|Home | About | Journals | Submit | Contact Us | Français|
Retinotectal map formation develops via topographically specific guidance and branching of retinal axons in their target area. This process is controlled, in part, by reverse signalling of ephrinAs expressed on retinal axons. As glycosylphosphatidylinositol-anchored molecules, ephrinAs require transmembrane co-receptors to exert this function, for which the two neurotrophin receptors, p75NTR and TrkB, were recently proposed.
We show here that the ligands for these receptors, the brain-derived neurotrophic factor precursor (proBDNF) and its processed form, BDNF, respectively, control the branching of retinal axons antagonistically, which they mediate by inducing the corresponding neurotrophin receptor-ephrinA complexes. Moreover, scavenging proneurotrophins, by adding antibodies specific for the pro-domain of proBNDF or a soluble extracellular domain of p75NTR, abolish repellent ephrinA reverse signalling in the stripe assay.
This indicates that retinal cells secrete proneurotrophins, inducing the ephrinA-p75NTR interaction and enabling repellent axon guidance. The antagonistic functions of proBDNF and BDNF raise the possibility that topographic branching is controlled by local control of processing of proneurotrophins.
The retino-tectal projection is a well suited model system to investigate the formation of topographic maps and the control of local axon branching. In this projection, retinal ganglion cell (RGC) axons grow into the tectum in a non-topographic manner and initially overshoot their future termination zones. Termination zones are formed through interstitial branching, with branching of axons from nasal retina in the caudal tectum and axons from temporal retina in rostral tectum. The map is refined by arborisations and pruning of overshoot axon segments. The final map is a product of both activity-independent and activity-dependent processes [1-3].
Some aspects of this mapping process are controlled by retinally expressed ephrinA molecules, with higher expression on nasal than on temporal retinal axons. This differential expression mediates a repulsion of nasal axons from parts of the target area expressing high(er) amounts of EphA molecules, that is, the anterior tectum . Recently, the neurotrophin receptors p75NTR and tropomyosin-related kinase (Trk)B were proposed as co-receptors for ephrinAs, which are glycosylphosphatidylinositol-anchored and therefore have no direct contact with the cytosol [5,6].
Ligands for these receptors are the brain-derived neurotrophic factor precursor (proBDNF) and its processed form, BDNF. proBDNF binds with high affinity to p75NTR, while BDNF binds with high affinity to TrkB [7-9]. Both the pro-form and the processed form are secreted from neurons, and their processing is controlled on various levels [10-17]. This control of processing is crucial, as the activation of either p75NTR or TrkB leads often to opposing biological effects ; for example, activation of TrkB results in cell survival, while activation of p75NTR leads to cell death . Similarly, in synapse function, TrkB and p75NTR are involved antagonistically in long-term plasticity versus long-term depression [20,21].
This study investigates whether the interaction of ephrinAs with either TrkB or p75NTR also results in antagonistic effects on axon guidance and branching of retinal axons . We have approached this by using various in vitro assays. Our findings on the antagonistic functions of proBDNF versus BDNF on axon guidance and branching fit well to data showing that a (conditional) inactivation of p75NTR results in a disturbance of the retinocollicular map with shifted and ectopic termination zones and an increase in non-topographic branching anterior to the termination zones .
We have recently shown that the interaction between ephrinAs and TrkB is promoted by the ligand BDNF . EphrinAs interact also with p75NTR in vivo, though a ligand-dependency was not investigated here . We have here addressed this question by co-transfection of ephrinA5 and p75NTR cDNAs into Chinese hamster ovary (CHO) cells and co-immunoprecipitations before and after activation of p75NTR (Figure (Figure1).1). Our results show that in the absence of p75NTR activation, only little p75NTR co-immunoprecipitates with ephrinA5, while after its activation the amount of p75NTR was substantially increased (Figure (Figure1).1). Data from four independent experiments showed an increase in co-immunoprecipitated p75NTR. A comparable induction of the p75NTR-ephrinA5 complex was observed after application of proBDNF or the precursor of nerve growth factor (proNGF; Additional files 1 and 2). Thus, both the ephrinA-p75NTR and the ephrinA-TrkB interaction are promoted by their respective ligands.
In stripe assay experiments using a substrate of alternating lanes of EphA7-Fc and Fc protein overlaid with laminin/merosin, nasal retinal axons are repelled from growing on EphA7-Fc lanes, leading to a striped outgrowth . This matches, in principle, their targeting behaviour in vivo (Additional file 3), where axons with high expression of ephrinAs avoid regions in the tectum with high EphA concentrations. In further support of the idea of repellent ephrinA reverse signalling involving p75NTR, we show now that RNA interference (RNAi)-mediated knockdown of p75NTR abolishes the striped outgrowth of retinal axons (Figure (Figure2)2) .
