Abstract
Nerve growth factor (NGF) is a key mediator of nociception, acting during the development and differentiation of dorsal root ganglion (DRG) neurons, and on adult DRG neuron sensitization to painful stimuli. NGF also has central actions in the brain, where it regulates the phenotypic maintenance of cholinergic neurons. The physiological function of NGF as a pain mediator is altered in patients with Hereditary Sensory and Autonomic Neuropathy type V (HSAN V), caused by the 661C>T transition in the Ngf gene, resulting in the R100W missense mutation in mature NGF. Homozygous HSAN V patients present with congenital pain insensitivity, but are cognitively normal. This led us to hypothesize that the R100W mutation may differentially affect the central and peripheral actions of NGF. To test this hypothesis and provide a mechanistic basis to the HSAN V phenotype, we generated transgenic mice harboring the human 661C>T mutation in the Ngf gene and studied both males and females. We demonstrate that heterozygous NGFR100W/wt mice display impaired nociception. DRG neurons of NGFR100W/wt mice are morphologically normal, with no alteration in the different DRG subpopulations, whereas skin innervation is reduced. The NGFR100W protein has reduced capability to activate pain-specific signaling, paralleling its reduced ability to induce mechanical allodynia. Surprisingly, however, NGFR100W/wt mice, unlike heterozygous mNGF+/− mice, show no learning or memory deficits, despite a reduction in secretion and brain levels of NGF. The results exclude haploinsufficiency of NGF as a mechanistic cause for heterozygous HSAN V mice and demonstrate a specific effect of the R100W mutation on nociception.
SIGNIFICANCE STATEMENT The R100W mutation in nerve growth factor (NGF) causes Hereditary Sensory and Autonomic Neuropathy type V, a rare disease characterized by impaired nociception, even in apparently clinically silent heterozygotes. For the first time, we generated and characterized heterozygous knock-in mice carrying the human R100W-mutated allele (NGFR100W/wt). Mutant mice have normal nociceptor populations, which, however, display decreased activation of pain transduction pathways. NGFR100W interferes with peripheral and central NGF bioavailability, but this does not impact on CNS function, as demonstrated by normal learning and memory, in contrast with heterozygous NGF knock-out mice. Thus, a point mutation allows neurotrophic and pronociceptive functions of NGF to be split, with interesting implications for the treatment of chronic pain.
- allodynia
- dorsal root ganglia
- hereditary sensory and autonomic neuropathy type V
- learning and memory
- nerve growth factor
- skin innervation
Introduction
Nerve growth factor (NGF; Levi-Montalcini, 1987), in addition to its classical neurotrophic actions, is an established mediator of pain pathophysiology. Indeed, NGF is a key molecule in the pain machinery, with powerful proinflammatory and sensitizing actions (Denk et al., 2017). NGF regulates pain perception in humans and pain-related behavioral responses in animals; injections of NGF in animals (Lewin et al., 1993) and humans (Svensson et al., 2003) elicit rapid and long-lasting hyperalgesia, while NGF-neutralizing molecules are effective analgesics in many models of persistent pain (McMahon et al., 1995; Ugolini et al., 2007). NGF is strategically positioned to regulate pain, acting on both nociceptors (Lewin et al., 2014) and attraction of immune cells at injury sites (Levi-Montalcini et al., 1996; Skaper, 2017). NGF levels increase in affected tissues in various experimental or pathological inflammatory conditions (Minnone et al., 2017). Based on this, anti-NGF antibodies are being clinically evaluated to alleviate chronic pain (e.g., due to osteoarthritis; Lane et al., 2010). On the other hand, its robust neurotrophic properties have led to testing NGF as a therapeutic candidate for different conditions including diabetic polyneuropathy (Apfel et al., 1994; Pittenger and Vinik, 2003) and neurodegenerative diseases (Mitra et al., 2019). However, these trials failed due to significant adverse effects, including NGF-induced pain (Apfel, 2002). Thus, elucidating NGF actions on pain transmission, also taking into account signaling through TrkA and p75NTR, would be a step toward overcoming these limitations.
In this regard, Hereditary Sensory Autonomic Neuropathies (HSANs) are rare congenital pain insensitivity diseases that offer an opportunity to dissect NGF roles in pain. Indeed, several mutations leading to pain insensitivity have been described in the genes encoding NGF and TrkA (tropomyosin receptor kinase A; Indo, 2001; Capsoni, 2014; Nahorski et al., 2015). For example, recessive mutations in the TrkA gene cause HSAN IV, characterized by insensitivity to pain, autonomic defects, and mental retardation (Indo, 2001). A Swedish family suffering from severe loss of pain, leading to frequent bone fractures and Charcot joints, was discovered to carry, instead, the 661C>T transition in the NGFb gene, resulting in the R100W mutation in mature NGF. This “painlessness” disorder was called HSAN V (Einarsdottir et al., 2004). Compared with HSAN IV patients, homozygous HSAN V patients display a similar congenital indifference to noxious stimuli, but no cognitive deficits (Einarsdottir et al., 2004). In contrast, heterozygous carriers, despite reduced skin innervation and unmyelinated fiber number, along with altered thermoception, do not present with readily detectable clinical signs and have been identified only through pedigree and genetic screening (Axelsson et al., 2009; Minde et al., 2009; Perini et al., 2016).
We and others have shown that the NGFR100W protein displays reduced binding to, and signaling via, p75NTR, whereas interaction with TrkA is unaffected (Covaceuszach et al., 2010; Capsoni et al., 2011; Sung et al., 2018). Thus, we proposed that NGFR100W, with its biased TrkA agonist receptor profile (Covaceuszach et al., 2010; Capsoni et al., 2011), might help in dissecting trophic and nociceptive actions of NGF. To elucidate how these molecular features concur to determine the clinical HSAN V phenotype, we describe here the characterization of a mouse knock-in line harboring the NGFR100W mutation. We focused on heterozygous NGFR100W/wt mice, since homozygous NGFR100W/R100W mice die by the first month of life (Testa et al., 2019).
We demonstrate that heterozygous NGFR100W/wt mice display impaired nociception, despite having normal dorsal root ganglion (DRG) neurons. The NGFR100W protein has a reduced capability to activate pain-specific signaling, correlating with a reduced ability to induce mechanical allodynia. Surprisingly, however, NGFR100W/wt mice, unlike heterozygous mNGF+/− mice, show no learning or memory deficits, despite reduced NGF secretion. Together, our results provide significant insights into the molecular pathogenesis of the HSAN V phenotype and demonstrate a specific effect of NGFR100W on nociception, with no impact on cognitive performance. These features make NGFR100W an attractive tool to manipulate pain sensitivity and to exert neurotrophic actions in the absence of pain sensitization effects.
Materials and Methods
Ethics statement on mouse experiments.
All animal procedures were approved by the Italian Ministry of Health and were fully compliant with Italian (Ministry of Health guidelines, Legislative Decree n°26/2014) and European Union (Directive n°2010/63/UE) laws on animal research. The experiments were performed in strict accordance with the ARRIVE guidelines (Animal Research: Reporting in Vivo Experiments). In addition, the principles of the Basel Declaration, including the “3R” concept, have been considered throughout the whole project. Both male and female mice were used in this study.
Generation of knock-in human NGFR100W/wt mice.
pCMV6-XL5-human NGFwt plasmid was obtained from OriGene (catalog #SC123827) and pCMV6-XL5-human NGFR100W was generated using site-specific mutagenesis PCR. The targeting constructs were generated using classical cloning technologies. Briefly, a BAC (bacterial artificial chromosome; clone RP24-160F12) containing the entire regions of interest flanking the NGF sequence, was used to generate intermediate plasmids by cloning in pBluescript SK(−) the 5′ homology arm (from 89489 to 94076, restriction site MfeI) and 3′ homology arm (from 94803 to 99710, restriction site HindIII). The human NGFR100W coding sequence was cloned in the pKO2.1 targeting vector carrying the DTA (diptheria toxin A) negative selection cassette (provided by Dr. L. Ronfani, San Raffaele Hospital, Milan, Italy). The targeting vector was transfected into R1p.15 cells (background SV129), and positive clones were selected using neomycin resistance.
Southern blot analysis.
Genomic DNA was extracted by means of phenol:chloroform:isoamyl alcohol from ∼350 cell clones electroporated with NGFR100W targeting vector. The purified DNA was first incubated with EcoRI (for 5′ screening), then positive clones were confirmed by XbaI digestion (for 3′ screening). Digestions were run in a 0.8% agarose gel overnight (O/N) at 50 V. After a mild depurination and denaturation, gels were blotted on nitrocellulose and filters were incubated with an external 5′ or 3′ probe. The 5′ probe labels a 10.7 kb EcoRI band in the wild-type (WT) allele and a 5.2 kb EcoRI band in recombinant allele (Fig. 1A). The 3′ probe labels an 8 kb XbaI band in the WT allele and a 12.7 kb XbaI band in the recombinant allele (Fig. 1B).
