HUMAN GENE THERAPY 12:2237–2249 (December 10, 2001)Mary Ann Liebert, Inc.

Continuous Delivery of Neurotrophin 3 by Gene Therapy Has a Neuroprotective Effect in Experimental Models of

Diabetic and Acrylamide Neuropathies

PIERRE-FRANÇOIS PRADAT,1,2 PHILIPPE KENNEL,3 SOUAD NAIMI-SADAOUI, 3 FRANÇOISEFINIELS,3 CECILE ORSINI,3 FRÉDÉRIC REVAH,4 PIA DELAERE,3 and JACQUES MALLET1

ABSTRACT

Neurotrophic factors (NFs) are promising agents for the treatment of peripheral neuropathies such as dia-betic neuropathy. However, the value of treatment with recombinant NF is limited by the short half-lives ofthese molecules, which reduces efficiency, and by their potential toxicity. We explored the use of intramus-cular injection of a recombinant adenovirus encoding NT-3 (AdNT-3) to deliver sustained low doses of NT-3. We assessed its effect in two rat models: streptozotocin (STZ)-induced diabetes, a model of early diabeticneuropathy characterized by demyelination, and acrylamide experimental neuropathy, a model of diffuse ax-onal neuropathy which, like late-onset human diabetic neuropathy, results in a diffuse sensorimotor neu-ropathy with dysautonomy. Treatment of STZ-diabetic rats with AdNT-3 partially prevented the slowing ofmotor and sensory nerve conduction velocities (p , 0.01 and p , 0.0001, respectively). Treatment with AdNT-3 of acrylamide-intoxicated rats prevented the slowing of motor and nerve conduction velocities (p , 0.001and p , 0.0001, respectively) and the decrease in amplitude of compound muscle potentials (p , 0.0001), anindex of denervation. Acrylamide-intoxicated rats treated with NT-3 had higher than control levels of mus-cle choline acetyltransferase activity (p , 0.05), suggesting greater muscle innervation. In addition, treatmentof acrylamide-intoxicated rats with AdNT-3 significantly improved behavioral test results. Treatment withAdNT-3 was well tolerated with minimal muscle inflammation and no detectable general side effects. There-fore, our results suggest that NT-3 delivery by adenovirus-based gene therapy is a promising strategy for theprevention of both early diabetic neuropathy and axonal neuropathies, especially late axonal diabetic neu-ropathy.

2237

OVERVIEW SUMMARY

The value of NT-3 delivery by gene therapy was tested inanimal models of neuropathy. Diabetic rats were given in-tramuscular injections of an adenovirus encoding NT-3(AdNT-3). This treatment was also given to acrylamide-in-toxicated rats, which develop a neuropathy with diffuse ax-onal degeneration responsible for sensorimotor deficits anddysautonomia, as in human late axonal diabetic neuropa-thy. Treatment with AdNT-3 partially prevented the elec-trophysiological abnormalities related to the neuropathy inthese two models. In addition, treatment of acrylamide-in-

toxicated rats improved behavioral test results and in-creased muscle choline acetyltransferase activity, suggest-ing increased muscle innervation. Therefore, continuous de-livery of NT-3 by gene therapy is a promising strategy forthe prevention of diabetic neuropathy.

INTRODUCTION

DIABETIC NEUROPATHY is a common complication of diabetesmellitus. It has been estimated that about 50% of individ-

uals who have had diabetes for 25 years suffer symptomatic pe-

1Laboratoire de Génétique Moléculaire de la Neurotransmission et des Processus Neurodégénératifs (LGN), UMR C9923, Centre National dela Recherche Scientifique, Hôpital de la Pitié-Salpétrière, 75651 Paris Cedex 13, France.

2Fédération de Neurologie Mazarin, Hôpital de la Pitié-Salpétrière, 75651 Paris Cedex 13, France.3Aventis Pharma Gencell, 94403 Vitry sur Seine, France.4Cerep, 92500 Rueil Malmaison, France.

ripheral neuropathy (Pirart, 1978). The most common type ofneuropathy is symmetrical polyneuropathy that is primarily sen-sorimotor and often includes the autonomic system. Sensorysigns and symptoms may be divided into those related to theloss of sensory large fibers, which are responsible for convey-ing vibratory sensation and proprioception, and those associ-ated with the loss of sensory small fibers, which are responsi-ble for conveying pain and temperature sensations. Diabeticneuropathy is caused by segmental and paranodal demyelina-tion (Behse et al., 1977) and axonal degeneration (Chopra etal., 1969; Chopra and Fannin, 1971).

Therapeutic agents to combat diabetic neuropathy would ob-viously have profound effects on neurological symptoms butalso on other diabetic complications associated with peripheralnerve disorders such as trophic troubles and foot ulcerations,which can lead to amputation. There are good arguments in fa-vor of using neurotrophic factor neurotrophin 3 (NT-3) as a po-tential therapeutic agent (Tomlinson et al., 1997). Indeed, NT-3 promotes the survival of large-fiber sensory neurons, motorneurons, and sympathetic neurons (Barbacid, 1994), all ofwhich are impaired in diabetic neuropathy. Furthermore, thepotential of NT-3 for the treatment or prevention of neu-ropathies is supported by evidence from animal models of sen-sory neuropathies, such as pyridoxine- and cisplatin-inducedneuropathies (Gao et al., 1995; Helgren et al., 1997; Pradat etal., 2001), and hereditary motor neuropathy (Haase et al., 1997).

However, the use of trophic factors is limited by their poorbioavailability after systemic administration. In particular, theplasma half-life of NT-3 after intravenous administration in ratis only 1.28 min (Poduslo and Curran, 1996). Because of theirshort half-life (Dittrich et al., 1994; Poduslo and Curran, 1996),neurotrophic factors might not reach the peripheral neurons insufficient quantities to exert a beneficial effect. The clinical useof neurotrophic factors thus requires the development of suit-able modes of delivery. The physiological condition may wellbe the continuous delivery of low doses of recombinant neu-rotrophic factor and, therefore, this type of administration maybe more effective and safer for therapeutic applications than re-peated injections of high doses (Yuen and Mobley, 1995; Gravelet al., 1997; Sendtner, 1997). Gene therapy is a promising strat-egy to achieve this goal, as shown by experiments in animalmodels of neuropathies, such as axotomy in newborn rats(Gravel et al., 1997) and mouse models of hereditary motorneuropathy (Haase et al., 1997) or cisplatin-induced sensoryneuropathy (Pradat et al., 2001).