Interestingly, these experiments (as well as comparable experiments with retinae from wild-type and p75NTR knockout mice ) were performed in the absence of added pro-neurotrophins, which is surprising in view of our data showing a ligand-induced/promoted interaction of ephrinA5 with p75NTR (see above; Figure Figure11).
We hypothesized that proneurotrophins might be secreted from retinal axons and, in an autocrine-paracrine loop, induce the formation of the ephrinA5-p75NTR receptor complex. It had been shown already that chicken RGCs express RNA coding for proBDNF/BDNF starting at embryonic day (E)6  and that BDNF protein is present in the RGC layer at E8.5 . However, the expression of proBDNF was not investigated. Here we have stained explant cultures from chick E6 retinae with a well characterised antibody raised against the pro-domain of proBDNF  to show the presence of this protein in axons and growth cones of RGCs (Figure (Figure33).
To test functionally the secretion of proBDNF from retinal cells (inducing an ephrinA-p75NTR complex), we added the antibody against the pro-domain of proBDNF (see above and ) to the medium of the stripe assay experiments as a scavenger to neutralise any secreted proBDNF. We found that the presence of this antibody led to the loss of striped outgrowth, indicating abolishment of repulsion of retinal axons from EphA7-Fc lanes (Figure (Figure4A;4A; Additional file 4). Control antibodies had no effect (Figure (Figure4A;4A; Additional file 4). In a second approach, we used a soluble extracellular domain of p75NTR to scavenge any p75NTR ligands contained in the growth medium . Again we observed a reduction in the level of repellent axon guidance (Figure 4B-G). These two independent sets of data suggest that striped outgrowth driven by ephrinA reverse signalling is dependent on local secretion of proneurotrophins, which induces an ephrinA-p75NTR complex competent to 'read' repellent guidance signals. This complex is likely to contain also the sortilin receptor necessary for proBDNF-mediated p75NTR functions . Similarly, p75NTR must be bound to its neurotrophin ligands to participate in Sema 3- and ephrinB2-mediated growth cone collapse of sympathetic neurons .
The ligand inducibility of this repulsion is of particular interest as it appears at first glance surprising that RGC axons can invade the tectum from the high end of the repellent EphA gradient (Additional file 3). In light of our findings, retinal axons conceivably become sensitive to the repellent gradient only after their ingrowth into the tectum, as possibly only then proBDNF is secreted from retinal axons leading to an induction of the ephrinA-p75NTR complex. This switch from storage to secretion might be developmentally regulated and/or linked to the fact that proBDNF secretion is regulated by neural activity [13,15]. Thus, a change, for example, in the pattern of spontaneous neural activity might lead to proBDNF secretion only after RGC axons have invaded the tectum, that is, after they have surmounted the high end of the repellent EphA gradient.
We have shown recently that a BDNF-promoted interaction between ephrinAs and TrkB results in an increase in retinal axon branching and that increasing the level of ephrinAs on retinal axons further increases the level of branching .
We have investigated now the role of the precursor form, proBDNF, on axon branching (Figure (Figure5).5). We performed these experiments in the presence of 5 ng/ml BDNF to provide a substantial level of axon branching. Co-application of 5 ng/ml proBDNF then resulted in a down-regulation of axon branching, in that it effectively neutralised the branch-promoting effect of BDNF. Proneurotrophins alone had no effect on branching. This downregulation was mediated by p75NTR as it was no longer observed after knock-down of this receptor (Figure (Figure5C).5C). A similar reduction in branching was observed when using a different proneurotrophin, proNGF, which does not interact with TrkB, neither in its pro- nor its processed form (Figure (Figure5A).5A). However, while mature NGF has no effect on RGC branching in vivo , we can not exclude that proNGF might be involved in suppressing RGC axon branching.
Thus, while BDNF application (via TrkB) promotes retinal branching, proBDNF application (via p75NTR) suppresses it. The map disturbance seen in (conditional) p75NTR knockout mice with increased axonal branching anterior to the termination zone  is in good agreement with the data shown here, offering a molecular explanation for the observed p75NTR phenotypes. Our data suggest that the local ratio between proBDNF and BDNF (and with that a differential activation of TrkB versus p75NTR) in the tectum/superior colliculus contributes to topographically specific branching. Thus, during map formation, proBDNF might be processed to BDNF only in those areas of the tectum where branching is topographically appropriate [21,29], leading to increased branching. Secretion and processing of proBNDF/BDNF from tectal cells might be involved here too.