Positive clones were injected into C57BL/6 mouse blastocysts, and chimeric animals were obtained.
Mice were genotyped by PCR. The following PCR primers were used: fw_human: 5′-TTTAGCACCCAGCCTCCCCGTGAAG-3′; fw_mouse: 5′-CAGAAGGAGACTCTGTCCCTG-3′; and rev_human-mouse: 5′-CACCTCCTTGCCCTTGATGTCTG-3′.
Band sizes are as follows: wild-type, 400 bp; and mutant 200 bp (Fig. 1C).
Behavioral analyses.
Experiments were performed on NGFwt/wt, NGFR100W/wt, mNGF+/+, and mNGF+/− mice. Mice were kept under a 12 h light/dark cycle, with food and water available ad libitum.
Hot plate test.
Mice were placed on a surface heated from 42°C to 54°C with 3°C steps. Animals were sequentially tested, allowing a 10 min resting period between each temperature step. The temperature threshold required to observe paw licking and the time required to observe this reaction at each temperature step were recorded.
Cold sensitivity test.
Mice were put in a plastic cage and habituated for 30 min. Acetone (50 μl; Sigma-Aldrich) was sprayed onto the plantar surface of the hindpaw using a Gilson pipette, and the responses were reported as a 4-point score: 0 = no response; 1 = brisk withdrawal or flick of the paw; 2 = repeated flicking of paw; and 3 = repeated flicking of the hindpaw with licking. Acetone was applied six times, alternating between paws, with an interval of 5 min between each application. The frequency of response, expressed as a percentage (number of trial characterized by a response/total number of trials) was evaluated.
Capsaicin injection test.
Mice were placed individually in a Plexiglas box for 15 min before drug injection to allow habituation. Capsaicin (catalog #141000, Abcam) was dissolved in dimethylsulfoxide (DMSO) and injected into the ventral surface of the right hindpaw using a Hamilton syringe at a concentration of 3 μg/μl in saline solution (total injection volume, 10 μl; DMSO final concentration, 0.1%). Control mice were injected with 10 μl of 0.1% DMSO in saline. Following the injection, mice were observed for 15 min and the amount of time spent licking and/or lifting the injected paw was measured.
Object recognition test.
The apparatus consisted of a PVC arena (60 × 60 × 30 cm) with white floor and black walls. The test was performed in 3 d. On day 1, mice were subjected to a habituation phase in which they received two 5 min sessions in the empty arena, separated by a 30 min interval. On day 2, mice were exposed to two identical objects for 7 min to evaluate the total time of exploration. On day 3, mice were placed back in the arena and exposed to a familiar object and another novel object (memory phase). The time spent exploring each object was recorded.
Morris water maze.
The test was performed in a water tank (120 cm diameter) filled with white opaque water. The platform, placed in the center of the southwest quadrant, was submerged 1 cm below the water surface. The 6-month-old mice were trained with two trials per day, with a 40 min interval, for 9 consecutive days. Mice were allowed up to 2 min to locate the platform, and the latency to reach it was recorded. If the mouse failed, the experimenter guided it onto the platform. Data were acquired and analyzed using an automated tracking system (Ethovision XT, Noldus).
Tape response assay.
Mice were habituated to a Plexiglas container for 5 min, and then a 3 cm piece of adhesive tape was applied to the back. Mice were observed for 5 min to measure the latency to the first tape removal attempt and the total number of attempts.
Cotton swab assay.
Mice were placed in an arena consisting of an elevated chamber with a grid floor and were allowed to habituate for 1 h. A cotton swab was stroked through the floor along the plantar paw surface five times, alternating between paws with a 10 s interval. The number of withdrawals was counted and expressed as a percentage of the total number of trials.
In vivo nociceptive assay.
As reported in the study by Capsoni et al. (2011), CD1 male mice (Charles River, Italy) were subjected to a mechanical allodynia behavioral test after the injection of either WT or R100W NGF in the hindpaw plantar surface at a 0.2 μg/μl concentration (corresponding to 4 μg in a total injection volume of 20 μl) in saline. Control mice were injected with 20 μl of saline. The von Frey test (Ugo Basile, Italy) was performed before treatment and 1, 3, 4, and 5 h postinjection.
NGF treatment.
Mouse wild-type NGF was administered daily at the dose of 1 μg/kg to pregnant dams by subcutaneous injection; treatment was protracted until 10 d after delivery. From postnatal day 10 (P10) to P60, pups received a daily subcutaneous injection (1 μg/kg) and intranasal administration (480 ng/kg) of NGF. A 21 d washout period was allowed before subsequent experiments.
Dorsal root ganglion neuron primary cultures.
DRG neurons were prepared from neonatal (5-d-old, P5) Wistar rats (Charles River, Italy) from both sexes, as reported in the studies by Bonnington and McNaughton (2003) and Taneda et al. (2010). Briefly, DRGs were collected, incubated for 1 h at 37°C with 0.125% collagenase (Sigma-Aldrich), mechanically dissociated, and plated onto coverslips or Petri dishes pretreated with 10 μg/ml poly-l-lysine (Sigma-Aldrich), at a density of 50,000 cells/well of a 48-well tissue culture plate. DRG cultures were maintained in serum-free medium, consisting of DMEM/F12 (Invitrogen) supplemented with 87.5 ng/ml 5-fluoro-2′-deoxyuridine, 37.5 ng/ml uridine, 50 U/ml penicillin, and 50 μg/ml streptomycin (all from Sigma-Aldrich) and 0.05% Invitrogen N2 supplement (Thermo Fisher Scientific) at 37°C in 5% CO2. The treatment with N2 supplement allows the presence of a physiological level of growth factors, thus preventing the occurrence of a neurotrophin withdrawal state. After 2–3 d in vitro, DRG cultures were stimulated for experimental procedures using either control human NGFwt or human NGFR100W (100 ng/ml) for 5 d or were maintained in basal medium conditions. At the end of this incubation period, B2R and phospho-transient receptor potential vanilloid-1 (pTRPV1) protein expression were measured by Western blot after 3 h of bradykinin (BK; 1 μm) application (antibodies used were as follows: rabbit anti-B2R, 1:1000, Alomone Labs; and rabbit anti-pTRPV1 and anti-TRPV1, both 1:1000, Millipore). Substance P (SP) was quantified in the culture medium using commercial enzyme immunoassay, according to the manufacturer instructions (Cayman Chemical).
To characterize neuronal viability, DRG cultures were fixed in 4% PFA for 10 min at room temperature (RT), incubated O/N at 4°C with mouse anti-NeuN (1:200; Sigma-Aldrich), and rabbit anti-Neurofilament 200 (1:200; Sigma-Aldrich) followed by goat anti-mouse secondary antibody (1:400; Sigma-Aldrich), goat anti-rabbit rhodamine-conjugated secondary antibody (1:1000; Sigma-Aldrich), and Hoechst 33258 (0.25 μg/ml) for 1 h and 5 min, respectively, at RT. Fluorescence images were acquired using an Olympus BX51 microscope and a 60× oil-immersion objective (numerical aperture, 1.4). The number of NeuN-immunoreactive cells was normalized on the total number of cells (i.e., Hoechst stained). At least 40 microscopic fields per coverslip, in four coverslips from three independent experiments, were quantified for each experimental group.
Human NGFR100W purification.
Human NGFR100W cDNA was cloned into the prokaryotic expression vector pET19b downstream the sequence of the human BDNF prodomain, to produce a chimeric human proBDNF/NGFR100W construct, and expressed in the Escherichia coli strain Rosetta(DE3)PLysS. The corresponding chimeric protein was refolded from inclusion bodies and purified using an adaptation of the protocol used for proNGF in the study by Paoletti et al. (2015). The purified proBDNF-NGFR100W was proteolytically processed with trypsin to produce mature NGFR100W, as previously described (Paoletti et al., 2015).
Immunohistochemistry.
NGFwt/wt, NGFR100W/wt, mNGF+/+, and mNGF+/− mice were transcardially perfused with 4% PFA in PBS, pH 7.4, and brains were dissected and postfixed O/N in the same solution, then cryoprotected in 30% sucrose in PBS for 36 h. The brains were sectioned with a sliding freezing microtome (Leica) to obtain 45-μm-thick coronal sections that were washed three times in TBS with 0.3% Triton X-100, then treated with 3.5% H2O2 in TBS to inactivate endogenous peroxidases. Sections were blocked for 30 min with 10% FBS and 0.3% Triton X-100 in TBS, followed by an O/N incubation at 4°C with 1:500 goat anti-choline acetyltransferase (ChAT; catalog #AB144P, Millipore). Biotinylated secondary antibodies (Vector Laboratories) were diluted in 10% FBS in PBS for 3 h at RT. Finally, sections were incubated in Vectastain ABC HRP Kit (Vector Laboratories) in PBS for 1 h, followed by another incubation in TBS solution containing diaminobenzidine (DAB; Sigma-Aldrich) and the enzyme Glucose Oxidase type VII (Sigma-Aldrich); the reaction was stopped after 10 min. Stained sections were mounted on glass slides using DPX Medium. Images were acquired with a Nikon Eclipse E600 Optical Microscope, and the density of immunoreactive cells was calculated using ImageJ.