Animal models of diabetic neuropathies can be used to testnew therapeutic strategies. The streptozotocin (STZ)-diabeticrat is the most commonly used animal model of diabetes. In theSTZ-diabetic rat, an early slowing of motor and sensory nerveconduction velocities develops about 2–4 weeks after the onsetof diabetes. The animals do not present neuropathic symptoms(Bravenboer et al., 1993) or axonal degeneration (Sharma andThomas, 1984). This is an appropriate model of early diabeticneuropathy because abnormalities in nerve conduction veloci-ties in the absence of significant axonal degeneration are anearly feature of human diabetic neuropathy and are frequentlyfound at the presymptomatic stage (Lamontagne and Buchthal,1970). Diabetic rats have a short life span, which precludes thedevelopment of axonal damage, a marked feature of human di-abetic neuropathy at the symptomatic stage (Chopra and Fannin,

1971). This highlights the need for preclinical studies to test theefficacy of treatments not only in STZ-diabetic rats but also inanimal models of diffuse acquired axonal neuropathies. Experi-mental acrylamide neuropathy is a classic model of diffuse pe-ripheral axonal degeneration in all animals in which it has beenstudied (mice, rats, cats, and monkeys) (Bradley and Asbury,1970; Hopkins, 1970; Schaumburg et al., 1974; Spencer andSchaumburg, 1977; Saita et al., 1995). An advantage of the acry-lamide neuropathy model is that animals present clear clinicalsigns. Therefore, the effect of therapeutic intervention on clini-cal markers may be predictive of the effects on symptoms of neu-ropathy in humans.

We used these two complementary animal models to evalu-ate the efficacy of NT-3 delivery to the bloodstream after intra-muscular injection of a recombinant adenovirus carrying thegene. We used electrophysiological, behavioral, and biochemi-cal markers as end points for assessing the efficacy of treatment.

MATERIALS AND METHODS

Animals

Male Sprague-Dawley rats (IFFA-Credo, L’Arbresle,France) were used for the acrylamide intoxication and STZ-in-duced diabetes experiments (weighing 280 and 160 g, respec-tively). Animals were maintained in a room with controlled tem-perature (21–22°C) and a light–dark cycle (12 hr/12 hr) withfood and water available ad libitum. All experiments were car-ried out in accordance with institutional guidelines.

Design of treatment of normal rats with AdNT-3

Rats of group 1 (n 5 10) received an intramuscular injectionof an adenovirus encoding NT-3 (AdNT-3, 1010 PFU [plaque-forming units]/animal). Rats of group 2 (n 5 10) received in-tramuscular injections of saline. Electromyography was per-formed 3 weeks after intramuscular injections.

Design of the streptozotocin-induced diabetes experiments

Diabetes was induced by a single intraperitoneal injection(50 mg/kg) of STZ (Sigma, St. Louis, MO) freshly dissolvedin 0.9% sterile saline (day 0). Glucose levels were measured ina blood sample obtained 72 hr later by tail prick, using a glu-cose oxidase-impregnated test strip and reflectance meter (Glu-costix and Glucometer; Bayer Diagnostics, Ames Department,Puteaux, France). An animal was considered diabetic if the glu-cose concentration was greater than 15 mM. Glucose levelswere measured, and the animals were weighed every weekthereafter for the entire study period. Rats were randomly as-signed to four groups. Three groups (groups 1, 2, and 3) con-sisted of diabetic animals. Rats of group 1 (n 5 16) were givenintramuscular injections of AdNT-3 (1010 PFU). Rats of group2 (n 5 16) were given intramuscular injections of an adenovi-rus encoding Escherichia coli b-galactosidase (Adbgal), andrats of group 3 (n 5 44) were given intramuscular injections ofsaline. All injections were given 7 days after the induction ofdiabetes. A fourth group (n 5 10) consisted of nondiabetic con-trol rats receiving intramuscular injections of saline.

PRADAT ET AL.2238

Electromyographic measurements were performed on day 5(48 hr before the intramuscular injection of adenovirus), andthen on days 21, 28, and 35. Plasma samples were randomlytaken from eight animals per group on day 3 (4 days before theinjection of adenovirus) and then on days 10, 17, 24, 31, and38. Rats were killed on day 38 and triceps brachii muscles wereimmediately dissected.

Design of the acrylamide intoxication experiments

Four experimental groups were formed. Day 0 was definedas the first day of behavioral tests (see below). Three groupswere treated with acrylamide (Sigma). Acrylamide was dilutedin 0.9% sterile saline and administered intraperitoneally (50mg/kg per injection) on day 7 and then three times per week for3 weeks. Forty-eight hours before initiation of acrylamide treat-ment (on day 5), rats of group 1 (n 5 20) were given intramus-cular injections of AdNT-3 (1010 PFU/animal); rats of group 2 (n 5 20) were given intramuscular injections of Adbgal(1010 PFU/animal); and rats of group 3 (n 5 35) received in-tramuscular injections of saline. A fourth group (n 5 20) of nor-mal control rats received intraperitoneal and intramuscular in-jections of saline.

Behavioral tests were performed on day 0 (basal value de-termination) and then each week during the intoxication period(on days 14, 21, and 28). Electromyography was performed atthe end of the intoxication period (day 31).

Intramuscular injections

Animals were anesthetized with pentobarbital sodium (Sanofi,Libourne, France) and intramuscular injections were given witha disposable insulin syringe equipped with a 27-gauge needle.Each animal was injected with adenoviral vectors at a concentra-tion of 1010 PFU in sterile saline. Muscles of the forelimbs, ex-tensor digitorum (100 ml/muscle), and triceps brachii (100 ml/mus-cle) were injected bilaterally. The total volume was injected into5 to 10 sites per muscle. Vector injection and animal care con-formed to L2 biosecurity conditions and institutional guidelines.

Adenoviruses

pXL3253 and pXL2822 are E1,E3-deleted adenoviral back-bones with an expression cassette inserted in place of the E1region. pXL3253 contains the mouse prepro and mature NT-3cDNAs under the transcriptional control of the cytomegalovirus(CMV) promoter. pXL2822 contains the E. coli lacZ gene witha nuclear localization signal (nls) under the control of the CMVpromoter as previously described (Crouzet et al., 1997). Briefly,these two Ad5-derived adenoviral backbones were obtained byrecombination in E. coli (Crouzet et al., 1997). PacI-restrictedbackbones were used to transfect 293 cells to obtain the corre-sponding adenoviruses, AdNT-3 and Adbgal. The adenoviruseswere amplified as previously described (Crouzet et al., 1997),purified from CsCl gradients, and titered on 293 cells accord-ing to standard methods. The viral titers were 5.5 3 1010

PFU/ml for AdNT-3 and 1.2 3 1010 PFU/ml for Adbgal.

ELISA

A sandwich enzyme-linked immunosorbent assay (ELISA)was used to evaluate NT-3 immunoreactivity in plasma and

muscle samples as described previously (Pradat et al., 2001).The sensitivity of the ELISA was 15 pg/ml with the lysis bufferused for the homogenization of muscles and 15 pg/ml in ratplasma.

