 1. Two follow-up studies with 24 mice per study were performed using the same treatment regimen but with tissue harvest performed 6 days post-surgery/3 days post-injection, 21 days post-surgery/7 days post-injection or 28 days post-surgery/14 days post-injection. The objective of the follow-up studies was to investigate possible microglia and astrocyte alterations at different time points relative to both SNI or sham surgery and injection of rBoNT/A1 or vehicle. Treatment was assigned randomly using the online tool Research Randomizer (www.randomizer.org, version 4.0) (Urbaniak 2013–2021). The treatment groups were: SNI + vehicle, SNI + rBoNT/A1, sham + vehicle and sham + rBoNT/A1. All experiments, not including surgery, were performed in a blinded fashion and by the same investigator.

Peripheral nerve injury

The animals were subjected to spared nerve injury (SNI), which is described in detail elsewhere (Richner et al. 2011). Briefly, the common peroneal and tibial branches of the sciatic nerve were unilaterally ligated with non-absorbable suture (polypropylene 6 − 0, Ethicon) and cut distally from the ligation, just distal to the branching point of the sural nerve, which was left untouched. For sham surgery, the sciatic nerve was visualized but left intact. Wounds were closed with tissue adhesive (Klinibond). Sevoflurane gas (AbbVie A/s, Copenhagen, Denmark) was applied with the UNO Basic Anaesthesia set-up for small rodents (UNObv). Analgesia: Lidocaine SAD (10 mg/ml; AstraZeneca, Cambridge, United Kingdom) applied on wound; buprenorphine (Temgesic, 0.3 mg/ml; INDIVIOR Limited, Berkshire, United Kingdom); and antibiotics (Ampicillin STADA, powder dissolved to 250 mg/ml in isotonic saline, STADA Nordic). Temgesic and Ampicillin were diluted and mixed 1:10 in isotonic saline (9 mg/ml, Fresenius Kabi, Bad Homburg, Germany), 0.1 ml was injected subcutaneously. Surgery was performed 14 days prior to injection of rBoNT/A1 or vehicle.

von Frey measurements

Mechanical sensitivity was assessed in the von Frey test, and mechanical allodynia reported as reductions in the paw withdrawal threshold (PWT), as described previously (Richner et al. 2011). Von Frey filaments (Semmes-Weinstein monofilaments, Stoelting; Wood Dale, USA) were applied to the lateral part of the ipsilateral (injured) hind paw in ascending order five times over the total period of 30 s starting with a 0.02 g filament. After repeating this procedure on all remaining mice on the test grid, the contralateral (non-operated) paw of the first mouse was tested with the same filament. Following, the next filaments were applied in the same way. A positive reaction, (i.e., the nociceptive threshold), was defined as either paw withdrawal, flinching or paw licking in response to three out of five repetitive stimuli. Baseline von Frey measurements were performed 1 day prior to SNI/sham-surgeries (day 0). To reduce person-to-person variation as well as sex and odour bias the same researcher performed the testing throughout the experiment (Sorge et al. 2014). Mice were allowed minimum 15 min of habituation before testing, which were always performed during the light phase.

Dissection and tissue preparation

Spinal cords and DRG were harvested by dissection after deep anesthetisation and decapitation (Richner et al. 2017). To keep the spinal cords as intact as possible, harvesting was performed using the hydraulic extrusion technique. After cutting the skin open the spinal column was isolated and trimmed to make the spinal cord visible both rostrally and caudally. A syringe holding ice-cold PBS with an adjusted pipette tip was inserted in the caudal end of the spinal column and the spinal cord was extruded into a petri dish containing PBS on ice. Next, the spinal column was cut open and the ipsilateral L4 DRG was carefully removed, and transferred to a petri dish on ice filled with PBS (Richner et al. 2017). Spinal cords and DRG were post fixed in 4% paraformaldehyde (PFA). The lumbar enlargement of the spinal cord was isolated and transversely cut in three equally sized sections. Sections from DRG and spinal cord were afterwards immersion-fixed in 4% phosphate-buffered formaldehyde, dehydrated in increasing concentrations of ethanol ending with xylene before subsequently being embedded in paraffin. An automated rotary microtome (RM2255, Leica) was used to cut paraffin blocks containing spinal cords and DRG in parallel series of 2 μm and 3 μm sections, before mounting two series of three sections each on a Superfrost Plus Microscope slide. Sciatic nerve on the ipsilateral side was dissected out and fixed in 2% glutaraldehyde, rinsed in PBS, and transferred to 1% osmium tetroxide for 1 h. The nerve tissue was dehydrated in increasing levels (50%, 70%, 90% and 99%) of dimethylformamide and embedded in reactive LR-white resin overnight. An ultramicrotome (Leica EM, UC6) was used to cut a transverse section at 70 nm and placed on 2 × 1 mm Cu grids.