Our data on the antagonism of BDNF and proBDNF in retinotectal (nasal) axon guidance strengthens the general model of neurotrophin receptor function that proposes that antagonistic functions of Trks and p75NTR control numerous biological processes [9,18]. Our data agree well, for example, with studies in the peripheral nervous system analysing the pruning of sympathetic axons after their projection to eye muscles. Here, correctly targeted ('winning') branches are maintained because of prevalent Trk signalling, while p75NTR activation in 'losing' axons causes axonal degeneration by suppressing TrkA-mediated signalling [30,31].
EphA7-Fc was from R&D Systems, Fc control from Calbiochem, BDNF from Promega, wild-type proBDNF and proNGF from Alomone Lab (Jerusalem, Israel). PanTrk203 antibody was a gift from Joan Comella, Lleida (Spain), and the antibody specific for the pro-domain of proBDNF (mAb287 ) was from Genecopoeia, Inc. (Rockville, MD, USA) Anti-FLAG and NGF was from Sigma and αHA from Abcam (Sigma: St. Louis, MO, USA; Abcam UK: Cambridge, U.K.). αHA and αFLAG agarose beads for immunoprecipitation of HA-tagged constructs were obtained from Sigma. The source of all other reagents is described in Marler et al. .
Outgrowth/branching assays and stripe assays were performed as described in Marler et al. . RGCs for both the stripe assay and the axon branching assay were grown in Neurobasal media with 2% B27 supplement, 1 mM glutamine, 5 μM forskolin and 1% penicillin/streptomycin; thus, in synthetic media containing, per se, no neurotrophins. For the axon branching assay the media were supplemented with proneurotrophins or neurotrophins as described in the respective figure legends. RGCs were identified by staining for markers expressed on retinal axons, such as TrkB, Brn3A and Thy-I. Routinely, those cells with the longest axons were positive for these markers. Scoring was performed after 3 days for the branching assay using proneurotrophin/neurotrophin treatment (Figure (Figure5).5). For the standard stripe assay, E6 retinae were used and scoring was performed after 2 days in culture. Here the cells were plated on a substrate of alternating lanes of EphA7-Fc and Fc, or Fc and Fc as a control. Protocols are described in Marler et al.  and Rashid et al. . To investigate axon guidance in the stripe assay, concentrations of 30 μg/ml each were used. Axons were scored as '0' if the majority of axons showed no preference for either stripe. They were scored as '1' if they showed some avoidance of the EphA7-Fc* stripe (or Fc*), and as '2' for a strong preference for growing on Fc stripes, that is, almost completely or fully avoided the EphA7-Fc* (or Fc*) stripes. The evaluation of all branching and stripe assays was done blind to the constructs transfected and ligands added. Stripe assay data shown are the result of three independently performed experiments.
For the proBDNF stripe assay experiments, single cells from E6 chick retina were electroporated with an eGFP expression plasmid and plated on alternating lanes of EphA7-Fc and Fc, or Fc and Fc. Experiments were performed in the presence of a proBDNF antibody (1:200)  or a control antibody (mouse monoclonal antibody for placental alkaline phosphatase; 1:200). The quantification of axon growth preferences shows an abolishment of repellent guidance in the presence of the proBDNF antibody. The number of axons analysed for each experimental condition was > 40.
For the axonal branching assays single cells were prepared from the nasal third of E8 chick retinae and electroporated with an eGFP expression construct to facilitate later analysis of neurite outgrowth. Cultures were plated on laminin (10 μg/ml) and merosin (2 μg/ml) coated dishes and cultured for 4 days.
Co-immunoprecipitation experiments were performed as described in Marler et al. . In brief, after transfection, cells were lysed in 25 mM Tris, pH 7.4, 150 mM NaCl, 1% (v/v) Triton X100, 1 mM EDTA pH 8.0, 10% (v/v) glycerol, 1 mM phenylmethyl-sulfonyl fluoride (PMSF), 1 mM sodium fluoride (NaF), 1 mM sodium vanadate, 1 mM sodium pyrophosphate, 1 mM glycerol pyrophosphate and Roche Complete Protease Inhibitor. (Roche UK: Hertfordshire, U.K.). After passing through a 26G needle, cell lysates were cleared by centrifugation for 20 minutes at 13,000 g. Approximately 1 mg of total protein from each sample was incubated with 20 μl of respective affinity gel overnight at 4°C on an orbital shaker. Immunocomplexes were washed four times with lysis buffer, suspended in 20 μl of Laemmli buffer, boiled and loaded onto a 10% SDS-polyacrylamide gel, and transferred onto nitrocellulose. Blots were immunodetected with suitable antibodies and developed using Western Lightning' Chemoluminescence Reagent from Perkin Elmer according to the manufacturer's instructions and it was exposed to an X-ray film as long as required (Perkin Elmer: Waltham, MA, USA).
