The Effects of Transmembrane Sequence and Dimerization on Cleavage of the p75 Neurotrophin Receptor by γ-Secretase*

Background: p75 neurotrophin receptor (p75NTR) signaling is modulated by dimerization and regulated intramembrane proteolysis (RIP). Results: Transmembrane sequence and TrkA but not ligands regulate p75NTR homodimerization and γ-secretase cleavage. Conclusion: Although γ-secretase does not require a dimeric substrate, p75NTR dimerization facilitates RIP. Significance: Structural change mediated by homo- and heterodimerization is more important than ligand for inducing RIP of p75NTR. Cleavage of transmembrane receptors by γ-secretase is the final step in the process of regulated intramembrane proteolysis (RIP) and has a significant impact on receptor function. Although relatively little is known about the molecular mechanism of γ-secretase enzymatic activity, it is becoming clear that substrate dimerization and/or the α-helical structure of the substrate can regulate the site and rate of γ-secretase activity. Here we show that the transmembrane domain of the pan-neurotrophin receptor p75NTR, best known for regulating neuronal death, is sufficient for its homodimerization. Although the p75NTR ligands NGF and pro-NGF do not induce homerdimerization or RIP, homodimers of p75NTR are γ-secretase substrates. However, dimerization is not a requirement for p75NTR cleavage, suggesting that γ-secretase has the ability to recognize and cleave each receptor molecule independently. The transmembrane cysteine 257, which mediates covalent p75NTR interactions, is not crucial for homodimerization, but this residue is required for normal rates of γ-secretase cleavage. Similarly, mutation of the residues alanine 262 and glycine 266 of an AXXXG dimerization motif flanking the γ-secretase cleavage site within the p75NTR transmembrane domain alters the orientation of the domain and inhibits γ-secretase cleavage of p75NTR. Nonetheless, heteromer interactions of p75NTR with TrkA increase full-length p75NTR homodimerization, which in turn potentiates the rate of γ-cleavage following TrkA activation independently of rates of α-cleavage. These results provide support for the idea that the helical structure of the p75NTR transmembrane domain, which may be affected by co-receptor interactions, is a key element in γ-secretase-catalyzed cleavage.

The pan-neurotrophin receptor p75 (p75 NTR ) 6 is capable of regulating a range of neuronal functions. p75 NTR is best known for its role in mediating neural cell death during development, as well as in the adult following injury, including that induced by pro-neurotrophins in conjunction with a co-receptor sortilin (1,2). However, p75 NTR can also regulate cell proliferation, migration, axon guidance, and survival. In particular, when interacting with TrkA receptors, p75 NTR can form high-affinity binding sites for NGF and enhance trophic signaling (2,3).
p75 NTR is one of more than 60 transmembrane receptors known to undergo a two-step process of regulated intramembrane proteolysis (RIP) (4,5). As inhibition of this process influences some aspects of p75 NTR -mediated signal transduction, including neuronal survival (6 -9), understanding the normal regulation of p75 NTR RIP could elucidate how specific p75 NTR signals are modulated.
During RIP, transmembrane receptors are first cleaved by one or more metalloproteases (␣or ␤-cleavage) to release their extracellular domains, with p75 NTR being cleaved five amino acids N terminally above the membrane by tumor necrosis factor ␣-convertase (TACE)/ADAM17 (10,11), a process that can be induced by Trk activity (9). The remaining C-terminal fragments, containing the transmembrane and intracellular domains, are subsequently cleaved by ␥-secretase, releasing the intracellular domain from its membrane tether. Removal of the extracellular domain of transmembrane proteins by ␣-secretase generates a free amine at the N-terminal stump that is recognized by one of the components of ␥-secretase, nicastrin (12). This interaction enables ␥-secretase to dock with its substrates and position the receptors for subsequent cleavage by the catalytic component, presenilin (12)(13)(14)(15)(16). Less is known about the regulation of this cleavage. However, conformation of the transmembrane domain can exert a major influence on the site of ␥-secretase cleavage in substrates other then p75 NTR (14,15).
p75 NTR has previously been reported to form dimers, most likely mediated through the transmembrane domain, as there is minimal contribution of the extra-and intracellular domains to its self-association (25,33,34). Furthermore, it has been suggested that the transmembrane 262 AXXXG 266 sequence within p75 NTR may contribute to its dimerization (24), and that dimers of p75 NTR might be stabilized by an intermolecular cysteine 257 (Cys 257 ) disulfide bridge within the membrane (24,35). In this study we tested whether p75 NTR uses its AXXXG motif as well as Cys 257 to facilitate homodimerization, and whether these residues influence the rate of basal and NGF-induced ␥-secretase cleavage.