Tissue preparation for histochemistry, immunohistochemistry, and immunofluorescence

Spinal cord and DRG: Iba1 and GFAP

Sections were dried at 60°C for 30 minutes. Next, the slides were deparaffinized: 2 × 15 minutes immersed in Histo-Clear, then in decreasing solutions of ethanol (99%, 96%, 70%), and finally rinsed shortly in distilled water. Endogenous peroxidase activity in the tissue was blocked by incubating the slides in tris-buffered saline (TBS) with 3% H₂O₂. TBS with H₂O₂ is made in own laboratory by following reagents: Tris-HCl (T5941, Sigma Aldrich) and NaCl (1367, Chemsolute). Antigen retrieval was performed using heat-induced epitope retrieval (HIER). After washing the slides in 0.2% milk and 0.3% Triton-X in TBS, the slides were incubated overnight in a moisturized incubation chamber at 4°C in a TBS plus 0.3% Triton-X solution with either rabbit anti-iba1 antibody (019-19741, Wako) at a 1:500 / 1:1000 (DRG / spinal cord), or polyclonal rabbit anti-GFAP antibody (Z0334, Dako) at 1:4000 concentration. The slides were then washed in TBS before incubation at room temperature for 75 minutes in a solution of TBS plus 0.03% Triton-X with HRP-conjugated goat anti-rabbit IgG antibody (PO448, Dako) at 1:400. Detection of the secondary antibodies was done after a TBS washing by incubating the slides with 3.3’-diaminobenzidin (DAB) dissolved in TBS and H2O, followed by washing with TBS with 0.03% Triton-X. To obtain a nuclear counterstain, the slides were submerged in Mayer’s haematoxylin for 2 min, with a 10-minute removal of excess haematoxylin in tap water afterwards. Following tissue dehydration in increasing solutions of ethanol (70%, 96%, 99%) and 2 × 10 min immersion in xylene, the slides were mounted with coverslips using Eukitt mounting medium.

Spinal cord: cSNAP25

Staining was performed as outlined in “Iba1 and GFAP’’ with a few differences. Buffer used for washing contained TBS plus 0.05% Tween 20 (pH 7.6). PT-Link (PT10126, Dako), a pre-treatment system, was used for deparaffinization, rehydration and epitope retrieval. The slides were placed in an EDTA buffer (Dako, K8004, pH 9) and received a pre-treatment program with an epitope retrieval step lasting 20 minutes at 121°C. Primary antibody solution contained polyclonal rabbit anti-cSNAP25 antibody (EF14007, Ipsen) (Périer et al. 2021) at 1:4000 concentration in an antibody diluent (2202, Dako). The next day the slides were reheated at 37°C for one hour before incubation with a secondary antibody solution containing the buffer and biotinylated goat anti-rabbit IgG antibody (BA-1000, Vector laboratories Inc.) at 1:400 concentration for 30 minutes. Before detection of the secondary antibodies with DAB, amplification was made by incubation with Vectastain^® Elite^® ABC Reagent (PK-6100, Vector laboratories Inc.) for 30 min at room temperature.

Spinal cord: KCC2

Staining was performed as outlined in “Iba1 and GFAP” with a few differences: The buffer used for washing contained 0.3% Triton-X in TBS, 5% goat serum and 1% bovine serum albumin (BSA). The buffer used to dilute the primary antibody, polyclonal rabbit anti-KCC2 antibody (07-432, Millipore) at 1:1000 concentration, contained 0.3% Triton-X in TBS and 1% BSA. After incubation in a dark moisturized chamber for 1 h with the secondary antibody, Alexa Fluor 568 conjugated goat anti-rabbit IgG antibody (A1011, Invitrogen) at 1:400 concentration, slides were washed in TBS and cover slipped using 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) and a fluorescence mounting medium (S3023, Dako).

Spinal cord: GABA_AR α1 and GABA_AR α3 subunits

Staining was performed as outlined in “Iba1 and GFAP” with the following differences: Primary antibody solution contained monoclonal mouse anti-GABA_AR α1 antibody (ab94585, Abcam) at 1:200 concentration and polyclonal rabbit anti-GABA_AR α3 antibody (ab224214, Abcam) at 1:500 concentration diluted in TBS with 0.3% Triton 0.2% milk. The slides were incubated for 2 h in a dark moisturized chamber with a secondary antibody solution containing Alexa Fluor 488 conjugated goat anti-mouse IgG antibody (A11001, Invitrogen) at 1:400 concentration and Alexa Fluor 568 conjugated goat anti-rabbit IgG antibody (A11011, Invitrogen) at 1:400 concentration. Following a wash in TBS the slides were cover slipped using 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) and a fluorescence mounting medium (S3023, Dako).

DRG: GFAP

Staining was performed as outlined in “Iba1 and GFAP” with a few important differences. Primary antibody solution contained polyclonal rabbit anti-GFAP antibody (Z0334, Dako) at 1:1000 concentration in 1% bovine serum albumin (BSA) plus 0.3% Triton-X solution. Incubated overnight. Next, the slides were washed in TBS before incubation in a dark moisturized chamber a room temperature for 1 h in a solution of BSA/Triton-X buffer with anti-rabbit 568 antibody (835724, Invitrogen) at 1:400 concentration. The slides were washed in distilled H2O with added DAPI for 10 min and the slides were mounted with coverslips using fluorescence mounting medium (S3023, Dako).

DRG: toluidine blue

Slides were dried and deparaffinized before immersion in 0.1% toluidine blue for 15 min, then in increasing grades of ethanol, and finally mounted with coverslips.

