Nerve Growth Factor Regulates the Firing Patterns and Synaptic Composition of Motoneurons

Development/Plasticity/Repair Nerve Growth Factor Regulates the Firing Patterns and Synaptic Composition of Motoneurons María A. Davis-Lo´pez de Carrizosa,* Camilo J. Morado-Díaz,* Sara Morcuende, Rosa R. de la Cruz, and A ´ngel M. Pastor Departamento de Fisiología y Zoología, Facultad de Biología, Universidad de Sevilla, 41012 Sevilla, Spain Target-derived neurotrophins exert powerful synaptotrophic actions in the adult brain and are involved in the regulation of different forms of synaptic plasticity. Target disconnection produces a profound synaptic stripping due to the lack of trophic support. Consequently, target reinnervation leads to synaptic remodeling and restoration of cellular functions. Extraocular motoneurons are unique in that they normally express the TrkA neurotrophin receptor in the adult, a feature not seen in other cranial or spinal motoneurons, except after lesions such as axotomy or in neurodegenerative diseases like amyotrophic lateral sclerosis. We investigated the effects of nerve growth factor (NGF) by retrogradely delivering this neurotrophin to abducens motoneurons of adult cats. Axotomy reduced the density of somatic boutons and the overall tonic and phasic firing modulation. Treatment with NGF restored synaptic inputs and firing modulation in axotomized motoneurons. When K252a, a selective inhibitor of tyrosine kinase activity, was applied to specifically test TrkA effects, the NGF-mediated restoration of synapses and firing-related parameters was abolished. Discharge variability and recruitment threshold were, however, increased by NGF compared with control or axotomized motoneurons. Interestingly, these parameters returned to normal following application of REX, an antibody raised against neurotrophin receptor p75 (p75 NTR ). In conclusion, NGF, acting retrogradely through TrkA receptors, supports afferent boutons and regulates the burst and tonic signals correlated with eye movements.Ontheotherhand,p75 NTR activationregulatesrecruitmentthreshold,whichimpactsonfiringregularity.Toourknowledge, this is the first report showing powerful synaptotrophic effects of NGF on motoneurons in vivo. Introduction In the adult brain, neurotrophins modulate a variety of functions, from regeneration to myelination and from pain to depression. Frequently, these actions involve synaptic reorganizations (McAllister et al., 1999). This synaptotrophic role of neurotrophins results in changes in the balance between excitation and inhibition and in the number and activity of synapses (Singh et al., 2006; Huh et al., 2008). Moreover, neurotrophins have a role regulating discharge properties at the cellular and network levels (Berardi et al., 2003; Caleo et al., 2003; Luther and Birren, 2009). However, given the multiplicity of pathways and signaling events in the brain, combined with its heterogeneous synaptic organization, no common rule for neurotrophin action emerges that could apply to every part of the brain. Therefore, an in vivo approach is also needed to unravel the actions of neurotrophins under physiological and pathological conditions (Chao, 2003). Extraocular motoneurons express TrkA, the high affinity receptor for nerve growth factor (NGF) (Benítez-Temin˜o et al., 2004). In contrast, other cranial and spinal motoneurons only express it transiently during development. Interestingly, the TrkA receptor reappears in spinal motoneurons after lesion or in response to pathologies such as amyotrophic lateral sclerosis (ALS), a disease that spares extraocular motoneurons (Oba´l et al., 2006; Ferrucci et al., 2010). In this case, surviving motoneurons switch their neurotrophin expression from brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3) to NGF (Nishio et al., 1998). NGF is expressed in developing skeletal muscle, but is downregulated in the adult (Ebendal et al., 1986; Ernfors et al., 1991; Koliatsos et al., 1993; Ip et al., 2001) and upregulated during muscle repair (Deponti et al., 2009) or in response to musculoskeletal pathologies (Ku¨st et al., 2002; Toti et al., 2003). The action of NGF in extraocular motoneurons and reasons for persistence of NGF neurotrophic responses into adulthood are unclear. Decreased retrograde trophic support after peripheral nerve injury leads to profound synaptic stripping of motoneuron membranes that reduces responsiveness to afferent signals (de la Cruz et al., 1996). These alterations revert to normal following reinnervation of natural or exogenous targets (Delgado-García et al., 1988; Benítez-Temin˜o et al., 2005) or when BDNF and NT-3 are exogenously applied (Davis-Lo´pez de Carrizosa et al., 2009). A possible role for NGF in regulating the morphophysiological characteristics of extraocular motoneurons and its afferent inputs has not yet been studied. To analyze this possibility, we induced alterations in the levels of NGF by either axotomy or Received Feb. 8, 2010; revised April 29, 2010; accepted May 7, 2010. This work was supported by MCYT-FEDER Grants BFU2009-07121 and PET2008-0226, Fundacio´n Eugenio RodríguezPascual,andJuntadeAndalucía(P06-CTS-01420andP09-CVI-4617).M.A.D.-L.C.wasascholaroftheFISSin Spain. Some experiments were performed in the Imaging and Biology central services (CITIUS). We thank Dr. Louis Reichardt, University of California, San Francisco, for generously providing the REX antibody. We also thank Dr. Francisco Alvarez, Wright State University, for critical reading of the manuscript. *M.A.D.-L.C. and C.J.M.