For analysis of superior cervical ganglia (SCGs), embryonic day 16.5 fetuses were extracted from pregnant dams, washed in PBS, and fixed by immersion in 4% PFA for 4 h, then cryoprotected in 30% sucrose in PBS and sectioned using a cryostat. The quantification of SCG cell number was performed on Nissl-stained sections, with the experimenter blind to the genotype of the animal, and representative images were obtained by staining for tyrosine hydroxylase (TH; 1:200; catalog #AB152, Millipore) O/N at 4°C (Crerar et al., 2019); subsequent steps were as described above.
Whole-mount staining was performed on internal organs dissected from P0.5 pups, O/N fixed in 4% PFA, then dehydrated in methanol series, followed by O/N quenching of endogenous peroxidases in 80% methanol, 17% DMSO, and 3% H2O2. After rehydration, samples were blocked in 4% BSA, 1% Triton X-100 in PBS, and incubated with 1:200 anti-TH antibody diluted in the same blocking solution for 72 h at 4°C. The signal was revealed by incubation with HRP-conjugated anti-rabbit antibody (1:200; catalog #sc-2004, Santa Cruz Biotechnology) diluted in blocking solution O/N at 4°C, followed by DAB processing. Finally, samples were cleared using a 2:1 solution of benzyl benzoate and benzyl alcohol (Crerar et al., 2019). Samples were imaged using a 4× objective, and optical density of the signal was quantified with the experimenter blind to the genotype of the animal.
Skin and DRG immunofluorescence.
DRGs from adult mice (2 and 6 months old) were collected in an Eppendorf tube containing cold PBS, then postfixed in 4% PFA for 30 min at RT, embedded in 2% agarose and sectioned at 50 μm thickness using a vibratome. Sections were washed twice with PBS-Triton 0.3%, then subjected to a 30 min blocking step in 5% NGS and 0.3% Triton X-100 in PBS, followed by an O/N incubation at 4°C with primary antibodies diluted as shown below. Alexa Fluor-conjugated secondary antibodies (Thermo Fisher Scientific) were diluted 1:1000 in 0.3% Triton X-100 and 5% NGS in PBS for 2 h at RT. Sections were mounted using Invitrogen Prolong Gold Medium (Thermo Fisher Scientific).
For immunofluorescence analysis, the hairy and glabrous skins were collected, allowed to dry, and postfixed in 4% PFA at 4°C O/N, then incubated in 30% sucrose in PBS and frozen in OCT (optimal cutting temperature) medium (Leica). Sections (50 μm thick for hairy skin, 20 μm thick for glabrous skin) were obtained using a cryostat. Immunostaining was performed as described above.
The antibodies and dilutions used were as follows: 1:500 mouse anti-NF200 (Sigma-Aldrich); 1:200 mouse anti-calcitonin gene-related peptide (CGRP; Rockland); 1:100 Invitrogen isolectin GS-B4-biotin conjugate (Thermo Fisher Scientific); 1:200 rabbit anti-B2R (Alomone Labs); 1:300 mouse anti-TRPV1 (Millipore); 1:200 Dako rabbit anti-PGP 9.5 (Agilent Technologies); and 1:300 rabbit anti-NGF M20 (Santa Cruz Biotechnology).
The M20 anti-NGF antibody was validated by measuring the immunofluorescence signal intensity obtained using different dilutions on glabrous skin samples from NGF+/+ and NGF+/− mice. A nonlimiting concentration of the primary antibody (0.67 μg/ml) was able to detect the lower skin NGF content in NGF+/− mice compared with NGF+/+ mice. Decreasing the antibody concentration (0.33 μg/ml) led to incomplete titration of NGF in the skin of NGF+/+ mice, whereas the signal in samples from NGF+/− mice was unaffected. Further dilution of the antibody (0.17 μg/ml) resulted in inefficient detection of NGF in samples from both NGF+/+ and NGF+/− mice. ANOVA-2 (antibody concentration × genotype interaction, F(2,34) = 3.563, p = 0.039) followed by Holm–Sidak post hoc test (0.67 μg/ml; NGF+/+ vs NGF+/−, p < 0.001; 0.33 μg/ml; NGF+/+ vs NGF+/−, p < 0.001, 0.17 μg/ml; NGF+/+ vs NGF+/−, p = 0.314; NGF+/+: 0.67 vs 0.33 μg/ml, p = 0.303; 0.67 vs 0.17 μg/ml, p < 0.001; 0.33 vs 0.17 μg/ml, p < 0.001; NGF+/−: 0.67 vs 0.33 μg/ml, p = 0.697; 0.67 vs 0.17 μg/ml, p = 0.047; 0.33 vs 0.17 μg/ml, p = 0.030; n = 7 for each group).
All images were acquired with a Leica SP5 Confocal Microscope and analyzed with Fiji (NIH).
RNA preparation for microarray analysis.
DRGs from wild-type and NGFR100W/wt mice (6 months of age) were isolated and collected. The total RNA was extracted with TRIzol reagent (Life Technologies) according to the manufacturer instructions, DNase treated, and purified using Qiagen columns. RNA content was determined on a NanoDrop UV-VIS Spectrophotometer. Only samples with an absorbance ratio of 1.8 < OD260/OD280 < 2.0 were further processed. Each sample was then quality checked for integrity using a BioAnalyzer 2100 (RNA 6000 Nano Kit, catalog #G2938C, Agilent Technologies).
Whole-genome expression profiling.
Gene expression profiling was performed using the Agilent Technologies one-color microarray system. Two hundred nanograms of Poly A+RNA were retrotranscribed using oligo-dT primers linked to the T7 promoter, and the resulting cDNA was used as a template for cyanine 3-CTP (cytidine triphosphate)-labeled cRNA preparation, using the Agilent Technologies Low Input Linear Amplification Kit. The labeled cRNA was purified with RNeasy Mini Spin Columns (Qiagen). To monitor both the labeling reactions and the microarray performance, Agilent Technologies Spike-In Mix was added to the mRNA samples before labeling reactions according the RNA Spike-In protocol. Cyanine 3-labeled cRNA was hybridized to Agilent Technologies 8x60K whole-mouse genome oligonucleotide microarrays (Grid ID, 028005). Microarray hybridizations were performed in SureHyb Hybridization Chambers (Agilent Technologies) containing 600 ng of cyanine 3-labeled cRNA per hybridization. The hybridization reactions were performed at 65°C for 17 h using the Gene Expression Hybridization Kit (Agilent Technologies). The hybridized microarrays were disassembled at RT in Gene Expression Wash Buffer 1 (Agilent Technologies). After disassembly, the microarrays were washed in Gene Expression Wash Buffer 1 for 1 min at RT, followed by washing with Gene Expression Wash Buffer 2 for 1 min at 37°C. Then, microarrays were treated with acetonitrile for 1 min at RT. Fluorescence signals of the hybridized Agilent Technologies Microarrays were detected using the Microarray Scanner System (catalog #G2564B, Agilent Technologies). The Feature Extraction Software (version 10.7.3.1, Agilent Technologies) was used to process the microarray image files.
Microarray data analysis.
Data filtering, normalization, analysis, and plotting were performed using R-Bionconductor (https://www.bioconductor.org/). In particular, differential expression was analyzed with the R package limma version 3.5 (Ritchie et al., 2015). All the features with the flag gIsWellAboveBG = 0 (too close to background) were filtered out and excluded from the following analysis. Filtered data were normalized by aligning samples to the 75th percentile. Differentially expressed genes were selected by a combination of fold-change and moderated t test thresholds (R limma test with FDR < 0.05; |Log2 fold change ratio| > 1.0). Pathway analysis and network plotting of differential gene lists was performed using the on-line tool StringDB (https://string-db.org; (Szklarczyk et al., 2015).
Electron microscopy.
