Null mutations in the gene encoding themajor myelin protein of the central nervous system, proteolipid protein1 (PLP1), cause an X-linked form of spastic paraplegia (SPG2) associated with axonal degeneration.While motorsymptoms are the best known manifestations of this condition, its somatosensory disturbances have been de-scribed but poorly characterized.We carried out a longitudinal study in an animalmodel of SPG2—mice carryinga deletion of the Plp1 gene (Plp-null mice). Plp-null mice exhibited severe early-onset thermal hyperalgesia, inthe absence of thermal allodynia. We first performed an electrophysiological testing which showed an early de-crease in peripheral and spinal conduction velocities in Plp null mice. Such as the abnormal sensitive behaviors,this slowing of nerve conduction was observed before the development of myelin abnormalities at the spinallevel, from 3 months of age, and without major morphological defects in the sciatic nerve. To understand thelink between a decrease in nerve velocity and an increased response to thermal stimuli before the appearanceof myelin abnormalities, we focused our attention on the dorsal horn of the spinal cord, the site of integrationof somatosensory information. Immunohistochemical studies revealed an early-onset activation of astrocytesand microglia that worsened with age, associated later in age with perturbation of the expression of the sensoryneuropeptides calcitonin-gene-related peptide and galanin. Taken together, these results represent complemen-tary data supporting the hypothesis that Plp-nullmice suffer fromganglionopathy associatedwith late onset cen-tral demyelination but with few peripheral nerve alterations, induced by the glial-cell-mediated sensitization ofthe spinal cord. The mechanism suggested here could underlie pain experiments in other leukodystrophies aswell as in other non-genetic demyelinating diseases such as multiple sclerosis.
The white matter (WM) of the central nervous system (CNS) is acomplex structure composed of a vast number of axons sheathed in acompact lipid-rich membrane synthesized by oligodendrocytes, themyelin sheath. The proteolipid protein gene (PLP1), located on the Xchromosome, undergoes alternative splicing to produce the most
ogy and neuroprotection of dis-biophysique — INSERM 1107,enri Dunant, 63001 Clermont-7 79 13.
ect.com).
ghts reserved.
abundant proteins of CNSmyelin: full length PLP and its shorter isoform,DM20 (Baumann and Pham-Dinh, 2001). PLP is almost completely con-served across mammalian species, and naturally occurringmutations inthe PLP1 gene are associated with dysmyelinating disorders both inhumans and animals (Duncan et al., 2011).
In humans, the deletion of the PLP1 gene leads to a dysmyelinatingdisorder associated with ascending axonopathy, called spastic paraple-gia type 2 (SPG2) (Cailloux et al., 2000; Inoue, 2005). In male patientswith SPG2, motor milestones of the first year of life are not affected,but progressive gait abnormalities appear during the five followingyears, leading to severe spasticity and the loss of ambulation after pu-berty. The dysmyelinating process is characterized by a normal numberof oligodendrocytes which, however, produce a less stable and compactmyelin sheath associated with extensive axonal degeneration (Garbernet al., 1997; Garbern et al., 2002).
56 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
There is compelling evidence for the presence of pain or somatosen-sory dysfunctions in patients with primary myelin defects. In patientswith multiple sclerosis (MS), motor dysfunctions, the main clinicalsigns observed, are often associated with increased sensitivity(Michalski et al., 2011; Solaro and Uccelli, 2011). This is also true ofSPG2 patients, based on clinical observations of pain sensitivity, notablyafter surgery or during aging (Boespflug-Tanguy, personal data), anddescriptions of peripheral neuropathy (Garbern et al., 1999). Defectsin axon–glia interactions are suspected to be one of the main causativefactors of the symptoms of MS as well as of SPG2.
In a genetically modified mouse model of SPG2 carrying Plp1 genedeletion (Plp-null mice) (Klugmann et al., 1997), the age- and length-dependent degeneration of long spinal tracts has also been observed,along with the swelling of small-diameter axons (Garbern et al., 2002;Griffiths et al., 1998).While CNSmyelin sheathswere initially describedas rather normal-looking, defects inmyelin compaction have since beenreported in older animals by neuropathological analysis (Rosenbluthet al., 2006) as well as in vivo imaging (Hassen et al., 2009). However,unlike these anatomical investigations, behavioral studies in Plp-nullmice are scarce and have only revealed late-onset defects in rotarodperformance (Griffiths et al., 1998; Yin et al., 2006).
In this study, therefore, we investigated somatosensory function inPlp-null mice. We first carried out longitudinal behavioral studies forthe evaluation of hypersensitivity to noxious heat stimuli or allodynia.Next, we undertook electrophysiological analyses to study peripheraland central conduction, as well as morphological examinations of thesciatic nerves and spinal cord. Lastly, we performed immunohistochem-ical labeling of the spinal dorsal horn to look for abnormalities of cellularorganization that could underlie the central sensitization in Plp-nullmice. Our results provide new insights into the relationship betweenmyelin defects and somatosensory dysfunctions.
Materials and methods
Animals
Male Plp-null mice and their wild-type (WT) littermates were akind gift from Pr K.A. Nave, Gottingen Germany. They were generat-ed as previously described (Klugmann et al., 1997), bred for 15generations into the C57Bl/6N background in the breeding colonyof the Max Planck Institute for Experimental Medicine, and thentransferred to our animal facility. Genotyping was performed as pre-viously described (Klugmann et al., 1997). Mice were housed eightper cage in a temperature-controlled environment (22 °C) under a12:12 light/dark cycle (light from 7:00 AM. to 7:00 PM.), with adlibitum access to pelleted food (Harlan, Gannat, France) and water.Experiments were performed in accordance with French andEuropean Economic Community guidelines for the care and use oflaboratory animals. All efforts were made to minimize the numberof animals used and their suffering (86/09/EEC—Council Directiveof 24 November 1986).
Behavioral testing
Thermal hyperalgesia: tail flick testHeat-pain sensitivity in Plp-null mice was evaluated using the tail
flick test. The mouse was held through a towel with only the tail ex-posed. The test consisted of immersing the distal part of the mousetail in hot water at 46 °C, 48 °C or 52 °C. Before testing, the miceunderwent a habituation period to familiarize themwith the immobili-zation technique and tail immersion, in order to avoid stress due toma-nipulation. The habituation period involved 4 trials each on 4consecutive days, during which the latency between immersion in46 °Cwater and the rapidwithdrawal of the tailwasmanually recorded.To avoid tissue lesioning, a cut-off time of 30 s was set for the tail im-mersion. After each trial, the tail was dried and the mouse restored to
its home cage for an inter-trial interval (ITI) of 15 min. Testing at46 °Cwas performed under the same conditions on the fifth day, imme-diately after the habituation session. Testing at 48 °C and 52 °Cwas car-ried out subsequently, and each required only one day of habituation.The test sessions also consisted of 4 trials each, with an ITI of 15 min,and results are expressed as themean of the 4 trials. For each genotype,14 mice were tested every three months from 3 to 15 months of age.
Thermal allodynia: thermal place preference testThe thermal place preference test enables the evaluation of thermal
allodynia inmice. Two hot plate analgesiameters (34 cm long and 8 cmwide) (Bioseb, Paris, France) were placed side by side such that the ad-jacent thermal surfaces were enclosed in a single chamber made oftransparent Plexiglas (34 cm long, 16 cm wide and 25 cm high). Thevariation in temperature of each plate was limited to±0.1 °C. The tem-peratures used were chosen in accordance with the literature (Noëlet al., 2009; Shimizu et al., 2005). In this study, we restricted our analy-sis to warm temperatures that corresponded to those used to evaluatehyperalgesia. To evaluate heat allodynia,micewere tested for 3 sessionsof 5 min each. In the habituation session, both plates were set at 34 °C.In the second session, Plate 1was set at 38 °C, while Plate 2 remained at34 °C. In the third session, Plate 1 was adjusted to 42 °C while Plate 2remained at 34 °C. The mice were allowed to move freely betweenthe two plates for 5 min, and their movements, including time spenton eachplate and distance traveled, recordedusing a video tracking sys-tem (Viewpoint, Lyon, France). We then calculated the length of timespent in Plate 2 (fixed or habituation temperature: 34 °C) as a percent-age of that spent in Plate 1 (variable temperature: 34 °C, 38 °C and42 °C depending on the session) and the cumulative distance traveled.For each genotype, 14 mice were tested every three months from 3 to15 months of age.
