Non-canonical Ret signaling augments p75-mediated cell death in developing sympathetic neurons

Apoptosis is a fundamental developmental process during organogenesis. In the nervous system developmental cell death, also known as programmed cell death (PCD), is an evolutionarily conserved process that allows an organism to match the size of the neuronal population with the size of its target tissue. In the peripheral nervous system, there is a widespread overproduction of neurons, with most populations producing twice the number of neurons that are present in adulthood (Oppenheim, 1991). Neurons project to their targets and compete for a limited supply of neurotrophic factors. Neurons that make appropriate or sufficiently extensive connections receive an adequate amount of target-derived neurotrophic factors and survive, whereas those that do not are eliminated through apoptotic signaling cascades (Levi-Montalcini, 1987; Oppenheim, 1991). Importantly, the mechanisms underlying PCD can be reactivated during nervous system injuries and neurodegenerative diseases (Ibáñez and Simi, 2012), underscoring the importance of understanding of these molecular mechanisms in detail.

PCD in the nervous system is perhaps best understood in sympathetic neurons of the superior cervical ganglion (SCG). Perinatally, these neurons are wholly dependent on target-derived NGF for their survival (Levi-Montalcini, 1987; Smeyne et al., 1994). NGF is the founding member of the neurotrophin family, also consisting of brain-derived neurotrophic factor (BDNF), neurotrophin (NT)–3, and NT-4 (Chao, 2003). NGF exerts its pro-survival functions through the receptor tyrosine kinase TrkA, which is ubiquitously expressed in sympathetic neurons. TrkB and TrkC, the cognate receptors for BDNF/NT-4 and NT-3, respectively, are not expressed in the SCG and, as such, these neurotrophins are dispensable for the survival of developing sympathetic neurons (Bamji et al., 1998).

In addition to the competition for survival factors, evidence also points to the presence of active pro-apoptotic signaling mechanisms through various death receptors within the TNF superfamily, including the p75 neurotrophin receptor and TNFR1 (Bamji et al., 1998; Barker et al., 2001). p75 is a promiscuous receptor that regulates several cellular functions through its interactions with other coreceptors. p75 can bind to all four neurotrophins (Gentry et al., 2004), and acts collaboratively with sortilin as the high-affinity receptor for the proneurotrophins (Nykjaer et al., 2004). In the SCG, p75 has been reported to have both pro-survival and pro-apoptotic functions (Gentry et al., 2004; Kraemer et al., 2014). p75 inhibits ligand-induced TrkA ubiquitination and subsequent internalization and degradation, thereby potentiating NGF-TrkA signaling (Makkerh et al., 2005). However, in the absence of NGF, or the presence of BDNF or proBDNF, p75 activation triggers apoptosis (Bamji et al., 1998; Lee et al., 2001; Kenchappa et al., 2010). Consistent with these studies, in p75^−/− mice, the number of sympathetic neurons is greatly increased, and the rate of apoptosis after NGF deprivation is strongly diminished (Bamji et al., 1998; Deppmann et al., 2008). Furthermore, coincident knockout of p75 in TrkA^−/− sympathetic neurons largely rescues neurons from apoptosis, consistent with a role for p75 in apoptosis after NGF withdrawal (Majdan et al., 2001). These and other studies have led to the proposal that there is competition between neurons during PCD. “Winning” neurons—those that receive adequate amounts of target-derived NGF, and are themselves protected from cell death—up-regulate and release pro-apoptotic p75 ligands such as BDNF, which induce apoptosis in nearby unprotected “losing” neurons (Deppmann et al., 2008). Although it remains unclear to what extent NGF withdrawal, pro-apoptotic competition, or a combination of both ultimately accounts for apoptosis mediated by p75 in the sympathetic nervous system, it is clear that multiple stimuli can induce p75-mediated apoptosis.

An additional neurotrophic factor receptor, Ret, is expressed in the SCG during the period of PCD, but its role has not been examined. Ret is a receptor tyrosine kinase that is activated by a family of four growth factors known as the glial cell line–derived neurotrophic factor (GDNF) family ligands (GFLs), which includes GDNF, neurturin, artemin, and persephin. These ligands do not bind directly to Ret, and instead bind to one of four cognate glycosylphosphatidylinositol-anchored coreceptors known as the GDNF family receptor–αs (GFRαs; Airaksinen and Saarma, 2002). Once this GFL-GFRα complex forms, it then binds to Ret, allowing for its dimerization and activation. Ret has two C-terminal splice variants, Ret9 and Ret51, each with unique signaling capabilities and function (de Graaff et al., 2001). Ret signaling has been shown to be critical for survival in several neuronal populations including subpopulations of sensory neurons of the dorsal root ganglia, enteric neurons, and spinal γ-motor neurons (Enomoto et al., 2001; Airaksinen and Saarma, 2002; Luo et al., 2007, 2009). Moreover, Ret signaling is required for sympathetic chain ganglia migration, coalescence of the ganglia, and early axon pathfinding (Enomoto et al., 2001). These functions result in early and severe morphological deficits in the SCG before the period of PCD in Ret^−/− mice, limiting investigation of the function of Ret in this process.

In this study, we investigated the function of Ret in PCD. We demonstrate that Ret expression is restricted to a subpopulation of apoptotic neurons that are rapidly eliminated. Ret and p75 form a molecular complex induced by pro-apoptotic stimuli, and Ret is required for p75-mediated apoptosis induced by multiple stimuli in vitro. Importantly, p75 deletion specifically within Ret-expressing neurons, and Ret deletion specifically during PCD, result in a significant abrogation in PCD in vivo. These studies collectively revealed a surprising non-canonical function of Ret in augmenting apoptotic signaling through p75 during PCD in vivo.