Microscopy and quantification

Number density estimation of Iba1- and GFAP-labelled cells in the spinal cord

A NanoZoomer 2.0 HT scanner (Hamamatsu, Japan) was used for acquiring super images (an image mosaic consisting of many individual field of views) of the Iba1- and GFAP stained slides using a 20X/0.75 lens before visualizing the stained tissue on a computer running VIS image analysis software (Visiopharm, Hoersholm, Denmark). Sampling and quantification were performed using the automatic physical disector module in VIS. The grey matter border of neighbouring sections was aligned either manually or by the software, and furthermore divided into left and right side to compare densities between the two sides. A counting frame measuring 96.4 μm x 72.6 μm was systematically distributed in a grid over the grey matter, with each frame spaced 125 μm apart in both X and Y dimensions (Fig. 2a). Sampling of fields of view was performed with the 40X lens. Counting frames were paired and were superimposed on the sampled images of both the reference section and the lookup section, thus allowing quantification in both sections after alignment of the counting frames. Criteria for counting an astrocyte or a microglia cell was the same. Only cells with a clearly defined nucleus and with a DAB-stained soma completely contained by the frame or touching the green inclusion line, and not present in the lookup section were counted. Cells either in contact with the red exclusion line or present in both reference and lookup sections were not counted (example in Fig. 2a). Counting frame corners were used as test points. The number density of microglia and astrocytes was estimated as:

Fig. 2.

Photomicrograph of (a) Spinal cord sections with Ibal-labelled microglia (left) and of GFAP-labelled astrocytes (right) with overlaid counting frames, (b) Type A and type B neurons with overlaid 2D unbiased counting frame and 3D nucleator (c) GFAP- labelled satellite glia with overlaid 2D unbiased counting frame and 3D nucleator

where gm equals grey matter. Nv (cells/gm) is the number density of either microglia or astrocytes per grey matter volume. ΣQ−(cells) is the total number of microglia or astrocyte counts. The area per test point, a/p, is the area of the counting frame divided with number of test points per frame. The number of test points hitting grey matter is denoted ΣP(gm). Quantification took place in both directions, hence, multiply by 2 and T is the section thickness.

Quantification of cSNAP25 in the spinal cord

A screening to trace and visualise the manifestation of rBoNT/A1 after intraplantar injection in the ipsilateral hind paw of the mice was performed. The expected visualisation of the toxin in the spinal cord following immunohistochemistry was unilateral staining in lumbar spinal cord after intraplantar injection in left hind paw, based on recent suggestions of retrograde axonal transport of the toxin (Antonucci et al. 2008; Marinelli et al. 2012; Matak et al. 2012).

Number density estimation of Iba1-labelled cells in DRG

Image acquisition was done using a microscope with an Olympus BX61VS 20X/0.40 lens connected to a computer running Olympus VS-ASW virtual slides system (Tokyo, Japan). Sampling, quantification, and counting were performed as summarised for Iba1- and GFAP-labelled cells in the spinal cord but with a 60X lens.

Volume estimation of type A and B neurons in DRG

Three sections per ganglion were used for estimation and nucleoli were used as the unique point. Volume estimation were obtained using the combination of modified disector-sampling and 3D nucleator (Braendgaard and Gundersen 1986; Møller et al. 1990; Tandrup 1993) assuming isotropy of neurons. Volume estimation was carried out with the help of a computer running VIS and a microscope system consisting of an Olympus BX51 (Tokyo, Japan) mounted with an Olympus XC50 camera and a high-precision Proscan III (Prior Scientific Instruments Ltd., UK) motorized microscope stage. Sampling and estimation were performed using the micro imager module in VIS with a counting frame area of 7000 µm², systematically distributed at random, with each frame spaced 125 μm apart in both X and Y dimensions (Fig. 2b). The objective used was a UpansApo 60X/1.35 oil. Neuronal types were distinguished based on published criteria (de Moraes et al. 2017; Tandrup 2004). Type A neurons were large, with irregular cytoplasmatic staining and large, heavily stained nucleus Type B neurons were smaller in size, with a darker and more evenly stained cytoplasm and a smaller nucleus.

GFAP-expressing SGC profiles per neuronal area in DRG

The slides were imaged on a microscopy setup consisting of a Leica DM6000B microscope with a CTR6000 motorized 4-slide stage and a MAC 5000 XY Stepper Motor Stage controller (Ludl electronic products ltd.) and a MAC 5000 XY joystick control unit (Ludl electronic products Ltd.). Image acquisition was done using a Leica N plan 60X/1.35 lens, connected to a computer running the virtual images microimager module in VIS and counting frame as described above. The 2D nucleator in VIS was used to measure the mean neuronal area, and expressed SGC profile number were counted (Fig. 2c) as:

where is the number of GFAP-expressing SGC profile number per neuronal area, and ΣQ(SGC) is the total number of expressed SGC profile number.

Quantification of GABA_AR α3 intensity in DRG

The used images were similar to those described in “Expressed SGC profiles per neuronal area”. Using the ROI tool, measurement of mean grey value was performed in the NIH freeware program ImageJ (version 2.0.0) to quantify fluorescence intensity.

