Delayed peripheral nerve regeneration and central nervous system collateral sprouting in leucocyte common antigen-related protein tyrosine phosphatase-deficient mice

Animals

The generation and genotype analysis of LAR-ΔP mice has been described in detail (Schaapveld et al., 1997a). Original F1 transgenic mice were on a C57BL/6 × 129 hybrid genetic background. F9 heterozygous mice, resulting from eight subsequent backcrosses onto C57BL/6 and therefore on an almost pure C57BL/6 background, were intercrossed. Resulting male homozygous LAR-ΔP and wildtype littermates were used, the peripheral regeneration group at the age of 2 months (22–25 g) and the collateral sprouting group at 4–5 months (25–30 g). Additional groups of wildtype and LAR-ΔP mice were used for the histological analysis of the sciatic nerves at 4 days post-crush lesion.

All procedures involving animals were approved by the Animal Care Committee of the University Medical Centre St. Radboud (Nijmegen, The Netherlands) and conformed to the Dutch Council for Animal Care and the NIH guidelines. All efforts were made to reduce the number of animals used.

Sciatic nerve crush lesion

The sciatic nerve of the right hind paw in wildtype (n = 8) and LAR-ΔP (n = 10) mice was subjected to a crush lesion. A standardized crush lesion method yielding reproducible sensorimotor deficits in the lesioned paw has been described for rats (De Koning et al., 1986), and needed only minor alterations to apply in mice. In short, mice were anaesthetized using a 1 : 1 : 2 mix containing Hypnorm^® (fentanyl citrate 0.315 mg/mL + fluanisone 10 mg/mL; Janssen Pharmaceutica), Dormicum^® (midazolam 5 mg/mL, Roche) and distilled water, with an intraperitoneally injected dose of 0.07 mL/10 g body weight. Following incision of the skin between the knee and the thigh, the sciatic nerve was carefully exposed and subsequently crushed for 60 s using a haemostatic forceps with a waffle-shaped mouth structure. The nerve was crushed at the sciatic notch point immediately distal from where it emerges from under the gluteus maximus muscle. The skin was sutured and the mice were kept on a heating pad for an extra hour to prevent body temperature loss before they returned to their home cage.

Sensory function recovery test

The return of sensory function following sciatic crush lesion was determined using the foot-withdrawal reflex test (De Koning et al., 1986; Van der Zee et al., 1991). Mice were immobilized by hand and the soles of their feet were facing the examiner. A range of small electric currents (0.1, 0.3 and 0.5 mA) was applied to the sole of the foot by a small connector containing two stimulation electrodes. Mice immediately retracted their uninjured paw upon sensing the electric stimulus. When the stimulus was applied to the sole of their lesioned paw, mice initially did not show the withdrawal reflex. During the regeneration/reinnervation process the reflex was restored. Absence of the withdrawal reflex upon stimulation at 0.5 mA was interpreted as no recovery, whereas mice responding with the reflex withdrawal at 0.1 mA current (for 3 consecutive days) were considered to be recovered. The reflex withdrawal was measured at postlesion day 3 and daily from day 10 onward until full recovery.

Motor function recovery test

The recovery of motor function following sciatic nerve crush lesion was monitored through analysis of the individual mouse free walking pattern. The walking test method has been described originally for rats (De Medinacelli et al., 1982) and calculations were modified by De Koning & Gispen (1987; see also van Meeteren et al., 1997). The progress of motor function recovery in the sciatic nerve was calculated using eight footprint parameters (distances in mm) and a correction factor (applicable to mice as well) according to the following formula:

with SFI, sciatic functional index (%); NTOF, normal to opposite foot; ETOF, experimental to opposite foot; NPL, normal print length; EPL, experimental print length; NTS, normal toe spreading; ETS, experimental toe spreading; NIT, normal innertoe spreading; EIT, experimental innertoe spreading. This formula provides an SFI value of ≈ −100% directly after the crush lesion (postlesion day 3; ETS and EIT are both set at 2 mm), and an SFI around zero for nonlesioned control mice and when full recovery of motor function is obtained.

In the test procedure each mouse was allowed to get used to the experimental environment by once letting the animal walk through an inclining (10°) alley (40 × 3.5 cm) that leads into a dark box. Then, a strip of photographic paper (Kodak, Polymax II RC semimatt) was placed on the bottom of the alley and, after dipping the animal's hind feet in photographic paper developer fluid (Kodak, Polymax RT), the animal was again placed at the beginning of the alley to let it walk into the dark box. Subsequently, after allowing the photographic strip to dry, mouse footprint parameters (indicated above) were measured. In this way, motor function recovery was determined at day 3, and then every second day starting from day 8 postlesion.