Studies analyzing Ret expression in the developing sympathetic nervous system using either a Ret^GFP/+ reporter line or in situ hybridization demonstrated Ret expression to be nearly ubiquitous throughout the ganglion by embryonic day (E) 11.5–E12.5, but then to decline significantly by E14.5–E15.5, correlating with its role in sympathetic neuron migration and coalescence of the sympathetic chain ganglia. Curiously, Ret expression then reemerges by E16.5, corresponding to the onset of PCD (Nishino et al., 1999; Enomoto et al., 2001). While Ret was expressed in many neurons perinatally, very few neurons expressed the GFRα coreceptors, which is surprising given that GFL signaling through Ret requires GFRα receptors. To determine the functional significance of Ret expression during PCD, we performed a tamoxifen (TMX) pulse experiment using a Cre-inducible tomato reporter line (Rosa26^LSL-tdTomato) crossed to Ret-Cre/ER^T2 mice (Luo et al., 2009). This experimental strategy (Fig. 1 A) enabled us to permanently mark the population of neurons expressing Ret at E16.5, thereby allowing the determination of whether these cells are eliminated during PCD. Embryos were collected at E17.5, E19.5, and postnatal day (P) 3. Strikingly, there was a significant 76.4% reduction in the number of Ret⁺ neurons present at E19.5 compared with E17.5 (Fig. 1, B and C; quantifications in Fig 1 E; 952.4 ± 49.1 vs. 224.3 ± 20.0 neurons; P < 0.0001), with a further loss of neurons by P3 (Fig. 1 D; 13.5 ± 2.3 neurons). These data indicate that the Ret⁺ neurons present at E16.5 are rapidly eliminated, and the number of Ret⁺ neurons labeled is strikingly similar to the number of apoptotic SCG neurons reported in previous studies investigating PCD in the SCG (Aloyz et al., 1998; Majdan et al., 2001). Additionally, we observed many examples of Ret⁺ neurons appearing atrophic (arrows in Fig 1, B and C), a characteristic associated with PCD in sympathetic neurons (Deckwerth and Johnson, 1993).

Given the well-established role of p75 in PCD in the SCG, we sought to determine whether Ret expression coincides with p75 expression during PCD. To this end, SCGs from Ret-Cre/ER^T2; Rosa26^LSL-tdTomato mice were immunolabeled for p75 and red fluorescent protein (RFP; tomato). As a control for the specificity of p75 immunolabeling, SCGs were taken from p75^−/− mice at P0 and immunolabeled for p75, which demonstrated a complete lack of staining (Fig. S1). p75 and RFP coimmunolabeling revealed that p75 is widely expressed throughout the SCG, while Ret expression is restricted to a subpopulation of SCG neurons, all of which express p75 (Fig. S1, B and C). Collectively, these results indicate that Ret is expressed within a subpopulation of p75-expressing neurons, and that Ret⁺ neurons undergo apoptosis during the period of PCD.

To investigate whether there is a functional interaction between p75 and Ret, we used the NIH/3T3 cell line, which does not express Ret9 or Ret51 and expresses only low levels of p75 (Calco et al., 2014). As shown in Fig. 2 A, cotransfection of Ret9 or Ret51 with p75 leads to robust formation of Ret9-p75 and Ret51-p75 receptor complexes, regardless of whether Ret or p75 immunoprecipitation (IP) was used. To demonstrate that this interaction is both specific and relevant in neurons in vivo, Ret^fx/fx (Ret-WT) mice were mated with UBC-Cre/ER^T2 mice, a TMX-inducible Cre line driven by the ubiquitously expressed Ubiquitin C promotor (Ruzankina et al., 2007), allowing for temporally controlled deletion of Ret. Ret^fx/fx; UBC-Cre/ER^T2 (Ret-cKO) mice were administered TMX (0.25 mg/g body weight) from E14.5–E18.5 and euthanized at E19.5 to collect embryos. Spinal cords were then dissected and lysed, and the detergent was extracted, followed by IP of Ret. Ret deletion was highly efficient, and this led to a corresponding loss in p75 that coimmunoprecipitated with Ret antibodies (Fig. 2, B–D). These findings were further confirmed through proximity ligation assays conducted on NIH/3T3 cells transfected with p75 and Ret51, or p75 and Ret51-HA. As expected, we observed strong colocalization of p75 and HA-tagged Ret when staining using anti-p75 and anti-HA antibodies, but not with untagged Ret (Fig. 2 E). These results indicate that this Ret-p75 interaction is present under physiological conditions in vitro and in vivo.

To test the hypothesis that p75-Ret association may be enhanced by pro-apoptotic p75 stimuli including NGF withdrawal, BDNF, and proBDNF, primary sympathetic neurons were generated from E18.5 rat embryos and cultured in the presence of 50 ng/ml NGF. After 2 d, NGF was removed and neurons were rinsed with medium twice, followed by the addition of an anti-NGF (αNGF) blocking antibody for the indicated times. The neurons were then washed, the detergent was extracted, and Ret IPs were performed. Although a basal level of association typically existed, blocking NGF signaling led to a striking increase in the interaction between p75 and Ret, which was statistically significant at all time points analyzed following NGF deprivation (Fig. 3, A and B). BDNF treatment led to a significant increase in p75-Ret association by 12 h as compared with low NGF (P < 0.01), although the extent of this interaction was smaller compared with NGF deprivation (Fig. 3, C and D). Similar to BDNF, stimulation with an uncleavable form of proBDNF (10 ng/ml) led to a striking increase in p75-Ret association by 6 h, and this increase was sustained for at least 24 h (Fig. 3, E and F). Despite the robust induction of p75-Ret association following stimulation with proBDNF, we were unable to detect co-IP of the proneurotrophin coreceptor, Sortilin, with Ret, despite reliable detection of sortilin in the IP supernatants (Fig. 3 E). These data suggest a potential sortilin-independent proBDNF induction of p75-Ret complex formation, although it cannot be ruled out that proBDNF cleavage occurs in vitro, or that the Ret antibodies used for the IP preclude sortilin co-IP is some manner.

To determine whether pro-apoptotic conditions could induce the formation of this receptor complex in vivo, Ret-cKO mice were used to ensure specificity of IP of the Ret-p75 complex. P0 mice were given TMX from P0–P4, followed by injection of vehicle (as a negative control) or a NGF-blocking antibody, as we have done previously (Tsui-Pierchala et al., 2002). SCGs were then collected, lysed, and subjected to Ret IP and immunoblotting. Ret immunoblotting verified efficient knockdown following TMX administration. Interestingly, we observed that anti-NGF administration led to a significant increase in Ret compared with vehicle-treated mice, with a corresponding increase in co-IP p75 (Fig. 3, G–I). Collectively, these data suggest that pro-apoptotic stimuli increase p75-Ret association and also suggest that Ret is up-regulated following exposure to pro-apoptotic stimuli in vivo.

p75-mediated apoptosis requires the function of several downstream signaling effectors, including early binding of TRAF6 and NRIF to full-length p75 (p75-FL) shortly after activation, an early wave of JNK/c-Jun pathway activation, cleavage of the p75-ECD by the TACE complex, liberation of the p75-ICD by the γ-secretase complex, and a late wave of activation of the JNK/c-Jun pathway, ultimately leading to terminal activation of executioner caspases. These final events lead to morphological and nuclear changes such as DNA fragmentation, chromosome condensation, and nuclear blebbing, all characteristic signs of apoptosis (Kraemer et al., 2014).