Immunohistochemistry

Muscles were fixed with 7% paraformaldehyde in phosphate-buffered saline (PBS), cryoprotected by incubation overnightin 30% sucrose, and then frozen in isopentane cooled in liquidnitrogen. Transverse cryostat sections (15 mm thick) of musclefibers were transferred onto Superfrost/Plus slides (Fisher, Pitts-burgh, PA) and studied by immunohistochemistry for NT-3 andby histochemistry for b-galactosidase.

A rabbit polyclonal antibody against NT-3 was used for NT-3 immunohistochemistry (R&D Systems, Abingdon, UK). Theoptimal conditions were an antibody dilution of 1:200 and anovernight incubation at room temperature. After washing, NT-3 was detected with an Envision kit (Dako, Carpinteria, CA)and diaminobenzidine (Sigma). Muscle sections were lightlycounterstained by a standard hematoxylin–eosin–safranin pro-cedure.

For b-galactosidase histochemistry, muscle sections werestained for 4–16 hr for b-galactosidase activity with 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside (X-Gal, 400 mg/ml;Sigma), 1 mM magnesium chloride, and 5 mM potassium fer-riferrocyanide in 0.1 mM PBS at 37°C. Muscle sections werelightly counterstained with fast red.

Electrophysiological measurements

All recordings were made blind with a standard electromyo-gram (EMG) apparatus (Keypoint; Dantec, Les Ulis, France) inaccordance with the guidelines of the American Association ofElectrodiagnostic Medicine (1992). Animals were anesthetizedwith pentobarbital before recording. During the procedure, theanimals were kept under a heating lamp to maintain the mus-cles at physiological temperature: the temperature was moni-tored on the surface of the tail with an electronic contact ther-mometer and maintained within 0.5°C.

For acrylamide-intoxicated rats, three markers were used.First, we measured the amplitude of the muscle action poten-tial recorded in the right gastrocnemius. The sciatic nerve wasstimulated through a needle electrode in the proximal thigh. Thecompound muscle action potential (CMAP) was recordedthrough a needle electrode inserted in the medial compartmentof the gastrocnemius. The CMAP reflects the number of func-tional motor units and therefore is an index of the axonal in-tegrity of motor fibers (Oh, 1993). Second, we measured themotor nerve conduction velocity (MNCV) of the right sciaticnerve. The sciatic nerve was stimulated as previously described,through a needle electrode placed first in the proximal thighand then 50 mm distally. The CMAP was recorded in the plan-tar muscle for both stimulations. Motor conduction velocity wascalculated as the distance between stimulation points dividedby the difference in latencies of responses to stimulation at thetwo locations. Third, sensory function was assessed by record-ing the latency of the sensory nerve action potential of the cau-dal nerve in the tail. The tail nerve was stimulated through apair of needle electrodes placed at the base of the tail. A pairof recording electrodes was positioned at a fixed distance of 80

GENE THERAPY IN DIABETIC NEUROPATHY 2239

mm distal to the stimulating electrodes. A substantial sensorypotential can be recorded when the stimulus to the caudal nerveis below the motor threshold. Sensory nerve conduction veloc-ity (SNCV) was calculated as the distance between the stimu-lating and the recording electrode (80 mm) divided by the sen-sory distal latency.

Behavior assessment in acrylamide-intoxicated rats

A clinical score was determined and two behavioral tests werecarried out. All were assessed blind to the status of the rats. Theclinical scale used was modified from Rivlin and Tator (1977).Each animal was attributed a score between 1 and 4: grade 1, com-plete paralysis of the legs; grade 2, weight bearing but unable towalk; grade 3, walking but not normally; grade 4, normal.

The two behavioral tests were the inclined plane and the land-ing foot spread. We used an inclined plane and tested the abilityof each rat to maintain its position on it (Rivlin and Tator, 1977;Fournier et al., 1993). The device consisted of two rectangularboards. One of the boards served as the base and the other as themovable inclined plane. For clinical assessment rats were placedin such a position that their body axis was perpendicular to theaxis of the inclined plane. To maintain itself on the plane the an-imal must use both fore and hind limbs. The maximum inclina-tion of the plane at which a rat could maintain itself for 5 secwas recorded. In practice, the angle was either increased by 5°or decreased by 5° to find the steepest angle at which the ratcould maintain its position for 5 sec without falling.

The landing foot spread distance (LFSD) was measured asdescribed previously (Edwards and Parker, 1977; Sporel-Oza-kat et al., 1990; Saita et al., 1995): the rats were dropped froma horizontal position at a height of 30 cm. The position of theforth digit of each hind limb on landing was marked and thedistance between the two was measured. This was repeated fourtimes for each rat, and the mean value was calculated.

Determination of choline acetyltransferase activity

To determine choline acetyltransferase (ChAT) activities,soleus muscles were dissected out and frozen at –80°C inisopentane. Muscles were thawed at room temperature and ho-mogenized in a glass–glass conical homogenizer in 10% (w/v)medium containing 100 mM Tris-HCl–2 mM EDTA–0.4% Tri-ton X-100, pH 7.4, all at 4°C. Fifty-microliter samples of ho-mogenate was transferred to Eppendorf tubes for protein assay.Homogenates were centrifuged at 20,000 3 g for 30 min at 4°C.ChAT activity was assayed in the supernatant by the methodof Fonnum (1975), using [14C]acetyl-CoA (Amersham, Ar-lington Heights, IL) as a substrate. In these conditions, theamount of acetylcholine produced was linearly related to theamount of homogenate used for the assay.

Statistical analysis

All markers were compared by an analysis of variance(ANOVA) with post hoc testing using Fisher’s protected leastsignificant difference test. StatView software (Abacus Con-cepts, Berkeley, CA) was used. The level of significance wasset at p , 0.05. The following symbols are used in the figuresto express the level of significance: *p , 0.05; **p , 0.01;***p , 0.001; ****p , 0.0001.

RESULTS

Intramuscular injections of AdNT-3 (1010 PFU/animal) weregiven before the development of symptomatic neuropathy inthe two animal models. In STZ-diabetic rats, AdNT-3 was in-jected 7 days after the induction of diabetes by intraperitonealinjection of STZ. In acrylamide experimental neuropathy, ratswere treated with AdNT-3 48 hr before the first intraperitonealinjection of acrylamide. To determine whether the effect ofAdNT-3 treatment was caused by transgene delivery, we mea-sured the effect of treatment with an adenoviral vector encod-ing a nontherapeutic gene, E. coli b-galactosidase (Adbgal).The effect of treatment with AdNT-3 was also tested in normalrats.