Entorhinal cortex lesion

Wildtype (n = 27) and LAR-ΔP (n = 27) mice were anaesthetized with Hypnorm^®/Dormicum^® (see above), placed in a Kopf stereotaxic apparatus with the skull in a horizontal position, and subjected to a unilateral aspirative lesion of the entorhinal cortex, which results in transection of the perforant path (Zhou et al., 1989; Steward, 1992; Deller et al., 1997). In brief, after an incision through the skin the skull was exposed and cleaned with 70% ethanol. A rectangular opening was created in the skull by drilling at the following coordinates related to bregma (Franklin & Paxinos, 1997): AP −4.2 mm, ML +3.0 mm; AP −4.2 mm, juncture of muscle to the lateral skull; AP −4.9 mm, ML +3.0 mm; AP −4.9 mm, juncture of muscle to lateral skull. The meninges over the occipital pole were incised and retracted, and the entorhinal area exposed by gently displacing the caudal pole of the cerebral hemisphere rostrally away from the cerebellum. The entorhinal cortex, including the presubiculum, parasubiculum and part of the subiculum, was removed by aspiration (using a fire-polished Pasteur pipette) along the caudal pole of the hemisphere beginning ≈ 3 mm from the midline and extending laterally to the parietal bone, from the dorsal surface down to the base of the skull. After the lesion, the incision through the skin was cleaned again and closed with a surgical suture, and the mouse was placed temporarily on a heating pad (to prevent hypothermia due to anaesthesia) before it returned to the home cage. At 1, 2 and 5 weeks following the entorhinal cortex lesion, mice (n = 9 for each genotype and time point) were killed and the brains processed for quantitative histological analysis (see below). The nonlesioned contralateral sides and two groups of nonlesioned mice (wildtype, n = 4; LAR-ΔP, n = 4) served as controls.

Perfusion procedure

Mice were perfused transcardially with 15 mL 0.1 m phosphate-buffered saline (PBS) followed by 30 mL 4% paraformaldehyde in 0.1 m phosphate buffer (pH 7.4). For the central nervous system (CNS) collateral sprouting experiment the brains were removed, postfixed in the same fixative overnight and cryoprotected for 24 h in 30% sucrose in 0.1 m phosphate buffer, all at 4 °C. Coronal sections (40 µm) were cut on a vibratome (Leica VT10005, Leica Instruments GmbH, Germany) and collected in PBS containing 0.06% sodium azide (Millonig's buffer). Brain sections were sampled starting just before the genu of the corpus callosum until the ventral hippocampal area (AP +1.1 mm to AP −2.92 mm from bregma; Franklin & Paxinos, 1997).

Peripheral nerve sectioning

To perform neurofilament (NF)-200 immunostaining on control and regenerating peripheral nerves, mice were perfused as described above and both sciatic nerves were dissected, postfixed overnight, and then stored in Millonig's buffer. Cryoprotection prior to cryostat cutting of the nerve tissue was obtained through graded sucrose solutions (7.5, 15 and 30%), each step at least 2 h. The sciatic nerves were then cut (8-µm-thick cryostat sections) at 1, 3 and 5 mm distance from the distal border of the crush site (at the time of the lesion marked by a small epineural suture). The control nonlesioned sciatic nerve also received a small epineural suture at a comparable position. For each nerve, and for each distance level (1, 3 and 5 mm distal from crush site), sections were mounted on Superfrost Plus miscroscope slides (Menzel Gläser, Braunschweig, Germany), and left to dry. Each slide contained a set of 10 sections (8 µm thick, 24 µm apart) and was stored at −80 °C until staining.

NF-200 immunostaining

With a DAKO pencil a repellant border around the sections was created (in order to be able to apply a volume of 500–600 µL). The slides were placed in a humid incubation box, moistened with 600 µL blocking buffer (PBS + 0.15% glycine + 1% Cold Water Fish Skin gelatin) and incubated for 30 min at room temperature (RT). The blocking solution was carefully removed by suction. The nerve sections were then incubated with the primary antibody, rabbit anti-NF-200 (NF-200, 1 : 1000; Sigma Chemical Co., St Louis, MO, USA) dissolved in blocking buffer, overnight at RT. The next day the slides were rinsed 3 × 5 min in PBS, 0.05% Tween-20 (PBST), and incubated for 1 h at RT, with a donkey antirabbit-biotinylated secondary antibody 1 : 250 (Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA) diluted in blocking buffer. Slides were again rinsed 3 × 5 min in PBST. The avidin–biotin complex solution (10 mL, 1 : 125, ABC Vectastain Elite; Sigma) was made 30 min prior to use, then added to the slides which were incubated for 30 min at RT. Slides were rinsed three times for 5 min in PBS. Finally, 3-amino-9-ethylcarbazole (AEC) substrate solution (4 mg AEC dissolved in 1 mL N,N-dimethylformamide, and mixed with 14 mL 0.1 m acetate buffer, pH 4.9, with added 15 µL 30% hydrogen peroxide) was applied to the nerve sections on the slides for 8 min at RT. The colouring reaction was stopped by rinsing the sections with distilled water. In some experiments, haematoxylin counterstaining was included for 2–3 min followed by a rinse with tap water for 10 min, and two washes with distilled water. Then, the tissue was entrapped in Kaiser's gelatin–glycerol and the slides coverslipped. Images were collected using a Dialux 20 microscope (Leitz) connected to a videocamera attached to the computer image analysis system.