-D. contributed equally to this work. Correspondenceshould be addressedto Dr. AngelM. Pastor, Departamentode Fisiología yZoología, Facultad de Biología, Avenida Reina Mercedes 6, 41012 Sevilla, Spain. E-mail: [email protected]. DOI:10.1523/JNEUROSCI.0719-10.2010 Copyright © 2010 the authors 0270-6474/10/308308-12$15.00/0 8308 •The Journal of Neuroscience, June 16, 2010 •30(24):8308–8319 externally applying NGF to the stump of the sectioned nerve (Davis-Lo´pez de Carrizosa et al., 2008). Then, we examined the actions of NGF through its two receptors, neurotrophin receptor p75 (p75 NTR ) and TrkA. Our data show that NGF has powerful synaptotrophic effects on excitatory and inhibitory synaptic inputs to motoneurons that control firing patterns and discharge characteristics. Materials and Methods Animals and surgical procedures. Experiments were performed on adult female cats weighing 2.0–2.5 kg obtained from authorized suppliers (Servicio de Animales de Experimentacio´n, Universidad de Co´rdoba, Co´rdoba, Spain). Animals were surgically prepared for chronic recording of eye movements and neuronal extracellular recordings in the abducens nucleus (n⫽6) or for histological purposes (n⫽6). The latter were just instrumented for chronic supply of NGF. Surgical and handling procedures for chronic experiments were in compliance with current legislation (R.D. 120/2005 BOE 252/34367-91, 2005). Animals instrumented for chronic recordings received a vagolytic injection of atropine sulfate (0.5 mg/kg, i.m.) and were then anesthetized with sodium pentobarbital (35 mg/kg, i.p.) and positioned in a stereotaxic frame to implant bipolar stimulating electrodes in the abducens nerve and the medial longitudinal fasciculus to antidromically activate abducens motoneurons and interneurons, respectively (Fig. 1A). Coils, made up of two turns of Teflon-insulated stainless-steel wire, were implanted in the sclera of both eyes. A recording chamber made of acrylic resin was constructed over a squared window (5 mm per side) drilled in the occipital bone to allow the transcerebellar passage of recording micropipettes. Finally, a pedestal made of dental acrylic with embedded bolts was constructed to immobilize the head during the recordings. Postoperative and continuous care. Postoperative care was provided daily throughout the experiment. During the initial 3 d, antibiotics (streptomycin and penicillin, 20.000 I.U./kg, i.m.), and analgesics (pirazolone, 0.1 g/kg, i.m.) were administered. The eyes and the skin around the implanted sockets were also inspected daily and cleaned if needed with sterile saline. During the postoperative period, the animal was adapted to the restraining system for periods of 20 min during the aseptic cleansing of the recording chamber with sterile saline and instillation with drops of antibiotics (0.3% gentamicin sulfate) and calibration of eye movements. For closure of the chamber, the cerebellar surface was protected with a silicone sheath and the chamber sealed with aseptic gauze and a cap. Chronic recordings. Control recording sessions started after 2 weeks of postoperative recovery. The animal was gently restrained in a fabric bag, wrapped with elastic bandages, and placed in a Plexiglas cradle inside the magnetic field search coil (Fuchs and Robinson, 1966). Extracellular recordings were performed with glass micropipettes pulled with long shanks (30 mm) and tips beveled to a resistance of 1–2 M⍀ when filled with 2 MNaCl. The left abducens nucleus was located with the aid of the antidromic field potential produced by electrical stimulation of the VIth nerve. The extracellular neuronal activity was amplified and filtered at a bandwidth of 10 Hz–10 kHz for display and digitalization purposes. Data storage and analysis. Horizontal and vertical eye position of both eyes and associated neuronal activity were digitally stored for off-line analysis (Power 1401, Cambridge Electronic Design). Computer programs written in Matlab 6.5 displayed instantaneous firing frequency (i.e., the reciprocal of the interval between two adjacent spikes) and the position of both eyes for the purpose of data selection. Relationships between neuronal firing rate (FR) [in spikes/s (sp/s)] and horizontal eye position (EP) [in degrees (deg)] were obtained by linear regression analysis to calculate the slope, i.e., the neuronal sensitivity to eye position (k, in sp/s/deg), the ordinate intercept (F 0 , in sp/s), and the abscissa intercept (the recruitment threshold, in deg). Firing rate during fixations was represented by the equation FR ⫽k⫻EP ⫹F 0 . Relationships between neuronal firing and eye velocity during saccades were obtained by linear regression after subtraction of the position component (k⫻EP) calculated from the previously known sensitivity to eye position. Thus, the equation used was FR ⫺k⫻EP ⫽ r⫻EV ⫹F 0 , where r(in sp/s/deg/s) is the neuronal sensitivity to eye velocity (EV) (in deg/s). Due to the impairment of eye movements of the affected eye, we used the noninjected, contralateral eye for computation of eye movements. Abducens nerve section and preparation of the implantable intraorbitary device. For the delivery of NGF or inhibitors of NGF receptors, we implanted the stump of the abducens nerve in a custom-made suction chamber that allowed the control of the extracellular environment that bathed the nerve stump. This also prevented axonal reinnervation of extraocular muscles or any other structure (Baker et al., 1981; Delgado- García et al., 1988). This so-called intraorbitary device consisted of a chamber with two pieces of tubing inserted for the push–pull of fluid Figure 1. A, Diagram of the experimental design for single-unit and eye movement recordings (Rec). The abducens nucleus (ABD) contains motoneurons (Mn), which innervate the lateral rectus muscle (LR), and internuclear neurons (Int), which has axons that contact the contralateral medial rectus muscle (MR) innervating motoneurons through the oculomotor nerve (IIIn) of the oculomotor nucleus (OCM).Abducensneuronsreceiveburstinputfromexcitatory(EBN)andinhibitory(IBN)reticularneurons,andtonicinputsfrombilaterally (i,ipsilateral;c,contralateral)projectingprepositushypoglossi(Ph)neurons.Bipolarstimulating(St.)electrodeswereimplantedontheVIth nerve(VIn;St.1)andonthemediallongitudinalfasciculus(MLF;St.2).Aftercontrolsessions,thelateralrectusmusclewasremovedandan intraorbitarydevice (IOD)was implantedin the VIthnerve. B,Diagram ofthe timecourseof treatments. Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing J. Neurosci., June 16, 2010 •30(24):8308–8319 • 8309 (Davis-Lo´pez de Carrizosa et al., 2008). The chamber was made with a standard polypropylene pipette tip shortened to ⬃15 mm, thus giving it a conical shape. The wider side of the cone contained the tips of two 6-cm-long pieces of thin-wall polytetrafluoroethylene (PTFE) capillary (26 gauge, Small Parts). The tubing was glued in place to the wide side of the chamber. The narrow part of the conical chamber was left open, ⬃1 mm wide, to tightly fit the stump of the abducens nerve. After a few control recording sessions to locate the abducens nucleus, the intraorbitary device was implanted under general anesthesia (dose as above) in a second surgical step. The eye was approached temporally through a skin incision over the zygomatic arch. The abducens nerve was dissected free and the lateral rectus muscle removed. The nerve stump was inserted in the chamber by applying suction through the tubing system. The tubing was passed under the skin and was connected into a capped plastic chamber affixed to the head restraint system with dental acrylic. Administration of NGF and the inhibitors K252a and REX. The protocol for NGF (Sigma) delivery consisted of single injections of 15 ␮ lofa100 ng/ ␮ l solution in saline into the nerve chamber. These were delivered on alternate days over a period of 4– 6 weeks (Fig. 1B). NGF, with or without the different inhibitors, was delivered during the recording session after rinsing the chamber thoroughly with sterile saline. The concentration used was similar to those used in other in vivo studies (Villa et al., 1996; Pizzorusso et al., 1999). K252a, a selective inhibitor of tyrosine protein kinase activity (Rutherford et al., 1997; Rattiner et al., 2004; Bozdagi et al., 2008), was delivered at 100 nMin saline with 2% dimethylsulfoxide mixed with NGF (Tapley et al., 1992). To block p75 NTR activity, the polyclonal antibody against the rat extracellular domain of p75 NTR (referred to as REX; Weskamp and Reichardt, 1991) was injected at a final concentration of 200 ␮ g/ ␮ l. This concentration is within the range used for in vivo experiments (Lucidi-Phillipi et al., 1996; Pizzorusso et al., 1999). The REX antibody was a gift from Dr. Louis Reichardt (University of California, San Francisco, San Francisco, CA). Finally, the vehicle control consisted of the administration of sterile saline. NGF treatment started on the day of axotomy (two animals), but administration of NGF⫹K252a (two animals) or NGF⫹K252a⫹REX (two animals) started 15 d after NGF treatment. Three additional animals from our previous study (Davis-Lo´pez de Carrizosa et al., 2009) that were only axotomized and had received saline injections were also used. Western blot analysis. Three cats were deeply anesthetized (sodium pentobarbital, 60 mg/kg, i.p.) upon termination of recording experiments to dissect extraocular muscles. This tissue was homogenized in cold lysis buffer containing a mixture of protease and phosphatase inhibitors, disrupted by sonication and centrifuged at 13,000 rpm for 30 min at 4°C. The supernatants were isolated and total protein concentration was determined by the Bradford method (Bradford, 1976), using bovine serum albumin (BSA) as a standard. Proteins were diluted in sample buffer, denatured at 95°C for 6 min, and then separated by 15% SDS PAGE (50 ␮ g/lane). They were then transferred to polyvinylidene difluoride membrane by electroblotting. Nonspecific binding sites were blocked with 5% BSA for 1 h and then the blots were incubated overnight at 4°C in a solution containing an antibody against NGF (1:200, sc-548, Santa Cruz Biotechnology) diluted in TBS-Tween 0.1% supplied with 5% BSA. The membranes were incubated in HRP-conjugated anti-rabbit antibody (1: 200) for1hatroom temperature. The immunoreaction was detected with an ECL Western blotting kit (GE Healthcare). Bands were visualized on Image Reader LAS-3000 (Fujifilm). Immunocytochemical procedures for fluorescent confocal microscopy. Six animals were prepared just for histology. Two per group received NGF, NGF⫹REX, or NGF⫹REX⫹K252a for 15 d after axotomy. Two additional animals from our previous study were just axotomized (Davis- Lo´pez de Carrizosa et al., 2009). After terminal anesthesia (sodium pentobarbital, 60 mg/kg, i.p.), animals were perfused intracardially with physiological saline followed by 4% paraformaldehyde in 0.1 Mphosphate buffer, pH 7.4. The brain was extracted and postfixed in the same solution for another 2 h. A vibratome was used to obtain 50- ␮ m-thick coronal sections containing the abducens nucleus. These were washed in 0.01 Mphosphate buffer, pH 7.4, with 0.9% saline and containing 0.1% Triton X-100 (PBS/TX), and then nonspecific binding sites were blocked with normal donkey serum (10% in PBS/TX) and left overnight at 23°C in different combinations of primary antisera. Triple immunofluorescence labeling was, in general, for choline acetyltransferase (ChAT) for motoneurons, synaptophysin (SYN) for synaptic boutons, and glial fibrillary acidic protein (GFAP) for astrocytes. Dual immunofluorescence was usually performed against ChAT and either TrkA (for NGF high-affinity receptor) or the low-affinity p75 NTR . Primary antisera cocktails were performed by mixing the following primary antibodies and dilutions: ChAT (goat pAb, 1:300; Millipore), SYN (rabbit pAb, 1:500; Zymed Laboratories), GFAP (mouse mAb G-A-5, 1:800; Sigma), TrkA (directed against a specific peptide mapping adjacent to the carboxy terminus of the TrkA receptor; rabbit pAb, 1:400; Santa Cruz Biotechnology) and p75 NTR (rabbit, 1:200; a gift from Dr. Louis Reichardt). VGAT immunoreactivity was used to calculate which fraction of the total number of boutons was inhibitory (Gonza´lez-Forero et al., 2004). VGAT is the vesicular transporter of GABA and glycine (rabbit pAb, 1:500; Millipore). After washing, sections were incubated for 1–1.5 h in a second mixture of antibodies fluorescently tagged as follows: donkey anti-rabbit IgG coupled to Cy2, donkey anti-goat IgG coupled to Cy3 and donkey anti-mouse IgG coupled to Cy5 (Jackson ImmunoResearch). Secondary antibodies were diluted 1:50 to 1:100 in PBS/TX. Finally, sections were washed in PBS and mounted on glass slides and coverslipped with Dako- Cytomation fluorescent mounting medium (Dako). Confocal microscopy (Leica TCS SP2) at high magnification was used to sequentially capture images at the same focal plane by means of different filters and merged in ImageJ (NIH). Gray scales were adjusted to maximize their dynamic range. Neurons included in the analysis were randomly sampled at high magnification (⫻63) in 1–4 different focal planes containing the nucleus and separated by 2–3 ␮ m along the z-axis. SYN- or VGAT-immunoreactive (IR) boutons were counted to calculate the linear density of boutons. The synaptic coverage was also measured as the percentage ratio between the perimeter occupied by SYN-IR, VGAT-IR terminals, or GFAP-IR profiles over the ChAT-positive somata of motoneurons (Pastor et al., 1997). The p75 NTR and TrkA receptor mean gray value was also determined within the somatic boundaries of ChAT-positive motoneurons and scaled to the somatic area. The mean gray value in the neuropil of SYN, VGAT, and GFAP was also measured using square boxes of 30.4 ␮ m per side (80 pixels per side), sampled through the neuropil of images captured at ⫻40. This measurement was considered an index of the optical density of immunolabeling in the nucleus and was computed as the average gray value within the selection after background level subtraction. That is, the sum of the gray values of all the pixels (8-bit resolution) divided by the number of pixels. For comparisons between groups, one-way ANOVA at a significance level of p⬍0.05 followed by post hoc comparisons were performed in Sigma Plot 11 (Systat Software). Results TrkA and p75 NTR expression in abducens motoneurons are modulated after axotomy Abducens motoneurons, identified by ChAT immunoreactivity (Fig. 2A–F, insets), expressed constitutively the p75 NTR (Fig. 2A) and the TrkA receptor (Fig. 2D) (Benítez-Temin˜o et al., 2004) in the form of a speckled immunoreactive pattern covering the cell body and axons (Fig. 2A,D,H). The mean gray value of both NGF receptor immunoreactivities at the somatic level increased 15 d after axotomy (ANOVA; p⬍0.001) (Fig. 2B,E,G). NGF administration to the cut nerve demonstrated a partial recovery of the p75 NTR receptor, although it was still higher than normal values, whereas TrkA immunoreactivity was similar to axotomy and therefore higher than control (Fig. 2C,F,G). NGF receptors have been shown to be upregulated after axotomy (Johnson et al., 1999; Cui et al., 2002; Yuan et al., 2006) and after activation of the TrkA receptor (Rankin et al., 2008). The presence of NGF in the extraocular muscles was demonstrated by Western blot, as illustrated for the lateral rectus (Fig. 2I). A dense band at ⬃15 kDa, corresponding to the mature form of NGF protein, and several 8310 •J. Neurosci., June 16, 2010 •30(24):8308–8319 Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing bands at ⬃32 and 40 kDa, indicative of NGF precursors (Buttigieg et al., 2007), were observed. NGF reinstates the firing pattern of axotomized abducens motoneurons The tonic and phasic firing of abducens motoneurons during spontaneous eye movements has been extensively described (Delgado-García et al., 1986). Control abducens neurons had a tonic and regular discharge for eye position during eye fixations (Fig. 3A) that increased proportionally to the degree of eye rotation directed toward the ipsilateral side of recording, namely the on-direction, represented by a leftward movement recorded in a motoneuron on the left side (Figs. 3A,4A, numbers depicted 1 through 4). Firing decreased proportionally, and eventually stopped whenever the eye moved toward the off-direction (Figs. 3A,4A, numbers indicated 5 through 7). The correlation of firing rate versus eye position allowed calculation of the eye position sensitivity (Fig. 4B). The slope of the linear regression line obtained between the firing rate and the eye position represents the neuronal sensitivity to eye position (k, measured in sp/s/deg) (Fig. 4B). The mean eye position sensitivity computed from 70 control abducens motoneurons was 6.4 ⫾0.26 sp/s/deg (Table 1). The abscissa intercept of the firing rate to eye position plot is considered the eye position at which the unit is recruited into activity (Fig. 4B). Mean recruitment threshold was ⫺7.1 ⫾0.86° in control motoneurons (Table 1). Motoneuronal activity was also related to eye velocity during saccades, thus motoneurons displayed highfrequency bursts and pauses during on- and off-directed saccades, respectively (Fig. 4A, red box). The slope of the linear regression analysis between firing rate and eye velocity was the sensitivity to eye velocity, namely r(Fig. 4C). In control motoneurons, r⫽0.73 ⫾ 0.04 sp/s/deg/s (Table 1). The firing pattern of abducens motoneurons (Fig. 3B), but not that of internuclear neurons (data not shown), was perturbed after VIth nerve axotomy (Delgado-García et al., 1988; Davis-Lo´pez de Carrizosa et at., 2009). Changes took place over the first week and did not recover during the entire survival time (Fig. 4D, green circles). Axotomized motoneurons showed an overall reduction in firing rate and a severe decay of modulation related to eye position and velocity (Fig. 3B, Table 1). By contrast, administration of NGF immediately after lesion not only prevented the decay in eye position and velocity sensitivities due to axotomy, but also quickly restored the firing pattern of motoneurons and enhanced their sensitivities to the parameters of eye movement (Fig. 3C,D). However, the effects of retrograde administration of NGF through the cut nerve produced other changes in the firing pattern that were apparent from the first week after treatment (around the second recording session) (Fig. 4D, blue triangles). NGF produced a significant rise in both eye position (9.37 ⫾0.30 sp/ s/deg) and velocity (0.93 ⫾0.03 sp/s/ deg/s) sensitivities with respect to control (Table 1, Fig. 4D, gray band labeled as NGF) (p⬍0.05). These were maintained at high levels throughout the course of the experiment (Figs. 3C,D,4D, blue triangles). Remarkably, motoneuronal recruitment threshold was shifted toward more eccentric eye positions (Table 1). We believe this may be indicative of higher levels of inhibition, despite the overall higher sensitivities to eye position and velocity compared with control. Results on motoneuronal recruitment threshold will be presented later. A concerted action of NGF through TrkA and p75 NTR receptors We tested whether the effects of NGF on abducens motoneurons were mediated via high- and/or low-affinity receptors by the sequential blocking of TrkA and p75 NTR by using