Fixed nerves were washed in phosphate buffer at RT (10 × 5 min), osmicated in 2% (w/v) OsO4 in H2O (2 h at 4°C), washed again (0.05 m maleate buffer, pH 5.15–10 × 5 min), then dehydrated in ethanol (70%, v/v) for 15 min; 80% (v/v) for 15 min; 90% (v/v) for 15 min; 100%, 4 × 15 min). Subsequently, nerves were infiltrated first with propylene-oxide (2 × 15 min) then with a mixture of 50% (v/v) propylene-oxide and 50% (v/v) resin catalyzed with 2% DMP30 overnight at RT. Embedding (100% resin catalyzed with 2% DMP30) was followed by 24 h heat treatment for proper polymerization of resin at 65°C. Myelinated axons were counted on semithin (1-μm-thick) sections stained with toluidine blue and imaged with a Zeiss Axioplan light microscope. Myelinated fibers were counted individually using MetaMorph software (Molecular Devices). In cases of multiple nerve components (i.e., separate bundles of fibers, with individual connective sheets), fiber counts were performed for each component and the corresponding values were summed up. Unmyelinated axons were counted with the aid of transmission electron microscopy at 20,000× magnification. Sectioning was performed using a Leica Ultracut Ultramicrotome. Ultrathin sections (90 nm thick) were collected on single-hole copper grids (Formvar Support Film Slots, 2 × 1 mm Cu grids; FF2010-CU). Images were obtained using a Jeol 1200EX II Transmission Electron Microscope equipped with a charge-coupled device Olympus Veleta Megaview camera covering 5% of the total cross-sectional area of the nerve. Based on the cross-sectional area, a total of 211 fields/NGFwt/wt and 146 fields/NGFR100W/wt samples were obtained covering central as well as peripheral portions of each nerve systematically. Image files were saved as a TIFF and transferred to MetaMorph, where axons were counted manually.
Nerve conduction velocity measurements.
Mice were killed using CO2 inhalation, and the saphenous nerve was dissected and placed in an organ bath (Wetzel et al., 2007). The chamber was perfused with a synthetic interstitial fluid buffer containing the following (in mm): NaCl 123, KCl 3.5, MgSO4 0.7, NaH2PO4 1.7, CaCl2 2.0, Na-gluconate 9.5, glucose 5.5, sucrose 7.5, and HEPES 10, pH 7.4, at 3 ml/min at 32°C. The distal part of the nerve was placed in the organ bath, while the proximal part was placed in an adjacent chamber filled with mineral oil for recording. An electric probe was used to stimulate the nerve, and a compound action potential was recorded and analyzed using LabChart4 software (AD Instruments, Australia). Each electrical stimulus elicited a response consisting of three peaks, corresponding to the Aβ, Aδ, and C fibers, respectively. To measure the conduction velocity of each fiber type, the distance between the electric probe and the recording electrode was divided by the time elapsed from the beginning of the stimulus to the appearance of the peak.
Electrophysiology on brain slices.
Acute brain slices were prepared from 5- to 6-month-old mice, following the protocol described in the study by Barone et al. (2018). Mice were killed by cervical dislocation, and the brain was quickly dissected in an ice-cold, O2-saturated cutting solution containing the following (in mm[scap]): NaCl 85, sucrose 75, glucose 25, NaHCO3 24, KCl MgCl2 4, 2.5, NaHPO4 1.25, and CaCl2 0.5, pH 7.4. The 350-μm-thick sections were prepared using a Leica VT-1200S vibratome while keeping the brain immersed in the same ice-cold cutting solution. Sections were transferred to a recovery chamber filled with artificial CSF (aCSF) containing the following (in mm): NaCl 119, glucose 10, HEPES 10, NaHCO3 6.2, KCl 2.5, CaCl2 2, MgCl2 1.2, and NaH2PO4 1, pH 7.4, and was held at 32°C for 30 min, then recovery was completed for an additional 60 min at RT. For recording, sections were transferred to a submerged chamber continuously perfused with aCSF at 32°C saturated with O2, and a concentric bipolar stimulating electrode was used to stimulate Schaffer collateral pathway fibers, while a glass pipette (1 MΩ impedance) filled with aCSF was placed in the CA1 stratum radiatum. Field EPSPs (fEPSPs) were recorded using a MultiClamp 700A Amplifier plugged to a Digidata 1322A interface (Molecular Devices). Current stimuli were delivered using a stimulus isolator (WPI). By using a stimulus intensity eliciting a response that amounted to 30–50% of the maximum, a stable baseline, using a 0.05 Hz test stimulus frequency, was obtained before delivering high-frequency stimulation (HFS; four trains of 1 s at 100 Hz, with an intertrain interval of 10 s) to induce long-term potentiation. The post-HFS fEPSP was monitored for at least 60 min. Data were analyzed using Clampfit (Molecular Devices). The experimenter was blind to the genotype of the animal.
Hek293 cells culture.
Hek293 cells were maintained at 37°C, 5% CO2, in Gibco DMEM/F-12 medium supplemented with 10% FBS, 1% l-glutamine, and 1% penicillin/streptomycin (Thermo Fisher Scientific). Hek293 cells were transfected with pCMV6-XL5-human NGFwt and pCMV6-XL5-human NGFR100W plasmids following the manufacturer instructions for Invitrogen Lipofectamine 2000 (Thermo Fisher Scientific). Forty-eight hours after transfection, the supernatants were immunoprecipitated and subjected to Western blotting as described below (see NGF immunoprecipitation and Western blot).
NGF immunoprecipitation and Western blot.
Cerebral cortices were isolated from adult mice and homogenized in lysis buffer (Tris-HCl 100 mm, NaCl 400 mm, SDS 0.1%, and Triton X-100 1%). The homogenates were sonicated, incubated in ice for 30 min, and centrifuged at 15,000 × g for 30 min at 4°C. Protein concentration in the supernatant was quantified using the Bradford method (Bio-Rad). Four milligrams of protein were immunoprecipitated with an excess of anti-NGF αD11 antibody in NET Gel Buffer (Tris-HCl, pH 7.5, 50 mm; NaCl 150 mm; 0.1% Nonidet P-40; EDTA, pH 8, 1 mm; 0.25% gelatin; and 0.02% NaN). After immunoprecipitation, total lysates were loaded on 12% acrylamide gels and blotted using nitrocellulose membranes. The primary antibody was anti-NGF M20 (1:500; Santa Cruz Biotechnology), the secondary antibody was goat anti-rabbit HRP-conjugated (1:500; Santa Cruz Biotechnology). Blot images were acquired using a ChemiDoc System (Bio-Rad), and the optical density was quantified using ImageJ (NIH).
Data analysis and statistics.
Statistical significance was assessed using SigmaStat 12 (Systat Software). Data are presented as the mean ± SEM; detailed statistics for every comparison are reported in the corresponding figure legend.
Results
Generation of NGFR100W/wt mice and characterization of their nociceptive phenotype
To shed light on the consequences of the HSAN V-associated NGFR100W mutation, we generated a knock-in mouse line harboring the human NGFR100W sequence (Fig. 1). Homozygous NGFR100W/R100W mice die by P30; Testa et al., 2019). On the other hand, heterozygous mice thrive normally and show no visible gross deficit. We analyzed the phenotype of heterozygous NGFR100W/wt mice in detail, during youth (2 months) and adulthood (6 months). Chemical nociception induced by capsaicin injection in the hindpaw was impaired at both ages (Fig. 2A). Thermal nociception was normal at 2 months of age and decreased at 6 months, with adult NGFR100W/wt mice displaying a higher latency to respond to a high-temperature stimulus (Fig. 2B). On the other hand, cold sensitivity, measured by topical acetone application on the hindpaw was reduced at both ages (Fig. 2C). In NGFR100W/wt mice, a non-noxious stimulus represented by a small piece of tape applied to the back (i.e., to the hairy skin) took more time to induce a removal reaction only at 6 months of age (Fig. 2D). On the other hand, the response to gentle stroking of the glabrous skin was normal at both time points (Fig. 2E).
Molecular strategy for the generation of R100W knock-in mice. A, Top, Endogenous mouse NGF locus with 5′and 3′ Southern blot probes and expected sizes of wild-type Southern blot bands. Middle, Targeting vector for site-specific recombination. Bottom, Targeted NGF locus with 5′ and 3′ Southern blot probes and expected sizes of recombinant Southern Blot bands. Color codes are as follows: pink, mouse NGF coding sequence; yellow, human NGF coding sequence; brown, NeoR+ selection cassette; orange, PGK promoter; blue, loxP sites; light green, left and right homology arms; dark green, DTA-negative selection marker. B, Representative image of embryonic stem cells Southern blot. C, PCR genotyping of NGFR100W/wt, NGFwt/wt, and NGFR100W/R100W mice; wild-type band, 400 bp; mutant band, 200 bp.