General effects of streptozotocin injection andinfluence of treatment with AdNT-3

All STZ-injected (diabetic) rats showed high glucose levelsand poor weight gain. Treatment with AdNT-3 did not have anyeffect on glucose levels or weight. At the end of the study (day38), the mean glucose level was 23.1 6 2.1 mM in the AdNT-3-treated diabetic rats (n 5 16), 22.4 6 2.5 mM in the Adbgal-treated diabetic rats (n 5 16, p 5 0.85), and 23.6 6 1.2 mM inthe vehicle-treated diabetic rats (n 5 44, p 5 0.83). Similarly,the weight of AdNT-3-treated diabetic rats (mean, 215.7 6 4.8g) was not significantly different from that of Adbgal-treateddiabetic rats (mean, 220.1 6 7.3, p 5 0.62) or vehicle-treateddiabetic rats (216.5 6 3.2, p 5 0.89). The mean weight of con-trol nondiabetic rats was 249.9 6 3.9 (p , 0.0001 comparedwith diabetic rats) and the mean glucose level was 4.4 6 0.16(p , 0.0001 compared with diabetic rats).

Analysis of transgene expression

The efficiency of transgene delivery to the STZ-diabetic ratswas assessed by measuring NT-3 concentrations in plasma andmuscle by ELISA. Significant levels of NT-3 were found in theplasma of AdNT-3-treated rats (Fig. 1). Plasma NT-3 concen-tration peaked 3 days after AdNT-3 injection and gradually de-creased to approximately 18% of the maximal value 1 monthpostinjection. In contrast, NT-3 was not detected in the plasmaof normal controls or diabetic rats treated with Adbgal or ve-hicle. The kinetics of NT-3 expression in plasma were similarin acrylamide-treated rats (data not shown).

Muscle NT-3 concentrations were measured at the end of thestudy (day 38), 1 month after AdNT-3 injection. High levels ofNT-3 were detected in AdNT-3-treated rats (20,137 6 1218pg/muscle, n 5 6). Low levels of muscle NT-3 were detectedin control rats (45 6 19 pg/muscle; n 5 6) and in diabetic ratstreated with Adbgal (73 6 45 pg/muscle, n 5 6) or vehicle(38 6 8 pg/muscle, n 5 10).

Expression of the transgene in muscle was also studied by his-tochemistry for b-galactosidase protein (day 38). Intense label-ing was observed in muscles injected with Adbgal (Fig. 2). Thepresence of b-galactosidase was demonstrated in both myofibernuclei and sarcoplasma, as expected for a protein containing anuclear localization signal. A slight inflammatory reaction waspresent around b-galactosidase-expressing myofibers. NT-3-im-munostained myofibers were detected in muscles injected withAdNT-3 (Fig. 3). NT-3 immunolabeling was scattered through-

PRADAT ET AL.2240

out the muscle fibers, as would be expected with a secreted pro-tein such as NT-3. There was no sign of necrotic myofibers orother evidence of severe muscle toxicity.

Treatment of STZ-diabetic rats with AdNT-3 improveselectrophysiological test results

Neuropathy in diabetic rats was assessed by measurement ofmotor (Fig. 4A) and sensory (Fig. 4B) nerve conduction ve-locities. As expected, in control rats there was an increase in

nerve conduction velocity during the experimental period. Thisis a well-known feature of this model and is due to maturationof the peripheral nervous system of these young adult rats.Three weeks after STZ injection, motor and sensory nerve con-duction velocities (NCVs) in diabetic rats were significantlyslower than those of normal controls (MNCV, p , 0.01; SNCV,p , 0.001). Compound muscle action potentials (CMAPs), anindex of axonal integrity, were not significantly different be-tween diabetic rats and normal controls.

Nerve conduction velocities were significantly higher in

GENE THERAPY IN DIABETIC NEUROPATHY 2241

FIG. 1. Plasma NT-3 levels in STZ-diabetic rats after intramuscular injection of AdNT-3 (n 5 8). The values represent means 6SEM.

FIG. 2. b-galactosidase histochemistry in Adbgal-injected muscles (day 38). Sections show myofibers expressing b-galactosidase:a strong b-galactosidase signal is present in both nuclei and sarcoplasma. A few inflammatory foci were observed (arrow).

STZ-diabetic rats treated with AdNT-3 than in untreated dia-betic rats. The percentage of prevention of the decrease inMNCV related to diabetes was 54% on day 28 (p , 0.01) and52% on day 35 (p , 0.05). For SNCV, the prevention was 62%on day 21 (p , 0.05), 58% on day 28 (p , 0.0001), and 55%on day 35 (p , 0.01).

We tested whether treatment of normal rats with AdNT-3 mod-ified the electrophysiological markers used. Motor and sensoryNCV, 3 weeks after intramuscular injection of AdNT-3 into nor-mal rats (n 5 10), did not differ from those determined after treat-ment with vehicle (MNCV: 35.8 6 0.9 vs. 36.8 6 1.36 m/sec,p 5 0.56; SNCV: 33.7 6 0.36 vs. 33.3 6 0.49 m/sec, p = 0.49).

General effects of acrylamide intoxication andinfluence of treatment with AdNT-3

The general toxicity of acrylamide was evaluated by moni-toring survival and body weight. Twelve of the 75 animals treated(16%) died during the acrylamide intoxication period, showingsigns of cachexia, severe neurological problems with paralysisof the hind limbs, and urinary retention related to autonomic in-volvement (3 in the second week of treatment and 9 during thethird week). Treatment with AdNT-3 did not affect survival: thedeath rate was 3 of 20 (15%) in the acrylamide/AdNT-3-treatedgroup, 3 of 20 (15%) in the acrylamide/Adbgal-treated group,and 6 of 35 (17%) in the acrylamide/NaCl-treated group. Intox-ication led to significant impairment of weight gain in all ani-mals treated. The weight of control animals increased by 37.4%whereas that of acrylamide-treated rats remained stable (11.9%).Treatment with Ad-NT3 did not modify the weight of animals.At the end of the study (day 30), the weight of acrylamide/AdNT-3-treated rats (mean, 298.8 6 7.6 g; n 5 17) was not significantlydifferent from that of acrylamide/Adbgal-treated rats (mean:295.8 6 6.1 g, n 5 17; p 5 0.77) or acrylamide/NaCl-treated rats(284.7 6 8.8 g, n 5 29; p 5 0.28). The weight of control rats nottreated with acrylamide was 401.5 6 7.8 g (n 5 20; p , 0.0001compared with acrylamide-treated rats).