K252a and REX, respectively, as selective inhibitors. Because axotomized motoneurons first showed significant differences from control cells after 14 d (Fig. 4D, green circle at 14 d) ( p⬍0.05), we initiated the pharmacological blockade of NGF receptors by this time. We administered NGF⫹K252a 15 d after the axotomized motoneurons had been treated with NGF (Fig. 3E). K252a added to NGF did not alter the motoneuronal firing modulation with eye position and eye velocity compared with control (Fig. 3E). Motoneuronal sensitivity decayed rapidly, and within ⬃5 d it stabilized at values that were different from the axotomy and the NGF paradigm (Fig. 4D, cyan squares), but not different from control values (Table 1). This result indicated that NGF acting through the p75 NTR alone was sufficient to restore the firing sensitivity of the motoneuron, but when p75 NTR acted together with the TrkA receptor, the effects were cooperative, resulting in sensitivities that rose above control. Moreover, the dual blockade of both Figure2. Changesontheexpressionofp75 NTR andTrkA.A–C,Motoneuronsillustratingp75 NTR immunostainingincontrol(A), 15 d after axotomy (B), and 15 d treated with NGF after axotomy (C). Insets show ChAT immunoreactivity for the same motoneurons. D–F, Same as A–C, but for TrkA immunostaining. G, Changes in cytoplasmic mean gray value (MGV in counts per pixel, cpp) afterp75 NTR (graybars)andTrkA(blackbars)immunoreactivity.Asterisksanddotsindicatedifferenceswithrespecttothecontrol andaxotomy groups,respectively (ANOVA,Holm-Sidak methodfor pairwisemultiple comparisons,p⬍0.01). H,Axons inthe rootletsof theabducensnerve(ABDn)demonstratingTrkAimmunoreactivity.I,ImmunoblottingshowingtheexpressionofNGFproteininthelateral rectusmuscle (LRm).Size markersunits are kDa.CON, Control;Ax, axotomy.Scale bars: F,20 ␮ m;H,2 ␮ m. Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing J. Neurosci., June 16, 2010 •30(24):8308–8319 • 8311 types of NGF receptor with K252a and REX led to the complete blockade of NGF actions (Fig. 3F). In this case, the overall firing rate of motoneurons resembled the behavior after axotomy, showing very low sensitivities (Fig. 4D, red diamonds) that were similar to the axotomy values (Table 1). It should be noted that, in contrast with axotomy, motoneurons that were doubly blocked fired in a continuous, low-modulation mode, as if they were disinhibited, but with very little excitatory drive (Fig. 3F). These results suggest that the sensitivities of axotomized motoneurons were enhanced by one receptor to control values and further boosted when both receptors were activated by NGF. Recruitment threshold changes with activation of p75 NTR Recruitment threshold, calculated as the abscissa intercept of the rate-position plot, changed after NGF treatment (supplemental Fig. 1A, available at www. jneurosci.org as supplemental material). Abducens motoneurons show orderly recruitment patterns that correlate with neuronal sensitivity (Pastor and Gonza´lez- Forero, 2003). Thus, when regression lines between firing rate and eye position are plotted for individual cells, the slopes (k) are higher as the cells become recruited further toward ipsilateral eye positions (supplemental Fig. 1B, available at www.jneurosci.org as supplemental material, for the 70 control motoneurons recorded). When we plotted eye position sensitivity (k) versus recruitment threshold for each neuron, an exponential correlation was found for each experimental group (supplemental Fig. 1C,D, available at www.jneurosci.org as supplemental material). Motoneurons treated with NGF not only had higher sensitivities to eye movements—that is, both the tonic sensitivity (k) during fixations and the phasic sensitivity (r) during saccades were enhanced—but they also showed a higher recruitment threshold (Table 1). The recruitment threshold, i.e., the eye position at which the motoneuron first enters into activity, shifted from a mean of ⬃⫺6to⫺7° in control and axotomized motoneurons to ⬃⫺2° in NGF-treated cells (Table 1; supplemental Fig. 1A, available at www.jneurosci.org as supplemental material). Interestingly, NGF⫹K252a-treated cells (block of the high affinity receptor) did not show this modification of the average recruitment threshold, indicating that the shift in recruitment threshold was due to p75 NTR . Furthermore, when we blocked both types of NGF receptor, the recruitment threshold returned to axotomized (similar to control) values (Table 1, supplemental Fig. 1A, available at www.jneurosci.org as supplemental material). Alterations in firing variability are due to p75 NTR activation It was noticed that the variability of firing changed in response to the treatments. This was best seen in instantaneous firing frequency traces, which showed larger variability compared with controls (Figs. 3, 5A–C). To analyze firing regularity, we selected periods with stationary firing rate (Fig. 5D–F). The histograms of the interspike intervals (ISI) were constructed with at least 80 intervals from the same eye fixation (Fig. 5G–I). ISI histograms followed normal distributions (Kolmogorov–Smirnov test, p⬎0.05), except those from NGF-treated cells, which displayed Poisson-like distributions that were highly skewed toward long intervals. Control cells had the smallest variance of ISIs (Fig. 5G). In contrast, NGF- treated motoneurons showed a distribution with both shorter and longer intervals (Fig. 5I). The values of axotomized neurons were slightly expanded toward longer intervals (Fig. 5H). Autocorrelation histograms showed that control motoneurons produced highly regular spike trains. In control and axotomized motoneurons, up to eight peaks were readily observed in the autocorrelogram (Fig. 5J,K). However, no repetitive pattern could be seen in the autocorrelograms of NGF-treated cells (Fig. 5L). The dependence of the SD of the mean firing rate on the mean firing frequency has been previously described in spinal and extraocular motoneurons (Powers and Binder, 2000; Gonza´lez- Forero et al., 2002). We selected the coefficient of variation (i.e., the percentage ratio of the SD to the mean), because it is insensitive to scale changes and thus adequate to establish comparisons between groups. The coefficient of