The R100W mutation is associated with decreased sensitivity to noxious stimuli. A, Decreased hyperalgesic response to capsaicin in juveniles and adults. Two months, ANOVA-1 (F(2,12) = 12.869, p = 0.002), followed by Bonferroni post hoc test (NGFwt/wt vs VEH, **p = 0.002; NGFwt/wt vs NGFR100W/wt, *p = 0.017; NGFR100W/wt vs VEH, n.s.); VEH, n = 3; NGFwt/wt, n = 5; NGFR100W/wt, n = 5. 6 months, ANOVA-1 (F(2,13) = 49.995, p < 0.001), followed by Bonferroni post hoc test (NGFwt/wt vs VEH, ***p < 0.001; NGFwt/wt vs NGFR100W/wt, ***p < 0.001; NGFR100W/wt vs VEH, p = 0.017); VEH, n = 3; NGFwt/wt, n = 5; NGFR100W/wt, n = 6. B, Normal sensitivity to hot stimuli at 2 months of age and impairment in adult HSAN V mice, with increased latency in NGFR100W/wt animals to respond to high temperatures. Left, 2 months, Student's two-tailed t test (t = 0.126, p = 0.901); NGFwt/wt, n = 11; NGFR100W/wt, n = 15. 6 months, Student's two-tailed t test (t = 4.743, p < 0.001); NGFwt/wt, n = 4; NGFR100W/wt, n = 9. Right, ANOVA-2 (“genotype” × “temperature” interaction, F(4,79) = 3.283, p = 0.017), followed by Bonferroni post hoc test, ***p < 0.001, **p = 0.003; NGFwt/wt, n = 8; NGFR100W/wt, n = 8. C, Impaired cold sensitivity in both juveniles and adults. 2 months, Student's two-tailed t test (t = 2.445, *p = 0.035); NGFwt/wt, n = 6; NGFR100W/wt, n = 6. 6 months, Student's two-tailed t test (t = 2.457, *p = 0.026); NGFwt/wt, n = 8; NGFR100W/wt, n = 10. D, Decreased hairy skin sensitivity in adult HSAN V mice. 2 months, Student's two-tailed t test (t = 1.261, p = 0.226); NGFwt/wt, n = 6; NGFR100W/wt, n = 11. 6 months, Student's two-tailed t test (t = 2.305, *p = 0.042); NGFwt/wt, n = 5; NGFR100W/wt, n = 8. E, Normal light touch sensitivity. 2 months, Student's two-tailed t test (t = 0.155, p = 0.879); NGFwt/wt, n = 5; NGFR100W/wt, n = 11. 6 months, Student's two-tailed t test (t = 0.050, p = 0.961); NGFwt/wt, n = 8; NGFR100W/wt, n = 7.
These behavioral data show that NGFR100W/wt mice have a reduced responsiveness to chemical and thermal noxious stimuli, are less sensitive to somatosensory inputs, but display normal light touch.
Phenotypic analysis of DRGs from NGFR100W/wt mice
To look for a cellular-functional correlate of the impaired nociceptive behavior of NGFR100W/wt mice, we first focused on the two main NGF-sensitive neuronal populations, defined by the expression of IB4 and CGRP, respectively (Harrison et al., 2004). We imaged DRGs from 2- and 6-month-old NGFR100W/wt mice and found no change in the total number of DRG neurons (2 months: NGFR100W/wt, 1040.4 ± 37.9 cells/mm2; NGFwt/wt, 1003.3 ± 52.7 cells/mm2; n = 3/group; Student's t test, p = 0.598, t = −0.571; 6 months: NGFR100W/wt, 1340.4 ± 36.8 cells/mm2; NGFwt/wt, 1391.1 ± 29.5 cells/mm2; Student's t test, p = 0.324, t = 1.044; NGFR100W/wt, n = 6; NGFwt/wt, n = 5) and in the percentages of cells expressing the neurofilament marker NF200, the nonpeptidergic nociceptor marker IB4, and the peptidergic nociceptor marker CGRP, respectively (Fig. 3A). We also analyzed DRGs from P5 mice, an early postnatal developmental stage during which segregation between peptidergic and nonpeptidergic neurons is in progress (Molliver et al., 1997) and found no differences between NGFR100W/wt mice and controls in the total cell number (NGFR100W/wt, 2134.3 ± 64.6 cells/mm2; NGFwt/wt, 2234.2 ± 116.9 cells/mm2; n = 5 and n = 4, respectively; Student's t test, p = 0.454, t = 0.793) or in the percentages of either NF200+, IB4+, or CGRP+ cells (Fig. 3B). This suggests that the neurotrophic potency of NGFR100W on DRG neurons is comparable to NGFwt. This was also demonstrated in an in vitro DRG neuron survival assay (NGFR100W, 158.2 ± 8.8%; NGFwt, 162.3 ± 18.03%; control, 100.0 ± 5.0% NeuN+/total Hoechst+ cells; ANOVA-1, F(2,17) = 8.621, p = 0.003; followed by Bonferroni post hoc test: NGFR100W vs NGFwt, p = 1.000; NGFwt vs control, p = 0.006; NGFR100W vs control, p = 0.01; n = 6 for each group). Given that the sympathetic nervous system is highly dependent on an intact NGF function for proper development (Levi-Montalcini and Booker, 1960; Glebova and Ginty, 2004), the unaltered neurotrophic actions of NGFR100W are also supported by the normal cell number of the SCG (Fig. 3C) and by the normal sympathetic innervation of key target organs (i.e., heart, stomach, kidney, spleen), as revealed by TH immunohistochemistry (Fig. 3D).
Analysis of DRG markers, sympathetic neurons and innervation, and DRG transcriptome of NGFR100W/wt mice. A, B, Normal expression of NF200, IB4, and CGRP in DRG neurons in both juveniles and adults (A) and in P5 mice (B). Scale bars, 100 μm. C, No significant difference in SCG cell numbers from NGFR100W/wt and NGFwt/wt mice; Student's two-tailed t test: t = 0.482, p = 0.650; NGFwt/wt, n = 4; NGFR100W/wt, n = 3. Scale bar, 100 μm. D, Normal sympathetic innervation of target internal organs in NGFR100W/wt mice with respect to NGFwt/wt controls; heart, Student's two-tailed t test: t = −0.268, p = 0.796; NGFwt/wt, n = 6; NGFR100W/wt, n = 4; stomach, Student's two-tailed t test: t = 0.216, p = 0.835; NGFwt/wt, n = 5; NGFR100W/wt, n = 4; kidney, Student's two-tailed t test: t = −0.002, p = 0.999; NGFwt/wt, n = 6; NGFR100W/wt, n = 4; spleen, Student's two-tailed t test: t = −0.165, p = 0.874; NGFwt/wt, n = 5; NGFR100W/wt, n = 3. Scale bars, 1 mm. E, Reduced expression of B2R and TRPV1 in DRG neurons of adult mice. Scale bar, 50 μm. B2R, Student's two-tailed t test: t = 6.219, **p = 0.003; NGFwt/wt, n = 3; NGFR100W/wt, n = 3. TRPV1, Student's two-tailed t test: t = 12.455, ***p < 0.001; NGFwt/wt, n = 4; NGFR100W/wt, n = 4. F, Volcano plot for differentially expressed genes in DRGs. The x-axis corresponds to log2FC (Log2FoldChange) differential expression, and the y-axis to −Log(FDR; −Log false discovery rate corrected p value). Red and green regions show significantly upregulated and downregulated genes, respectively. The log2FC and −Log(FDR) thresholds are shown as horizontal (−Log(FDR)<0.05) and vertical (|log2FC|>2.0) dashed lines (see Figure 3-1). G, Pathway analysis of differentially expressed genes in DRGs. The gene–gene network was obtained by StringDB tool (https://string-db.org). Colors indicate genes belonging to the main over-represented pathways and functional categories: immune response and chemokines (red and green), phagocytosis (blue and purple), killer cells mediated cytotoxicity (light blue), and rho GTPase (yellow).
Figure 3-1
However, when we analyzed BK receptor 2 (B2R), whose expression is upregulated by NGF (Petersen et al., 1998), and TRPV1, an NGF-responsive detector of heat- and capsaicin-induced pain (Zhang et al., 2005), we found a significantly decreased immunoreactivity for both receptors (Fig. 3E).
To detect subtler changes in DRGs from NGFR100W/wt mice, we performed a transcriptomic analysis to identify differentially expressed genes with respect to DRGs from wild-type mice. The data showed only a small number of differentially expressed genes in DRGs from NGFR100W/wt versus wild-type mice (Fig. 3F and Fig. 3-1), which is in line with the globally normal appearance of DRG neuronal subpopulations. Among the regulated genes, TyroBP (DAP12) and toll-like receptor 2 (TLR2) are noteworthy for their involvement in glial function (Liu et al., 2012; Shboul et al., 2019). Indeed, gene ontology and network analysis of differentially expressed genes in DRGs revealed that the R100W mutation significantly affects pathways involved in immune response, phagocytosis and Rho GTPase cycle (Fig. 3G), suggesting a major effect of the R100W mutation on DRG glial and microglial cells, rather than on DRG neurons.
These findings demonstrate that DRG neurons from NGFR100W/wt mice show only specific molecular changes directly linked to nociception, alongside an intriguing modulation of glia-related genes.