RNAi knockdown experiments were performed as described in Marler et al. . Target sequences for p75NTR RNAi(2) were ACCTCCGATGCTGAGTGCAGA and for p75NTR RNAi(3) were CTACGGCTACTTCCAGGATG.
The extracellular domain of p75NTR was obtained by PCR using the forward primer CC TGC CTG GAC AGT GTG, and reverse primer TCA GGT GCC ACG GCT CAC, covering the amino acids PCLDS.....VSRGT, using the chick p75NTR cDNA as a template. The resultant PCR fragment was cloned into a CMV promoter-containing expression vector with 5' sequences encoding a signal peptide and a triplicate FLAG tag. Clones were sequence-verified. For stripe assay experiments involving p75NTR, sol, FLAG, two 6 cm2 plates with 5 × 105 CHO cells each were transfected per construct. Two days later the media were harvested and 500 μl of transfected or mock transfected media was added to each of the stripe assay cover slips to give a 1:1 ratio. A further 1 ml of the media was reserved and analysed for the presence of p75NTR, sol, FLAG.
BDNF: brain derived neurotrophic factor; CHO: Chinese hamster ovary; E: embryonic day; eGFP: enhanced green fluorescent protein; NGF: nerve growth factor; NTR: neurotrophin receptor; RGC: retinal ganglion cells; RNAi: RNA interference; Trk: tropomyosin-related kinase.
The authors declare that they have no competing interests.
KM planned and performed experiments, to which SP, EB and CW contributed. CW characterised the p75 RNAi vectors. UD planned the experiments, supervised the project and wrote the paper.
Supplemental Figure Figure1.1. The ephrinA5-p75NTR interaction is promoted in a proneurotrophin-dependent manner. (A) CHO cells were transfected with plasmids encoding ephrinA5FLAG and TrkAHA or p75NTR;HA. One day later cells were lysed and subjected to immunoprecipitation using a αFLAG antibody and analysed in western blots as indicated. At low expression levels, ephrinA5 co-immunoprecipitates with TrkA but not p75NTR. IB, immunoblot; IP, immunoprecipitation. (B) CHO cells were transfected with p75NTR, FLAG and ephrinA5HA. A day later cells were serum starved and treated for 30 minutes with pro/neurotrophins as indicated. Subsequently, cells were lysed and immunoprecipitated using a αHA antibody. Western blots showed that co-immunoprecipitation of p75NTR with ephrinA5 is increased in the presence of ligand. Quantification is shown in Additional file 2.
Supplemental Figure Figure2.2. Quantification of ligand-induced co-immunoprecipitation of p75NTR with ephrinA5HA. Quantification of co-immunoprecipitation experiments as exemplified in Figure Figure11 and Additional file 1. For experimental details see legends for Figure Figure11 and Additional file 1. Concentrations used for the co-immunoprecipitations are given. The numbers of independently performed experiments were: for NGF, n = 4; for proNGF n = 4; and for proBDNF n = 3. For quantification, the intensity of bands corresponding to immunoprecipitated p75NTR was normalised using the intensity of bands corresponding to ephrinA5HA. Then ratios were determined between values obtained for presence versus absence of ligand. In the absence of ligand (control) the value is 1. The standard error of the mean is shown.
Supplemental Figure Figure3.3. The projection patterns of RGC axons from the retina to the tectum, the differential expression patterns of EphAs and ephrinAs in retina and tectum, as well as the uniform expression the neurotrophin receptors TrkB and p75NTR and their ligands in the retina.
Supplemental Figure Figure4.4. Abolishment of striped outgrowth of RGC axons on an EphA7-Fc/Fc matrix in the presence of a proBDNF antibody. (A) In the presence of a control antibody, a RGC axon (green) avoids a lane containing EphA7-Fc (in red). (B) In the presence of the proBDNF antibody, a RGC axon freely crosses EphA7-Fc (red) and Fc (unlabelled) lanes. Details of the experimental conditions are described in Figure Figure4A.4A. Scale bar = 25 μm.
We would like to thank Helen Smith and Milan Makwana for help with experiments, and Ivo Lieberam and Phillip Gordon-Weeks for valuable comments on the manuscript. This work was supported by grants from the Wellcome Trust and the BBSRC.