Electrophysiology
All electrophysiological tests were carried out under standard condi-tions in a sound-attenuated and electrically shielded recording cham-ber. Recording was performed in vivo in mice under generalanesthesiawith amixture of ketaminehydrochloride and chlorobutanol(Imalgene®, Merial, Lyon, France) and diazepam (Valium®, Roche,Basel, Switzerland) (1 mL/kg of a solution containing 11.25 mgImalgene and 0.375 mg of Valium, administered intraperitoneally),a procedure previously shown to not significantly affect peripheralnerve velocities (Ho and Waite, 2002). During the tests, core bodytemperature was continuously monitored with a rectal probe (Mi-croprobe Thermometer, BAT-12, WPI, Sarasota, USA) and main-tained constant at 37 ± 1 °C with a thermostat-controlled heatingpad (Homeothermic Blanket System—Harvard Apparatus, Holliston,USA). Electroneuromyograms (ENMGs), sensory nerve conductionvelocities (NCVs) and spinal somatosensory evoked potentials(SEPs) were recorded by stimulating the caudal nerve (for NCVsand spinal SEPs) and the plantar nerve (for ENMGs), using aNeuropack μ MEB-9100 recording device (Nihon Koden, Tokyo,Japan) and needle electrodes (stainless steel, diameter: 0.4 mm,Medtronic Xomed Inc., Minneapolis, USA). For each genotype, 7mice each were evaluated at 3 and 15 months of age.
Caudal sensory nerve conduction velocityAnimalswere positioned on their ventral surface so as to let their tail
hang completely loose, and the tail cleaned with 70% alcohol. A pair ofrecording needle electrodes was inserted at the base of the tail, whilea pair of stimulating needle electrodes was placed 60 mm caudally. Aground needle electrode was inserted between the stimulating and re-cording electrodes (as illustrated in Fig. 3). Sensory NCVswere recordedin the orthodromic direction (duration: 0.1 ms; intensity 0–1 mA insteps of 0.1 mA, high-pass filter: 2 Hz; low-pass filter: 5 kHz). CaudalNCVs were calculated by measuring the latency between the stimulus
57B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
artifact and the peak of the elicited action potential, divided by the dis-tance (60 mm) between the recording and stimulating electrodes. Fiveresponses were averaged for each recording.
Electroneuromyographic recordings: the H reflexFor H-reflex recording, we employed the noninvasive recording
method introduced byHo andWaite (2002), which consists of stimulat-ing the tibial nerve at the ankle. Compound muscle action potentials(CMAPs) of the hindpaw (plantar muscles), including the direct muscleresponse (M-wave) and the monosynaptic reflex response (H-wave),were recorded bymeans of needle electrodes. For this, an active record-ing needle electrode was positioned in the plantar muscles (betweenthe second and third cuneiform bones) and a reference recording elec-trode inserted subcutaneously in the third toe. A ground electrode wasinserted at the base of tail (as illustrated in Fig. 4). Signals from the re-cording electrode were filtered (2 Hz high-pass and 5 kHz low-pass).The plantar nerve was stimulated percutaneously through a pair of nee-dle electrodes placed in the ankle (at the Achilles tendon). CMAPs wereevoked by 0.1 ms-long square electrical pulses. First, a test stimulationof 1 Hz (1 per second) at gradually increasing intensities (in steps of0.1 mA up to 1 mA) was used to determine the thresholds for M- andH-waves (threshold protocol). The average of 5 consecutive responseswas recorded for each stimulus intensity. Then, CMAPs were testedwith a 1 mA electrical stimulation to determineM- andH-wave thresh-olds at gradually increasing frequencies (0.1, 1, and 5 Hz) (adaptive pro-tocol). As before, the average of 5 consecutive responses was recordedfor each stimulation frequency: the latencies of M- and H-waves weremeasured as the time elapsed between the stimulus artifact and thepeak of each waveform. The amplitudes of M- and H-waves were mea-sured between the baseline before stimulation and the peak of eachwaveform.
Spinal somatosensory evoked potentialsWe adapted the protocol described by Chandran et al. (1994) to use
a stimulation site at the base of the tail instead of bilaterally at the tibialnerves (Chandran et al., 1994). In ventrally-positioned animals, the backwas shaved and disinfected. A longitudinal midline incision was madein the skin. Electrical stimulation (square pulses, 0.1 ms, 1 mA, 1 Hz)was applied through a pair of stimulating needle electrodes at thebase of the tail. Spinal SEP recording (average of 20 consecutive re-sponses, high-pass filter: 2 Hz; low-pass filter: 5 kHz) was carried outthrough needle electrodes gently inserted in the paravertebral muscle,40 mm rostral to the stimulating electrodes (upper lumbar level). Theground electrodewas inserted subcutaneously between the stimulatingand recording electrodes (as illustrated in Fig. 5). The latencies of N1-and N2-waves were measured as the time elapsed between the stimu-lus artifact and the peak of each waveform. The amplitudes of N1- andN2-waves were measured between the baseline before stimulationand the peak of each waveform.
Morphological analysis of the sciatic nerves and cervical spinal cord
Four Plp-null andWTmice each were sacrificed at 3 and 15 monthsof age under pentobarbital anesthesia by perfusion through the left ven-tricle with 0.9% normal saline followed by freshly prepared 4% parafor-maldehyde (PFA) and 2.5% glutaraldehyde in 0.1 M phosphate bufferedsaline (PBS), using a peristaltic pump (rate: 20 mL/min for 5 min). Thesciatic nerves and spinal cordwere removed and post-fixed overnight at4 °C in the same fixative solution, and then stored at 4 °C in 4% PFA forless than a month, until use. To obtain semi-thin sections, the sciaticnerve and 2 mm sections of the cervical spinal cord were extensivelywashed with sodium cacodylate and then post-fixed in 1% osmium te-troxide for 1 h. Sections were then dehydrated through a series of alco-hols and acetone. After embedding in epoxy resin, semi-thin coronalsections (700 nm in thickness) were cut and stained with toluidineblue for examination under an optical microscope (BX51; Olympus,
Southend-on-Sea, UK) at 100× magnification. Digital images were ob-tained as described for histological sections. Ultrathin sections, 70 nmin thickness, were stained with uranyl acetate and lead citrate and ex-amined under a transmission electron microscope (Hitachi H7650,Tokyo, Japan) at 80 kV.
Immunohistochemistry
Eight 3- and 15-month-old Plp-null andWTmicewere anesthetizedwith pentobarbital and sacrificed by transcardiac perfusion with freshlyprepared 4% PFA in 0.1 M PBS. The spinal cord was entirely removed;the cervical part was dissected and embedded in 6% agarose. 20 μm-thick coronal slices were cut using a Vibratome HM 650 V (Microm,Francheville, France). Immunohistochemical labeling was carried outon histological sections (6–10 spinal cord dorsal horns per mouse).Cell distribution was evaluated using the following primary antibodies(4 °C overnight): goat anti-calcitonin gene-related peptide (CGRP;1:600, Abcam, Cambridge, UK), rabbit anti-galanin (1:1500, Millipore,Temecula, USA), goat anti-ionized calcium-binding adapter molecule 1(Iba1; 1:1000, Abcam, Cambridge, UK), rabbit anti-glial fibrillary acidicprotein (GFAP; 1:4000, Millipore, Temecula, USA), mouse anti-myelinbasic protein (MBP; 1:1500, Millipore, Temecula, USA) and mouseanti-neuronal nuclei (NeuN; 1:1000,Millipore, Temecula, USA). Follow-ing 3 consecutive 15-minute washes in PBS with 0.3% Triton-X100, sec-tions were incubated in the appropriate biotinylated secondaryantibodies (1:5000, Jackson ImmunoResearch, West Grove, USA) for2 h at room temperature. After washing, sections were processed for1.5 h using a standard Vectastain ABC kit (Vector laboratories, Burlin-game, USA), and staining visualized with diaminobenzidine (DAB, Vec-tor laboratories, Burlingame, USA). Both genotypes were handledconcomitantly using the same solutions. Control slices were incubatedas described in the absence of primary or secondary antibodies. Noimmunolabeling was seen in controls (data not shown).