Myelin staining

For morphometric analysis of the peripheral sciatic nerve, another group of wildtype (n = 4) and LAR-ΔP (n = 3) mice were anaesthetized and perfused transcardially with 15 mL 0.1 m PBS followed by 30 mL 0.5% paraformaldehyde and 1.5% glutaraldehyde in 0.1 m phosphate buffer (pH 7.4). The sciatic nerves on both sides were dissected and postfixed in the same fixation buffer for 1 h. Tissues were washed for 2 × 1 h, and overnight, in 0.1 m phosphate buffer. Postfixation was then continued for 1 h in 1% osmium tetroxide in 0.1 m phosphate buffer. Following a 2 × 1-h wash in 0.1 m phosphate buffer, tissues were dehydrated in an ascending series of aqueous ethanol, and subsequently transferred via a mixture of propylene oxide and Epon to pure Epon 812 as embedding medium. Semithin 1-µm sections were cut and mounted on glass slides, and subsequently counterstained with toluidine blue. Digitized light microscopy (Leitz Dialux 20 microscope) images were used for quantitative analysis to determine the number of myelinated axons per 10⁴ µm², the average axon area, and the average myelin area, at 4 days post crush lesion.

Acetylcholinesterase (AChE) histochemical staining of cholinergic fibres

AChE staining was performed according to Hedreen et al. (1985) on free-floating 40-µm brain sections in 6-wells plates, on a shaker, at RT. Five coronal dorsal hippocampus sections, each 160 µm apart (every fourth section), were selected.

Sections were rinsed in 0.1 m sodium acetate (pH 6.0, adjusted with acetic acid; 3 × 10 min) and then incubated in 650 mm sodium acetate (pH 6.0) containing 0.05% acetylthiocholine iodide (Sigma), 4 mm sodium citrate (Merck, Darmstadt, Germany), 3 mm cupric sulphate (Merck) and 0.1 mm potassium ferricyanide (Sigma) for 30 min. After five 1-min washes in 0.1 m sodium acetate (pH 6.0), sections were incubated in 1% ammonium sulphide (pH 7.5, adjusted with HCl) for 1 min. Subsequently, sections were incubated in 0.1% silver nitrate (pH 5.5, without adjustment) for 1 min. Before and after the latter step, sections were rinsed (5 × 1 min) in 0.1 m sodium nitrate (pH 5.9, without adjustment). Finally, sections were rinsed in 0.1 m sodium acetate (pH 6.0), mounted on Superfrost Plus glass slides (Menzel Gläser), air-dried overnight, dehydrated in graded ethanol series and xylene, and coverslipped with Eukitt (Boom, Meppel, The Netherlands).

Choline acetyltransferase (ChAT) immunostaining of basal forebrain cholinergic neurons

ChAT immunostaining was performed on free-floating brain sections in 24-well plates, on a shaker, at RT (unless stated otherwise) as described previously (Van Lieshout et al., 2001; Van der Zee & Hagg, 2002). Every third 40-µm coronal section through the basal forebrain area (medial septum/vertical diagonal band of Broca) was processed for immunocytochemical detection of the cholinergic marker enzyme ChAT (eight sections in total; comprising the majority of the basal forebrain cholinergic neurons). The sections were sampled starting at the decussation of the anterior commissures (‘point 0’) and counting backwards −3, −6, −9, −12, −15, −18, −21 and −24. After rinsing in Tris-buffered saline (TBS; 150 mm NaCl, 10 mm Tris-HCl, pH 7.5; 3 × 10 min) and blocking buffer (TBS/1% bovine serum albumin; 30 min; Sigma), sections were incubated with goat antihuman ChAT primary antibody (Chemicon International Inc., Temecula, CA, USA) diluted 1 : 2000 in blocking buffer for 2 h at room temperature and continued overnight at 4 °C. Biotinylated donkey antigoat Ig (1 : 500 in blocking buffer for 1.5 h; Jackson ImmunoResearch) served as secondary antibody. Then, sections were incubated in an avidin–biotin complex solution (1 : 300; in TBS for 1.5 h; Elite Vectastain Kit, Burlingame, CA, USA). Sections were rinsed in TBS (3 × 10 min) between successive incubation steps. Finally, sections were washed in 0.1 m Tris–HCl (pH 7.4; 10 min). Detection was performed by incubating in 0.1 m Tris-HCl (pH 7.4) containing 0.05% diaminobenzidine (DAB; Sigma), 0.0008% glucose oxidase (Sigma), 0.11% ammonium chloride (Merck), 0.5% d-(+)-glucose (Merck) and 3.33% ammonium nickel(II) sulphate (Sigma) for 20 min. After rinsing in TBS (2 × 10 min), sections were mounted on Superfrost Plus glass slides and air-dried overnight at RT. Following dehydration in graded ethanol series and xylene, slides were coverslipped with Eukitt.

Quantitative analysis

A Sprynt-based PC-image digital analysis system (Bos Inc., Waddinxveen, The Netherlands) containing a videocamera connected to a light microscope (Leitz Wetzlar Germany dialux 20) was used for morphometry and quantitative analysis of nerve sections on coded slides.