To confirm previous studies indicating that p75 is involved in apoptosis in sympathetic neurons, we cultured E18.5–P0 SCG neurons from p75^−/− mice and p75^+/+ mice (Bogenmann et al., 2011). These neurons were deprived of NGF (or maintained in NGF, as a control) for 12, 24, or 48 h. Importantly, apoptosis-related nuclear morphological changes (pyknosis) do not begin to become evident until ∼20 h following NGF deprivation (Deckwerth and Johnson, 1993). 12 h following these treatments, neurons were fixed and stained for phospho-c-Jun (p-c-Jun), and the number of neurons with nuclear p-c-Jun accumulation was quantified to assess the extent initiation of apoptosis. Numerous studies investigating PCD in the SCG describe p-c-Jun as one of the earliest molecular events that trigger apoptosis, detectable before caspase activation and nuclear pyknosis (Deshmukh and Johnson, 1997; Bamji et al., 1998; Werth et al., 2000). As demonstrated in Fig. S2 A, there were fewer p75^−/− neurons displaying phosphorylated c-Jun compared with p75^+/− controls. As a later indicator of apoptosis, apoptotic neurons were quantified by counting pyknotic nuclei. Similar to the p-c-Jun data, p75^−/− neurons had reduced apoptosis at 24 h, but not 48 h, when compared with p75^+/+ neurons (Fig. S2 B). These data are consistent with previous reports indicating that redundant death receptor signaling mechanisms are present and can mediate apoptosis after extended periods of NGF withdrawal in p75-deleted neurons (Deppmann et al., 2008; Kraemer et al., 2014).

To test the hypothesis that Ret augments p75-mediated apoptosis, primary SCG neurons were produced and transfected with siRNA targeted against Ret, or a non-targeting scrambled siRNA as a control. In addition, for all experiments, a non-targeting siGLO siRNA was included to verify transfection efficiency, which demonstrated that >90% of neurons were transfected. 48 h following siRNA transfection, neurons were lysed followed by immunoblotting to determine the efficacy of siRNA-mediated knockdown of Ret. Transfection with Ret siRNA, but not scrambled siRNA, was effective in reducing Ret levels by >65% (P < 0.05; Fig. 4, A and B). To assess whether Ret is required for p75-mediated apoptosis, 48 h after transfection, scrambled and Ret-siRNA–transfected neurons were subjected to four conditions for 12 h: high NGF (100 ng/ml), low NGF (1 ng/ml), BDNF (200 ng/ml) in the presence of low NGF (1 ng/ml), or αNGF (as described in Fig. 3). Neurons were then fixed, and the number of neurons with nuclear p-c-Jun accumulation was assessed (Fig. 4, D–K) and quantified (Fig. 4 C). We observed very few examples of nuclear p-c-Jun accumulation in high NGF– (Fig. 4, D and E) or low NGF–treated conditions (Fig. 4, F and G), regardless of siRNA or Ret-siRNA transfection. As expected, BDNF and αNGF treatment led to increased nuclear accumulation of p-c-Jun in scrambled siRNA-treated neurons, with NGF deprivation producing more robust effects (Fig. 4, H and J). In marked contrast, Ret siRNA-treated neurons had significantly reduced p-c-Jun⁺ nuclei for both treatment groups (Fig. 4, I and K). Quantifications indicated that statistically significant reductions were observed between scrambled siRNA and Ret siRNA neurons treated with BDNF (Fig. 4 C; 46.42% reduction; 31.76 ±1.877 vs. 17.02 ± 1.84%; P < 0.05) and αNGF (38.24% reduction; 67.95 ± 4.75 vs. 41.97 ± 6.11%; P < 0.0001). These results demonstrate that Ret is required for p75-mediated apoptotic signaling mediated by both ligand stimulation (BDNF) and trophic factor deprivation (αNGF).

While these results indicate that Ret augments early p75-mediated activation of the pro-apoptotic signaling cascade, we sought to determine whether there is a specific threshold of NGF deprivation required to trigger apoptotic p75-Ret signaling by performing a NGF dose response curve. Due to the limited window of transfection efficacy via the siRNA knockdown approach, we applied a permanent means of deleting Ret by using P0 Ret-WT and Ret-cKO SCG neurons maintained in the presence of NGF and 4-OH-TMX. Following Cre-mediated deletion of Ret, Ret-WT and Ret-cKO neurons were treated for 24 h with concentrations of NGF ranging from 0 to100 ng/ml. Neuronal apoptosis was then assessed by the presence of pyknotic nuclei. As expected, and consistent with previous 24-h death assays investigating NGF-mediated survival under similar culture conditions (Putcha et al., 2001), we observed fewer than 50% of Ret-WT neurons were able to survive 24 h following complete deprivation of NGF (Fig. 5 A), with increasing concentrations of NGF improving survival of these neurons in a dose-dependent manner (Figs. 5 C, E, G, and I). In contrast, Ret-cKO neurons had fewer apoptotic profiles with complete NGF deprivation (Fig. 5 B; P < 0.01) and 0.1 ng/ml NGF (Fig. 5 D; P < 0.05), but had statistically similar numbers of pyknotic nuclei at higher doses (1, 10, 100 ng/ml NGF; Figs. 5 F, H, J, and K). Immunoblotting confirmed effective deletion of Ret in Ret-cKO compared with Ret-WT mice (>95%; Fig. 5, L and M). These data suggest that Ret antagonizes NGF signaling through TrkA, and that Ret is required for p75-mediated apoptosis induced by NGF withdrawal in a dose-dependent manner.

To further confirm the role of Ret in p75-mediated apoptosis, BDNF and proBDNF death assays were conducted in several different culture models: (1) Ret-cKO neurons (Ret-WT as a control), wherein Ret is deleted as above; (2) p75^fx/fx; Ret-Cre/ER^T2 (p75-RC) neurons (neurons from littermate p75^fx/fx mice as a control), wherein p75 is deleted specifically within Ret⁺ neurons; and (3) p75^fx/fx; UBC-Cre/ER^T2 (p75-cKO) neurons (neurons from littermate p75^fx/fx mice as a control), wherein p75 is deleted in all neurons. Neurons were cultured from the above mice at E18–P0 and maintained for 5 d in the presence of NGF (50 ng/ml) and 4-OH-TMX (5 μg/ml), rinsed, and treated with low NGF (1 ng/ml), or low NGF in the presence of BDNF (200 ng/ml) or proBDNF (10 ng/ml). Neurons were fixed and analyzed for nuclear pyknosis as described above 48 h after stimulation. Interestingly, both BDNF and proBDNF stimulation led to a substantial increase in apoptosis in control neurons, and this effect was significantly reduced in all cKO models analyzed (P > 0.0001; Fig. 5 N).