EXPERIMENTAL PROCEDURES
Construct Design-p75 NTR expression constructs used a modified pCDNA3 (Invitrogen) backbone. The rat p75 NTR signal peptide including a Kozak sequence (nucleotides Ϫ29 to ϩ87) was inserted between the KpnI and EcoRV restriction sites, generating the vector pCDNA3-SP. To generate fluorescently tagged p75 NTR , the fluorophores CFP and YFP were next amplified by PCR from peCFP-N1 and peYFP-N1 (Clontech), respectively, using primers incorporating 5Ј EcoRV and NheI restriction sites and a 3Ј stop codon and XhoI site. Fluorophores were cloned in-frame between the EcoRV and XhoI restriction sites of pCDNA3-SP, generating vectors pCDNA3-CFP and pCDNA3-YFP, respectively. p75 NTR peptide coding sequences were finally amplified under standard PCR conditions with 5Ј EcoRV and 3Ј NheI restriction sites incorporated into the respective primers. Subsequently, p75 NTR coding sequences were cloned between the EcoRV and NheI restriction sites of the pCDNA-CFP or pCDNA3-YFP vector to generate in-frame FRET-capable fusion proteins. Table 1 provides a list of p75 NTR constructs used in this study, with the altered amino acids noted. The TrkA and kinase-inactive TrkA K538R constructs have been previously described (7).
Cell Culture and Transfection-HEK293 or HEK293T cells, which do not endogenously express p75 NTR , were cultured in RPMI medium (Invitrogen) supplemented with 10% FCS (JRH Biosciences) at 37°C in a humidified atmosphere with 5% CO 2 . Unless otherwise noted, 40 -50,000 cells were plated at ϳ60% confluence in a 24-well plate (Nunc) and transiently transfected using FuGENE6 (Roche Applied Science). 250 -500 ng of DNA per well was used for cleavage experiments, and 125 ng of CFP DNA and 250 ng of YFP DNA was used for transfection for FRET experiments. For cross-linking and immunoprecipitation experiments, 6-well plates were used, with the cell number and transfections scaled up 4-fold.
Wild type and p75 NTR -deficient PC12 cells were cultured in DMEM containing 1% nonessential amino acids (Invitrogen), 1% Glutamax (Invitrogen), 10% horse serum, and 5% FCS. p75 NTR -deficient PC12 cells, stably transfected with shRNA against rat p75 NTR and cultured under G418 selection, were kindly provided by Carlos Ibanez (Karolinska Institute). PC12 cells were transfected using the Amaxa system (Lonza; program U-029) and plated at a density of 1 ϫ 10 4 cells per well in a 96-well plate for acid phosphatase viability assay, and at 3 ϫ 10 5 cells per well in a 12-well plate for cleavage assays. To induce p75 NTR expression, PC12 cells were treated for 15 min with a 30-watt UV-C light bulb or 20 M oligomeric human amyloid ␤ peptide (36).
Acid Phosphatase Experiments-Survival of PC12 cells transfected with p75 NTR expression constructs was measured by acid phosphatase viability assay. Cells were washed gently 3 times in PBS in the tissue culture plate. Cells were collected by centrifugation (360 relative centrifugal force for 5 min) between washes to prevent cell loss. Following addition of 100 l of acid phosphatase buffer (0.1 M sodium acetate, 5 mM 4-nitrophenyl phosphate (Sigma N4645-1G), 0.1% Triton X-100, pH 5.0), plates were incubated for 30 min at 37°C before 10 l of 1 M NaOH was added per well to develop the reaction. Absorbance was measured at 405 nm on a POLARstar Optima plate reader.
Cross-linking and Co-immunoprecipitation Experiments-Transiently transfected HEK293 cells were treated with the thiol-reducible protein cross-linker dithiobis(succinimidyl propionate) or the non-thiol reducible cross-linker disuccinimidyl suberate (Pierce), according to the manufacturer's instructions. After washing, cells were lysed on ice for 10 min in TNE buffer (20 mM Tris, pH 8, 10 mM EDTA) supplemented with 1% Nonidet P-40 and complete protease inhibitor mixture (Roche Molecular Biochemicals). The results of immunoprecipitation experiments using cell lysates that had not been cross-linked with dithiobis(succinimidyl propionate) prior to lysis were equivalent to those following cross-linking. For coimmunoprecipitation studies, samples were immunoprecipitated overnight at 4°C using Gammabind G-Sepharose beads (Amersham Biosciences) coupled with anti-GFP (11814460001, Roche Applied Science). Nonspecific binding was removed by washing beads 5 times in Tris-buffered saline supplemented with 0.05% Tween 20, and proteins were eluted by boiling the samples in reducing sample buffer (20 mM DTT, 2.5% SDS). Protein samples were resolved under reducing conditions on 4 -12% SDS-PAGE gels prior to Western blotting.