Statistics

Data obtained from wildtype and LAR-ΔP mice were presented as means ± SEM. All data measurements were obtained blind as to genotype and subsequently analysed using the appropriate statistical tools (one-way anova, anova with repeated measures, post hoc unpaired Student's t-test, paired t-test; SPSS 9.0 statistics software). Statistical significance was set at P < 0.05.

Delayed sensory function recovery after sciatic nerve crush in LAR-ΔP mice

To investigate the role of LAR in peripheral nerve regeneration, a unilateral crush lesion of the sciatic nerve was applied in wildtype (n = 8) and LAR-ΔP (n = 10) mice. Gradual recovery of sensory function was daily assessed by applying a small current stimulus and scoring the subsequent occurrence or absence of the foot-withdrawal reflex. In wildtype mice, the sensory function response of the nonlesioned sciatic nerve in the contralateral paw showed a fast foot-withdrawal reflex upon stimulation with a 0.1-mA current stimulus to the sole of the foot. This control response was equally fast in the nonlesioned contralateral paw of LAR-ΔP mice. Also, there were no other indicators of sensory loss, such as self-mutilation or skin lesions, observed in the LAR-ΔP cohort.

Wildtype animals showed the first signs of sensory function recovery in their lesioned hindpaw as early as postlesion day 12, with the entire group reaching full recovery by day 16 (Fig. 1). LAR-ΔP mice, however, demonstrated a much delayed sensory function recovery starting at day 13, and reaching full recovery for the entire group only at postlesion day 22 (Fig. 1). Consequently, the average sensory function recovery time was, compared to wildtype animals (14.8 ± 0.5 days), significantly longer for LAR-ΔP mice (19.2 ± 1.1 days; P < 0.003).

Sciatic nerve crush did not affect motor function recovery in LAR-ΔP mice

In the same animals, a gradual recovery of motor function following unilateral sciatic crush lesion was assessed by monitoring the animal's gait in a walking alley. By measuring the footprint parameters of the regenerating hindpaw and the contralateral (nonlesioned) side the sciatic function index could be determined. Before the lesion, both wildtype and LAR-ΔP mice showed a normal gait, indicating normal motor function with regard to walking and the use of leg, paw and toe muscles.

After sciatic nerve crush lesion, wildtype mice showed a gradual return of motor function of the regenerating paw in time (Fig. 2). At postlesion day 3, a sciatic function index value of ≈ −100% established the completeness of the crush lesion. The first signs of motor function recovery became apparent at days 8 and 10 (index value −85 ± 8 and −45 ± 10%, respectively). Subsequently, motor function continued to improve over time, showing an index value of ≈ −20% at 2 weeks postlesion. Full recovery for wildtype animals was reached at postlesion day 18 with index value of −2 ± 5%.

LAR-ΔP mice showed a gradual return of motor function in their gait, similar to that seen in the wildtypes, and also displayed full recovery by postlesion day 18 (sciatic funtion index value −5 ± 5%; Fig. 2). Therefore, motor function recovery after sciatic nerve crush appeared not to be significantly affected in LAR-ΔP mice.

Diminished number of NF-200-positive axons in LAR-ΔP regenerating nerves

NF-200 immunostaining of nonlesioned sciatic nerves from both wildtype (Fig. 3A) and LAR-ΔP (Fig. 3B) mice demonstrated a similar staining pattern, with NF-200-positive axons appearing as small, medium-sized or large red dots (Fig. 3A and B). At 4 days following sciatic nerve crush lesion, and at 3 mm distal to the crush site, wildtype (Fig. 3C) and LAR-ΔP (Fig. 3D) mice showed NF-200-positive staining of newly formed axonal sprouts, seen as small red dots.

Quantitative analysis comprised counting of the number of NF-200-positive axonal sprouts at 1, 3 and 5 mm distal to the crush site, and in three nerve sections per distance for each mouse. The average axon number per 6250 µm² area, at all three distances, are presented in Fig. 3E for both genotype groups. When compared to lesioned wildtype nerves (n = 4), the number of newly outgrowing NF-200-positive axon sprouts in LAR-ΔP nerves (n = 4) was significantly less at all three distances (1, 3 and 5 mm) distal to the crush (anova repeated-measures, between-subjects effects: F_1,6 = 24.83, P < 0.002; Fig. 3E).

Interestingly, the quantitative analysis revealed an increase in the number of NF-200-positive axon sprouts at increasing distance from the crush site in wildtype, but not in LAR-ΔP, nerves (within-subjects contrasts: group × distance, F_1,6 = 5.95, P < 0.05; Fig. 3E). In wildtype nerves the axon count at 5 mm outnumbered that at 3 mm, which in turn exceeded that at 1 mm. This increase in axon numbers more distal to the crush site was not observed in LAR-ΔP nerves (Fig. 3E), corroborating the diminished regenerative sprouting capacity in LAR-ΔP mice.