To determine the extent to which p75 is required for apoptotic signaling initiated by NGF deprivation, P0 p75^fx/fx (p75-WT) and p75-RC neurons were maintained for 5 d in the presence of NGF (50 ng/ml) and 4-OH-TMX (5 μg/ml). Neurons were then rinsed and treated with NGF or deprived of NGF for 12 h to assess p75-mediated p-c-Jun activation. Neurons were then fixed and stained, and the number of neurons with nuclear p-c-Jun accumulation was quantified (Fig. 6, A and B). While nuclear p-c-Jun was only rarely present in NGF-treated neurons from either p75-WT or p75-RC mice, NGF deprivation led to a substantial increase in the number of p-c-Jun⁺ nuclei in p75-WT neurons. αNGF-treated p75-RC neurons, in contrast, had significantly fewer p-c-Jun⁺ nuclei (68.6% reduction; 59.45 ± 8.73 vs. 18.70 ± 2.05%; P < 0.0001).

Given that only a subset of sympathetic neurons express Ret during PCD, and that these in vitro data indicate that p75-mediated apoptosis appears to be augmented by Ret, we hypothesized that deletion of p75 within Ret⁺ neurons would be sufficient to impair PCD. To test the requirement of p75 in Ret⁺ neurons during PCD in vivo, we used p75-RC mice (or p75^fx/fx mice as a control; Fig. 6 C). In brief, mice were given TMX (0.25 mg/g body weight) once per day for 4 d, beginning at E14.5, at which time Ret-dependent SCG migration and coalescence is largely complete (Enomoto et al., 2001). Immunoblotting of the spinal cords from E19.5 mice confirmed the successful deletion of p75 from Ret⁺ neurons using this TMX dosing regimen (Fig. 6, C–E). SCGs were then stained for cleaved caspase-3 (cc3), βIII-Tubulin (TuJ1), tyrosine hydroxylase (TH), and DAPI to quantify the number of apoptotic SCG neurons (Fig. 6, F and G). Because caspase-3 is an irreversible executioner caspase, cc3 staining is a highly sensitive marker of neurons undergoing apoptosis. Interestingly, we observed a statistically significant reduction in cc3⁺ neurons in p75-RC SCGs compared with the SCGs of p75-WT mice (Fig. 6 G; 50.1% reduction; 96.04 ± 12.84 vs. 48.18 ± 12.77 cc3⁺/mm³; P < 0.05).

To examine whether Ret is directly involved in p75-mediated PCD in vivo, we used Ret-cKO mice with the TMX dosing strategy described previously (Fig. 6 C) to avoid the deleterious effects of Ret deletion during SCG coalescence, while also avoiding perinatal lethality described by other studies involving perturbations of Ret signaling (Uesaka et al., 2007; Uesaka and Enomoto, 2010). SCGs were immunostained for cc3, TH, and DAPI, and the number of cc3⁺ cells was quantified to compare apoptosis in Ret-WT and Ret-cKO mice. Ret-cKO mice had significantly fewer cc3⁺ neurons compared with Ret-WT mice (Fig. 6, H and I; 34.2% reduction; 268.62 vs. 176.85 cc3⁺/mm³; P < 0.05). Interestingly, this reduction compares similarly to the reduction observed in p75-RC mice. Immunoblotting of spinal cord lysates from Ret-WT and Ret-cKO mice confirmed the efficacy of Ret deletion (Fig. 6 J). To further confirm these findings, total cell counts were performed on SCGs collected from E19.5 Ret-cKO and p75-RC mice administered TMX as described in Fig. 6 C. We observed a significant increase in total cell counts in Ret-cKO (compared with Ret-WT) and p75-RC (compared with p75-WT) SCGs regardless of whether counts were performed using TuJ1 (left side) or TH (right side) as neuronal markers (Fig. 6 K). The magnitude of the increase in neuron numbers was similar between Ret-cKO and p75-RC SCGs (14.51% increase in TuJ1⁺ counts in Ret-cKO mice compared with 15.55% in p75-RC SCGs; 20.01% increase in TH⁺ counts in Ret-cKO SCGs compared with 17.79% in p75-RC SCGs). To determine what proportion of p75-mediated apoptosis requires Ret, total cell counts were performed on SCGs collected from E19.5 p75^−/− (p75-KO) and p75^+/+ (p75-WT). As expected, we observed a highly significant increase in total cell counts in p75-KO ganglia compared with p75-WTs, regardless of whether TuJ1 or TH was used (Fig. 6 K). Additionally, we found a small but significant increase in TuJ1⁺ neurons (13.42%) in p75-KO compared with p75-RC mice, with a trend (P = 0.0513) toward increased numbers of TH⁺ neurons (11.56%) in p75-KO compared with p75-RC mice, suggesting the existence of a Ret-independent p75-mediated apoptotic pathway. When taken together with the in vitro functional assays, these data provide compelling in vivo evidence suggesting that Ret augments p75-mediated apoptosis during PCD.