Pain sensitization by NGFR100W protein
To investigate whether the R100W mutation might also affect DRG function, we explored the ability of the NGFR100W protein to activate and sensitize wild-type DRGs. Incubation of DRG neuronal cultures with NGFR100W, followed by acute administration of the inflammatory neuropeptide BK (Chuang et al., 2001), led to reduced SP release (Fig. 4A), bradykinin B2R receptor expression (Fig. 4B), and TRPV1 phosphorylation (Fig. 4C), respectively. The reduced ability of NGFR100W to sensitize neurons in vitro was paralleled by in vivo experiments, using acute injection of NGFR100W in the hindpaw of CD1 mice. This treatment resulted in a significantly lower mechanical allodynia with respect to mice injected with NGFwt (Fig. 4D). Thus, NGFR100W diminishes the propensity of DRGs to sensitize and to increase transmission of noxious stimuli.
Reduced in vitro and in vivo sensitization capability of NGFR100W. A, Bradykinin-induced SP release in DRG cultures is reduced by NGFR100W cotreatment compared with NGFwt. ANOVA-1 (F(2,16) = 10.501, p < 0.002) followed by Student–Newman–Keuls post hoc test, ***p < 0.001, *p = 0.03; hNGFwt, n = 5; hNGFR100W, n = 6; control, n = 6. B, Downregulation of B2R expression by NGFR100W in DRG cultures stimulated with bradykinin. ANOVA-2 (NGF × bradykinin interaction, F(2,45) = 3.371, p = 0.044) followed by Bonferroni post hoc test, ***p < 0.001; hNGFwt-vehicle, n = 7; hNGFR100W-vehicle, n = 8; control-vehicle, n = 8; hNGFwt-bradykinin, n = 8; hNGFR100W- bradykinin, n = 7; control-bradykinin, n = 8. C, Reduced phosphorylation of TRPV1 by NGFR100W in DRG cultures stimulated with bradykinin. ANOVA-2 (NGF × bradykinin interaction, F(2,47) = 9.346, p < 0.001) followed by Bonferroni post hoc test, ***p < 0.001, **p = 0.008, *p = 0.02; n = 8 for each group. D, Human NGFR100W intraplantar injection causes reduced mechanical sensitization compared with human NGFwt. ANOVA-2 repeated-measures test (treatment × time interaction, F(8,144) = 5.785, p < 0.001) followed by Bonferroni post hoc test, hNGFwt versus saline, ***p < 0.001; hNGFwt versus hNGFR100W, ###p < 0.001, ##p = 0.002, hNGFR100W vs saline, n.s. p = 1.000; saline, n = 10; hNGFwt, n = 11; hNGFR100W, n = 9.
Exploring nociceptive information routes in NGFR100W/wt mice
The above-described in vitro and in vivo results prompted us to analyze the nociceptive input path. We first ruled out that the observed behavioral effects (Fig. 2) were due to a nerve conduction deficit; indeed, the conduction velocities of the three main sensory fiber populations, Aβ, Aδ and C fibers, were normal (Fig. 5A).
Reduced C-fiber density and skin innervation in NGFR100W/wt mice and rescue of nociceptive deficits by NGFwt treatment. A, No alteration in conduction velocity of Aβ, Aδ, and C fibers in adult mice. Aβ fiber peak, Student's two-tailed t test (t = 0.435, p = 0.669); Aδ fiber peak, Student's two-tailed t test (t = 0.737, p = 0.470); C fiber peak, Student's two-tailed t test (t = 1.629, p = 0.120); NGFwt/wt, n = 10; NGFR100W/wt, n = 11 nerves. B, Electron microscopy analysis of adult sciatic nerve reveals: significant reduction of cross-section area, Student's two-tailed t test (t = 3.810, **p = 0.004); NGFwt/wt, n = 5; NGFR100W/wt, n = 6; unaffected number of myelinated axons, Student's two-tailed t test (t = 0.983, p = 0.351); NGFwt/wt, n = 5; NGFR100W/wt, n = 6, and significant reduction of slow nonmyelinated C axons, Student's two-tailed t test (t = 3.295, **p = 0.009); NGFwt/wt, n = 5; NGFR100W/wt, n = 6. C, Representative images and quantification of PGP9.5 expression show normal innervation at 2 months and a significant reduction at 6 months of age. Scale bar, 50 μm. 2 months, Student's two-tailed t test (t = 0.340, p = 0.751); NGFwt/wt, n = 3; NGFR100W/wt, n = 3; 6 months, Student's two-tailed t test (t = 4.779, **p = 0.004); NGFwt/wt, n = 4; NGFR100W/wt, n = 3. D, Age-dependent reduction in glabrous skin innervation. Scale bar, 50 μm. 2 months, Student's two-tailed t test (t = 0.792, p = 0.473); n = 3 for both groups. 6 months, Student's two-tailed t test (t = 5.800, **p = 0.002); NGFwt/wt, n = 4; NGFR100W/wt, n = 3. E, Decreased NGF levels in the glabrous skin of adult mice. Scale bar, 50 μm. Student's two-tailed t test (t = 3.169, *p = 0.034); n = 3 for both groups. F, Rescue of the sensitivity to capsaicin after treatment with mouse NGFwt from gestation until 2 months of age. NGFwt/wt, Student's two-tailed t test (t = 0.323, p = 0.754); saline, n = 5; NGF, n = 7; NGFR100W/wt, Student's two-tailed t test (t = 2.764, *p = 0.033); saline, n = 5; NGF, n = 4.
We then analyzed the ultrastructure of the sciatic nerve using transmission electron microscopy. Compared with controls, the cross-sectional area of the whole nerve in NGFR100W/wt mice was significantly decreased (Fig. 5B). In this regard, the number of myelinated axons contained in the nerve was unaffected, whereas a significant decrease in the number of nonmyelinated axons in NGFR100W/wt mice was observed (Fig. 5B), which might determine the smaller overall nerve section. These neurophysiological and ultrastructural data match analogous findings in HSAN V heterozygous human carriers (Minde et al., 2009).
Finally, we analyzed target tissues of sensory fibers, namely hairy and glabrous skin (Zimmerman et al., 2014). The area of PGP9.5-immunoreactive terminals was not significantly different at 2 months of age, whereas, at 6 months of age, NGFR100W/wt mice showed a significant reduction in both hairy and glabrous skin sensory innervation (Fig. 5C,D). This is in keeping with the reduction in the skin innervation of individuals with HSAN V (Axelsson et al., 2009). Moreover, the age-dependent decrease in PGP9.5-immunoreactive fibers in the hairy skin correlates with the lower performance of NGFR100W/wt mice in the tape removal test (Fig. 2D).
Since target-derived NGF modulates innervation (Davies et al., 1987; Purves et al., 1988), we analyzed NGF expression in the glabrous skin by immunofluorescence and found a significantly reduced signal intensity in NGFR100W/wt mice, compared with controls (Fig. 5E). To validate this measurement, we used the same anti-NGF primary antibody to titrate the different levels of NGF in the glabrous skin of NGF+/+ and NGF+/− mice. Serial dilutions were used to demonstrate that the dose of antibody used for the experiment shown in Figure 5E can indeed detect the relative differences in NGF levels corresponding to the two genotypes, whereas, at a higher dilution, no significant difference was observed in NGF signal intensity (see Materials and Methods).
Taking into account the NGF reduction in the peripheral target sites of sensory fibers, we reasoned that exogenous administration of wild-type NGF could rescue the nociceptive deficit of NGFR100W/wt mice. To this aim, we performed a long-term NGF treatment, from embryonal life until 2 months of age, and analyzed the sensitivity of treated mice to capsaicin. A 21 d washout period after the last treatment was allowed (see Materials and Methods) to exclude an acute sensitizing action of NGF. This strategy proved successful in fully rescuing the nociceptive impairment of NGFR100W/wt mice (Fig. 5F).
This evidence indicates that the R100W mutation is responsible for multiple alterations in the pathway carrying nociceptive inputs, and that early treatment with wild-type NGF can restore pain perception.
NGFR100W affects the secretion of wild-type NGF
The R100W mutation has been described to cause an impairment in the secretion of mature NGF in PC12 and COS cells (Larsson et al., 2009), but whether this is also true in human cells is not known. NGF is secreted as a homodimer. It is not known, however, whether NGFR100W affects the secretion of wild-type NGF when the two are coexpressed, as it happens in heterozygous NGFR100W/wt mice. We confirmed the secretion deficit of NGFR100W in human HEK cells (Fig. 6A). Strikingly, the secretion of wild-type, mature NGF was also impaired when coexpressed with NGFR100W (Fig. 6A). This suggested that a similar phenomenon might occur in vivo in NGFR100W/wt mice. Consistently, we found reduced total NGF levels of in plasma and brain from NGFR100W/wt mice (Fig. 6B,C).