Histological sections were examined using an optical microscopeBX51 (Olympus, Southend-on-Sea, UK) at 4× and 10× magnifications.Digital imageswere capturedwith a CCD digital camera (Roper Scientif-ic GmbH, Ottobrunn, Germany) using Metamorph version 6.2 software(Roper Scientific GmbH, Ottobrunn, Germany). All image analyses andquantifications were performed using ImageJ software (ImageJ 1.44,National Institutes of Health, USA, http://rsbweb.nih.gov/ij/). The dorsalhorn was manually delimited and positive pixels in this region of inter-est were determined using the “threshold” tool of the software. Thetotal number of positive pixels (i.e. the area of immunoreactivity) wascalculated as a percentage of the total number of pixels for each sliceand averaged per animal (mean ± standard error of the mean, SEM).The distinct aspect of NeuN immunoreactivity allowed us to quantifythe number of positive cells using the “analyze particles” tool in ImageJ.For precise comparisons betweenWT and transgenic mice for each typeof labeling, we verified that the manually delineated surface of the dor-sal horn was statistically similar in size between genotypes (data notshown).
Statistical analysis
For behavioral studies, datawere analyzed using two-wayanalysis ofvariance (ANOVA) for repeated measures, with the age (in months) asthe within-subjects (i.e. repeated) factor, and the genotype (Plp-nullvs. WT mice) as the between-subjects factor. Post hoc comparisonswere carried out with Bonferroni tests for multiple comparisons be-tween Plp-null and WT mice at each time point.
For electrophysiological studies, a non-parametric analysis was per-formed using the Mann–Whitney U test to evaluate the effect of geno-type at the different time points or the effect of time on the differentgenotypes.
For immunohistochemical studies, a Student's t-test was carried outto compare the percentage of immunoreactive area or the number of
58 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
immunolabeled cells between Plp-null and WT mice at a given age, forthe different antibodies used.
All statistical analyses were carried out using StatView software(SAS institute Inc., Cary, NC). A p-value of b0.05 was considered thethreshold for statistical significance.
Results
Plp-null mice exhibit thermal hyperalgesia
To determine if Plp-null mice exhibited enhanced sensitivity topain, similar to patients with primary myelin defects or somatosen-sory dysfunctions, we assessed whether they displayed any changesin the heat pain sensitivity in the tail flick test. At 46 °C, Plp-null miceexhibited a significant decrease in their latency to tail withdrawalcompared to that of WT mice. Two-way ANOVA revealed a signifi-cant main effect of age [F(4,130) = 4.52, p b 0.01] and genotype[F(1,130) = 17.48, p b 0.001] but no significant age × genotype in-teraction [F(4,130) = 0.37, p N 0.05]. Although withdrawal laten-cies tended to decrease at all ages tested, post hoc analysisrevealed a significant decrease only in 6-month-old Plp-null micewhen compared to WT mice (p b 0.05; Bonferroni test) (Fig. 1A).
Similarly, at 48 °C and 52 °C, two-way ANOVA revealed a significantmain effect of age [F(4,130) = 14.22, p b 0.001 and F(4,130) = 30.68,p b 0.001, respectively] and genotype [F(1,130) = 82.99, p b 0.001and F(1,130) = 67.68, p b 0.001, respectively] but no significantage ×genotype interaction [F(4,130) = 0.60, p N 0.05 andF(4,130) = 0.48, p N 0.05, respectively]. Post hoc analysis revealedthat Plp-null mice exhibited a significant decrease in their withdrawallatencies at all ages tested, when compared to WT mice (p b 0.01;Bonferroni test), except at 3 months at 52 °C, where a slight but non-significant decrease was observed (Figs. 1B and C).
Fig. 1. Evaluation of thermal hyperalgesia inwild-type (WT) and Plp-nullmice using the tail flickages, although this was only significant at 6 months of age. (B) At 48 °C, Plp-null mice exhibite(C) At 52 °C, Plp-null mice exhibited a significant decrease in tail withdrawal latency at all agep b 0.001; Bonferroni test).
Plp-null mice exhibit no thermal allodynia
Because somatosensory dysfunctions are associated not only with in-creased pain sensitivity but also with allodynia, Plp-null mice were sub-jected to thermal place preference tests to determine whether theydisplayed allodynia in response to heat stimuli. In the habituation session,when the 2 plates were set at the same temperature (34 °C), two-wayANOVA revealed no difference between genotypes [F(1,130) = 0.23,p N 0.05] or ages [F(4,130) = 1.56, p N 0.05] and no age × genotype in-teraction [F(4,130) = 1.10, p N 0.05] in the time spent on each plate(Fig. 2A). When the variable-temperature plate was heated to 38 °C,whatever the age tested, Plp-null and WT mice showed a slight prefer-ence for the 34 °C plate, and this preference became more markedwhen the variable temperature plate was heated to 42 °C. Two-wayANOVA revealed no difference between genotypes [F(1,130) = 1.64,p N 0.05 and F(1,130) = 0.02, p N 0.05 for 34 °C vs. 38 °C and 34 °C vs.42 °C respectively] and no age × genotype interaction [F(4,130) = 2.26,p N 0.05 and F(4,130) = 0.26, p N 0.05 for 34 °C vs. 38 °C and 34 °C vs.42 °C respectively]. A slight difference between ages was observed onlyunder 34 °C vs. 38 °C conditions [F(4,130) = 4.06, p b 0.01] (Figs. 2B–C).
Differences in locomotor activity between genotypes are a well-known confounding factor in behavioral tests evaluating somatosen-sory functions. Locomotor activity during the thermal place prefer-ence test was therefore evaluated. Statistical analysis showed thatthere was a significant effect of genotype [F(1130) ≥ 15.99,p b 0.001 for all conditions] during all thermal allodynia tests.ANOVA also revealed a significant effect of age only during the habit-uation phase [F(4130) = 3.712, p b 0.01], and a genotype × age in-teraction only under 34 °C vs. 38 °C conditions [F(1130) = 3.477,p b 0.01]. The significant increase in the locomotor activity of Plp-null mice compared to that of WT mice was observed, regardless ofthe temperature of the plates, from the age of 9 months onward
test. (A) At 46 °C, Plp-nullmice exhibited a slight decrease in tail withdrawal latency at alld a significant decrease in tail withdrawal latency compared toWTmice at all ages tested.s except 3 months. Results are expressed as means ± SEM (*: p b 0.05; **: p b 0.01; ***:
Fig. 2. Evaluation of thermal allodynia in wild-type (WT) and Plp-null mice using the thermal place preference test. (A) Habituation session with the two plates at same temperatureshows that,whatever the age or genotype,mice showednopreference for any plate. (B) Under 34 °C vs. 38 °C conditions, therewas nodifference in thermal allodynia betweengenotypes.(C) Under 34°c vs. 42 °C conditions, there was no difference in thermal allodynia between genotypes. Results are expressed as means ± SEM.
59B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
(p b 0.05; Bonferroni test) (Table 1), i.e. after the onset of the firstsomatosensory dysfunctions. These data indicate that the somato-sensory dysfunctions observed in Plp-null mice are not related tohyperactivity.