Having established that Ret augments p75-mediated apoptosis, we examined whether Ret deletion altered p75 enhancement of NGF/TrkA signaling, p75-mediated activation of apoptotic effectors, or both. To this end, primary neurons were cultured from P0 Ret-WT and Ret-cKO mice maintained in NGF and 4-OH-TMX. Neurons were then maintained in NGF (50 ng/ml), or deprived of NGF for 12 h, followed by detergent extraction and immunoblotting for Ret. This confirmed that the deletion efficacy using this strategy was greater than 90% for both treatment groups (Fig. 7, A and B). To explore the possibility that Ret deletion results in a reduction of total levels of p75, immunoblotting was performed for p75 (fp75-FL; Fig. 7, A and C). NGF-maintained Ret-WT neurons had a significant reduction in p75 following NGF deprivation (Fig. 7 C; P < 0.05), likely as a result of regulated intramembrane proteolysis (RIP) cleavage of p75. Interestingly, there was no significant difference in p75 levels in NGF vs. αNGF-treated Ret-cKO neurons. To determine directly whether Ret deletion impairs p75 cleavage, thereby impairing downstream apoptotic signaling, Ret-WT and Ret-cKO neurons were treated with NGF or were deprived of NGF for 12 h in the presence of the degradation pathway inhibitors MG-132 (5 µM) and epoxomicin (5 µM). These inhibitors are necessary to prevent the rapid degradation of the C-terminal fragment p75 (p75-CTF) generated by RIP cleavage (Kanning et al., 2003). Neurons were then lysed, and p75 cleavage was assessed using an antibody against the p75-CTF, which detects both uncleaved p75 (p75-FL) and the 28 kD p75-CTF. In this culture system, only the p75-CTF fragment was observed, and we were not able to detect the p75-ICD fragment. Interestingly, deletion of Ret led to a drastic reduction in the amount of p75 cleavage induced by NGF deprivation in Ret-cKO neurons compared with Ret-WT neurons (Fig. 7, D and E). Importantly, the antibody used to detect the p75 cleavage fragments (Promega) has epitopes directed at the CTF, and binding affinity of the antibody to the p75-FL and p-75-CTF is likely dissimilar. Thus, direct comparison of these p75 fragments is not possible. Finally, to determine whether this effect was secondary to a reduction in the amount of TACE or presenilin-1 (PSN-1), immunoblotting was performed, and their levels were quantified. As shown in Fig. S3 (A–C), no statistically significant differences were observed in TACE or PSN-1 in Ret-cKO neurons as compared with Ret-WT neurons.

To determine whether downstream apoptotic signaling was altered, we performed immunoblotting for p-c-Jun, activated JNK (pJNK), total JNK, and actin in neurons that were either maintained in NGF or deprived of NGF. As expected, NGF deprivation led to a significant activation of p-c-Jun in Ret-WT neurons compared with neurons maintained in NGF. However, in Ret-cKO neurons, this activation of p-c-Jun following NGF deprivation was impaired, and αNGF-treated Ret-cKO neurons had significantly lower levels of p-c-Jun compared with Ret-WT neurons (Fig. 7, F and G; P < 0.01). Correspondingly, a similar reduction in pJNK/JNK levels was observed as well (Fig. 7, F and H; P < 0.0001).

A previous study demonstrated that p75-mediated apoptosis requires interaction with TRAF6 (Kanning et al., 2003). To examine whether TRAF6/p75 association was altered in the absence of Ret, p75 was immunoprecipitated from Ret-WT and Ret-cKO neurons treated with NGF or αNGF, and TRAF6 immunoblotting was performed. No significant differences were observed in TRAF6 association with p75 in any of the conditions (Fig. S3, D and E). Collectively, these data suggest that Ret augments apoptotic signaling by facilitating RIP cleavage of p75.

Because Ret antagonizes NGF-mediated survival in a dose-dependent manner (Fig. 5), we sought to test the hypothesis that Ret antagonizes NGF/TrkA signaling. Ret-WT and Ret-cKO neurons were deprived of NGF overnight, followed by treatment with NGF at the concentrations indicated for 15 min. Neurons were then detergent-extracted, TrkA was immunoprecipitated, and immunoblotting was performed for phosphotyrosine, TrkA, phosphorylated AKT (pAKT), total AKT, and actin (as a loading control). When analyzing the effects of Ret deletion on TrkA signaling, we observed that total levels of TrkA (TrkA/actin) were not affected by either genotype or by the experimental treatment (Fig. 7, I [lower band] and J). In contrast to TrkA levels, we observed a significant increase in NGF-mediated activation of TrkA (phosphorylated TrkA [pTrkA]/TrkA/actin) at 0 ng/ml (likely due to residual TrkA activation) and 1 ng/ml, but not 50 ng/ml (Fig. 7, I and K), in Ret-cKO neurons compared with Ret-WT neurons. Additionally, we observed a significant increase in NGF/TrkA-mediated activation of pAKT in Ret-cKO neurons compared with Ret-WT neurons, with a non-significant trend toward residually increased pAKT in NGF-deprived neurons (Fig. 7, I and L).

Based on these findings, we hypothesized that removal of Ret potentiates NGF/TrkA signaling by altering activation-induced TrkA degradation. To test this hypothesis, Ret-WT and Ret-cKO neurons were deprived of NGF overnight, followed by treatment with NGF as indicated (Fig. 7, M and N). Lysosomal and proteasomal degradative pathways were inhibited by using concanamycin and epoxomycin, respectively, to inhibit the rapid loss of ubiquitinated receptors, explaining why TrkA levels did not change in these experiments. As expected, NGF treatment led to a significant increase in TrkA ubiquitination in Ret-WT neurons by 15 min, which was significantly impaired in Ret-cKO neurons (Fig. 7, M and N). To determine whether the loss of TrkA-mediated ubiquitination leads to functional increases in surface levels of TrkA following removal of Ret, cell surface biotinylation experiments were performed. Cell surface proteins were biotinylated using NHS-LC-Biotin (or not biotinylated, as a control) and precipitated using neutravidin agarose, followed by immunoblotting for TrkA, and Ret and transferrin as controls. Interestingly, we observed a substantial increase in surface levels of TrkA in Ret-cKO neurons compared with Ret-WT neurons (Fig. 7, O and P). Collectively, these data suggest that during PCD, Ret pushes SCG neurons toward apoptosis both by inhibiting pro-survival signaling through TrkA and simultaneously enhancing p75-mediated apoptosis.

In this study we identified Ret as a novel component of the cell death machinery in sympathetic neurons that acts in concert with p75 during PCD. Ret expression is restricted to neurons that are rapidly eliminated through apoptosis, and pro-apoptotic stimuli induce formation of a Ret-p75 complex. Ret potentiates p75-mediated activation of downstream signaling effectors in response to apoptotic cues, and acts to augment p75-mediated apoptosis in a dose-dependent manner following NGF withdrawal. The removal of p75 specifically within Ret⁺ neurons is sufficient to diminish PCD, and this is mirrored following deletion of Ret in vivo. Ret potentiates apoptosis through two unique mechanisms that are ultimately connected by p75. First, Ret inhibits TrkA activation by promoting receptor ubiquitination, thereby reducing survival signaling. Second, Ret directly enhances RIP cleavage of p75 in response to pro-apoptotic cues, thereby inducing the late activation of the JNK/c-jun pathway that is necessary for downstream activation of pro-apoptotic effectors. Given the functions of Ret in developing neurons to date have been exclusively trophic in nature, these finding raise several important future questions.