Impaired secretion and reduced NGF levels in the NGFR100W condition. A, Impaired secretion of human NGFR100W in HEK293 cells transfected with the corresponding plasmid. Cotransfection of HEK293 cells with human NGFwt and human NGFR100W mimics the heterozygous condition, and shows a decrease in the secretion of human NGFwt. ANOVA-1 (F(5,24) = 23.529, p < 0.001) followed by Student–Newman–Keuls post hoc test, ***p < 0.001, **p < 0.01, *p < 0.05; NGF, n = 5; hNGFwt 0.6 μg, n = 4; hNGFR100W 0.6 μg, n = 5; hNGFwt + hNGFR100W 0.3 μg/each, n = 3; hNGFwt 0.3 μg, n = 4;hNGFR100W 0.3 μg, n = 5; mock, n = 4. B, Reduced plasma NGF levels in NGFR100W/wt mice. Student's two-tailed t test (*p = 0.031); NGFwt/wt, n = 6; NGFR100W/wt, n = 7. C, Lower abundance of NGF in brain extracts from NGFR100W/wt mice. Student's two-tailed t test (t = 2.465, *p = 0.031); NGFwt/wt, n = 6; NGFR100W/wt, n = 7.
NGFR100W/wt mice show no learning and memory deficits
NGF is not only involved in the development and survival of sensory neurons, but is also a key regulator of brain development, in addition to being required for learning and memory processes, via its actions on CNS neuronal target cells such as forebrain cholinergic neurons (Chao, 2003).
We studied the learning and memory phenotype of NGFR100W/wt mice. Interestingly, NGFR100W/wt mice, tested in the Morris water maze (MWM), showed no difference in the learning performance, compared with wild-type controls (Fig. 7A), despite the lower brain NGF levels (Fig. 6B,C). A similar conclusion was drawn when challenging NGFR100W/wt mice and controls in the object recognition test. Indeed, NGFR100W/wt mice showed no significant differences in the exploratory activity and in the preference index (Fig. 7C,D), which indicate unaffected visuospatial recognition memory. Moreover, Schaffer collateral-CA1 long-term potentiation (LTP), a well established electrophysiological correlate of learning and memory (Davis et al., 1992), did not significantly differ in NGFR100W/wt mice from NGFwt/wt animals, thus supporting an intact hippocampal function (Fig. 7E,F).
Normal cognitive function, hippocampal synaptic plasticity, and ChAT+ neuron density in NGFR100W/wt mice. A, Normal Morris water maze learning curve for adult NGFR100W/wt mice. ANOVA-2 with repeated measures (main effect of “training day”: F(8,112) = 15.600, p < 0.001). NGFwt/wt, n = 5; NGFR100W/wt, n = 8. B, Delayed Morris water maze learning curve for mNGF+/− mice. ANOVA-2 with repeated measures (“training day” × “genotype” interaction, F(8,123) = 2.836, p = 0.007), followed by Bonferroni post hoc test, *p < 0.05, ***p < 0.001; mNGF+/+, n = 7; mNGF+/−, n = 7. C, Unimpaired Novel Object Recognition (NOR) sample phase for NGFR100W/wt and mNGF+/− mice. 2 months, Student's two-tailed t test: t = 0.385, p = 0.704; NGFwt/wt, n = 7; NGFR100W/wt, n = 15; 6 months, Student's t two-tailed test: t = 1.094, p = 0.289; NGFwt/wt, n = 10; NGFR100W/wt, n = 9; Student's two-tailed t test: t = 1.303, p = 0.215); mNGF+/+, n = 6; mNGF+/−, n = 9. D, Unimpaired NOR test phase for NGFR100W/wt and visual recognition memory deficit for mNGF+/− mice. 2 months, Student's two-tailed t test: t = 0.282, p = 0.780; NGFwt/wt, n = 7; NGFR100W/wt, n = 15; 6 months, Student's t two-tailed test: t = 1.453, p = 0.166; NGFwt/wt, n = 10; NGFR100W/wt, n = 8; Student's two-tailed t test: t = 4.450, ***p < 0.001; mNGF+/+, n = 6; mNGF+/−, n = 9. E, No significant differences in the time course of LTP at CA3–CA1 synapses induced by high-frequency stimulation (HFS) of the Schaffer collateral pathway between NGFR100W/wt and control mice. ANOVA-2 repeated measures (main effect of time, F(69,966) = 21.579, p < 0.001; genotype, F(1,966) = 0.026, p = 0.875; genotype × time interaction, F(69,966) = 0.156, p = 1.000); n = 9 for each group. Representative traces show superimposed pre-HFS and post-HFS fEPSPs. F, Comparable levels of LTP (obtained from averaging the last 10 min of recording) in NGFR100W/wt and NGFwt/wt mice. Student's two-tailed t test: t = −0.001, p = 0.999; n = 9 for each group. G, H, Normal septal ChAT+ neuron density in basal forebrain and striatum in juvenile and adult NGFR100W/wt mice, and decreased density in mNGF+/− mice. Scale bar, 200 μm. Basal forebrain: 2 months: Student's two-tailed t test: t = 0.958, p = 0.375; NGFwt/wt, n = 4; NGFR100W/wt, n = 4; 6 months: Student's two-tailed t test: t = 1.271, p = 0.278; NGFwt/wt, n = 3; NGFR100W/wt, n = 4; Student's two-tailed t test: t = 3.529, **p = 0.010; mNGF+/+, n = 5; mNGF+/−, n = 4. Striatum: 2 months: Student's two-tailed t test: t = 1.144, p = 0.296; NGFwt/wt, n = 4; NGFR100W/wt, n = 4; 6 months: Student's two-tailed t test: t = 0.180, p = 0.864; NGFwt/wt, n = 3; NGFR100W/wt, n = 4; Student's two-tailed t test: t = 3.212, *p = 0.015; mNGF+/+, n = 5; mNGF+/−, n = 4.
Heterozygous NGF knock-out mice (mNGF+/) are also characterized by lower brain levels of wild-type NGF (Chen et al., 1997), similar to NGFR100W/wt mice. In this case, as expected, mNGF+/− mice showed delayed learning in the MWM when compared with their controls (Fig. 7B). No significant differences in the latency to locate the platform where detected among the four experimental groups on day 1 (NGFwt/wt, 100.560 ± 11.742 s; NGFR100W/wt, 98.438 ± 9.919 s; NGF+/+, 117.714 ± 2.286 s; NGF+/−, 113.286 ± 3.887 s; ANOVA-1: F(3,26) = 1.565, p = 0.225). This indicates a uniform initial performance in this task across different genotypes. In addition, mNGF+/− mice had impaired object recognition memory, despite unaffected exploratory behavior (Fig. 7C,D).
Cholinergic neurons are an NGF target population with an essential role in modulating learning and memory (Li et al., 1995). Thus, we analyzed the density of ChAT-immunoreactive neurons in the medial septum and striatum, and found them to be normal in NGFR100W/wt mice compared with controls (Fig. 7G,H). On the other hand, the density of ChAT-immunoreactive neurons in the same brain structures was decreased in mNGF+/− mice (Fig. 7G,H). These neuroanatomical data nicely correlate with the corresponding learning and memory behavioral phenotypes. Thus, the NGFR100W mutation, unlike ngf gene deletion, does not affect spatial learning and memory processes, in keeping with heterozygous R100W human carriers, who are reported to be cognitively normal (Einarsdottir et al., 2004).
Discussion
Regulation of nociception and pain responses by NGF has long been a key area of research on this neurotrophin (Denk et al., 2017). This has spurred a significant translational interest for treating chronic pain conditions (Norman and McDermott, 2017). However, current understanding of these processes is still incomplete. Studying congenital insensitivity to pain disorders, such as HSAN V, offers a unique opportunity to fill this gap, through an alternative viewpoint on the NGF–TrkA axis involvement in chronic pain. In this regard, HSAN V has intriguing features, namely (1) the severe insensitivity to pain of homozygous patients, as opposed to the often clinically silent phenotype of heterozygous carriers; and (2) the absence of cognitive deficits, as opposed to HSAN IV patients, carrying TrkA mutations (Minde et al., 2004; Minde et al., 2009). We and others have previously elucidated that NGFR100W (1) is a TrkA biased agonist, failing to effectively engage p75NTR and activate the nociception-specific PLC-γ pathway (Covaceuszach et al., 2010; Capsoni et al., 2011; Sung et al., 2018); (2) retains full neurotrophic activity via TrkA in a variety of cellular assays (Capsoni et al., 2011); and (3) disrupts processing of proNGF in cultures, leading to decreased secretion of mature NGF (Larsson et al., 2009; Fig. 6A). Moreover, topic injection of NGFR100X mutants does not induce acute pain (Capsoni et al., 2011; Sung et al., 2018; Fig. 4D). We have exploited these findings to obtain a “painless NGF” molecule for therapeutic purposes (Capsoni et al., 2017; Cattaneo and Capsoni, 2019), but how the features of NGFR100W contribute to the HSAN V phenotype is still far from clear. In particular, what are the consequences of this mutation on the overall architecture of pain-sensing structures and on nociceptive responses?