Electrophysiology
Decrease in caudal nerve conduction velocity in Plp-null miceSince changes in sensitivity are often associated with electrophysio-
logical modifications (Arezzo et al., 2011), we next assessed electro-physiological responses in Plp-null mice, first focusing on peripheralNCVs (Fig. 3). We showed a decrease in the caudal NCV in Plp-nullmice when compared to WT mice; however this decrease reachedsignificativity at 15 but not 3 months of age (p = 0.013 andp = 0.073; Mann–Whitney U test, respectively). Whatever the geno-type, there was no significant change in the NCV between 3 and 15months (p N 0.05; Mann–Whitney U test). These data indicate thatthe deletion of the Plp1 gene leads to a delay in the response time ofthe caudal nerve at least in older mice.
Table 1Evaluation of total distance traveled by wild-type (WT) and Plp-null mice during each phase o
NS: non-significant; Bonferroni test.⁎ p b 0.05.⁎⁎ p b 0.01.
Increase in M- and H-wave latencies in electroneuromyograms ofPlp-null mice
ENMG recording permits the neurophysiological measurement ofsensorimotor function and is complementary to caudal NCV evaluation.The latency of the muscular response (M-wave) was significantly in-creased in Plp-nullmicewhen compared toWTmice at 3 and15months(p = 0.007 and p = 0.013 respectively; Mann–Whitney U test),whereas the latency of the H-wave was significantly higher in Plp-nullmice than in WT mice only at 15 months (p = 0.002; Mann–WhitneyU test) (Fig. 4). A statistically significant increase in theM-wave latencybetween 3 and 15 months of age was observed only in WT mice(p = 0.009; Mann–Whitney U test). In contrast, no difference in theamplitude of M- or H-waves was observed between genotypes or be-tween ages (p N 0.05;Mann–WhitneyU test) (Table 2). These data indi-cate that the ENMG dysfunction in Plp-null mice is due to a delay inwave latencies with no effect on their amplitudes.
The increase inwave latencies observed in Plp-null mice, is probablydue to conduction defects. However, desynchronization or delays in theinitiation of the reflex could also have the same consequences. In order
Fig. 3.Caudal nerve conduction velocity (NCV) inwild-type (WT) and Plp-nullmice at 3 and 15 months. Top left: rawdata obtained from15-month-old Plp-null andWTmice. Bottom left:schematic representation of the experimental setup, with black elements (+ and−) representing the stimulating electrodes, white elements (+ and−) representing the recording elec-trodes and the gray element representing the ground electrode. Right: compared toWTmice, Plp-null mice exhibited a statistically significant decrease in the caudal NCV at 15 months ofage, but not at 3 months. Results are expressed as means ± SEM (*: p b 0.05; Mann–Whitney U test).
60 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
to discriminate between these two hypotheses, we carried out a“threshold protocol” with stimulations of gradually increasing intensi-ties, but did not observe any difference in the response betweengenotypes at 3 or 15 months. The first evoked responses (M- and H-waves) were observed in both genotypes at 0.2 mA (3 months) and0.3 mA (15 months), respectively, and all mice exhibited evoked re-sponses at a stimulation of 0.5 mA stimulation. These data indicatethat Plp-null mice do not suffer from a defect in the initiation thresholdof the reflex, but only in its conduction.
Under normal conditions, the magnitude of the H-wave is attenuat-ed by repeated activation. This attenuation has been thought to reflectpresynaptic mechanisms mediated by descending inhibitory pathways(Hosoido et al., 2009) or post-activation depression i.e. the depletionof neurotransmitters at the synapse between Ia afferents and spinalmo-toneurons (Palmieri et al., 2004). To determine involved-wave attenua-tion, we carried out the “adaptive protocol” by stimulating at increasingrepetition rates. Plp-null and WT mice did not show any significant
Fig. 4. Latencies of the muscular (M) and sensitive component (H) of the electroneuromyM-waves shown here represented the average of 5 consecutive responses evoked by a 1 mmice. Bottom left: schematic representation of the experimental setup, with black elementhe recording electrode in the plantar muscle (−) and the reference recording electrode in3 months, only the latency of M-waves was increased in Plp-null mice, whereas at 15 momeans ± SEM (*: p b 0.05; **: p b 0.01; Mann–Whitney U test).
difference in their response at either 3 or 15 months of age (p N 0.05;Mann–Whitney U test). At 3 months, the H-wave amplitudes evokedby 1 Hz and 5 Hz stimulations were 56.4 ± 10.1% and 44.9 ± 11.1%of those evoked by a 0.1 Hz stimulation in WT mice, and 79.0 ± 8.8%and 32.4 ± 5.1% of the same in Plp-null mice. At 15 months of age,these percentages were 82.7 ± 9.4% and 50.2 ± 5.2% in WT mice vs.97.0 ± 8.2% and 57.5 ± 2.5% in Plp-null mice. These results suggestthat presynaptic mechanisms are not involved in the increase in the la-tency of M- and H-waves.
Alterations in spinal somatosensory evoked potentials in Plp-null micePLP is the main protein in CNS myelin, and axonal degeneration is
known to occur in both the descending motor and ascending sensorypathways in SPG2 (Edgar et al., 2004; Griffiths et al., 1998; Hassenet al., 2009). We therefore completed our electrophysiological explora-tionswith an analysis of central conduction by recording spinal SEPs. Toevaluate spinal SEPs, we adapted the protocol described by Chandran
ogram (ENMG) in wild-type (WT) and Plp-null mice at 3 and 15 months. The H- andA stimulation at 1 Hz. Top left: raw data obtained in 15-month-old Plp-null and WTts (+ and −) representing the stimulating electrodes, white elements representingthe third toe (+), and the gray element representing the ground electrode. Right: atnths, the latencies of both M- and H-waves were increased. Results are expressed as
NS: non-significant; Mann-Whitney U test.⁎ p b 0.05.
61B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
et al. (1994) with a minor modification: the stimulating electrode wasplaced at the base of the tail (in order to correlate the electrophysiolog-ical response with the tail flick reflex) and not bilaterally at the ankle(Chandran et al., 1994). Because of this change, we were able to recordonly the N1- and N2-waves (but not N3) but with an interpeak interval(0.60 ± 0.03 ms in 3-month-oldWTmice) very close to that describedby Chandran et al., (0.67 ± 0.04 ms in 12-week-oldmice), thus validat-ing our protocol.
Spinal SEP analysis revealed a significant increase in the latencyto the N1 peak in 15-month-old Plp-null mice when compared toWT mice (p = 0.02; Mann–Whitney U test), whereas at 3 months,this difference did not reach significance (p = 0.09; Mann–WhitneyU test). There was also an increase in N2 peak latency in Plp-nullmice at 3 and 15 months of age (p = 0.007 and p = 0.008 respec-tively; Mann–Whitney U test) when compared to WT mice. Whilethe peak latencies of the N1 and N2 components remained un-changed between 3 and 15 months in WT mice, they increased dur-ing the same period in Plp-null mice (N1: p = 0.002 and N2:p = 0.003; Mann–Whitney U test) (Fig. 5). These data indicate thatPlp-null mice exhibit a decrease in central nerve conduction.
There was no difference in the amplitude of the N1 componentof the spinal SEP between genotypes. The amplitude of the N2 com-ponent was increased in Plp-null mice only at the age of 3 months(p = 0.02; Mann–Whitney U test). When comparing mice of a sin-gle genotype at different ages, the only significant difference ob-served was a decrease in the amplitude of the N2 component inPlp-null mice between 3 and 15 months of age (p = 0.006;Mann–Whitney U test) (Table 3). Thus, defects in central nerveconduction are mainly restricted to a decrease in velocity in Plp-null mice.
Fig. 5. Latencies of the N1 and N2 components of spinal somatosensory evoked potentials in wmonth-old Plp-null and WT mice. Bottom left: schematic representation of the experimental sments (+ and −) represent the recording electrodes and the gray element representing theN2-wave, whereas at 15 months of age, the latencies of both the N1- and N2-waves were incrU test).