variation was higher at lower frequencies but decreased at higher frequencies, as exemplified Figure3. EffectsofNGFandNGFreceptorblockersonthedischargepatternofabducensmotoneurons.A,Firingrate(FR)profile of a control abducens motoneuron recorded during spontaneous eye movements. The single-unit discharge of action potentials (AP) is shown in the middle trace. EP, Eye position; L, movement to the left; R, movement to the right. B, Same as A, but for a motoneuron recorded 20 d after axotomy without reinnervation. Note reduced firing during both saccades and spontaneous fixations.C,D,NGFtreatmentcausesrecoveryoftheburst-tonicfiringpattern;however,thedischargevariabilityincreasedduring eyefixations.E,MotoneuronstreatedwithK252astillshowedpartoftheeffectsproducedbyNGF,indicatingthatpartoftheeffects of NGF could also be mediated via the p75 NTR .F, Motoneurons treated with K252a and the p75 NTR blocker REX demonstrated an axotomy-like firing profile although they were firing continuously. 8312 •J. Neurosci., June 16, 2010 •30(24):8308–8319 Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing for 2097 periods obtained from 70 control motoneurons. The coefficient of variation versus firing rate was fitted by a double exponential function (Fig. 6A). The firing variability was slightly higher than the control in the axotomized motoneurons, but extraordinarily larger in those treated with NGF at all frequencies tested (ANOVA, p⬍0.05) (Fig. 6B). We compared the variability across groups in the frequency range where the coefficient of variation was relatively stable and motoneurons of all groups were represented, that is, between 30 and 80 sp/s (Fig. 6C). Firing variability was slightly higher than the control in the axotomy group, much larger in the NGF group, and slightly less when the TrkA receptor was blocked, with all these differences being significant ( p⬍0.05). The variability returned to axotomy level only when both NGF receptors were blocked (ANOVA, Holm- Sidak, p⬍0.05) (Fig. 6C, red bar). Therefore, firing variability was mainly controlled through the activation of p75 NTR . NGF synaptotrophism prevents synaptic stripping Functional alterations, as described above, could arise because of changes in the synaptic composition of inputs to the motoneuron. The somatic surface coverage of synaptic and glial processes was measured by means of immunoreactivity against SYN and GFAP, respectively. To determine the ratio of inhibition to excitation, we measured the synaptic coverage of inhibitory boutons by using immunostaining against VGAT. In control abducens motoneurons, 46.1 ⫾2.0% of the somatic perimeter was surrounded by SYN-IR boutons (Figs. 7Aand 8) and only 8.7 ⫾ 1.4% was covered by astroglial filaments (supplemental Fig. 2A, available at www.jneurosci.org as supplemental material). Approximately 70% of the somatic boutons were VGAT-IR in controls (Fig. 7B). Axotomy reduced the coverage of SYN-IR boutons to 21.4 ⫾2.1% (ANOVA, p⬍0.05) (Fig. 8); a large proportion of the remaining boutons were VGAT-positive. In contrast, astrocyte processes proliferated to cover up to 40.1 ⫾ 1.6% of the cell surface (supplemental Fig. 2, available at www. jneurosci.org as supplemental material, Axotomy). The administration of NGF to axotomized motoneurons prevented the drop in both SYN-IR and VGAT-IR boutons (Figs. 7 and 8, NGF). NGF also reduced the GFAP reaction induced by axotomy to an intermediate level (14.1 ⫾1.1%) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material, NGF). Blockade of NGF action via p75 NTR with the REX antibody also prevented SYN-IR synaptic stripping (Fig. 8, NGF⫹REX), indicating that the synaptotrophic action of NGF occurred through the TrkA receptor. Finally, the dual blockade of NGF receptors produced an axotomy-like elimination of boutons (Fig. 8, NGF⫹K252a⫹REX). Changes in the density of SYN-IR and VGAT-IR boutons occurred in parallel in the different experimental groups except for the NGF⫹REX treatment (Fig. 8). The neuropil of the abducens nucleus showed similar quantitative changes as in the perimeter of motoneurons. Mean gray value, an index of the optical density, measured in the neuropil, demonstrated a similar trend in synaptic coverage with SYN-IR and VGAT-IR profiles (Fig. 7,C,D, supplemental Fig. 2B, available at www.jneurosci.org as supplemental material). Remarkably, the ratio of VGAT-IR to SYN-IR boutons changed from 0.75 in control to 0.6 in the doubly blocked NGF- treated motoneurons, indicating a reduction in the proportion of inhibitory boutons. However, the ratio with respect to control did not change in axotomy, NGF-treated, or NGF⫹REX groups ( ␹ 2 test, p⬎0.05). The fact that the ratio of inhibition to excitation was larger in the control and NGF-treated cells than in double-blockade cells was in agreement with the physiological Figure 4. NGF produces recovery of the discharge characteristics of axotomized motoneurons. A, Control motoneurons burst for on-directed saccades and fire tonically during eye fixations. Panels show from top to bottom: right horizontal eye position (RH EP), velocity (RH EV), andthefiringrate(FR).L,Movementtotheleft;R,movementtotheright.B,Regressionplotof the motoneuron shown in Aobtained during spontaneous fixations. Some of the fixations are labeled with numbers both in Aand B. The slope of the regression line is the eye position sensitivity of the motoneuron (k). In this case, k⫽7.0 sp/s/deg (r⫽0.89; p⬍0.001). The interceptwiththeabscissa representsthe eyepositionforneuronal recruitment. Th,Threshold. C,Partialregressionplotduringspontaneoussaccades(liketheboxedareainredinA)obtained after plotting firing rate, minus the position component, versus eye velocity. The slope of the regression line is the eye velocity sensitivity of the motoneuron (r). In this case, r⫽0.75 sp/s/deg/s (r⫽0.74; p⬍0.001). D, Time course of changes in eye position sensitivity of motoneuronsthat received thedifferent treatments.The yellow star indicates thecontrol population mean. NGF-treated cells are indicated with blue triangles and axotomized ones with green circles. The dual NGF-receptor blocking treatments started 15 d after NGF treatment and consisted on NGF⫹K252a (cyan squares) and NGF⫹K252a⫹REX (red diamonds). The dashed lines and gray bands indicate the mean and SE of the control, axotomized, and NGF-treated populations. Asterisks and dots indicate differences with the control and axotomy group, respectively (ANOVA; Holm-Sidak; p⬍0.05). Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing J. Neurosci., June 16, 2010 •30(24):8308–8319 • 8313 recordings. These motoneurons showed much higher levels of inhibition in overall firing than after NGF treatment. In contrast, inhibition was scarce after double blockade (Fig. 3, notice the large periods of inhibition in Cand Dvs the almost continuous low-rate discharge in F). Thus, it appears that if NGF-treated cells showed a larger excitation, they also must have an equally large inhibition, whereas the contrary can be applied for the doubleblockade group (i.e., the more reduced the excitation, the less inhibition was present). Changes in the inhibition to excitation balance To check this idea, we measured k-on and k-off as an index of the excitation and the inhibition received by the motoneurons during firing activity. The k-on value was measured as the slope of the regression line between firing rate and fixations attained after on-directed saccades. The opposite held for the k-off parameter. Axotomy reduced both the k-on and k-off values and NGF treatment elevated both above the control level (ANOVA test, p⬍0.05) (Fig. 9A). The progressive blockade of NGF receptors demon- Table 1. Effects of NGF and receptor blocking on motoneuronal sensitivities and recruitment threshold Treatment Nk(sp/s/deg) r(sp/s/deg/s) Th (deg) Control 70 6.40 ⫾0.26 0.73 ⫾0.04 ⫺7.10 ⫾0.86 Axotomy 53 3.54 ⫾0.25* 0.47 ⫾0.03* ⫺6.62 ⫾1.00 NGF 134 9.37 ⫾0.30* † 0.93 ⫾0.03* † ⫺2.22 ⫾0.31* † NGF⫹k252a 46 5.31 ⫾0.34 †‡ 0.65 ⫾0.04 †‡ ⫺1.59 ⫾1.51* † NGF⫹k252a⫹REX 39 4.34 ⫾0.27* ‡ 0.53 ⫾0.04* ‡ ⫺5.28 ⫾0.73 Values are means ⫾SEM. Symbols indicate significant differences between control (*), axotomy ( † ), and NGF ( ‡ ) groups (one-way ANOVA, Holm–Sidak method for post hoc comparisons, p⬍0.05). Th, Threshold. Figure5. NGF effectson thedischarge variability ofaxotomized motoneurons.A–C, Firingpatterns of control(A), axotomized(B), andaxotomized butNGF-treated (C) motoneurons.The firing rate(FR)incontrolandaxotomizedmotoneuronsshowedahighlyregularprofileduringspontaneouseyemovements(EP).Thesingle-unitdischargeofactionpotentials(AP)isshowninthemiddle trace.D–F,Periodsof1sofinstantaneousfiringrateinacontrol(D),axotomized(E),andNGF-treated(F)abducensmotoneuron.Stationaryperiodswereselectedashavingsimilarmeanfiringrate acrossgroups. Note,however, thelarge increaseof boththe SD and the coefficient of variation (CV) in F.G–I, First-order ISI histogramsfrom thecorresponding cellsshown inA–C, butconstructed usinga2ssample(binsof0.1ms).DistributionswerenormalexceptinI(Kolmogorov–Smirnovtest;p⬎0.05),whichshowedahighskewtotheright.J,Thecontrolautocorrelogram(1msbinsize) showed one clear peak, isolated from the base, and up to three clear peaks, fused in the base (same cell as in A,D, and G). K, The autocorrelogram of the axotomized motoneuron shown in Band E demonstrated a somewhat higher variability than the control motoneuron. L, The autocorrelogram of the NGF-treated motoneuron illustrated in Cand Fshowed a complete absence of spike predictability. 8314 •J. Neurosci., June 16, 2010 •30(24):8308–8319 Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing strated that both parameters decayed partly with the K252a treatment and totally when the dual block of NGF receptors was used, resulting in values similar to axotomy (Fig. 9A). The overall consequences of NGF treatment and blockade are summarized in Figure 9B. The overall effect of NGF treatment is a counterclockwise rotation of the regression line computed as higher slope (Fig. 9B, blue line). Conversely, the overall effect of the NGF dual blockade is a clockwise rotation (lower slope) of the line (Fig. 9B, red line) to values similar to the axotomy state. However, in this case, the ratio of inhibition to excitation was smaller than what was expected from a simple clockwise rotation of the control group (compare the different angles between the solid and dashed red lines and the corresponding solid and dashed blue lines in the lower left quadrant of Fig. 9B). Thus, motoneurons with less activity, i.e., the group treated with NGF⫹K252a⫹REX, also showed proportionally less inhibition. Discussion Our results show that NGF is a trophic factor for adult abducens motoneurons that restores afferent innervation and discharge characteristics after axotomy. NGF treatment resulted in the prevention of the lesion-induced stripping of synaptic boutons and reestablished the excitability and neuronal sensitivities to eye movements by acting through both p75 NTR and TrkA receptors. This is the first in vivo evidence that complementary retrograde signaling through the two different NGF receptors results in the regulation of firing patterns through changes in the excitability and composition of input to motoneurons. Effects of NGF on synaptic coverage Treatment with NGF did not induce synaptic stripping of motoneuron membranes caused by axotomy on motoneurons. This effect was mainly due to the action of NGF via TrkA, since blocking p75 NTR did not induce synaptic stripping in axotomized animals treated with NGF, and only when both NGF receptors were blocked did synaptic loss occur. It is well known that axotomy induces considerable synaptic stripping, and that the degree of deafferentation depends on manipulations of target reinnervation (Sumner, 1975). Thus, if neurons reinnervate any target, the effects of axotomy wane (Delgado-García et al., 1988; Bra¨nnstro¨m and Kellerth, 1999), but neurons remain partly deafferented when target reinnervation is prevented (Pinter and Vanden Noven, 1989). Our NGF treatment results in a high responsiveness of axotomized motoneurons to this neurotrophin. Both the phasic and tonic firing increased, and only after the complete blockade of NGF receptors did the sensitivities return to axotomized levels. Phasic firing during saccades is provided by excitatory and inhibitory reticular neurons located in the paramedian pontomedullary reticular formation (Bu¨ttner and Bu¨ttner-Ennever, 2006). The tonic component