Treatment of acrylamide-intoxicated rats with NT-3improves behavioral test results

Acrylamide-treated rats developed clear clinical signs of neu-ropathy. To quantify these abnormalities, we used a semiquan-titative neurological score and behavioral tests in which acry-lamide neuropathy is known to affect the results (Fig. 5). Therewere no differences between the performances of the animalsof the four groups before the intoxication period (baseline val-ues). Differences between acrylamide-treated rats and normalcontrols were significant for all behavioral markers after 1 weekof acrylamide intoxication (day 16). As assessed by the neuro-logical score, animals treated with acrylamide developed in-creasing locomotor problems with a combination of ataxia andmotor deficit predominating in the hind limb and, to a lesserextent, in the fore limbs (p , 0.001 compared with normal rats).The maximum angle of the inclined plane at which a rat couldmaintain itself was shallower for acrylamide-intoxicated thancontrol rats (p , 0.0001). The LFSD (landing foot spread dis-tance) was greater in acrylamide-intoxicated than control rats(p , 0.01). As expected, the LFSD increased with time in allgroups because of the growth of the animals.

Acrylamide-treated rats that received intramuscular AdNT-3 showed less marked locomotor deficits than those receivingNaCl or Adbgal. The difference was significant for neurologi-cal score from day 23 (p , 0.001) to the last point tested onday 30 (p , 0.01). Concerning the maximal angle on the in-clined plane, the difference was significant from day 16 (p ,

0.001) to day 30 (p , 0.01). The LFSD was also less abnormalin AdNT-3-treated rats from day 16 (p , 0.05) to day 24 (p ,

0.05), which was the last time point tested.

Treatment of acrylamide-intoxicated rats with AdNT-3improves electrophysiological test results

As electromyographical procedures, including anesthesia,may interfere with behavioral test performance, we measured

PRADAT ET AL.2242

FIG. 3. NT-3 immunohistochemistry in AdNT-3-injected muscles (day 38). Sections show myofibers expressing NT-3. TheNT-3 immunolabeling is scattered throughout the muscle fibers.

GENE THERAPY IN DIABETIC NEUROPATHY 2243

A

B

FIG. 4. Effect of AdNT-3 treatment on streptozotocin-induced neuropathy as assessed by electrophysiological tests: motor nerveconduction velocities of the sciatic nerve (A) and sensory nerve conduction velocities of the caudal nerve of the tail (B). Onlycomparisons of diabetic rats treated with AdNT-3 to diabetic rats treated with Adbgal or vehicle are represented. Differences be-tween normal control and diabetic rats were significant (see text). The n values indicate the number of animals per group. Thevalues represent means 6 SEM (*p , 0.05; **p , 0.01; ****p , 0.0001).

EMG only at the end of the experimental protocol (day 31).The results are shown in Fig. 6. Acrylamide intoxication im-paired the activity of motor fibers, as demonstrated by the slow-ing of motor and sensory NCVs and reduction of CMAP am-plitudes (p , 0.0001 with reference to controls for all thesemarkers). All electrophysiological markers were significantlyless impaired in acrylamide-intoxicated rats treated with AdNT-3. Prevention of acrylamide-induced electrophysiologicalchanges, expressed as a percentage, was 51% for MNCV (p ,

0.001), 53% for SNCV (p , 0.0001), and 63% for CMAP (p ,

0.0001).

Treatment of acrylamide-intoxicated rats with AdNT-3partially prevents the decrease in muscle cholineacetyltransferase activity

We measured muscle choline acetyltransferase (ChAT) ac-tivity, which is a functional index of muscle innervation (Han-tai et al., 1990; Ribaric et al., 1991; Blondet et al., 1992). At

the end of the study (day 31), total muscle ChAT activity inacrylamide-intoxicated rats was 31% of that in normal controls(p , 0.0001). Treatment with AdNT-3 prevented 45% of thedecrease in muscle ChAT activity (p , 0.05) (Fig. 7). We alsoinvestigated the effects of treatment with AdNT-3 on muscleChAT activity in normal rats: the treatment of normal controlrats (n 5 10) with AdNT-3 did not increase enzyme activitiesin muscles (Fig. 7). This suggests that treatment of acrylamide-intoxicated rats with AdNT-3 improves the density of motor in-nervation but does not increase the expression of ChAT in mo-tor neurons.

DISCUSSION

This study demonstrates a beneficial effect of AdNT-3 treat-ment in an animal model of early diabetic neuropathy, STZ-di-abetic neuropathy, and in an animal model of diffuse axonaldegeneration, experimental acrylamide neuropathy.

PRADAT ET AL.2244

FIG. 5. Effect of AdNT-3 treatment on acrylamide-induced neuropathy as assessed by behavioral tests: neurological score (A),the maximum inclination of the plane at which a rat could maintain itself for at least 5 sec (B), and landing foot spread distance(C). Only comparisons between acrylamide-intoxicated rats treated with AdNT-3 and acrylamide-intoxicated rats treated withAdbgal or vehicle are represented. Differences between normal control and acrylamide-intoxicated rats were significant (see text).The n values indicate the number of animals per group on day 30. The values represent means 6 SEM (*p , 0.05; **p , 0.01;****p , 0.0001).

A

Clinical and electrophysiological findings in STZ-diabeticneuropathy were consistent with published findings (Braven-boer et al., 1993; Apfel et al., 1994). STZ-diabetic rats wereasymptomatic and showed abnormalities of motor and sensorynerve conduction velocities, suggesting demyelination, withoutfeatures of axonal degeneration. Furthermore, as previouslymentioned (Willars et al., 1988), preliminary experimentsshowed that the total ChAT activity of the soleus was not di-minished in diabetic rats, suggesting a normal density of mo-tor innervation (data not shown).

The effect of treatment with AdNT-3 on nerve conduction

velocities suggests that this treatment is effective in the earlypresymptomatic stage of diabetic neuropathy, when axonal le-sions are not yet present. Other preclinical studies of treatmentwith NT-3 in diabetic neuropathy have been limited to the STZmodel (Tomlinson et al., 1997). It has been shown that repeatedsubcutaneous administration of high doses of recombinant NT-3 to STZ-diabetic rats prevents the slowing of sensory nerveconduction velocity (Tomlinson et al., 1997). However, in con-trast to our report, treatment with recombinant NT-3 did notimprove motor nerve conduction velocity. Thus continuous de-livery of low levels of NT-3 in diabetic neuropathy seems to

GENE THERAPY IN DIABETIC NEUROPATHY 2245

FIG. 5. Continued.

B

C

be more effective for peripheral nerve motor involvement thanthe systemic administration of high doses.