Raremorphological abnormalities in the sciatic nerves and late-onsetmyelinand axonal degeneration in the cervical spinal cord of Plp-null mice
We then assessed whether the functional alterations detectedwith neurophysiological studies were supported by the presence ofmorphological changes in the PNS or CNS. Neuropathological analy-ses were first performed in sciatic nerves. Semi-thin coronal sectionsof sciatic nerves stained with toluidine blue revealed that myelinthickness and axonal organization as well as the density of myelinat-ed fibers were similar in Plp-null mice and WT mice, regardless ofage (Fig. 6). At lower magnifications, sciatic nerve sections showedwell preserved nerve architecture in both genotypes and at bothages, with no fiber loss or increase in the endoneurial space due toedema or reduced myelin content (data not shown). The sciaticnerve of Plp-null mice exhibited axons with less regular crosssections than WT mice, as observed at 3 months, more strikingly, at15 months of age (Fig. 6, arrows). Electron micrographs of the sciaticnerve revealed the presence of both myelinating and non-myelinating Schwann cells at the same density in both WT and Plp-null mice. Myelin thickness and axonal organization seemed to bepreserved in Plp-null mice, and there were no obvious signs of de-generation or macrophage infiltration (Fig. 6). In older mice of bothgenotypes, some axonal swelling and large whorls of myelin mem-brane, which were often associated with degenerating axons orSchwann cells, were sometimes observed (data not shown). Togeth-er, these findings show that neuropathological abnormalities in thesciatic nerves of Plp-null mice are rare even in older mice, and arelimited to the abnormal morphology of the axons, i.e. a cross-section that is less regular than in WT mice. Given their rarity,
ild-type (WT) and Plp-null mice at 3 and 15 months. Top left: raw data obtained in 15-etup, with black elements (+ and −) representing the stimulating electrodes, white ele-ground electrode. Right: at 3 months, Plp-null mice exhibited increased latency only foreased. Results are expressed as means ± SEM (*: p b 0.05; **: p b 0.01; Mann–Whitney
Fig. 6. Section of the sciatic nerve inwild-type (WT) and Plp-null mice aged of 3 and 15 months. 700 nm-thick epoxy-embedded sciatic nerve sections stainedwith toluidine blue from 3-and 15-month-oldWT and Plp-null mice, photographed at 100Xmagnification.Myelin thickness and axonal organization seemed to be preserved in Plp-null mice. Arrows indicate axonswith abnormal less regular cross Sections. 70 nm-thick epoxy-embedded sciatic nerve sections stained with uranyl acetate and lead citrate from 15-month-old WT and Plp-null mice,photographed at 12000X magnification. Myelinating and non-myelinating Schwann cells were observed at the same density in both WT and Plp-null mice. Myelin thickness and axonalorganization seemed to be preserved even at this magnification.
62 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
morphological abnormalities in the Plp-null sciatic nerve are proba-bly not responsible for the somatosensory disorders described earli-er in this study.
Fig. 7. Expression ofmyelin basic protein (MBP) in the spinal dorsal horn of 3- and 15-month-olsion in the dorsal horn ofWT and Plp-nullmice at 4× and 10×magnifications, at 3 months (A) a15 months: MBP expression is significantly lower in Plp-null mice than in WT mice only at 15
Neuropathological analyses of the cervical spinal cordwere also per-formed, and revealed dysmyelination and axonal degeneration in 15-month-old but not in 3-month-old Plp-null mice (data not shown), as
d Plp-null andwild-type (WT)mice. Photomicrographs showing examples of MBP expres-nd 15 months (B). (C) Quantification ofMBP immunoreactivity in the dorsal horn at 3 andmonths of age. Results are expressed as means ± SEM (***: p b 0.001; Student's t-test).
63B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
previously reported both by others and ourselves (Garbern et al., 2002;Griffiths et al., 1998; Hassen et al., 2009). Because the abnormalsensitive behaviors as well as the decrease in nerve conduction velocityappear as soon as 3 months of age, it is very likely that the late-appearing myelin and axonal abnormalities in the spinal cord of Plp-null mice are not the causative events of these defects.
Immunohistochemical study
Decrease in the amount of myelin in Plp-null miceHaving established that behavioral and electrophysiological sensory
processes are significantly affected in Plp-null mice without any clearneuropathological substrates in either the PNSor CNS,we decided to ex-amine the cellular mechanisms potentially underlying these changes inthe dorsal horn of the spinal cord: the site of integration of somatosen-sory information. First, wemeasured the expression ofmyelin basic pro-tein (MBP), a major marker of myelination, in the spinal dorsal horn of3- and 15-month-old Plp-null mice. While there was no difference be-tween genotypes in relative MBP immunolabeling in 3-month-oldmice (p N 0.05; Student's t-test) (Figs. 7A and C), at 15 months of agemice with a deletion of the Plp1 gene exhibited a large and significant
Fig. 8. Expression of glial fibrillary acidic protein (GFAP) in the spinal dorsal horn of 3- andples of GFAP expression in the dorsal horn of WT and Plp-null mice at 4× and 10×immunolabeling in the dorsal horn at 3 and 15 months of age: Plp-null mice exhibit a larages. Results are expressed as means ± SEM (**: p b 0.01; ***: p b 0.001; Student's t-test
decrease in MBP immunoreactivity compared to WT mice (p b 0.0001;Student's t-test), reflecting a dysmyelinated state (Figs. 7B and C).
Astrocytic and microglial proliferation in the spinal dorsal horn ofPlp-null mice
The first response to demyelination is the activation of microglia(Jack et al., 2005) and astrocytes (Nair et al., 2008), and numerous stud-ies have demonstrated their influence on pain sensitivity in a variety ofinjurymodels (DeLeo and Yezierski, 2001; Scholz andWoolf, 2007).Wetherefore examined the spinal dorsal horn of 3- and 15-month-old Plp-null and WT mice for the presence of reactive astrocytes (using GFAPimmunolabeling) and microglia (using Iba1 immunolabeling).
The density of GFAP immunolabeling in the spinal dorsal horn wassignificantly enhanced in Plp-null mice compared to WT mice at both3 and 15 months of age (p = 0.005 and p b 0.0001, respectively;Student's t-test) (Figs. 8A–C).
Microglial reactivity, as revealed by the density of Iba1immunolabeling, was also significantly enhanced in Plp-null micewhen compared to WT mice at 3 and 15 months in the spinal dorsalhorn (p = 0.0007 and p = 0.002, respectively; Student's t-test)(Figs. 9A–C). These findings suggest that astrocytic and microglial
15-month-old Plp-null and wild-type (WT) mice. Photomicrographs showing exam-magnifications, at 3 months (A) and 15 months (B). (C) Quantification of GFAPge and significant increase in GFAP expression when compared to WT mice, at both).
Fig. 9. Expression of ionized calcium binding adapter molecule 1 (Iba1) in the spinal dorsal horn of 3- and 15-month-old Plp-null and wild-type (WT) mice. Photomicrographs showingexamples of Iba1 expression in the dorsal horn ofWT and Plp-null mice at 4× and 10×magnifications, at 3 months (A) and 15 months (B). (C) Quantification of Iba1 immunolabeling inthe dorsal horn at 3 months and 15 months of age: Plp-null mice exhibited a significant increase in Iba1 expression when compared to WT mice at both ages. Results are expressed asmeans ± SEM (**: p b 0.01; ***: p b 0.001; Student's t-test).
64 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
proliferation, similar to the reactive gliosis observed in other injurymodels, is also observed in the spinal dorsal horn of Plp-null mice.
Up-regulation of sensory neuropeptides in the spinal dorsal horn ofPlp-null mice
To identify potential differences in the total number of neurons inthe dorsal horn of Plp-null and WT mice at 3 and 15 months of age,we carried out immunolabeling for a neuron-specific nuclear protein,NeuN. No statistical difference in the percentage of NeuN-labeled areawas observed between the genotypes at 3 or 15 months (p N 0.05;Student's t-test) (Figs. 10A–C). An evaluation of the total number ofneurons revealed that at 15 months, Plp-null mice tend to have fewerneurons in the spinal dorsal horn than WT mice (p = 0.03; Student'st-test), although no difference was observed between the genotypes at3 months (p N 0.05; Student's t-test) (Fig. 10D).