Decreased axon area, but not myelin area, in LAR-ΔP regenerating nerves

Nonlesioned myelinated nerves

Because there might be a small chance that the expression of cytoskeletal-associated proteins or axonal transport rates are affected by the absence of LAR phosphatase activity, nerve morphometric measurements that are not dependent on axonal marker expression levels and/or transport velocities were analysed as well. Using osmium tetroxide–toluidine blue myelin staining of 1-µm semithin sciatic nerve sections we determined the number of myelinated axons per 10⁴ µm², and measured the axonal area, the axonal area frequency distribution, the myelin area and the myelin : axon ratio in sciatic nerves of wildtype and LAR-ΔP mice (Fig. 4). In nonlesioned sciatic nerves, the overall appearance of the cross-section was similar for wildtype (Fig. 4A) and LAR-ΔP (Fig. 4B) mice, with a total nerve bundle area of 170 546 ± 3530 µm² and 167 853 ± 5039 µm², respectively. Quantitative analysis of the sections of nonlesioned nerves demonstrated no significant differences between wildtype and LAR-ΔP mice regarding the axon area frequency distribution (Fig. 4C), the total number of myelinated axons per 10⁴ µm² (Fig. 4D), the mean myelin area (Fig. 4G) or the mean axon area (Fig. 4H). In addition, the ratio of myelin area to axon area was not different either between nonlesioned wildtype and LAR-ΔP mice (1.19 ± 0.10 and 1.03 ± 0.10, respectively). Concluding, these morphological observations did not reveal any gross abnormalities in sciatic nerve development in LAR-ΔP mice.

Assessing septohippocampal cholinergic collateral sprouting as a model for axonal outgrowth in the mouse central nervous system

The evidence that LAR plays a role in peripheral nerve regeneration prompted us to investigate the impact of LAR on axonal outgrowth in the central nervous system. With axonal regeneration in the adult central nervous system being very limited, or not occurring at all, due to inhibiting factors (Schwab & Bartholdi, 1996; Qiu et al., 2000), we decided to investigate axonal outgrowth during the process of collateral sprouting, a phenomenon that occurs frequently in the peripheral (Diamond et al., 1992) and in the central (Deller & Frotscher, 1997) nervous system. Because LAR is found to be expressed in neuronal cells, including basal forebrain cholinergic neurons (Longo et al., 1993; Schaapveld et al., 1998), the well-described entorhinal cortex lesion model for rats (Lynch et al., 1977; Scheff et al., 1980; Van der Zee et al., 1992; Deller & Frotscher, 1997) was adapted to study collateral sprouting in our mice (see also Zhou et al., 1989; Steward, 1992). The entorhinal cortex glutamatergic neurons are projecting through the perforant path to the outer two-thirds of the molecular layer of the fascia dentata (Amaral & Witter, 1989). Upon lesioning of the entorhinal cortex the basal forebrain cholinergic neurons, which project via the fimbria fornix to the dentate gyrus OML as well, start to form collateral sprouts.

AChE staining of the cholinergic fibres in entorhinal cortex-lesioned mice indeed showed a specific and localized increase in AChE density and fibre number in the OML 5-7), and not in other layers of the dentate gyrus such as the SG, GC or the PL/hilus (Fig. 5, Table 1).

Entorhinal cortex lesion also results in a shrinkage of the denervated dentate gyrus (see review Deller & Frotscher, 1997). Therefore, the width of the nonlesioned contralateral and lesioned ipsilateral dentate gyrus dorsal blade molecular layer was determined (Table 2). A reduction in width of the ipsilateral molecular layer (expressed as percentage decrease) was seen at 1, 2 and 5 weeks following entorhinal cortex lesion in wildtype mice and, to a similar extent, in LAR-ΔP mice (Table 2).

Collateral sprouting in the central nervous system was delayed in LAR-ΔP mice

Wildtype mice which received a unilateral entorhinal cortex lesion showed a clear and augmenting increase in AChE staining density in the ipsilateral OML at 1, 2 and 5 weeks (Fig. 5C, E and G), when compared to the nonlesioned OML (Fig. 5A). In comparison with the contralateral side (Fig. 5B), LAR-ΔP mice also demonstrated an increase in AChE density in the ipsilateral OML at 1, 2 and 5 weeks postlesion (Fig. 5D,F and H), albeit to a lesser extent.

Using quantitative computer-assisted image analysis, the AChE density levels of the ipsilateral and contralateral OML were measured at 1, 2 and 5 weeks following entorhinal cortex lesion (n = 8, 9 and 9, respectively, for both groups; Fig. 7A). Two groups of nonlesioned wildtype (n = 4) and LAR-ΔP (n = 4) animals served as extra controls (0 weeks; Fig. 7A). In wildtype animals, already at 1 week following entorhinal cortex lesion, a significant increase was measured regarding AChE staining density of the OML at the lesioned side, when compared to the contralateral nonlesioned side (Fig. 7A). A further augmented increase in AChE density was seen at 2 weeks, and again a higher percentage increase at 5 weeks postlesion (Fig. 7A). Importantly, the nonlesioned contralateral OML of wildtype and LAR-ΔP mice had similar AChE density levels to those observed in nonlesioned wildtype or LAR-ΔP animals.