Quantification of KCC2 intensity in spinal cord

The slide were photographed as outlined in “Expressed SGC profiles per neuronal area”, and image acquisition was done using a Leica N plan 20X/0.40 lens. The microscope was connected to a computer running VIS analysis software. Image analysis was performed using the NIH freeware program ImageJ (version 2.0.0) (Rasband 1997–2018) (Supplementary Fig. 1a). To quantify fluorescence intensity of immunolabeled KCC2 in the superficial dorsal horn of the lumbar spinal cord, laminae I and II were outlined bilaterally as described in (Paxinos G 2009) using the ROI tool of ImageJ. Mean fluorescence intensity (MFI) was calculated as previously reported (Little et al. 2012) using the following equation:

where i is the mean grey value, pp is the positive pixel area, and p is total pixel area. Before the analysis, the images received background threshold corrections through the automated ImageJ default threshold function. To describe the potential change in fluorescence intensity MFI was calculated as mean fold change (MFC) by dividing the fluorescence intensity of the ipsilateral superficial dorsal horn with the fluorescence intensity of the contralateral side. Data are expressed as arbitrary units.

Quantification of GABA_AR subunits in spinal cord

Photo and image acquisition is outlined in “Quantification of KCC2 intensity” (Supplementary Fig. 1b). Determination and quantification of the fluorescence level of the immunolabeled GABA_AR subunit α3 were done by adaptation of a method based on an original protocol, published from QBI, The University of Queensland, Australia (Fitzpatrick 2014). The following equation was applied to calculate the adapted corrected total cell fluorescence (aCTCF):

where IntDen is the integrated density (the product of area and mean grey value), area is either the left or right half of the spinal cord grey matter utilising the ROI tool of ImageJ, and mean_NegCont is the mean fluorescence of the negative control. Initially smooth continuous background noise was removed from the images by the background subtraction function of ImageJ. aCTCF is reported as MFC (ipsilateral half divided by the contralateral half).

Measurement of sciatic nerve parameters using transmission electron microscopy

Image acquisition was done using a Jeol JEM-1400 plus TEM (Tokyo, Japan) setup. Systematic uniform random sampling of field of views (FOVs) was performed using SerialEM software (Mastronarde 2005) and measurements of axon area, area myelin thickness and C-fiber area were performed in microimager module of VIS.

Statistical analysis

Data are expressed as the Mean ± SD for all tests and observations unless otherwise stated. Data were analysed using one-way ANOVA and two-way repeated measures ANOVA with post hoc test as appropriate. The significance level was set at p < 0.05.

Results

von Frey measurements

To investigate the analgesic effect of rBoNT/A1 on induced neuropathic pain we performed SNI-or sham surgery, a gold standard of experimentally induced mechanical allodynia, before administering either rBoNT/A1 or vehicle. The development of mechanical allodynia was assessed through von Frey measurements. SNI operated mice demonstrated a robust decrease in PWT following surgery reflecting development of mechanical allodynia compared to sham operated mice [p < 0.0001, (day 2)] (Fig. 3). rBoNT/A1 treatment reversed allodynia, as SNI operated animals treated with rBoNT/A1, showed increased PWT in comparison with vehicle treatment starting from day 15 up to day 28 [p = 0.008 (day 15), p = 0.0009 (day 16), p = 0.01 (day 20), p = 0.006 (day 24), p = 0.002 (day 28)]. In sham operated animals rBoNT/A1 did not alter PWT in comparison with vehicle treatment. There was no effect of treatment on PWT on the naïve contralateral paw (data not shown). We observed no difference in body weight between the treatment and vehicle groups throughout the experiment (Fig. 4).

Fig. 3.

Mechanical sensitivity (PWT) of mice that were exposed to SNI or were sham-operated and treated with either vehicle (GPB) or rBoNT/A1. ***p < 0.001, **p < 0.01 SNI+rBoNT/A1 vs SNI+Veh. p < 0.0001 comparison of PWT after SNI or sham operation. Two way repeated measures ANOVA with post hoc Tukeys test. Data are presented as Mean ± SEM. Scissors: SNI or sham surgery was performed on day 0. Syringe: injection rBoNT/A1 or vehicle was performed on day 14. GPB, gelatin phosphate buffer; SNI, Spared Nerve Injury; PWT, paw withdrawal threshold

Fig. 4.

Mean percent (%) body weight change in rBoNT/A1-injected vs. vehicle-injected mice exposed to SNI or sham treatment. Independent t-test. Data are presented as Means ± SEM. Day 0 = SNI or sham surgery. Injection = intraplantar injection of rBoNT/A1 or vehicle (day 14). SNI, spared nerve injury

DRG biomarkers

As cell volume regulation is an important homeostatic function, defining not only cell shape, but also cell growth, cell migration, and cell death (Hoffmann et al. 2009), it reflects pathophysiological changes following e.g. nerve damage (Lang 2007). Type A and type B neurons correspond to two groups of DRG neuronal somas differentiated according to thier morphology (Lawson 1979). Type A neurons are mainly proprio- and mechanosensory, whereas type B neurons are mainly nociceptive (Harper and Lawson 1985). To determine whether the spared nerve injury and following rBoNT/A1 injection was reflected in DRG type A and B neurons, estimation, and comparison of cell volume across the four treatment groups were made. There was a significant difference in the volume of type B neurons, but not type A neurons across the four groups (Fig. 5a and b). SNI operated vehicle injected mice exhibited a significant increase in type B neuron volume compared to sham operated mice [p = 0.03]. No significant difference in DRG type B neuron volume was found when comparing vehicle to rBoNT/A1 injected SNI operated mice (Fig. 5b).