The expression pattern of some sensory neuropeptides, such asCGRP and galanin, is up-regulated in the dorsal horn in neuropathicpain models (Liu and Hökfelt, 2002; Scholz and Woolf, 2007). Thus,we also examined the expression of these two neuropeptides in the spi-nal dorsal horn of Plp-null mice at both ages. The density of CGRPimmunolabeling was significantly up-regulated in the dorsal horn ofPlp-null mice when compared to WT mice at 15 months (p = 0.01;Student's t-test) (Figs. 11B and C), while there was no difference be-tween genotypes at 3 months (p N 0.05, Student-t-test) (Figs. 11A andC). Although, a slight increase in galanin expression was observed inthe dorsal horn of Plp-null mice at both 3 and 15 months of age
(Figs. 12A-C), this up-regulation was not statistically significant(p = 0.11 and p = 0.07, respectively; Student's t-test).
Discussion
This study shows for the first time that Plp-null mice exhibit se-vere early-onset thermal hyperalgesia that is not associated withchanges in thermal allodynic behavior. Electrophysiological andneuropathological findings indicate an early decrease in peripheraland spinal conduction velocities with no clear morphological sub-strate in the sciatic nerves, and only late-appearing axonal and mye-lin abnormalities in the spinal cord. However, immunohistochemicalanalyses of the spinal dorsal horn, the site of integration of somato-sensory information, reveal an early-onset activation/proliferationof astrocytes andmicroglia that worsens with age, leading to the per-turbation of sensory neuropeptide expression. Taken together, theseresults strongly suggest that Plp-null mice suffer from early-onsetganglionopathy associated with late-onset central demyelinationbut with few peripheral nerve alterations, probably induced byglial-cell-mediated spinal cord sensitization.
The behavioral characterization of Plp-null mice has been re-stricted up to now to dysfunctions in rotarod performance in oldmice (Griffiths et al., 1998). Here, we describe for the first time the
Fig. 10. Expression of the neuron-specific nuclear protein, NeuN, in the spinal dorsal hornof 3- and 15-month-old Plp-null andwild-type (WT)mice. Photomicrographs showing examplesof NeuN expression in the dorsal horn ofWT andPlp-nullmice at 4× and10×magnifications at 3 months (A) and 15 months (B). (C) Quantification of the percentage of theNeuN-labeledarea of the dorsal horn at both ages and (D) the total number of NeuN-labeled dorsal horn neurons per section in Plp-null and WT mice at both ages (D). Plp-null mice exhibited fewerneurons than WT mice at 15 months of age. Results are expressed as means ± SEM (*: p b 0.05; Student's t-test).
65B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
early appearance of a behavioral abnormality in response to somato-sensory stimuli: these mice exhibit thermal hyperalgesia, as shownby their shorter response times to thermal pain in the tail-flick testin animals as young as 3 months. In contrast, no modification inheat allodynia was observed in the thermal place preference test.Preliminary experiments, evaluating mechanical allodynia usingVon Frey test, showed reduced withdrawal threshold in Plp nullmice compared to WT mice. However, this mechanical allodynia ob-served in 3-month-old mice was not found in 9-month-old mice(data not shown). Taking together these results showed a strongand early hypersensitivity of Plp-null mice which is mainly restrictedto painful stimuli. To identify the cellular substrate of this hypersen-sitivity we then performed electrophysiological and histologicalanalyses.
Plp-null mice exhibit electrophysiological abnormalities characteristicsof a ganglionopathy
Schaumburg et al., in a very meticulous study, have shown that thestimulation of the distal part of the caudal nerves leads to the activation
of sensory axons that mediate pain and thermal sensitivity, and that arealmost exclusively unmyelinated or thinly myelinated (Schaumburget al., 2010). In our study, this protocol, which is well adapted for corre-lating electrophysiological datawith tail flick behavior, has demonstrat-ed that Plp-null mice develop a slowing of sensory conductionwith age.
ENMGs recorded during the Hoffman reflex (H-reflex) typicallyreveal two responses: an initial M-wave elicited by the direct activa-tion of motor axons, and a later H-wave elicited by the transsynapticactivation of motoneurons by group Ia muscle afferents through thespinal reflex pathway (Hosoido et al., 2009). In this study, Plp-nullmice exhibit an increase in the latency of the M-wave regardless ofthe age tested, but an increase in the latency of the H-wave only at15 months. There is no difference between the genotypes with re-spect to the amplitudes of the M- and H-waves. Protocols to deter-mine the threshold (stimulation with increasing intensities) andadaptation of responses (stimulation with increasing frequencies)reveal no differences between the genotypes, suggesting that the in-creased latencies of the M- and H-waves in Plp-null mice are not as-sociated with a defect in the initiation of the reflex but only itsconduction.
Fig. 11. Expression of the sensory neuropeptide calcitonin gene-related peptide (CGRP) in the spinal dorsal horn of 3- and 15-month-old Plp-null and wild-type (WT) mice. Photomicro-graphs showing examples of CGRP expression in the dorsal horn of WT and Plp-null mice at 4× and 10×magnifications at 3 months (A) and 15 months (B). (C) Quantification ofCGRP immunolabeling in the dorsal horn at 3 and 15 months of age: Plp-null mice exhibited a significant increase in CGRP expression at 15 months. Results are expressed asmeans ± SEM (*: p b 0.05; Student's t-test).
66 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
The evaluation of spinal SEP in Plp-null mice also reveals anincrease in the latencies of the N1 (only in old mice) and N2 peaks,but little modification of wave amplitudes. Previous reports havesuggested that the N1 peak is a result of conduction through large-diameter myelinated muscle afferents, while N2 is carried by fiberswith lower conduction velocities, notably large-diameter myelinat-ed cutaneous afferents (Chandran et al., 1994). These data highlightthe fact that central conduction is slowed in the myelinated fibers ofPlp-null mice.
The observed slowing of conduction velocities in the absence of aclear decrease in wave amplitudes or polyphasic CMAPs suggests a de-fect that is predominantly myelination-related, with little axonal loss,in Plp-null mice (Schaumburg et al., 2010; Zielasek and Toyka, 1999).Furthermore, the alteration of both peripheral and central conductionsupports the idea that both peripheral and central axons of dorsal rootganglion cells are affected, as observed in human ganglionopathies(Lauria et al., 2003).
Plp-null mice show no major sciatic nerve abnormalities, while spinal corddysmyelination and axonal defects are observed only in older mice
Numerous studies have associated hyperalgesia and allodynia inneuropathic models with demyelination (Bruna et al., 2010; Furushoet al., 2009). We therefore assessed whether the functional alterations
detected by our neurophysiological studies were confirmed by thepresence of clear morphological changes in the PNS or CNS. In ourstudy, the sciatic nerve of Plp-null mice reveals nomajor morphologicalabnormalities except with regard to the profiles of myelinated axons,which tend to bepolygonal rather than round in cross section, especiallyin oldermice. In the cervical spinal cord, dysmyelination and axonal de-generation are observed in 15-month-old Plp-null mice but not in 3-month-old mice. Very strikingly, while sensitive behaviors and electro-physiological dysfunctions are very early, and do not strongly evolvedwith age, neuropathological defects appear only in aged Plp null mice;suggesting that dysmyelination and axonal degeneration are not theprimary cause of sensitive defects. However, electrophysiological re-sponses can be altered by non-structural or very subtle structuralchanges linked, for example, to 1) ion channel composition and distri-bution along the axons, 2) the dysfunction of motoneurons ratherthan the loss of motoneuron cell bodies, or 3) the dysfunction of axonaltrafficking (Arezzo et al., 2011). Strikingly, changes in the shape of axo-nal cross sections have been related to a decrease in the expression ofneurofilaments and microtubules (Parhad et al., 1995; Solomon andZurn, 1981), and defects in axonal transport have already been de-scribed in Plp-null mice (Edgar et al., 2004). Because PLP/DM-20 areexpressed in the PNS, although levels are lower than in oligodendro-cytes (Patzig et al., 2011), these proteins may be functional in Schwanncells, and be involved in the change in the axonal cross section.