To shed light on these issues, we generated a new knock-in mouse line carrying the sequence encoding human NGFR100W. In homozygosity, this mutation causes early postnatal lethality, rescued by NGF administration (Testa et al., 2019). This situation matches the effect of complete deletion of both Ngf alleles (Crowley et al., 1994) and its rescue by overexpression of Ngf (Harrison et al., 2004), thus pointing to reduced NGF bioavailability due to haploinsufficiency as the prevalent mechanistic explanation for the lethal phenotype of homozygous NGFR100W/R100W mice. Heterozygosity might mitigate the reduction in NGF bioavailability and reduce lethal developmental effects. Moreover, the vitality of NGFR100W/wt mice allows to look for further effects of NGFR100W, including those possibly arising from peculiar signaling properties.
Heterozygous mice thrive normally and show no gross deficits, but display a reduction in nociception, accompanied by reduced skin innervation and altered density of nonmyelinated fibers, correlating with reduced NGF content in the skin.
The main DRG neuronal populations of NGFR100W/wt mice were unaffected and displayed a normal distribution in adults and newborns, which is in line with the full neurotrophic power of NGFR100W. This rules out the possibility of an effect of the mutation on DRG development. However, the expression of key pain transduction-related molecules was altered. Triggered by this finding, we analyzed the global gene expression profile of DRGs by performing a transcriptomic study. NGFR100W/wt mice presented only subtle alterations, with a few hundred genes, belonging to gene categories involved in immune response, phagocytosis, and Rho-GTPase cycling, changed to a significant extent. Among the most affected genes were TyroBP/DAP12 and TLR2, which modulate production of proinflammatory cytokines in neuropathic pain (Liu et al., 2012; Kobayashi et al., 2016). These interesting points deserve further investigation in the future, also considering the recent finding that microglia is a NGF target (Rizzi et al., 2018).
Another well established target of NGF, critically requiring TrkA signaling for proper development, is the sympathetic system (Levi-Montalcini and Booker, 1960; Glebova and Ginty, 2004; Kuruvilla et al., 2004). The SCG of NGFR100W/wt mice had a normal cell count, and peripheral sympathetic innervation of internal organs was similarly unaffected, further supporting the conclusion that NGF–TrkA signaling is intact in NGFR100W/wt mice.
The reactivity to algogens (capsaicin) and the sensitivity to non-noxious stimuli were decreased in NGFR100W/wt mice, correlating with the reduction in glabrous and hairy skin innervation, and in nonmyelinated fibers. These alterations fit with the lower NGF content in the target tissue (i.e., skin). Consistently, classical data show that NGF is not involved in establishing skin innervation during development, but is required for its maintenance in the adult (Davies et al., 1987), and the treatment of NGFR100W/wt mice with NGFwt was able to rescue their insensitivity to capsaicin-induced pain.
DRG neurons primed with NGFR100W are less prone to sensitization and show lower nociception-related biochemical responses than NGFwt. In addition, NGFR100W was unable to induce mechanical allodynia in vivo. This is in line with the failure of NGFR100W application to potentiate H+-evoked responses in DRG cultures (Sung et al., 2018) and extends data showing that NGFR100W impacts nociceptive, but not trophic functions of NGF (Capsoni et al., 2011; Sung et al., 2018). Binding of NGF to p75NTR is required for the upregulation of bradykinin receptor expression (Petersen et al., 1998) and NGFR100W has a reduced affinity to this receptor (Covaceuszach et al., 2010; Sung et al., 2018). Interestingly, reduced p75NTR binding can explain the common features of NGFR100W/wt mice with p75NTR−/− mice, which have reduced nociception and skin innervation, despite normal DRG structure (Bergmann et al., 1997). This strongly suggests that the molecular mechanism for the reduced activation of pain sensitization pathways by NGFR100W is a differential engagement of TrkA and p75NTR receptors.
Congenital insensitivity to pain is usually explained by defective nociceptor development or function (Chen et al., 2015; Minett et al., 2015; Nahorski et al., 2015), whereas nociceptors develop normally in HSAN V mice. Therefore, the reduced sensitizing activity of NGFR100W may contribute to explain the HSAN V phenotype.
Reduced NGF levels can explain the lethality of homozygous NGFR100W/R100W mice (Testa et al., 2019), thus mimicking the lethality of mNGF−/− mice (Crowley et al., 1994). Under heterozygosity, the mutation has a milder impact, both in human carriers (Perini et al., 2016) and mice (our data), despite reduced plasma and brain NGF (Fig. 6B,C).
HSAN IV, differently from HSAN V, is characterized by mental retardation and anhidrosis (Indo, 2001; Capsoni, 2014), showing that mutations in the ligand and in its receptor are not completely overlapping. Indeed, heterozygous mice (our data) and humans (Minde et al., 2009) show normal cognitive performance and normal density of cholinergic neurons, in striking contrast to mNGF+/− mice, further demonstrating that NGFR100W retains a full neurotrophic potential in the CNS and suggesting a specific effect of the R100W mutation on nociceptive pathways. In this regard, it is worth noticing that conditional deletion of Ngf or TrkA in nestin+ neurons causes a reduction of ChAT+ neurons in the medial septum, but not in the striatum (Müller et al., 2012). Likewise, deletion of TrkA from dlx5/6-expressing neurons caused a reduction in ChAT expression in the basal forebrain, but not in the striatum, along with mild cognitive deficits (Sanchez-Ortiz et al., 2012). We observed no cholinergic deficit in NGFR100W/wt mice, whereas this was detected in NGF+/− mice not only in the medial septum, but also in the striatum. These differences could be explained either by (1) the unaltered neurotrophic activity of NGFR100W or by (2) the reduction in glial NGF/TrkA production/signaling caused by the global heterozygous deletion in NGF+/− mice, which in conditional knock-out mice occurred only in the neuronal lineage. This suggests a contribution of microglia to NGF actions in the CNS, which is in agreement with the recent finding that microglia is an NGF target cell in the brain (Rizzi et al., 2018). On the other hand, nociception and the associated pathways were not analyzed in detail in neuron-specific Ngf and TrkA conditional knock-out mice (Müller et al., 2012; Sanchez-Ortiz et al., 2012).
Since NGF is a homodimer, it is not known whether heterodimers between NGFwt and NGFR100W protomers can be formed, and what are their peripheral and central functional properties. This possibility is suggested by the decreased secretion of NGFwt, when coexpressed with NGFR100W, and, therefore, it is tempting to speculate that the NGFR100W chain exerts a dominant-negative effect on wild-type chains. Considering that the expression of NGFR100W alone also results in reduced secretion, it is conceivable that the presence of this particular mutant, leading to either the formation of NGFwt-NGFR100W heterodimers or to a mixture of NGFwt-NGFwt and NGFR100W-NGFR100W homodimers can affect the secretion/processing quality control systems, thus resulting in reduced extracellular release. An antibody against the mutant NGF chain would be required to demonstrate the presence of heterodimers, which will be the focus of future research efforts.
By generating a new mouse model for the human NGFR100W mutation in the heterozygous state, we have provided new insights into the HSAN V “painlessness” condition, which call for future study of its potential as a new tool to dissect the multifaceted roles of NGF in the nervous system. In addition, the “painless” properties of NGFR100W hold promising potential to design much needed strategies to treat chronic pain states.
Footnotes
This work was supported by the EU FP7 PAINCAGE project (Grant 603191, to A.C.), by Fondazione Telethon (Grant GGP11179, to A.C.), by European Union's Horizon 2020 research funds under the Marie Skłodowska-Curie (Grant 674901, to A.S. and E.S.), and by MIUR_PRIN17 project (Grant 2017HPTFFC_001, to A.C.). We thank Lorenza Ronfani, Rosanna Rinaldi, and Ivana Benzoni (San Raffaele Hospital, Milan, Italy); Mara D'Onofrio (Rita Levi-Montalcini European Brain Research Institute); laboratory members Maria Antonietta Calvello, Vania Liverani, Nicola Maria Carucci, Francesco Gobbo, Caterina Rizzi, Alessandro Viegi, and Alexia Tiberi (BioSNS, Scuola Normale Superiore) for their help and support; Elena Novelli (Institute of Neuroscience, CNR) for valuable technical help; Nicola Origlia (Institute of Neuroscience, CNR) for support in electrophysiology experiments; Enrico Pracucci and Gian Michele Ratto [NEST (National Enterprise for nanoScience and nanoTechnology) Laboratory, Scuola Normale Superiore] for support in DRG sectioning; Irene Perini and India Morrison (Linköping University, Linköping, Sweden) or useful discussions; and Moses W. Chao (New York University, New York, NY) for useful discussion and critical reading of the manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to Antonino Cattaneo at antonino.cattaneo{at}sns.it or Simona Capsoni at simona.capsoni{at}sns.it