Fig. 5.

Comparison of DRG markers across four treatment groups (a) A-type neuron volume (p =0.669), (b) B-type neuron volume (p = 0.028), (c) Number density of Ibal positive cells (p = 0.006) and (d) Expressed satellite glia cells per neuronal area (p = 0.138) +# are significantly different from +, #. One-Way ANOVA. post hoc: LSD, Data are expressed as Mean±SD. n = 8, rBoNT/Al injection: 14 days post nerve injury period. Sample collection: 14 days post injection period

DRG macrophages have been described to contribute to the mechanical hypersensitivity that characterizes the neuropathic pain phenotype (Raoof et al., 2020), (Yu et al. 2020). To explore this contribution of the SNI model to the allodynia, and the antiallodynic effect of rBoNT/A1, the number density of Iba1-labelled cells in the DRG was estimated. SNI significantly increased the number of Iba1-labelled cells in the DRG in comparison to sham treatment [p = 0.006] (Fig. 5c). Iba1 number density did not differ between rBoNT/A1- or vehicle-injected SNI operated mice.

There is good evidence that the SGC play an active role in the initiation and maintenance of neuronal changes that underlie neuropathic pain (Matsuka et al. 2020; Ohara et al. 2009), as SGC alterations have been documented in response to both injury and inflammation (Gazerani 2021) in different animal models of chronic pain (Costa and Moreira Neto 2015), including increased GFAP expression, as sign of SGC activation, following peripheral nerve injury (Hanani 2012; Stephenson and Byers 1995; Woodham et al. 1989). To study GFAP expression following SNI and subsequently rBoNT/A1 injection, GFAP positive SGCs per neuronal area in DRG was estimated. No significant differences could however be demonstrated among the four treatment groups (Fig. 5d).

GABA_AR in the DRG are proposed to play a significant part in the development of neuropathic pain following peripheral nerve injury, through weakening of the peripheral GABAergic mechanisms. To investigate the effect of SNI and rBoNT/A1 injection, type A, and type B GABA_AR α3 subunit expression intensity in DRG were measured. Comparison of mean gray value of GABA_A α3 receptors of both type A and type B DRG neurons in the four treatment groups demonstrated no significant differences (Supplementary Figs. 2a and 2b).

Spinal cord biomarkers

The participation of microglia and astrocytes in CNS are crucial to the etiology and maintenance of neuropathic pain (Scholz and Woolf 2007). To assess the involvement of microglia and astrocytes in BoNT/A mechanism of action the number density of Iba1- and GFAP-labelled cells in the spinal cord was estimated. The density of Iba1-labelled microglial cells was, however, similar across the four treatment groups (Fig. 6a). The four groups of mice had similar mean number density of GFAP-labelled astrocytes (Fig. 6b). A comparison of the ipsilateral and contralateral sides of the spinal cord presented, as a trend, higher number density of microglia and astrocytes on the operated, without significant side-to-side differences in either of the treatment groups (data not shown). Two follow-up experiments in which tissues were harvested 6 days post-surgery/3 days post-injection and 21 days post-surgery/7 days post-injection or 28 days post-surgery/14 days post-injection were performed. A significantly higher number of activated microglia in SNI operated mice compared to sham operated mice after early tissue harvest (6 days post-surgery) were demonstrated [p = 0.005] (Supplementary Fig. 4). No significant findings were observed following rBoNT/A1 injection. von Frey results in follow-up experiments were similar to the primary experiment (data not shown).

Fig. 6.

Comparison of spinal cord markers across four treatment groups (a) Mean number density of Ibal labelled microglia (p =0.207), (b) Mean number density of GFAP labeled astrocyte (p = 0.903). (c) KCC2 Mean fluorescence intensity (MFI) (p = 0.753) and (d) GABA_a R sub-units a3 adapted corrected cell fluorescence (aCTCF) p = 0.598;One-Way ANOVA. Data are expressed as Mean±SD

Disinhibition in the spinal dorsal horn has been portrayed as a key element of neuropathic pain (Colloca et al. 2017). The potassium chloride cotransporter 2 (KCC2) in the spinal cord has been linked to the development of neuropathic pain (Coull et al. 2003) by regulating the expression of GABA_A-receptor (GABA_AR)- subunit α3 (Ortinski et al. 2004) affecting GABA-neurotransmission. To explore the effect of rBoNT/A1 on these receptors their expression in the spinal cord were quantified by measuring and comparing fluorescence intensity and mean fold-change. However, no difference in mean fold-change or fluorescence intensity of KCC2 or GABA_AR subunits across the groups were demonstrated (Fig. 6c and d).