Fig. 12. Expression of the sensory neuropeptide galanin in the spinal dorsal horn of 3- and 15-month-old Plp-null and wild-type (WT) mice. Photomicrographs showing examples ofgalanin expression in the dorsal horn of WT and Plp-null mice at 4× and 10× magnifications at 3 months (A) and 15 months (B). (C) Quantification of galanin immunoreactivity inthe dorsal horn at 3 and 15 months of age: there was no significant difference between the genotypes in galanin expression even if there was a slight increase in Plp-null mice at bothages. Results are expressed as means ± SEM.
67B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
Plp-null mice showmorphofunctional alterations of the spinal dorsal horn,indicating central neuropathy
The ganglionopathy indicated by our electrophysiological studies, inthe absence of distinct early morphological changes in the peripheralnerves or spinal cord, is not sufficient to explain the development ofthe shorter response times to thermal pain in the tail-flick test in Plpnull mice. However, it suggests a defect in the integration of sensory in-formation.We thus focused our attention on the spinal cord dorsal horn,where somatosensory inputs are first integrated. Plp-nullmice display astrong increase in immunolabeling for both GFAP and Iba1 (markers forreactive astrocytes and microglia, respectively) that worsens with age,indicating increased proliferation of these glial cells. Reactive glial cellscan modulate spinal excitability and mediate hypersensitivity to painthrough the production and release of cytokines and growth factors,such as brain-derived neurotrophic factor, which can sensitize nocicep-tive neurons (Coull et al., 2005; Ulmann et al., 2008). We hypothesizedthat this increased activity of glial cells in the superficial dorsal horncould be responsible for the sensitization of spinal pain pathways andthen the consequent behavioral hypersensitivity observed in Plp-nullmice.
At 15 months, this increase in gliosis is associated with a significantdecrease in the number of NeuN-labeled neurons, but only to a trend intheir relative surface area. In addition, at the same age, the expression of
sensory neuropeptides in the dorsal horn is modified, with a significantup-regulation in CGRP and a slighter increase in galanin. These resultssuggest neuronal changes with age in Plp null mice with induction ofsensory neuropeptide expression. Enhanced expression of CGRP hasbeen described in numerous models of central neuropathic pain(Hulsebosch et al., 2009; Olechowski et al., 2009; Yu et al., 2009), andits observation in 15-month-old Plp null mice could be involved in theworsening with age of the hypersensitivity to noxious thermal stimuli.
Finally, despite cumulative evidences arguing for a glial-cell-mediated sensitization of the spinal cord, they do not rule out apossible involvement of changes at the peripheral receptor levelin the development of sensitive dysfunctions in Plp null mice. Nota-bly, acid-sensing ion channel 1 and transient Receptor PotentialVanilloid Type 1, two receptor already involved in the physiopa-thology of MS animal models, constitute interesting candidates(Musumeci et al., 2011; Vergo et al., 2011).
Abnormal spinal integration of sensory information could sustained thepain experiments observed in genetic and inflammatory myelin disorders
Plp-null mice reproduce many of the features observed in SPG2 andMS patients. In bothmyelin diseases, (a) sensitivity to pain is frequentlyassociatedwithmotor symptoms (MS:Michalski et al., 2011; Solaro andUccelli, 2011; SPG2: Cailloux et al., 2000; Inoue, 2005), (b) the neuro-pathological defects observed combine CNS demyelination, axonal
68 B. Petit et al. / Neurobiology of Disease 65 (2014) 55–68
defects and reactive gliosis (MS: Reynolds et al., 2011; SPG2: Garbernet al., 2002; Sima et al., 2009), and (c) both central and peripheral elec-trophysiological abnormalities are observed (MS: Gartzen et al., 2011;SPG2: Garbern et al., 1997, 1999). Thus, the abnormal processing of sen-sory information at the spinal level reported in mice in the presentstudy could also underlie the hypersensitivity observed in these myelindisorders.
It has been proposed that the cognitive defects observed in whitematter diseases are due to the dysmyelination of long axons and thesubsequent reduction in their conduction velocity, which limit the dis-tance over which neurons can fire in synchrony, perturb synaptic plas-ticity and degrade signal integration (Nave, 2010). Our results suggestthat a similar mechanism at the level of the spinal cord dorsal hornmay underlie the sensitivity defects observed in SPG2 patients.
Funding
This work was supported by the Neurodis Foundation, the EU-FP7Leukotreat project, the Regional Council of Auvergne and the EuropeanRegional Development Fund.
Acknowledgments
We thank the personnel of the Center for Medical Cell Imaging(CICS) of Auvergne University for their technical assistance concerningthe histological and morphological study, and Said Ghandour, AuroraPujol and Homa Adle-Biassette for the very interesting discussionsabout brain morphology, as well as, Klaus Nave who gave us access tothe transgenic mice.
References
Arezzo, J.C., Litwak, M.S., Zotova, E.G., 2011. Correlation and dissociation of electrophysi-ology and histopathology in the assessment of toxic neuropathy. Toxicol. Pathol. 39,46–51.
Baumann, N., Pham-Dinh, D., 2001. Biology of oligodendrocyte and myelin in the mam-malian central nervous system. Physiol. Rev. 81, 871–927.
Bruna, J., Udina, E., Alé, A., Vilches, J.J., Vynckier, A., Monbaliu, J., et al., 2010. Neurophysi-ological, histological and immunohistochemical characterization of bortezomib-induced neuropathy in mice. Exp. Neurol. 223, 599–608.
Cailloux, F., Gauthier-Barichard, F., Mimault, C., Isabelle, V., Courtois, V., Giraud, G., et al.,2000. Genotype–phenotype correlation in inherited brain myelination defects dueto proteolipid protein gene mutations. Clinical European Network on BrainDysmyelinating Disease. Eur. J. Hum. Genet. 8, 837–845.
Chandran, A.P., Oda, K., Shibasaki, H., Pisharodi, M., 1994. Spinal somatosensory evokedpotentials in mice and their developmental changes. Brain Dev. 16, 44–51.
Coull, J.A.M., Beggs, S., Boudreau, D., Boivin, D., Tsuda, M., Inoue, K., et al., 2005. BDNF frommicroglia causes the shift in neuronal anion gradient underlying neuropathic pain.Nature 438, 1017–1021.
DeLeo, J.A., Yezierski, R.P., 2001. The role of neuroinflammation and neuroimmune activa-tion in persistent pain. Pain 90, 1–6.
Duncan, I.D., Kondo, Y., Zhang, S.-C., 2011. The myelin mutants as models to study myelinrepair in the leukodystrophies. Neurotherapeutics 8, 607–624.
Edgar, J.M., McLaughlin, M., Yool, D., Zhang, S.-C., Fowler, J.H., Montague, P., et al., 2004.Oligodendroglial modulation of fast axonal transport in a mouse model of hereditaryspastic paraplegia. J. Cell Biol. 166, 121–131.
Furusho, M., Dupree, J.L., Bryant, M., Bansal, R., 2009. Disruption of fibroblast growth fac-tor receptor signaling in nonmyelinating Schwann cells causes sensory axonal neu-ropathy and impairment of thermal pain sensitivity. J. Neurosci. 29, 1608–1614.
Garbern, J.Y., Cambi, F., Tang, X.M., Sima, A.A., Vallat, J.M., Bosch, E.P., et al., 1997. Proteolipidprotein is necessary in peripheral as well as central myelin. Neuron 19, 205–218.