In LAR-ΔP mice an increase in AChE staining density of the ipsilateral OML, as compared to the contralateral layer, was also measured (Fig. 7A). At 1 and 2 weeks following entorhinal cortex lesion, an increase in AChE density was visible; this reached a plateau at 5 weeks. However, LAR-ΔP mice demonstrated significantly lower density levels over the whole period when compared to wildtype animals (anova repeated measures, F_1,16 = 8.10, P < 0.012). Post hoc t-test comparisons between the wildtype and LAR-ΔP mice revealed a significantly lower AChE density in LAR-ΔP mice at 1 week (P < 0.05) and at 2 weeks (P < 0.04) postlesion (Fig. 7A). When measured after 5 weeks, the average AChE density level in LAR-ΔP mice was still lower, but not significantly different from the wildtype animals, suggesting that the LAR-ΔP mice were slowly catching up.

In order to investigate whether the entorhinal cortex lesion-induced increase in AChE density is correlated with an increase in cholinergic fibre number, and thus indeed represents collateral sprouting of the cholinergic septohippocampal fibres, the actual number of AChE-positive fibres along the entire width of the ipsi- and contralateral molecular layer was determined. Figure 6 shows a 400× magnification, at which the individual fibres could be counted, of the contralateral (Fig. 6A and C) and ipsilateral (Fig. 6B and D) dentate gyrus molecular layer, at 2 weeks following entorhinal cortex lesion. The number of AChE-positive fibres present at the contralateral sides appeared to be similar for wildtype (Fig. 6A) and LAR-ΔP (Fig. 6C) mice. Ipsilaterally, both wildtype (Fig. 6B) and LAR-ΔP (Fig. 6D) mice demonstrated more AChE-positive fibres, compared to their contralateral side. However, LAR-ΔP mice (Fig. 6D) appeared to contain fewer fibres than observed in wildtype animals (Fig. 6B).

Quantitative measurements, at 2 weeks following entorhinal cortex lesion, confirmed that the actual number of AChE-positive fibres in the ipsilateral molecular layer (523 ± 17), when compared to the contralateral side (462 ± 15), was significantly increased in wildtype animals (n = 9, Fig. 7B). The percentage of entorhinal cortex lesion-induced increase in AChE-positive fibre number (13.4 ± 2.7%, Fig. 7B) was similar to the percentage increase in AChE staining density (13.3 ± 0.9%, Fig. 7A), at 2 weeks postlesion. This suggested that an increased AChE density, like the increase in fibre number, indicated collateral sprouting of the cholinergic septohippocampal fibres. The LAR-ΔP mice (n = 9) showed, at 2 weeks postlesion, a more subtle increase in the number of AChE-positive fibres ipsilaterally (7.0 ± 1.3%, Fig. 7B), which was significantly less than the increase in number of AChE-positive fibres counted in the wildtype animals (P < 0.05; Fig. 7B), indicating a lesser extent of collateral sprouting in LAR-ΔP mice.

Based on the fibre numbers and the time-course of AChE density findings, we concluded that in LAR-ΔP mice the collateral sprouting response of the septohippocampal fibres following entorhinal cortex lesion is delayed compared to wildtype animals.

Entorhinal cortex lesion-related reduction in size (and number) of a subset of ChAT-positive basal forebrain neurons in LAR-ΔP mice

To ascertain that the collateral sprouting is a feature of healthy undamaged neurons, which respond after damage to neighbouring cells, the total number and average size of basal forebrain cholinergic neurons was determined at 2 weeks following entorhinal cortex lesion, at the ipsilateral and the contralateral side of the medial septum/vertical diagonal band (Fig. 8). Surprisingly, a small decrease in the average number of ChAT-positive neurons was noted ipsilaterally in both wildtype (10 ± 3%) and LAR-ΔP (15 ± 3%) mice (Table 3). In addition, the average size of ipsilateral ChAT-positive neurons was significantly decreased in LAR-ΔP mice (5 ± 1%, paired t-test, P < 0.008), but not in wildtype animals (2 ± 1%), when compared to the contralateral neurons. A size–frequency distribution of wildtype and LAR-ΔP ipsilateral cholinergic neurons revealed a general shift to slightly smaller sized neurons in the last group (not shown). Because a small portion of the basal forebrain cholinergic neurons form septo-entorhinal projections (Alonso & Köhler, 1984), it is likely that, upon disruption of their entorhinal connection, this subset of cholinergic cells shows shrinkage and/or loss of their ChAT enzyme staining.

Delayed PNS regenerative sprouting in LAR-ΔP mice

Evidence has been provided showing that morphometrical aberrations are detectable in the CNS of LAR mutant mice (Yeo et al., 1997; Van Lieshout et al., 2001). In addition, LAR mRNA levels change following peripheral nervous system (PNS) sciatic crush lesion in rats (Haworth et al., 1998). This prompted us to investigate whether any phenotypic consequence of the LAR phosphatase deficiency in our LAR-ΔP mice could be noted. No differences in sensory and motor function, nor in the sciatic nerve morphometry (including fibre density, myelin area, axonal area and frequency distribution) were found between nonlesioned wildtype and LAR-ΔP animals. However, similar to what Xie et al. (2001) described for the ST534 LAR mutant mouse, we observed a delayed sensory function recovery and a reduced outgrowth of axonal fibres in the period following sciatic nerve crush lesion in LAR-ΔP mice. The diminished axon outgrowth was demonstrated, both by a reduced number of NF-200-positive axons and by a decreased average axon area of the myelinated axons, in regenerating sciatic nerves. These findings support the regulation of regenerative sprouting by LAR.