BoNT/A generates a novel epitope known as cleaved SNAP25 (cSNAP25) (Rossetto and Montecucco 2008). We stained for cSNAP25 by utilizing an antibody specific for this epitope. Staining was present in the spinal cord of rBoNTA1-injected animals, while being absent in vehicle-injected animals. Specifically, we detected unilateral staining of cSNAP25 in the ipsilateral dorsal horn (laminae II and III) of the lumbar spinal cord following intraplantar injection in the left hind paw (Fig. 7).

Fig. 7.

Lumbar spinal cord with cSNAP25 labelling in the dorsal horn. Left: 10X magnification. Right: 40X magnification (blue box of left image)

Sciatic nerve parameters

Degeneration of axons and demyelination are hallmarks of exposure to chemical toxins, mechanical trauma, and inadequate perfusion. Demyelination and sprouting of sensory fibers have been demonstrated together with occurrence of allodynia (Ueda 2011). To determine whether this contributes to SNI induced allodynia and is affected by rBoNT/A1 injection we measured sciatic nerve parameters including axon area, area myelin thickness and C-fiber area using transmission electron microscopy (TEM). When comparing the results of all the treatment groups, no significant differences in the axon area, area of the myelin thickness and C-fibers were found (Table 1, Supplementary Fig. 3).

Table 1.

Pain markers across four experimental groups. Columns are treatment groups and rows are pain markers. MFC = mean fold change (ipsilateral/contralateral). MGV = mean grey value. Group with the same letter = statistically significant difference. Comparison through one-way ANOVA analysis

	SNI + vehicle	SNI + rBoNT/A1	Sham + vehicle	Sham + rBoNT/A1
Iba1 cells (10− 3 mm3)	22.5	18.2	20.2	20.5
GFAP cells (10− 3 mm3)	38.3	39.4	38.8	40.1
KCC2 (MFC)	1.06	1.05	1.02	0.99
GABAAR alfa 1 (MFC)	1.46	1.35	1.36	1.21
GABAAR alfa 3 (MFC)	1.07	1.17	1.09	1.23
Volume of type A neurons (102 µm3)	14.6	13.9	12.9	12.8
Volume of type B neurons (102 µm3)	3.29 (ab)	3.00	2.32 (a)	2.33 (b)
Iba1 (10− 3 mm3)	5.48 (ab)	4.77	4.22 (a)	4.31 (b)
SGC (10− 3 µm2)	4.36	2.48	2.58	2.67
GABAAR alfa 3 type A neurons (MGV)	8.26	7.93	6.72	6.89
GABAAR alfa 3 type B neurons (MGV)	8.03	7.42	6.33	6.80
Axon area without myelin (µm2)	23.2	19.3	24.1	25.0
Myelin thickness area (µm2)	25.2	20.5	25.4	23.5
Axon area with myelin (µm2)	73.6	60.2	74.8	72.0
C-fiber area (µm2)	8.93	8.49	10.27	8.83

Discussion

Early studies suggest peripheral analgesic mechanisms of BoNT/A (Durham et al. 2004), while subsequent ones point at central effects (Aoki 2005) and cSNAP25 presence in the spinal cord (Matak et al. 2014, 2019). The pain biomarkers explored in this study offer sensitive tools to investigate central vs. peripheral activity of BoNT/A. The antiallodynic effect of rBoNT/A1 was seen starting 1 day after its injection. Analysis of microglia and astrocyte numbers in the spinal cord did not demonstrate their activation following SNI surgery. Thus, activity of rBoNT/A1 activity on glial cells could not be demonstrated. Levels of GABA_AR subunits in DRG and in the spinal cord, together with KCC2, remained largely unchanged in the four treatment groups. In SNI operated mice the volume of type B neurons, and the number of DRG macrophages and were significantly increased compared to those in sham-operated animals. However, no reduction following rBoNT/A1 injection were demonstrated. The number of GFAP-expressing DRG remained largely unchanged. Lastly, no differences in sciatic axon area, area of the myelin thickness in C-fibers were found.

Mechanical sensitivity

The rapid and significant reduction in ipsilateral PWT of SNI operated mice compared to sham operated mice, confirms development of mechanical allodynia. The rapid action of rBoNT/A1 is supported by its early neuronal uptake, inhibition of excitatory signalling, and modulation of peripheral and central pathways. In line with previous reports (Guida et al. 2020) mechanical allodynia was sustained throughout the experiment (28 days). LC/E-BoNT/A (a BoNT/A gene fused to the light chain (LC) of BoNT/E) reduced mechanical hyperalgesia in a rat SNI model (Wang et al. 2017), while BoNT/A reduced allodynia in the chronic constriction injury (CCI) models in mice and rat models (Vacca et al. 2013; Marinelli et al. 2010; Mika et al. 2011). Here, rBoNT/A1 increased PWT compared to vehicle, indicative of antiallodynic effects, lasting until day 28 (14 days post-injection). Similarly, BoNT/A effects lasting 12, 21 and up to 81 days have been demonstrated in CCI mouse and rat models (Vacca et al. 2013; Marinelli et al. 2010; Mika et al. 2011), and up to 14 days in a SNI rat model (Wang et al. 2017), albeit at higher doses (75 and 15 pg for rats and mice, respectively). While previous studies have reported contralateral analgesia following unilateral intraplantar BoNT/A, attributed to retrograde transport and central action (Bach-Rojecky et al. 2010; Marinelli et al. 2010; Favre-Guilmard et al. 2017), we observed no such effect in our neuropathic model. Although cSNAP25 was detected in the spinal cord, suggesting central engagement, we found no corresponding changes in contralateral mechanical sensitivity or spinal pain markers. This may reflect differences in model type, toxin formulation (recombinant vs. native), or timing. Our findings suggest that central effects of rBoNT/A1 may be context-dependent and not universally expressed across pain models. Here, the mice receiving rBoNT/A1 showed normal weight gain, confirming absence of confounding factors and no systemic spread of the toxin in the current study.