Garbern, J.Y., Cambi, F., Lewis, R., Shy,M., Sima, A., Kraft, G., et al., 1999. Peripheral neurop-athy caused by proteolipid protein genemutations. Ann. N. Y. Acad. Sci. 883, 351–365.
Garbern, J.Y., Yool, D.A., Moore, G.J., Wilds, I.B., Faulk, M.W., Klugmann, M., Nave, K.A.,Sistermans, E.A., van der Knaap, M.S., Bird, T.D., Shy, M.E., Kamholz, J.A., Griffiths,I.R., 2002 Mar. Patients lacking the major CNS myelin protein, proteolipid protein 1,develop length-dependent axonal degeneration in the absence of demyelinationand inflammation. Brain 125 (Pt 3), 551–561.
Gartzen, K., Katzarava, Z., Diener, H.-C., Putzki, N., 2011. Peripheral nervous system in-volvement in multiple sclerosis. Eur. J. Neurol. 18, 789–791.
Griffiths, I., Klugmann, M., Anderson, T., Yool, D., Thomson, C., Schwab, M.H., et al., 1998.Axonal swellings and degeneration in mice lacking the major proteolipid of myelin.Science 280, 1610–1613.
Hassen, W.B., Bégou, M., Traore, A., Moussa, A.B., Boehm, N., Ghandour, M.S., et al.,2009. Characterisation of spinal cord in a mouse model of spastic paraplegia re-lated to abnormal axono-myelin interactions by in vivo quantitative MRI.Neuroimage 46, 1–9.
Ho, S.M., Waite, P.M.E., 2002. Effects of different anesthetics on the paired-pulse depres-sion of the h reflex in adult rat. Exp. Neurol. 177, 494–502.
Hosoido, T., Motoyama, S., Goto, M., Mori, F., Tajima, T., Hirata, H., Wada, N., 2009. Char-acteristics of H- and M-waves recorded from rat forelimbs. Neurosci. Lett. 450,239–241.
Inoue, K., 2005. PLP1-related inherited dysmyelinating disorders: Pelizaeus–Merzbacherdisease and spastic paraplegia type 2. Neurogenetics 6, 1–16.
Jack, C., Ruffini, F., Bar‐Or, A., Antel, J.P., 2005. Microglia and multiple sclerosis. J. Neurosci.Res. 81, 363–373.
Klugmann, M., Schwab, M.H., Pühlhofer, A., Schneider, A., Zimmermann, F., Griffiths, I.R.,et al., 1997. Assembly of CNS myelin in the absence of proteolipid protein. Neuron18, 59–70.
Lauria, G., Pareyson, D., Sghirlanzoni, 2003. Neurophysiological diagnosis of acquired sen-sory ganglionopathies. Eur. Neurol. 50, 146–152.
Liu, H.-X., Hökfelt, T., 2002. The participation of galanin in pain processing at the spinallevel. Trends Pharmacol. Sci. 23, 468–474.
Michalski, D., Liebig, S., Thomae, E., Hinz, A., Bergh, F.T., 2011. Pain in patients with mul-tiple sclerosis: a complex assessment including quantitative and qualitative measure-ments provides for a disease-related biopsychosocial pain model. J. Pain Res. 4,219–225.
Musumeci, G., Grasselli, G., Rossi, S., De Chiara, V., Musella, A., Motta, C., et al., 2011. Tran-sient receptor potential vanilloid 1 channels modulate the synaptic effects of TNF-αand of IL-1β in experimental autoimmune encephalomyelitis. Neurobiol. Dis. 43,669–677.
Nair, A., Frederick, T.J., Miller, S.D., 2008. Astrocytes in multiple sclerosis: a product oftheir environment. Cell. Mol. Life Sci. 65, 2702–2720.
Nave, K.-A., 2010. Myelination and support of axonal integrity by glia. Nature 468,244–252.
Noël, J., Zimmermann, K., Busserolles, J., Deval, E., Alloui, A., Diochot, S., et al., 2009. Themechano-activated K+ channels TRAAK and TREK-1 control both warm and coldperception. EMBO J. 28, 1308–1318.
Olechowski, C.J., Truong, J.J., Kerr, B.J., 2009. Neuropathic pain behaviours in a chronic-relapsing model of experimental autoimmune encephalomyelitis (EAE). Pain 141,156–164.
Palmieri, R.M., Ingersoll, C.D., Hoffman, M.A., 2004. The Hoffmann reflex: methodologicconsiderations and applications for use in sports medicine and athletic training re-search. J. Athl. Train. 39, 268–277.
Parhad, I.M., Scott, J.N., Cellars, L.A., Bains, J.S., Krekoski, C.A., Clark, A.W., 1995. Axonal at-rophy in aging is associated with a decline in neurofilament gene expression.J. Neurosci. Res. 41, 355–366.
Patzig, J., Jahn, O., Tenzer, S., Wichert, S.P., de Monasterio-Schrader, P., Rosfa, S., et al.,2011. Quantitative and integrative proteome analysis of peripheral nerve myelinidentifies novel myelin proteins and candidate neuropathy loci. J. Neurosci. 31,16369–16386.
Reynolds, R., Roncaroli, F., Nicholas, R., Radotra, B., Gveric, D., Howell, O., 2011. The neu-ropathological basis of clinical progression in multiple sclerosis. Acta Neuropathol.122, 155–170.
Rosenbluth, J., Nave, K., Mierzwa, A., Schiff, R., 2006. Subtle myelin defects in PLP‐nullmice. Glia 54, 172–182.
Schaumburg, H.H., Zotova, E., Raine, C.S., Tar, M., Arezzo, J., 2010. The rat caudal nerves: amodel for experimental neuropathies. J. Peripher. Nerv. Syst. 15, 128–139.
Scholz, J., Woolf, C.J., 2007. The neuropathic pain triad: neurons, immune cells and glia.Nat. Neurosci. 10, 1361–1368.
Shimizu, I., Iida, T., Guan, Y., Zhao, C., Raja, S.N., Jarvis, M.F., et al., 2005. Enhanced thermalavoidance in mice lacking the ATP receptor P2X3. Pain 116, 96–108.
Sima, A.A.F., Pierson, C.R., Woltjer, R.L., Hobson, G.M., Golden, J.A., Kupsky, W.J., et al.,2009. Neuronal loss in Pelizaeus–Merzbacher disease differs in various mutations ofthe proteolipid protein 1. Acta Neuropathol. 118, 531–539.
Solaro, C., Uccelli, M.M., 2011. Management of pain in multiple sclerosis: a pharmacolog-ical approach. Nat. Rev. Neurol. 7, 519–527.
Solomon, F., Zurn, A., 1981. The cytoskeleton and specification of neuronal morphology.Neurosci. Res. Program Bull. 19, 100–124.
Ulmann, L., Hatcher, J.P., Hughes, J.P., Chaumont, S., Green, P.J., Conquet, F., et al., 2008. Up-regulation of P2X4 receptors in spinal microglia after peripheral nerve injury medi-ates BDNF release and neuropathic pain. J. Neurosci. 28, 11263–11268.
Vergo, S., Craner, M.J., Etzensperger, R., Attfield, K., Friese, M.A., Newcombe, J., et al., 2011.Acid-sensing ion channel 1 is involved in both axonal injury and demyelination inmultiple sclerosis and its animal model. Brain 134, 571–584.
Yin, X., Baek, R.C., Kirschner, D.A., Peterson, A., Fujii, Y., Nave, K.-A., et al., 2006. Evolutionof a neuroprotective function of central nervous system myelin. J. Cell Biol. 172,469–478.
Yu, L.-C., Hou, J.-F., Fu, F.-H., Zhang, Y.-X., 2009. Roles of calcitonin gene-related peptideand its receptors in pain-related behavioral responses in the central nervous system.Neurosci. Biobehav. Rev. 33, 1185–1191.
Zielasek, J., Toyka, K.V., 1999. Nerve conduction abnormalities and neuromyotonia in ge-netically engineered mouse models of human hereditary neuropathies. Ann. N. Y.Acad. Sci. 883, 310–320.