As the STZ model did not reproduce the axonal lesions typ-ical of symptomatic late diabetic neuropathy, we decided to testour strategy in an animal model of diffuse axonal neuropathy.The acrylamide experimental neuropathy model was chosen be-cause, as in human diabetic neuropathy, axonal lesions are dif-fuse and the sensitivity of the various fiber types in acrylamideneuropathy resembles that in diabetic neuropathy. Indeed, acry-lamide neuropathy is sensitive before the onset of motor in-

volvement and impairs the autonomic nervous system (Schaum-burg et al., 1974; Ralevic et al., 1991). Administration of highdoses of acrylamide induces somatofugal axonal atrophy, whichprecedes the development of axonal degeneration (Gold et al.,1992). Therefore, the experimental neuropathy does notprogress as a distal axonopathy, which is a frequent feature ofdiabetic neuropathy (Said et al., 1983). However, lesions in hu-man diabetic neuropathies are heterogeneous, multifocal, anddo not necessarily progress as a distal neuropathy, as in prox-imal diabetic neuropathies (Said et al., 1994). Experimental di-

PRADAT ET AL.2246

A

B C

FIG. 6. Effect of AdNT-3 treatment on acrylamide-induced neuropathy as assessed by electrophysiological tests (day 31): mo-tor nerve conduction velocities of the sciatic nerve (A), sensory nerve conduction velocities of the caudal nerve of the tail (B),compound muscle action potential in the gastrocnemius (C). Only comparisons between acrylamide-intoxicated rats treated withAdNT-3 and acrylamide-intoxicated rats treated with Adbgal or vehicle are represented. Differences between normal control andacrylamide-intoxicated rats were significant (see text). The n values indicate the number of animals per group on day 30. Thevalues represent means 6 SEM (**p , 0.01; ***p , 0.001; ****p , 0.0001).

abetic and high-level acrylamide-induced neuropathies alsohave several characteristics in common. Indeed, abnormalitiesof retrograde and anterograde axonal transport are consideredto be an important pathogenic mechanism of both experimen-tal acrylamide (Sidenius and Jakobsen, 1983; Miller andSpencer, 1984) and diabetic neuropathy (Jakobsen and Side-nius, 1980; Jakobsen et al., 1981). Neurotransmitter modifica-tions in rat enteric nerve, common to both acrylamide and di-abetic experimental neuropathies, have suggested a commonpathogenic mechanism (Belai and Burnstock, 1996).

To our knowledge, this is the first report of the effects of theadministration of a neurotrophic protein in an experimental neu-ropathy induced by acrylamide. AdNT-3 treatment had markedeffects on behavioral disorders, sensory and motor electro-physiological abnormalities, and muscle ChAT activity. NT-3did not affect sensory or motor nerve conduction or muscleChAT activity in normal animals, indicating that this neu-rotrophic factor acts directly on the pathological process. In par-ticular, the effect on CMAP and muscle ChAT activity showsthat treatment partially improved the density of muscle inner-vation. This could be due to the slowing of nerve degenerationand/or the enhancement of nerve regeneration resulting fromcollateral axonal sprouting. Analysis of terminal innervation inAdNT-3-treated pmn mice demonstrated axonal sprouting(Haase et al., 1997). Evidence from animal models of diabetes

suggests that the reduced availability of NT-3 may contributeto the pathogenesis of diabetic neuropathy (Rodriguez-Pena etal., 1995; Ihara et al., 1996; Fernyhough et al., 1998), andtherefore, delivery of exogenous NT-3 may slow nerve de-generation by restoring available NT-3. Slowing of nerve de-generation could also result from a nonspecific neuroprotec-tive effect on peripheral nerves, because NT-3 favors thesurvival of peripheral nerves after various challenges (McMa-hon and Priestley, 1995; Yuen and Mobley, 1995). NT-3 treat-ment probably acts downstream from the molecular lesions re-sponsible for the development of these diseases. Further studiesare necessary to determine the underlying mechanisms for thesubstantial protection of peripheral nerves by AdNT-3 treat-ment.

Intramuscular injection of recombinant adenovirus results intransgene expression in the cell bodies of the motor and sen-sory neurons that innervate the inoculated muscle (Finiels etal., 1995; Ghadge et al., 1995). However, our results implicatethe liberation of NT-3 into the bloodstream, with secondary up-take by nerve terminals or directly by the cell body of neurons.Indeed, the preventive effects of NT-3 on electrophysiologicalchanges were observed in motor and sensory neurons other thanthose that innervate the injected muscles.

The small amounts of NT-3 in plasma and the absence ofsystemic and local side effects in our model suggest that simi-

GENE THERAPY IN DIABETIC NEUROPATHY 2247

FIG. 7. Total ChAT activities in soleus muscles 1 month after the start of the experiment. Only comparisons between acry-lamide-intoxicated rats treated with AdNT-3 and acrylamide-intoxicated rats treated with Adbgal or vehicle are represented. Dif-ferences between normal control and acrylamide-intoxicated rats were significant (p , 0.01). The n values indicate the numberof animals per group on day 30. The values represent means 6 SEM (*p , 0.05).

lar doses may be well tolerated in humans. The gene transferstrategy may thus avoid a major limitation of treatment with re-combinant neurotrophic factors: the doses of neurotrophic fac-tors shown to be effective in animal studies often have side ef-fects in humans (Yuen and Mobley, 1995). For example, in aphase III trial of recombinant nerve growth factor (NGF) in di-abetic neuropathy (Apfel et al., 2000), the dose used (0.3 mg/kgper week, subcutaneously) was lower than that shown to pro-duce a response in diabetic rats by a factor of 30,000 (9 mg/kgper week, subcutaneously) (Apfel et al., 1994); the treatmenthad no beneficial effects on diabetic-related polyneuropathybut, nevertheless, 67% of patients suffered injection site hy-peralgesia.

Our results show that the continuous delivery of low levelsof NT-3 by gene therapy to be a promising treatment for dia-betic neuropathy. Indeed, the effect of treatment in STZ-dia-betic rats suggests that it may partially prevent the progressionof early diabetic neuropathy. Although the acrylamide modeldoes not perfectly mimic all forms of axonal diabetic neuropa-thy, the efficacy of treatment in this model suggests that it maydelay or slow the development of axonal lesions of diabeticneuropathy. In particular, this treatment may be useful in a sub-set of diabetic patients with predominant large-fiber symptomsor in whom small-fiber symptoms, particularly pain, are con-trolled by symptomatic treatments. Furthermore, a combinedtreatment with NT-3 and NGF, receptors for which are ex-pressed by small-fiber sensory neurons (Mu et al., 1993; McMa-hon et al., 1994), could be evaluated, as such treatment maypotentially alleviate all neuropathic symptoms.

ACKNOWLEDGMENTS

We thank Isabelle Loquet, Estelle Arnould, and DelphineCasanova for excellent technical support; and Leïla Houhou,Sylvie Berrard, and Catherine Roditi for critical reading of themanuscript.

REFERENCES

AMERICAN ASSOCIATION OF ELECTRODIAGNOSTIC MEDI-CINE (1992). Guidelines in electrodiagnostic medicine. MuscleNerve 15, 229–253.

APFEL, S.C., AREZZO, J.C., BROWNLEE, M., FEDEROFF, H., andKESSLER, J.A. (1994). Nerve growth factor administration protectsagainst experimental diabetic sensory neuropathy. Brain Res. 634,7–12.