The in vitro and in vivo data presented here support the notion that Ret, acting with p75, augments PCD in the developing SCG. Several lines of evidence support this assertion. We demonstrate that knockdown of Ret in vitro reduces apoptosis in response to apoptotic cues, and does so in a dose-dependent manner in the case of NGF withdrawal (Figs. 4 and 5). Additionally, and most critically, deletion of p75 specifically within Ret⁺ neurons is sufficient to diminish apoptotic signaling (Fig. 5 N and Fig 6, A, B, F, G, and K), and deletion of Ret in vivo quantitatively duplicates this result (Fig. 6, H–K), suggesting that Ret mediates the majority of p75-dependent apoptosis in SCG neurons. Also lending support to this model, we show that Ret expression in the SCG is limited to a subpopulation of neurons that are rapidly eliminated during the window of PCD (Fig. 1, B–F). While this Ret⁺ subpopulation of neurons is relatively small when a single TMX pulse is performed (∼1,000 neurons), these numbers quantitatively compare with the proportion of SCG neurons undergoing apoptosis at any given time during PCD (Aloyz et al., 1998; Majdan et al., 2001). Additionally, our data suggest that in vivo, trophic factor withdrawal induces the up-regulation of Ret (Fig. 3, G–I). Lastly, it is important to consider that completion of PCD within the SCG occurs over a period of several days (Deppmann et al., 2008), and thus, the proportion of neurons undergoing apoptosis at any specific age is a snapshot of the overall PCD process.

It remains unclear precisely which pro-apoptotic stimulus chiefly drives PCD through p75, and to what extent p75 accounts for PCD in the SCG. In this study, we observed that deletion of p75 and Ret resulted in substantially impaired apoptosis, but deletion of neither receptor fully blocked apoptosis. These data are consistent with previous reports analyzing apoptosis in the SCG in p75^−/− mice, as well as the in vivo data presented here, with p75-RC and Ret-cKO mice having an incomplete loss of cc3⁺ neurons (Fig. 5, E–J). These data indicate that p75-mediated apoptosis only represents a portion of the PCD occurring in sympathetic neurons, suggesting that other death receptors, or other apoptotic pathways, must play a significant role. Thus, while p75 may represent a critical conduit for sympathetic neurons to undergo PCD, prolonged atrophic or pro-apoptotic conditions may bring forth additional death receptor mechanisms or other pro-apoptotic mediators. Furthermore, our data provide compelling evidence that Ret augments p75-mediated apoptosis. However, based on our observation that neither Ret-cKO nor p75-RC mice completely phenocopy p75 KO mice (Fig. 6 K), it is likely that both Ret-dependent and Ret-independent p75-mediated apoptotic mechanisms exist.

The finding that Ret augments p75-mediated apoptosis implies a highly unusual non-canonical function for Ret. Several neurotrophic factor receptors have been suggested to serve as dependence receptors, whereby in the absence of ligand a pro-apoptotic signal is generated by default. Using transfected cell lines in vitro, Bordeaux et al. (2000) reported that Ret expression causes apoptosis through a signaling pathway involving caspase cleavage of full-length Ret, and this effect could be ameliorated by the addition of GDNF. Importantly, we did not find evidence of Ret cleavage in response to pro-apoptotic stimuli, nor do we find evidence for a dependence-receptor mechanism of cell death via Ret. The pro-apoptotic function of Ret described here, supported by both in vitro and in vivo evidence in neurons, argues for a non-canonical enhancement of p75-mediated apoptosis by Ret reliant exclusively on the presence of apoptotic cues that trigger the activation of p75, and does not involve deprivation of GDNF or other GFLs. Regardless, the investigation of whether this receptor complex serves as a dependence receptor in the absence of GFLs in other systems represents an interesting future direction.

In sympathetic neurons, the functions of p75 can be divided into two main categories: the trophic function of p75 in potentiating NGF/TrkA signaling, and the degenerative functions of p75. Given these opposing functions, a key question that emerges is how p75 can promote both survival and apoptosis. As p75 and TrkA are both ubiquitously expressed throughout the developing SCG, how do individual neurons determine whether p75 will potentiate TrkA signaling or actively promote apoptosis? It is possible that Ret acts as one such determinant, whereby upon its expression, p75 is pushed toward a pro-apoptotic signaling role, while also acting to dampen NGF/TrkA pro-survival signaling. To this end, Deppmann et al. (2008) demonstrated that a series of feedback loops regulate PCD in sympathetic neurons: high levels of NGF/TrkA signaling in “winning neurons” reinforce TrkA expression while also inducing up-regulation of pro-apoptotic p75 ligands. These trophically supported neurons then release these factors to act on neighboring atrophic neurons, which have down-regulated TrkA, activating p75 death signaling. In this model, it is likely that Ret is positioned to support this feedback mechanism, whereby expression of Ret antagonizes TrkA activation, expediting its down-regulation, while also creating a highly active death receptor complex with p75, ultimately enhancing p75 cleavage and downstream activation by apoptotic effectors.

Based on the coexpression of p75 and Ret in many neuronal populations, we speculate that the pro-apoptotic p75-Ret receptor complex discovered here may be of physiological significance in other populations, and may have varied functions depending on the cell type. For example, p75 acts to enhance GFL-Ret signaling in subpopulations of dorsal root ganglia sensory neurons, leading to the emergence of non-peptidergic nociceptors (Chen et al., 2017). Additionally, while Ret has been assumed to function as a pro-survival, pro-growth receptor tyrosine kinase, this non-canonical function of Ret in augmenting p75-mediated cell death may be of importance in the pathophysiology of nervous system injuries and neurodegenerative diseases.

All experiments were performed in compliance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International and were approved by the Institutional Animal Care and Use Committee of the University of Michigan.

UBC-Cre/ER^T2 (Ruzankina et al., 2007), Ret^fx/fx (Luo et al., 2007), p75^fx/fx and p75^−/− (Bogenmann et al., 2011), Ret^−/− (Schuchardt et al., 1994), Ret-Cre/ER^T2 (Luo et al., 2009), and Rosa26^LSL-tdTomato mice (Madisen et al., 2010) have all been previously described, and all mice were maintained in mixed genetic backgrounds, except for Rosa26^LSL-tdTomato, which was maintained in a C57BL/6J background. For timed matings, noon of the day on which a vaginal plug was detected was considered as E0.5. For the experiments tracing Ret expression, Rosa26^LSL-tdTomato mice were crossed to Ret-Cre/ER^T2 mice and given one i.p. injection of TMX (0.25 mg/g body weight) at E16.5 and euthanized for analysis at E17.5, E19.5, or P3. Additionally, Ret-cKO or p75-RC were given i.p. injections of TMX (0.25 mg/g body weight) consecutively for 4 d beginning at E14.5 and euthanized at E19.5.