It has previously been reported in rats that distal to the crush site more sprouts will be encountered, temporarily, than in the proximal area, consistent with the process of individual regenerating axons forming multiple regenerating fibres, many of which will be aborted during axon maturation (Murray, 1982; Jenq & Coggeshall, 1984; De Medinacelli, 1995). The fact that this temporary increase distal to the crush site is not detected in LAR-ΔP nerves further corroborates the diminished regenerative sprouting capacity in LAR-ΔP mice.

The LAR-ΔP mice result from gene targeting experiments in which the complete genomic information encoding the two tyrosine phosphatase domains of the molecule was deleted. Therefore, in theory, normal levels of a C-terminally truncated LAR protein might be present on cell surfaces. However, exploiting a monoclonal antibody against the extracellular part of native LAR in immunofluorescent flow cytometry we were able to show that LAR levels, at least on haematopoietic cells, are reduced to background values in LAR-ΔP mice (Terszowski et al., 2001). The LAR-ST534 mutant mice studied by Longo and colleagues, on the other hand, were generated through gene trap insertion (Skarnes et al., 1995), and the wildtype LAR protein may still be produced in some parts of the LAR-ST534 mice due to random splicing events (Yeo et al., 1997). It is comforting therefore that our findings confirm and extend those of Longo and coworkers (Xie et al., 2001), and emphasize that the cell-adhesive and/or catalytic activity of the LAR molecule facilitates the process of axonal outgrowth.

As mentioned, Stoker's group found a 50% reduction in LAR mRNA levels 3 days after sciatic nerve crush in rats (Haworth et al., 1998). Longo and coworkers, however, noted that LAR protein expression was increased about twofold at 2 weeks following the crush (Xie et al., 2001). This apparent discrepancy might be caused by the postlesion time points examined (3 vs. 14 days) or the difference in experimental readout (mRNA vs. protein level). Especially relevant for the latter is the notion that LAR protein levels are effectively regulated at the post-transcriptional level through cell contacts (Symons et al., 2002).

Interestingly, the additionally evaluated motor function recovery of the regenerating peripheral sciatic nerve was not affected in LAR-ΔP mice. The validity of our assay is backed by findings for inbred C57BL/6J mice and C57BL/6J-129 mixed background mice (Siconolfi & Seeds, 2001). First signs of ‘normal’ walking (holding the leg instead of dragging, supporting body on injured leg) were reported on day 10, and initial return of the toe reflex on day 11 postlesion. In our assay we used more detailed parameters [footprint length, interdistance, and (inner) toe spreading] that were collected every 2 days during a 3-week period, and found a similar time for return of proper motor function. The demonstration of affected sensory (see above), but not motor, function recovery after sciatic crush lesion perfectly matches studies reporting expression of LAR mRNA and protein in dorsal root ganglia sensory neurons, but the absence of LAR in spinal cord ventral horn motor neurons (Longo et al., 1993; Haworth et al., 1998; Zhang et al., 1998).

Notably, a recent report by McLean et al. (2002) demonstrated an enhanced rate of peripheral nerve regeneration, but also axonal outgrowth directional errors, after sciatic nerve injury in RPTPσ-knockout mice. A faster initial recovery (day 7–10) of normal gait in the walking pattern analysis, a higher nerve fibre count at 1 and 2 weeks post crush lesion, and a more rapid return (day 14) of electrophysiological muscle function was observed in the mutant animals (McLean et al., 2002). Their finding of a higher incidence of extrafascicular regenerating sprouts following nerve transection and immediate repair in RPTPσ mice pointed at abnormal axonal guidance, like that reported for Drosophila DLar loss-of-function mutants (Krueger et al., 1996). For the moment it remains completely enigmatic how LAR and RPTPσ, proteins that are highly homologous and share many biochemical and cell biological properties (Streuli, 1996; den Hertog, 1999; Stoker, 2001), can exert such opposing effects in peripheral nerve regeneration.