Immunofluorescence and Microscopic FRET Analysis-For immunofluorescence imaging, HEK293 cells were plated at ϳ25% confluence onto polyornithine-coated 12-mm coverslips and transiently transfected with wild type or YFP-tagged p75 NTR expression constructs as described above. After 2 days, cells were fixed with 4% paraformaldehyde dissolved in PBS. The coverslips were mounted onto microscope slides using Dako fluorescence mounting medium (Dako Corporation).
The FRET between p75 NTR proteins in single cells was measured as sensitized acceptor emission using the three-cube method, as previously described (37,38). Images in donor (excitation 405 nm, emission 470 -500 nm), acceptor (excitation 514 nm, emission 530 -600 nm), and FRET (excitation 405 nm, emission 530 -600 nm) channels were acquired on a Zeiss LSM 510 Meta confocal microscope using a ϫ63 oil immersion objective, with a numerical aperture of 1.4. Living HEK293T cells were imaged in PBS containing 1 mM CaCl 2 . The 512 ϫ 512 images recorded at 12-bit depth were batch converted from a Zeiss LSM format into the tif format using ImageJ and processed further in a custom-written procedure in IgorPro5 (Wavemetrics). This procedure performed background subtraction, cross-talk correction, shifting of correlated FRET channel images, if required, and thresholding. The FRET ratio (FR ϭ F AD /F A , where F AD and F A are the fluorescence intensity of the acceptor in the presence and absence of the donor, respectively) was then calculated on a pixel by pixel basis.
Measurement of FRET in Single Cells by Flow Cytometry-FRET analysis using flow cytometry was carried out as described previously (38,39). In brief, the FRET efficiency was calculated per HEK293 cell using an adapted sensitized acceptor emission method. Using custom-written algorithms in Igor-Pro5, calibration for approximate acceptor surface concentration, FRET efficiency, and the donor-acceptor ratio were calculated. Only cells with a donor-mole fraction x D ϭ 0.5 Ϯ 0.1, corresponding to a ϳ1:1 donor-acceptor ratio, were analyzed. The characteristic E max value was determined by iterative fitting as previously described (38,39). A mutant of the membrane targeting sequence of H-Ras, CTH-7A-L (39), was used as a control for random interactions (E max was ϳ10%).
Cell Population FRET Assay-FRET between p75 NTR molecules within a population of transiently transfected HEK293T cells was determined 48 h post-transfection of triplicate wells, using untransfected cells to determine background fluorescence. Cells from individual wells were washed once before being harvested in PBS (pH 7.4). Cell pellets were resuspended in 200 l of PBS and loaded into single wells of 96-well black microtiter plates (Greiner) for immediate analysis. Fluorescence was recorded simultaneously for the donor fluorophore (excitation 430 Ϯ 10 nm and emission 480 Ϯ 10 nm) and the acceptor fluorophore (excitation 485 nm and emission 530 Ϯ 10 nm), together with FRET (excitation 430 Ϯ 10 nm and emission 530 Ϯ 10 nm), on a POLARstar OPTIMA multidetection microplate reader (BMG Labtech). FRET was calculated as a FRET ratio (FR), which is the fractional increase in YFP emission due to FRET (37,40), according to the formula, where D ϭ donor filter set, A ϭ acceptor filter set, F ϭ FRET filter set, a ϭ acceptor only, d ϭ donor only, ƒ ϭ donor and acceptor, S 1 ϭ Da/Fa, S 2 ϭ Ad/Dd, S 3 ϭ Ad/Fd, S 4 ϭ Fa/Aa, Primary Neuronal Cultures and FRET-Experiments using animal tissue were approved by the Animal Ethics Committee at the University of Queensland and were conducted according to the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. Primary neuronal cultures were generated from dissociated dorsal root ganglia derived from newborn p75 NTR -deficient (NGFR Ϫ/Ϫ C57Bl6/J) (41) mice as previously described (7). Dissociated neurons (60,000) were transfected with 5 g of total plasmid DNA using a Basic Neuron Nucleofector Kit (Lonza), and electroporated with program G-013. Neurons were plated onto polyornithine (0.15%) and laminin (2.5 g/ml)-coated glass bottom 3-cm dishes (Matek) in MonoMed II medium (CSL) containing 1% FCS, 10 6 units/ml of leukemia inhibitory factor (Millipore), and 20 mM potassium chloride. After 36 -48 h, FRET images of neurons were captured on an inverted LSM510 confocal microsope (Zeiss) at 37°C under 5% CO 2 , immediately before and after the addition of 5 ng/ml of NGF. Where indicated, cultures were pretreated with compound E (200 nM) for 4 h or TAPI-2 (20 M) overnight as well as 4 h prior to imaging.