Ret-WT or Ret-cKO mice were directly administered TMX (0.40 mg/g body weight with maximum volume of 50 µl) via i.p. injection daily for 5 d beginning at P0. At P5, mice were delivered 50 µl of vehicle alone (PBS) or 50 µl of function blocking anti-NGF serum, as we have used previously (Tsui-Pierchala et al., 2002). Mice received four doses of PBS or αNGF (once every 12 h). At P7 the mice were euthanized, and the SCGs were collected and pooled from two mice of identical genotypes per condition. The SCGs were then lysed, detergent extracted, and subjected to Ret IP and immunoblotting.

SCG were fixed with 4% paraformaldehyde at 4°C for 2–3 h, washed in PBS three times for 10 min, and cryoprotected at 4°C in 1× PBS containing 30% sucrose overnight. Tissues were embedded in optimal cutting temperature compound (Tissue Tek), frozen, and stored at −80°C until use. SCGs were serially sectioned at 7 µM on a cryostat (CM1950; Leica Biosystems). Tissue sections were washed with PBS and blocked with 5% normal goat serum in PBS-T (0.1% Triton X-100 in 1× PBS) for 1 h, followed by incubation with primary antibody (diluted in blocking solution) in a humidified chamber overnight at 4°C. Sections were washed with PBS-T, and incubated with secondary antibody (1:500) for 2 h using donkey anti-goat, anti-mouse, anti-rabbit, or anti-sheep 488, 543, or 633 fluorochromes obtained from Biotium. Sections were washed again with PBS-T and mounted in fluoromount-G with DAPI (Southern Biotech). Images were taken at room temperature using an inverted fluorescence microscope (Axiovert 200M; Zeiss Microsystems) at a magnification of 20× with a digital zoom of 1.0–2.0. To image entire SCGs, tile scan was used with 15% overlap between adjacent images. Images were subsequently stitched together using the AxioVision software (Zeiss). Antibodies used include α-cc3 (1:300; Cell Signaling Technology), α-TH (1:1,000; Millipore), α-TuJ1 (1:200; Sigma-Aldrich), and α-p75 (NGFr; 1:200; Advanced Targeting Systems). For quantification of apoptotic cells, the number of cc3 or TuJ1-positive neurons, respectively, were counted on every third section by an observer naive to the genotypes of the mice. Area measurements of SCGs were performed in the AxioVision software (Zeiss) using the “Outline” function. For total cell counts, SCGs were serially sectioned at a thickness of 20 µM and stained for TuJ1, TH, and DAPI, and all neurons were quantified using either TuJ1 or TH as a surrogate marker as described in the figure legend. Images were exported as high-resolution tagged image files, and all figures were created using Adobe Creative Suite.

Spinal cords were harvested separately from P0 Ret-cKO mice, and then placed in a 2.0-ml tube with 250 µl IP buffer lacking NP-40, along with a steel grinding ball (5 mm; Qiagen). The spinal cords were then mechanically homogenized using the TissueLyzer II (Qiagen). The homogenates were mixed with 250 µl of 2% NP-40–containing IP buffer and incubated for 1 h at 4°C under gentle agitation. Homogenates were centrifuged for 10 min at 16,100g and subjected to an initial preclearing step with protein A and protein G alone at 4°C for 2 h under gentle agitation, followed by preclearing with protein A, protein G, and a species-matched non-specific control IgG for 2 h under gentle agitation. Following preclearing, IPs were performed as described in the "IPs, cell surface biotinylation, and quantitative immunoblotting" section.

NIH/3T3 cells were maintained in DMEM supplemented with 10% FBS, 2 mM glutamine, and 1% penicillin-streptomycin (Invitrogen). Cells were plated on 6-well tissue culture plates (Falcon) and allowed to proliferate until an approximate density of 50% confluence was obtained before transfection. Transfections were performed using Lipofectamine 2000 according to the manufacturer’s instructions (Invitrogen). A total of 5 µg plasmid DNA was added per well, using a plasmid encoding EGFP to keep the total amount of DNA constant between treatments. The plasmid encoding p75 was provided by Phil Barker (University of British Columbia-Okanagan, Kelowna, British Columbia, Canada). For proximity ligation assays, the DuoLink Proximity Ligation Assay (Sigma-Aldrich) was used according to the manufacturer’s instructions. NIH/3T3 cells were transfected with p75 and Ret51, or p75 and HA-tagged Ret51. Antibodies used for proximity ligation assays were p75 (1:500; Advanced Targeting Systems) and HA (1:500; Sigma-Aldrich), with DuoLink In Situ Orange Mouse/Rabbit kit used for secondary antibody and amplification. DAPI was used for nuclear localization.

SCGs were surgically dissected from E19 Sprague-Dawley rats (Charles River) or P0 Ret-WT or p75^fx/fx mice, and enzymatically dissociated via incubation in type I collagenase (Worthington) and a 1:1 ratio of HBSS:TrypLE (Invitrogen). Neurons were plated on gas-plasma–treated 35-mm² dishes (Harrick Plasma) coated with type I collagen (BD Biosciences). For all biochemical experiments using rat neurons, cells were plated as mass cultures at a density of three ganglia per plate, while experiments using mouse neurons were plated at a density of two ganglia (one animal) per plate. For death assays, neurons were plated as a droplet at a density of one ganglion per plate. Neurons were maintained in MEM containing 50 ng/ml NGF (Harlan), 10% FBS, anti-mitotic agents aphidicolin (3.3 µg/ml) and 5-fluoro-2-deoxyuridine (20 µM; Sigma-Aldrich), 2 mM glutamine, and 1% penicillin-streptomycin (Invitrogen). Neurons were maintained at 37°C with 8% CO₂ with medium changes every 3–4 d until experimental treatment. Treatments with growth factors and pharmacological agents were performed as described in the figure legends.

Neurons (2 DIV) were transfected with siRNA using a scrambled control (siGENOME Non-Targeting siRNA Pool #1; Dharmacon) or Ret (ON-TARGETplus SMARTpool; Dharmacon) siRNA at a concentration of 100 nM via i-Fect (Neuromics) according to the manufacturer’s instructions. Transfection efficiency was determined in all experiments by the cotransfection of a fluorescently labeled non-targeting control siRNA (siGLO RISC-free siRNA; Dharmacon). 48 h after transfection, at which time expression of siGLO was maximal, neurons were treated as described in the figure legends. For immunocytochemistry experiments, neurons were fixed with 4% paraformaldehyde for 5 min, washed, and stained with primary and secondary antibodies and imaged as described in the Fixation, sectioning, and immunostaining of SCG section, with the addition of α-p-c-Jun (9261; 1:500; Cell Signaling Technology). Samples were imaged using 40× magnification with a digital zoom of 1.0 using an Axiovert 200M microscope, with tile scans of 5 × 5 fields to randomly sample non-overlapping areas. AxioVision software was used to stitch images together using 15% overlap between adjacent images. Images were again exported as high-resolution tagged image files, and all figures were created using Adobe Creative Suites.