arises from the output of a velocity-to-position integrator connected to motoneurons (Aksay et al., 2001). It is likely that NGF exerts a trophic effect through synaptic contacts by these inputs to the abducens motoneurons. In fact, NGF is known to have a robust synaptotrophic role, regulating aspects of synaptic transmission such as synaptophysin expression, neurotransmitter release, expression of synaptic receptors, and changes in synaptic efficacy (Mendell and Arvanian, 2002; Cotrufo et al., 2003; Hazary et al., 2007; Huh et al., 2008; Tarsa and Balkowiec, 2009). Expression of p75 NTR and TrkA receptors Unlike other cranial and spinal motoneurons (Lindsay, 1994; Ferri et al., 2002), extraocular motoneurons are peculiar in that they express TrkA receptors (Benítez-Temin˜o et al., 2004). Therefore, the modulation and expression of both p75 NTR and TrkA receptors in the presence or absence of NGF was also distinct. There is variability in the occurrence of the Trk receptors, such that cranial and spinal motoneurons do contain Trk recep- Figure 6. Blockade of NGF receptors restores firing variability of axotomized motoneurons. A, Plot of the coefficient of variation (CV) versus the mean firing frequency of 2097 stationary periods obtained from 70 control motoneurons. Note that higher CV accompanies lower firing frequencies.B,Firingvariability(CV)ofallexperimentalgroupsshowedanexponentialdecayas the discharge rate increased (color codes as shown in C). Fitted lines are double exponentials with R 2 ⬎0.95 in all cases ( p⬍0.001). Values are represented as mean ⫾SEM of CV in percentage of different firing frequencies spaced every 5 Hz. All comparisons between groups were significantly different (ANOVA test, Tukey test for post hoc comparisons, p⬍0.001). C,ComparisonofCVintheinterval30–80sp/s.NotethatNGFtreatmentdramaticallyincreased the firing variability, whereas blockade of TrkA receptors partially decreased variability, and blockadeofbothp75 NTR andTrkAreceptorsrestoredfiringvariabilitytoaxotomylevels(ANOVA test,Holm–Sidakmethod, p⬍0.05). Symbols indicate significant differences between control (*; Con.), axotomy (●; Ax.), and NGF treatment (‡) experimental groups. #, NGF⫹K252a treatment. Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing J. Neurosci., June 16, 2010 •30(24):8308–8319 • 8315 tors, but only TrkB and TrkC are present in adults (Merlio et al., 1992; Henderson et al., 1993; Koliatsos et al., 1993; Piehl et al., 1994). During development, spinal motoneurons contain TrkA receptors, but their expression is downregulated afterward (Ernfors et al., 1989). Motoneurons affected by ALS shift their expression of Trk receptors from TrkB and TrkC to TrkA (Nishio et al., 1998). In the adult, the expression of TrkA receptors is reinduced by axotomy in retinal ganglion cells (Cui et al., 2002) and in motoneurons (Omura et al., 2005), and lost again after the reinnervation of the target muscle (Koliatsos et al., 1993). NGF has also been shown to affect axotomy responses via the p75 NTR (Ferri et al., 2002), which, as in the present data, increases after lesion (Johnson et al., 1999; Yuan et al., 2006). The presence of NGF signaling in abducens neurons might confer greater resistance to the effects of lesions and neurodegenerative diseases (Benítez-Temin˜o et al., 2004). Complementary effects via p75 NTR and TrkA receptors It is noteworthy that motoneuronal sensitivities after NGF treatment become higher compared with control. NGF (Hazari et al., 2007), like BDNF and NT-3, also enhances synaptic transmission (Munson et al., 1997; Sherwood and Lo, 1999; Arvanian et al., 2003, Davis-Lo´pez de Carrizosa et al., 2009). Our data indicates that the maintenance of synaptic coverage (i.e., the lack of synaptic stripping) was dependent exclusively on TrkA receptors. NGF increases in nicotinic transmission that are dependent on TrkA receptor activation have also been demonstrated in parasympathetic motoneurons (Hazari et al., 2007). Moreover, NGF acting through TrkA receptors has been shown to enhance the firing activity over control values in other central cholinergic neurons (Wu and Yeh, 2005). However, the neuronal sensitivities were partly dependent on both NGF receptors, because the blockade of TrkA receptors decreased sensitivities from the high values of the NGF treatment and the dual receptor blockade produced a further decay to axotomy levels. NGF treatment led to an increase in the ratio of inhibition to excitation, even if TrkA receptor was blocked. These cells showed ample inhibition, as demonstrated by higher recruitment thresholds and longer firing pauses during fixation. In parallel with these response changes, the ratio of VGAT to synaptophysin increased with respect to the dual blockade of NGF receptors, although it did not differ from control ratios. NGF-treated cells also demonstrated more excitation, as shown by larger kvalues. These larger synaptic drives might be made possible by greater synaptic efficacy, since there was no vacant synaptic space (Huh et al., 2008). Changes in the amount of inhibition appeared to compensate for changes in the excitatory drive in a homeostatic Figure 7. NGF prevents synaptic stripping of motoneurons. A–D, Confocal images of motoneurons in control, axotomy, and NGF-treated groups illustrating the innervation (green; A,C)orthe inhibitory boutons labeled with antibody against VGAT (green; B,D), or the ChAT-labeled motoneuronal cell bodies (red; A–D). All sections were obtained from animals 15 d after either axotomy or axotomy and treatment. Note the marked reduction of boutons after 15 d of axotomy. Scale bars: B,20 ␮ m; D,80 ␮ m. Figure 8. NGF promotes maintenance of afferent innervation on motoneurons. The linear density of synaptophysin-IR (SYN) and VGAT-IR (VGAT) boutons over the somatic perimeter of motoneurons changes after axotomy and the different treatments. Bars represent mean ⫾ SEM for 17–38 motoneuronal profiles in each group. Symbols indicate differences between control (*; Con.), axotomy (●; Ax.), and NGF (‡) groups. ANOVA, Holm–Sidak method for multiple pairwise comparisons ( p⬍0.05). 8316 •J. Neurosci., June 16, 2010 •30(24):8308–8319 Davis-Lo´pez de Carrizosa et al. •NGF Effects of Motoneuron Firing