APFEL, S.C., SCHWARTZ, S., ADORNATO, B.T., FREEMAN, R.,BITON, V., RENDELL, M., VINIK, A., GIULIANI, M., STEVENS,J.C., BARBANO, R., and DYCK, P.J. (2000). Efficacy and safetyof recombinant human nerve growth factor in patients with diabeticpolyneuropathy: A randomized controlled trial. JAMA 284, 2215–2221.

BARBACID, M. (1994). The Trk family of neurotrophin receptors. J.Neurobiol. 25, 1386–1403.

BEHSE, F., BUCHTHAL, F., and CARLSEN, F. (1977). Nerve biopsyand conduction studies in diabetic neuropathy. J. Neurol. Neurosurg.Psychiatry 40, 1072–1082.

BELAI, A., and BURNSTOCK, G. (1996). Acrylamide-induced neu-ropathic changes in rat enteric nerves: Similarities with effects ofstreptozotocin-diabetes. J. Auton. Nerv. Syst. 58, 56–62.

BLONDET, B., BARLOVATZ-MEIMON, G., FESTOFF, B.W., SORIA, C., SORIA, J., RIEGER, F., and HANTAI, D. (1992). Plas-minogen activators in the neuromuscular system of the wobbler mu-tant mouse. Brain Res. 580, 303–310.

BRADLEY, W.G., and ASBURY, A.K. (1970). Radioautographic stud-ies of Schwann cell behavior. I. Acrylamide neuropathy in the mouse.J. Neuropathol. Exp. Neurol. 29, 500–506.

BRAVENBOER, B., KAPPELLE, A.C., VAN BUREN, T., ERKE-LENS, D.W., and GISPEN, W.H. (1993). ACTH4–9 analogue ORG2766 can improve existing neuropathy in streptozotocin-induced di-abetic rats. Acta di Abetol. 30, 21–24.

CHOPRA, J.S., and FANNIN, T. (1971). Pathology of diabetic neu-ropathy. J. Pathol. 104, 175–184.

CHOPRA, J.S., CONNON, J.J., and BANERJI, N.K. (1969). Diabeticneuropathy. Clinical and electrophysiological study in synalbumin-positive and synalbumin-negative subjects. Diabetologia 5, 413–415.

CROUZET, J., NAUDIN, L., ORSINI, C., VIGNE, E., FERRERO, L.,LE ROUX, A., BENOIT, P., LATTA, M., TORRENT, C., BRANEL-LEC, D., DENEFLE, P., MAYAUX, J.F., PERRICAUDET, M., andYEH P. (1997). Recombinational construction in Escherichia coli ofinfectious adenoviral genomes. Proc. Natl. Acad. Sci. U.S.A. 94,1414–1419.

DITTRICH, F., THOENEN, H., and SENDTNER, M. (1994). Ciliaryneurotrophic factor: Pharmacokinetics and acute phase response.Ann. Neurol. 35, 151–163.

EDWARDS, P.M., and PARKER, V.H. (1977). A simple, sensitive,and objective method for early assessment of acrylamide neuropa-thy in rats. Toxicol. Appl. Pharmacol. 40, 589–591.

FERNYHOUGH, P., DIEMEL, L.T., and TOMLINSON, D.R. (1998).Target tissue production and axonal transport of neurotrophin-3 arereduced in streptozotocin-diabetic rats. Diabetologia 41, 300–306.

FINIELS, F., RIBOTTA, M., BARKATS, M., SAMOLYK, M.L.,ROBERT, J.J., PRIVAT, A., REVAH, F., and MALLET, J. (1995).Specific and efficient gene transfer strategy offers new potentialitiesfor the treatment of motor neuron diseases. Neuroreport 7, 373–378.

FONNUM, F. (1975). A rapid radiochemical method for the determi-nation of choline acetyltransferase. J. Neurochem. 24, 407–409.

FOURNIER, J., STEINBERG, R., GAUTHIER, T., KEANE, P.E.,GUZZI, U., COUDE, F.X., BOUGAULT, I., MAFFRAND, J.P.,SOUBRIE, P., and LE FUR, G. (1993). Protective effects of SR57746A in central and peripheral models of neurodegenerative dis-orders in rodents and primates. Neuroscience 55, 629–641.

GAO, W.Q., DYBDAL, N., SHINSKY, N., MURNANE, A.,SCHMELZER, C., SIEGEL, M., KELLER, G., HEFTI, F.,PHILLIPS, H.S., and WINSLOW, J.W. (1995). Neurotrophin-3 re-verses experimental cisplatin-induced peripheral sensory neuropathy.Ann. Neurol. 38, 30–37.

GHADGE, G.D., ROOS, R.P., KANG, U.J., WOLLMANN, R., FISH-MAN, P.S., KALYNYCH, A.M., BARR, E., and LEIDEN, J.M.(1995). CNS gene delivery by retrograde transport of recombinantreplication-defective adenoviruses. Gene Ther. 2, 132–137.

GOLD, B.G., GRIFFIN, J.W., and PRICE, D.L. (1992). Somatofugalaxonal atrophy precedes development of axonal degeneration in acry-lamide neuropathy. Arch. Toxicol. 66, 57–66.

GRAVEL, C., GôTZ, R., LORRAIN, A., and SENDTNER, M. (1997).Adenoviral gene transfer of ciliary neurotrophic factor and brain-de-rived neurotrophic factor leads to long term survival of axotomizedmotor neurons. Nat. Med. 7, 765–770.

HAASE, G., KENNEL, P., PETTMANN, B., VIGNE, E., AKLI, S.,REVAH, F., SCHMALBRUCH, H., and KAHN, A. (1997). Genetherapy of murine motor neuron disease using adenoviral vectors forneurotrophic factors. Nat. Med. 3, 429–436.

HANTAI, D., RAO, J.S., and FESTOFF, B.W. (1990). Rapid neuralregulation of muscle urokinase-like plasminogen activator as definedby nerve crush. Proc. Natl. Acad. Sci. U.S.A. 87, 2926–2930.

HELGREN, M.E., CLIFFER, K.D., TORRENTO, K., CAVNOR, C.,

PRADAT ET AL.2248

CURTIS, R., DISTEFANO, P.S., WIEGAND, S.J., and LINDSAY,R.M. (1997). Neurotrophin-3 administration attenuates deficits ofpyridoxine-induced large fiber sensory neuropathy. J. Neurosci. 17,372–382.

HOPKINS, A. (1970). The effect of acrylamide on the peripheral nervoussystem of the baboon. J. Neurol. Neurosurg. Psychiatry 33, 805–816.

IHARA, C., SHIMATSU, A., MIZUTA, H., MURABE, H., NAKA-MURA, Y., and NAKAO, K. (1996). Decreased neurotrophin-3 ex-pression in skeletal muscles of streptozotocin-induced diabetic rats.Neuropeptides 30, 309–312.