DRG

A significant increase in number density of Iba1-positive DRG macrophages of SNI operated mice indicates their involvement in mechanical allodynia, as shown previously (Cobos et al. 2018; Vega-Avelaira et al. 2009; Yu et al. 2020), however no indications of rBoNT/A1 interaction were found. Studies have shown that nerve injury induces different changes in SGC (Gazerani 2021) in different animal models of chronic pain (Costa and Moreira Neto 2015; Stephenson and Byers 1995; Woodham et al. 1989). Nevertheless, our results do not imply GFAP-expression of the SGC to be part of the SNI-induced allodynia or the mechanisms of rBoNT/A1 as the number did not differ significantly across groups. SNI operated mice also exhibited a significantly increased volume of type B neurons, but not type A, suggesting cellular alterations. As altered cell volume regulation generally contributes to the pathophysiology of pain (Lang 2007; Plesnila N 2017), this is potentially another mechanism behind the mechanical allodynia measured - a mechanism, however, not involved in the effect of rBoNT/A1. Type A and B neurons also contain GABA_AR which are proposed to be important in the development of neuropathic pain after peripheral nerve injury (Naik et al. 2008). Experimental data demonstrated weakening of spinal (Zhang et al. 2018; Peirs et al. 2015) and peripheral GABAergic circuits in models of pain (Wang et al. 2021). The observed presynaptic intracellular expression of GABA_A receptor α3 subunits is similar in the four treatment groups, hence no modulatory role of rBoNT/A1 was found.

Spinal cord

In general, a higher number density of microglial cells and astrocytes was exhibited on the injured side compared to the naïve side. Yet, this tendency was the same in both SNI- and sham-operated animals. We found no significant difference in microglia and astrocyte numbers after rBoNT/A1 injection. Follow-up experiments implied early activation of microglia in SNI mice, but no attenuation of astrocyte or microglia activation following rBoNT/A1 injection was shown. Decreased Iba1 levels in spinal cord (Zychowska et al. 2016), both unchanged (Zychowska et al. 2016) and reduced GFAP-levels (Vacca et al. 2013) are described effects of BoNT/A injection on mechanical allodynia related to CCI. The activity of rBoNT/A1 on allodynia of the SNI model does not indicate involvement of reduced microglia and astrocyte activation. One explanation to this discrepancy could be the different time points of toxin administration relative to surgery; SNI activates both astrocytes and microglia (De Luca et al. 2016; Luongo et al. 2010; Costigan et al. 2009) and early microglia activation was indicated in follow-up experiments, but not apparent in the main experiment, potentially masking an attenuating effect of rBoNT/A1.

No difference of KCC2 or GABA_AR subunits expression in spinal cord was seen across four treatment groups. GABA_AR is controlled by KCC2 through regulation of GABA_AR-subunits (Ortinski et al. 2004). Rescuing or upregulation of KCC2 expression, and selective GABA_AR-agonists (Malan et al. 2002) has been shown to reduce mechanical allodynia (Mapplebeck et al. 2019; Sánchez-Brualla et al. 2018). As GABA_AR has been suggested to be involved in the central antinociceptive effect of BoNT/A (Drinovac et al. 2014, Drinovac Vlah et al. 2016), further investigation of this proposed connection might prove insightful.

Here, we showed presence of cSNAP25 in the dorsal horns of the spinal cord upon peripheral injection of rBoNT/A1. Retrograde axonal transport of BoNT/A has been suggested after tracing the toxin with an antibody specific for the generated epitope cSNAP25, (Antonucci et al. 2008; Marinelli et al. 2012; Matak and Lacković 2014; Oehler et al. 2022; Périer et al. 2021). An unilateral injection of BoNT/A in an infraorbital nerve constriction rat model, demonstrated bilateral reduction of allodynia, indicative of a central effect (Filipovic et al. 2012). No significant difference in contralateral PWT between the four treatment groups were demonstrated in our experiment following injection. However, there was no significant difference between PWT contralaterally at the time of injection, hence impairing assessment of a possible contralateral effect of rBoNT/A1 in the SNI model.

Sciatic nerve

No differences in sciatic axon area, area of the myelin thickness and C-fibers were found, thus no indications of rBoNT/A1 affecting remyelination following nerve injury. BoNT/A reduces mechanical hyperalgesia but also improves functional recovery in CCI mouse models (Marinelli et al. 2010, 2012). However, these studies didn’t measure myelin thickness directly, but instead quantified the expression of regeneration proteins, as early as 24 h after BoNT/A administration. Hence, there is a possibility that this stage of remyelination has passed before our samples were collected.