Delayed CNS collateral sprouting in LAR-ΔP mice

The current study addressed for the first time the role of LAR in central nervous system axonal outgrowth. With basal forebrain cholinergic neurons expressing LAR mRNA (Longo et al., 1993), the entorhinal lesion-induced collateral sprouting of septohippocampal cholinergic fibres appeared well suited for this purpose. This model has been commonly used in rats (Lynch et al., 1977; Scheff et al., 1980; see also review Deller & Frotscher, 1997), and in a few reports in mice (Zhou et al., 1989; Steward, 1992; Jensen et al., 2000; Del Turco et al., 2002), where the increase in cholinergic markers such as AChE activity or histochemistry was interpreted as cholinergic collateral sprouting. The entorhinal cortex lesion-induced sprouting response in rats has been reported to increase until day 14 postlesion, after which it levelled off and reached a plateau (Deller & Frotscher, 1997). Several reports have questioned the usefulness of AChE density measurements for demonstrating this cholinergic sprouting, because other cholinergic markers were not altered (Aubert et al., 1994), or because shrinkage of the OML would account for the increase in density of ChAT-positive boutons (Henderson et al., 1998). However, 192 IgG-saporin-induced loss of medial septum cholinergic neurons abolished the cholinergic sprouting after entorhinal cortex lesion (Naumann et al., 1997), and ‘absolute’ AChE-positive fibre counts over the entire width of the molecular layer showed that the increased AChE staining density was truely representing a higher number of AChE-positive fibres (Van der Zee et al., 1992; this study). Determining the number of AChE-positive fibres over the entire width of the molecular layer did take into account the shrinkage of the dentate gyrus molecular layer, which occurs due to the lesion-induced retraction and degeneration of the (glutamatergic) entorhinal projection (Deller & Frotscher, 1997). Indeed, a reduction in width of the (lesion side) ipsilateral molecular layer, compared to the (nonlesioned) contralateral side, was observed at all time points following entorhinal cortex lesion in wildtype mice and, to a similar extent, in LAR-ΔP mice.

This study contained a time course with 5 weeks of augmenting AChE staining density and an increase in AChE-positive fibre number in the wildtype mice, indicating collateral sprouting of the cholinergic septohippocampal fibres. Interestingly, collateral sprouting following entorhinal cortex lesion did occur in LAR-ΔP mice as well, but the extent of the sprouting response was lower than that observed in wildtype animals, taking the whole experimental period into account. Similar to the PNS regenerative sprouting, the CNS collateral sprouting was not inhibited but merely delayed, with the level of sprouting in LAR-ΔP mice lagging behind that seen in wildtype animals.

Because previous work, in line with findings for the LAR-ST534 mice, has demonstrated that LAR-ΔP animals show a slightly reduced cholinergic fibre innervation of the dentate gyrus (Yeo et al., 1997; Van Lieshout et al., 2001), one might be tempted to blame the diminished CNS collateral sprouting on this decreased hippocampal innervation. However, the most apparent reduction in cholinergic innervation was reported for the supragranular layer (≈18%) and not so much in the molecular layer (≈4%). The LAR-ΔP cohort used in the current study showed only an ≈2% reduction in AChE density of the nonlesioned OML, which cannot explain the difference in sprouting extent at 1 and 2 weeks. Moreover, 5 weeks after entorhinal cortex lesion the collateral sprouting extent was even approaching that seen in wildtypes. Taken together, the data emphasise that the sprouting response just takes longer, and that the collateral axon sprout formation is merely delayed.

Our data demonstrate that CNS collateral sprouting can be measured reliably and reproducably in wildtype mice (on an almost pure C57BL/6 background), and as such can be applied to different (mutant) mouse strains to reveal in vivo CNS sprouting capacities.

Interestingly, an entorhinal cortex lesion-related reduction in the number of basal forebrain ChAT-positive cells was observed in wildtype mice. It is likely that the affected cells represent the septo-entorhinal projection (Alonso & Köhler, 1984) showing loss of (ChAT) immunostaining due to the lesion. In addition, in lesioned LAR-ΔP mice, not only a reduction in number but also a decrease in the average size of the cholinergic neurons was noted. We hypothesize that this cell body atrophy of the septo-entorhinal projection may reflect a greater sensitivity of these LAR-ΔP neurons to deprivation of target-derived trophic factors. Additional in vitro and in vivo experiments will be required to shed light on this matter.

CAM-like RPTP signalling

Although a well-established role for the CAM-like subfamily of RPTPs in the regulation of neural development and axonal outgrowth has been provided, the underlying signalling mechanisms are still poorly understood. This is reflected by the fact that physiological ligands and substrates for most RPTPs still have to be mapped. For a non-neuronal LAR splice variant the extracellular matrix laminin–nidogen complex has been identified as a ligand (O'Grady et al., 1998), and for RPTPσ some heparan sulphate proteoglycans have been proposed (Aricescu et al., 2002). RPTPδ, however, appears involved in homophilic interactions reminiscent of the type IIB RPTPs mu and kappa (Wang & Bixby, 1999). Alternatively, RPTPs may convey signals coming from other cell surface molecules such as cadherin complexes. Indeed, LAR-like RPTPs can associate with plakoglobin and α-catenin (Kypta et al., 1996), and E-cadherin complexes stabilize LAR protein levels (Symons et al., 2002). The RPTP intracellular regions of course are instrumental in regulating phosphotyrosine levels at the cell membrane, which are intrically linked to adhesive and locomotive processes. In addition, the cytosolic portions of LAR-like RPTPs have the potency to function as docking sites for proteins such as Trio (Lin & Greenberg, 2000; Stoker, 2001) and Liprins (Serra-Pages et al., 1998; Kaufmann et al., 2002; Wyszynski et al., 2002), all known to determine crucial steps in neural pathfinding and cytoskeletal dynamics. LAR thus seems to be in the proper company to serve in a protein machine dedicated to orchestrating extracellular cues and intracellular cytoskeletal movements necessary for efficient growth cone movement and axonal outgrowth.