JAKOBSEN, J., and SIDENIUS, P. (1980). Decreased axonal transportof structural proteins in streptozotocin diabetic rats. J. Clin. Invest.66, 292–297.

JAKOBSEN, J., BRIMIJOIN, S., SKAU, K., SIDENIUS, P., andWELLS, D. (1981). Retrograde axonal transport of transmitter en-zymes, fucose-labeled protein, and nerve growth factor in streptozo-tocin-diabetic rats. Diabetes 30, 797–803.

LAMONTAGNE, A., and BUCHTHAL, F. (1970). Electrophysiolog-ical studies in diabetic neuropathy. J. Neurol. Neurosurg. Psychiatry33, 442–452.

MCMAHON, S.B., ARMANINI, M.P., LING, L.H., and PHILLIPS,H.S. (1994). Expression and coexpression of Trk receptors in sub-populations of adult primary sensory neurons projecting to identifiedperipheral targets. Neuron 12, 1161–1171.

MCMAHON, S.B., and PRIESTLEY, J.V. (1995). Peripheral neu-ropathies and neurotrophic factors: Animal models and clinical per-spectives. Curr. Opin. Neurobiol. 5, 616–624.

MILLER, M.S., and SPENCER, P.S. (1984). Single doses of acry-lamide reduce retrograde transport velocity. J. Neurochem. 43, 1401–1408.

MU, X., SILOS-SANTIAGO, I., CARROLL, S.L., and SNIDER, W.D.(1993). Neurotrophin receptor genes are expressed in distinct pat-terns in developing dorsal root ganglia. J. Neurosci. 13, 4029–4041.

OH, S.J. (1993). Clinical Electromyography: Nerve Conduction Stud-ies, 2nd edition. (Williams & Wilkins, Baltimore, MD).

PIRART, J. (1978). Why don’t we teach and treat diabetic patients bet-ter? Diabetes Care 1, 139–140.

PODUSLO, J.F., and CURRAN, G.L. (1996). Permeability at theblood–brain and blood–nerve barriers of the neurotrophic factors:NGF, CNTF, NT3, BDNF. Mol. Brain Res. 36, 280–286.

PRADAT, P.F., FINIELS, F., KENNEL, P., NAIMI, S., DELAERE,P., REVAH, F., and MALLET, J. (2001). Partial prevention of cis-platin-induced neuropathy by electroporation-mediated nonviral genetransfer. Hum. Gene Ther. 12, 367–375.

RALEVIC, V., ABERDEEN, J.A., and BURNSTOCK, G. (1991).Acrylamide-induced autonomic neuropathy of rat mesenteric vessels:Histological and pharmacological studies. J. Auton. Nerv. Syst. 34,77–87.

RIBARIC, S., STEFANOVSKA, A., BRZIN, M., KOGOVSEK, M.,and KROSELJ, P. (1991). Biochemical, morphological, and func-tional changes during peripheral nerve regeneration. Mol. Chem.Neuropathol. 15, 143–157.

RIVLIN, A.S., and TATOR, C.H. (1977). Objective clinical assessmentof motor function after experimental spinal cord injury in the rat. J.Neurosurg. 47, 577–581.

RODRIGUEZ-PENA, A., BOTANA, M., GONZALEZ, M., and REQUEJO, F. (1995). Expression of neurotrophins and their recep-tors in sciatic nerve of experimentally diabetic rats. Neurosci. Lett.200, 37–40.

SAID, G., SLAMA, G., and SELVA, J. (1983). Progressive centripetaldegeneration of axons in small fibre diabetic polyneuropathy. Brain106, 791–807.

SAID, G., GOULON-GOEAU, C., LACROIX, C., and MOULONGUET,A. (1994). Nerve biopsy findings in different patterns of proximaldiabetic neuropathy. Ann. Neurol. 35, 559–569.

SAITA, K., OHI, T., HANAOKA, Y., FURUKAWA, S., FURUKAWA,Y., HAYASHI, K., and MATSUKURA, S. (1995). Effects of 4-methylcatechol, a stimulator of endogenous nerve growth factor syn-thesis, on experimental acrylamide-induced neuropathy in rats. Neu-rotoxicology 16, 403–412.

SCHAUMBURG, H.H., WISNIEWSKI, H.M., and SPENCER, P.S.(1974). Ultrastructural studies of the dying-back process. I. Periph-eral nerve terminal and axon degeneration in systemic acrylamide in-toxication. J. Neuropathol. Exp. Neurol. 33, 260–284.

SENDTNER, M. (1997). Gene therapy for motor neuron disease. Nat.Med. 3, 380–381.

SHARMA, A.K., and THOMAS, P.K. (1984). Animal models: Pathol-ogy and pathophysiology. In Diabetic Neuropathy Pathology andPathophysiology—Human and Experimental. P.J. Dyck, P.K.Thomas, A.K. Asbury, A.I. Windegrad, and D. Portes, editors. (W.B.Saunders, Philadelphia, PA) pp. 237–252.

SIDENIUS, P., and JAKOBSEN, J. (1983). Anterograde axonal trans-port in rats during intoxication with acrylamide. J. Neurochem. 40,697–704.

SPENCER, P.S., and SCHAUMBURG, H.H. (1977). Ultrastructuralstudies of the dying-back process. III. The evolution of experimen-tal peripheral giant axonal degeneration. J. Neuropathol. Exp. Neu-rol. 36, 276–299.

SPOREL-OZAKAT, R.E., EDWARDS, P.M., VAN DER HOOP, R.G.,and GISPEN, W.H. (1990). An ACTH(4–9) analogue, Org 2766, im-proves recovery from acrylamide neuropathy in rats. Eur. J. Phar-macol. 186, 181–187.

TOMLINSON, D.R., FERNYHOUGH, P., and DIEMEL, L.T. (1997).Role of neurotrophins in diabetic neuropathy and treatment withnerve growth factors. Diabetes 46(Suppl. 2), S43–S49.

WILLARS, G.B., TOWNSEND, J., TOMLINSON, D.R., COMPTON,A.M., and CHURCHILL, R.D. (1988). Studies on peripheral nerveand lens in long-term experimental diabetes: Effects of the aldose re-ductase inhibitor statil. Metabolism 37, 442–449.

YUEN, E.C., and MOBLEY, W.C. (1995). Therapeutic applications ofneurotrophic factors in disorders of motor neurons and peripheralnerves. Mol. Med. Today 1, 278–286.

Address reprint requests to:Dr. Jacques Mallet

LGN, Bâtiment CERVI, Hôpital de la Pitié-Salpétrière47-83 boulevard de l’Hôpital

75651 Paris Cedex 13, France

E-mail: mallet@infobiogen.fr

Received for publication May 9, 2001; accepted after revisionNovember 9, 2001.

Published online: November 29, 2001.

GENE THERAPY IN DIABETIC NEUROPATHY 2249