Limitations

We saw no change in pain-related biomarkers when comparing SNI and sham groups, which made it challenging to interpret the lack of rBoNT/A1 activity. Previous reports on the SNI model suggest that we might simply have missed the right window of microglia (Zhou et al. 2014) and astrocyte (Guida et al. 2015) activation and KCC2 downregulation (Sánchez-Brualla et al. 2018) in the spinal cord. Demonstration of allodynia development in CCI and SNI models without depletion of GABA_A-receptors (Polgár et al. 2004; Polgár and Todd 2008) correlates however with our findings. The window of activation of the biomarkers in SNI mice models might last less than 4 weeks, possibly explaining why we did not observe similar results. Pain-related behaviour was only evaluated via mechanical sensitivity through von Frey testing. Incorporating non-reflexive pain assays (e.g., facial grimace, burrowing, conditioned place preference) in future studies may enrich the behavioural characterization of BoNT/A.

Spinal cord and DRG experiments

Plausible involvement of DRG targets in SNI-induced mechanical allodynia was primarily demonstrated in significant increase in volume of type B neurons and number density of Iba1-positive cells. However, no clues to a possible antiallodynic effect of rBoNT/A1 injection were found in either spinal cord or DRG targets. While it is difficult to assess rBoNT/A1 activity regarding our spinal cord aims and two of our DRG aims (GABA_A-R and DRG), it is interpretable that rBoNT/A1 effect does not involve volume of type B neurons and number of Iba1-positive cells in DRG. Recently, in a model of inflammatory pain, cSNAP25 was found in the spinal cord but not in DRG following rBoNT/A1 injection, suggesting central mechanisms (Oehler et al. 2022). Hence, any potential modulation of DRG pain-related biomarkers might be indirect. Although our findings do not fully support the involvement of central mechanisms, the detection of cleaved SNAP25 (cSNAP25) in the spinal cord confirms that rBoNT/A1 can undergo retrograde axonal transport and reach central sites. This highlights the importance of further studies to clarify the relative contributions of peripheral and central pathways in rBoNT/A1-induced analgesia. Additionally, we examined microglial and astrocytic activation at 4 weeks post-SNI to assess neuroinflammatory status during the maintenance phase of neuropathic pain. While microglial activation has been shown to persist for over 3 months and remains functionally significant during chronic pain states (Echeverry et al. 2017; Kohno et al. 2022), we acknowledge that this time point may be beyond the peak of glial activation, potentially contributing to the lack of significant changes observed. Future studies using earlier or longitudinal assessments, or employing models with prolonged neuroinflammatory responses (e.g., diabetic or chemotherapy-induced neuropathy), may better capture glial dynamics and help unmask central mechanisms of rBoNT/A1 action.

Conclusion

The study found an increase in ipsilateral PWT in SNI operated mice following rBoNT/A1 injection compared to vehicle injection, indicating an alleviating effect of rBoNT/A1 on mechanical allodynia lasting up to 14 days. The presence of cSNAP25 in the spinal cord hints to activity preceded by retrograde axonal transport. However, quantitative analysis of the other pain markers in the spinal cord did not reveal any cellular or molecular changes further establishing rBoNT/A1s potential central activity. As mechanical allodynia was not reflected in all pain-related biomarkers explored in this experiment, possible interactions between rBoNT/A1 and these biomarkers were undetectable, hence further investigation of these aims at different time points relative to SNI surgery and/or injection(s) of rBoNT/A1 is important.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Hammer R. E.: Drafted of the article. Substantial contributions to the design, as well as acquisition, analysis, and interpretation of the data and the final approval of the article to be published. Omoniyi A. A.: Acquisition, analysis, and interpretation of the data, as well as substantial contributions to drafting of the article and the final approval of the article to be published. Richner M.: Acquisition of the data, substantial contributions to revising of the article and final approval of the article to be published. Lesmi S.: Acquisition of the data, substantial contributions to revising of the article and final approval of the article to be published. Josefsen S. S.: Analysation and interpretation of the data, substantial contributions to revising of the article and final approval of the article to be published. Vaegter C. B.: Interpretation of the data, substantial contributions to revising of the article and final approval of the article to be published. Kalinichev M.: Substantial contributions to revising of the article, interpretation of the data and final approval of the article to be published. Karlsson P.: Substantial contributions to the design and revising of the article and final approval of the article to be published. Nyengaard J. R.: Substantial contributions to the design, drafting and revising of the work, as well as interpretation of the data and final approval of the work to be published. All authors discussed the results and commented on the manuscript.

Funding

This work was supported by Ipsen Innovation, Les Ulis/Saclay, France. Recombinant BoNT/A (rBoNT/A1; IPN10260) was provided by Ipsen Bioinnovation (Milton Park, UK). We would like to thank Vincent Martin, employee of Ipsen, who was involved in the generation of some data presented in the article (c-SNAP25 IHC staining in the spinal cord).

Declarations

Disclosures

S. Lezmi and M. Kalinichev were employees of Ipsen at the time the research for this project was conducted.