Cells were stimulated as indicated in the figure legends. Following stimulation, dishes were placed on ice, gently washed with PBS, and lysed with IP buffer (TBS, pH 7.4, 1% NP-40, 10% glycerol, 500 µM sodium vanadate, and protease inhibitors) as described previously (Tsui and Pierchala, 2008). Antibodies for α-p75 (5 µl; 07–476; Millipore), α-Trk (8 µl; C-14; Santa Cruz Biotechnology), and α-Ret51 and/or α-Ret9 (8 µl; C-20 and C-19-G, respectively; Santa Cruz Biotechnology) were added along with protein A and protein G (Invitrogen) and incubated overnight at 4°C under gentle agitation. IPs were then washed three times with IP buffer and prepared for SDS-PAGE by adding 2× sample buffer (TBS, pH 6.8, 20% glycerol, 10% β-mercaptoethanol, 0.1% bromophenol blue, and 4% SDS) and boiling the samples for 10 min.

Cell surface biotinylation experiments were performed as described previously (Chen et al., 2017). In brief, neurons were washed with ice-cold PBS and labeled with 2 mmol EZ-Link NHS-LC-Biotin (in PBS; Pierce) for 40 min. Cells were washed and residual NHS-LC-Biotin inactivated with two 20-min TBS incubations. Cells were washed again and detergent extracted using the aforementioned IP buffer. Biotinylated proteins were precipitated with immobilized Neutravidin (Pierce) in an identical manner as the IPs described above, and IPs were then prepared for immunoblotting.

Samples for Western blotting were subjected to SDS-PAGE followed by electroblotting onto polyvinylidene difluoride membranes (Immobilin P; Millipore). Western blot analysis was performed using the following antibodies at the indicated concentrations: α-Ret51 (1:500–1:1,000; C-20, Santa Cruz Biotechnology), α-Ret9 (1:1,000; C19R; Santa Cruz Biotechnology), α-Ret (1:1,000; AF482, R&D Systems), α-phosphotyrosine (1:2,000-1:3,000; 4G10; Millipore), α-p75 (1:1,000; Advanced Targeting Systems; or 1:2,000; Promega), α-actin (1:2,000; JLA-20; Iowa Hybridoma), pTrkA (1:1,000; 9141; Cell Signaling Technology), α-TrkA (1:1,000; C-14; Santa Cruz Biotechnology), α-p-c-Jun (1:300; 9261; Cell Signaling Technology), α-phospho-JNK (1:1,000; 9251; Cell Signaling Technology), α-JNK (1:1,000; 9252; Cell Signaling Technology), α-Sortilin (1:1,000; ab16640; Abcam), α-TRAF6 (1:1,000; HPA020599; Sigma-Aldrich), α-TACE (1:1,000; sc-6416; Santa Cruz Biotechnology), α-PSN-1 (1:1,000; sc-7860; Santa Cruz Biotechnology), α-transferrin (1:2,000; T2027; Sigma-Aldrich), α-pAKT (1:1,000; 4058; Cell Signaling Technology), and α-AKT (1:2,000; 2920; Cell Signaling Technology). Blots were developed using a chemiluminescent substrate (Supersignal; Thermo Fisher Scientific). For quantifications, scanned images of x-ray films were imported into ImageJ (National Institutes of Health) and processed using the gel analysis tool. Integrated density values obtained from immunoblots were reported as mean values ± SEM, with arbitrary units on the vertical axis. Values were normalized to the appropriate control: for co-IP studies, values were normalized to the precipitated protein; for phospho-specific signaling effectors, values were normalized to total levels of these effectors; and values were normalized to actin (used as a loading control) for all other samples. All biochemical experiments were performed at least three times with similar results.

All results are expressed as the mean ± SEM. All statistical tests were performed using two-tailed parameters with a significance level of P ≤ 0.05 to test for statistical significance. For all statistical tests involving more than two variables, a two-way ANOVA with multiple comparisons was used with a Tukey post hoc test to adjust for multiple comparisons. A two-tailed Student’s t test was used for all comparisons between two treatment groups. A two-tailed Student’s t test was also used to compare each time course treatment group with the corresponding NGF-maintained control for these experiments. The data were originally entered into Excel and imported into GraphPad Prism, which was used for all statistical tests. Column or row statistics (GraphPad Prism) was used to confirm the data were normally distributed, allowing parametric tests to be used. The presence of asterisks indicates statistical significance: *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. The sample sizes are indicated in the figure legends. No sample sizes of fewer than three independent experiments were used. For the SCG counts, when possible, each animal represents the average count of two SCGs to increase statistical accuracy.

Fig. S1 shows validation of p75 immunostaining. Fig. S2 shows that p75 promotes sympathetic neuron apoptosis through a p-c-Jun–dependent pathway. Fig. S3 shows that removal of Ret does not alter TRAF6 association with p75 or up-regulation of the p75 cleavage enzymes PSN-1 and TACE.

We thank Drs. David Ginty and Joseph Savitt for providing Ret^fx/fx mice, and Drs. Wenqin Luo and Hideki Enomoto for providing Ret-Cre/ER^T2 mice.

Support was provided to C.R. Donnelly through the Rackham Merit Fellowship (University of Michigan), National Institute of Dental and Craniofacial Research grant T32 DE007057, and National Institute of Dental and Craniofacial Research grant F30 DE023479. These experiments were supported by National Institute of Neurological Disorders and Stroke grant R01 NS089585 and National Institute on Deafness and Other Communication Disorders grant R01 DC015799 awarded to B.A. Pierchala.

The authors declare no competing financial interests.

Author contributions: C.R. Donnelly and B.A. Pierchala designed experiments, interpreted data, and wrote the manuscript. C.R. Donnelly performed all in vitro experiments with assistance from M. Chowdhury and conducted all in vivo experiments utilizing Ret-Cre animals with the assistance of E.R. Suh. N.A. Gabreski performed in vivo experiments utilizing UBC-Cre animals. B.A. Pierchala was responsible for the overall direction and communication of the experiments.