Modeling p75 NTR Transmembrane Domain Dimerization-The solution nuclear magnetic resonance (NMR) structure of glycophorin A (PDB code 1AFO)(42) was used as a template to model the homodimeric transmembrane domain structure of p75 NTR . The structure of the transmembrane domain was built as a canonical ␣-helix, with and angles of Ϫ57°and Ϫ47°, respectively. The sequences used for each helix were p75 NTR residues 251-273 NLIPVYCSILAAVVVGLVAYIAF, and p75 NTR -LXXXL mutant NLIPVYCSILALVVVLLVAYIAF. The two helices in the homodimer were then aligned and oriented with a crossing angle of Ϫ40°, as observed experimentally for glycophorin A (42). The models were prepared using PyMOL (43).

p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage
Statistical Analysis-Experimental results were analyzed using one-way ANOVA unless otherwise stated. When significant ANOVA was detected, planed comparisons were made to detect between group differences. Alpha was controlled using a Bonferroni adjustment. p Ͻ 0.05 was considered significant.

p75 NTR Forms Homodimers Primarily at the Cell Surface-
Formation of p75 NTR protein dimers has been shown to have a strong regulatory effect on the activation of the cell death signals of the receptor (24,35). Similarly, the ability of p75 NTR to undergo RIP has a major influence on p75 NTR -mediated functions (2). To investigate the relationship between dimerization and RIP, we first assayed HEK293 cells transfected to express either full-length p75 NTR (p75 FL ) or p75 NTR with YFP fused to its C terminus (p75 FL-YFP ; Table 1). p75 NTR expression in these cells was compared with the levels, per g of protein, of p75 NTR expressed endogenously in PC12 cells. Although the expression level of p75 NTR in transfected cells was higher than basal PC12 cell expression (Fig. 1a), it was a similar level to that of cells under stress conditions known to induce p75 NTR expression, i.e. amyloid ␤ treatment and UV exposure (44). Furthermore, overexpression of p75 FL-YFP , like overexpressed wild type p75 FL (6,7), induced significant PC12 cell death compared with that obtained from control cells transfected with YFP ( Fig. 1b), demonstrating that the YFP motif does not affect the death signaling function of p75 NTR .
Using our constructs we then measured the size of self-associating p75 NTR oligomers in live HEK293 cells that do not express endogenous p75 NTR or Trk co-receptors, and have negligible levels of the p75 NTR co-receptor sortilin. Following cross-linking, the main complex size in both cases was double that of the monomer, indicating that the complexes were likely to be dimeric (Fig. 1c).
To study the regulation of these homo-associations, we used FRET analysis, which allows the inference of a specific interaction between proteins based on a FRET signal (40). In the current study donor CFP and acceptor YFP proteins were fused to putative interaction partners. CFP and YFP give 50% FRET efficiency when they are within 4.92 nm of each other (45).
To determine the spatial location of p75 NTR molecules that undergo FRET, we performed our analyses using confocal microscopy, in which FRET was measured as the FRET ratio, which describes the sensitized acceptor emission (37). p75 FL-YFP was most abundantly located at the cell surface (Fig.  1d), and the highest FRET between full-length p75 NTR molecules was found at this location (Fig. 1e). However, lower FRET was seen throughout the cytoplasm, presumably in the membrane of endosomal vesicles traveling toward or away from the cell surface (Fig. 1e). To examine FRET specifically related to self-association rather than any change in intracellular domain movement that might move the C terminally located fluorophores further from each other upon dimerization, we used a form of p75 NTR in which the fluorophore replaced the intracellular domain and was located immediately C-terminal to the transmembrane domain (p75 ECD ). This construct was similarly expressed on the surface of the cells and produced a significant FRET ratio in this location (Fig. 1, d and f). To confirm these results we used a flow cytometry-based FRET method in which FRET can be measured in live cells on a per cell basis (39). In these experiments, the dependence of FRET on the membrane density of labeled donor and acceptor proteins was taken into account. The FRET efficiency reached at high acceptor concentrations, E max , was extracted for cells that had a ϳ1:1 donor: acceptor expression ratio, as described previously (39). We found FRET values that were consistent with a significant proportion of these truncated p75 NTR molecules being present as homodimers (p75 ECD , E max: 21.8 Ϯ 0.7; Fig. 1g), if compared with the FRET associated with the nonspecific interactions of p75 NTR with a membrane-tethered control CFP, which has an E max of ϳ10 (38,39). Together these results suggest that p75 NTR forms dimeric complexes that can be measured by FRET methods that are not influenced by the intracellular domain.
The Transmembrane Domain Mediates p75 NTR Dimerization-To study the regulation of p75 NTR dimerization, we next validated a high-throughput population-based FRET assay using a multimode plate reader. A construct of CFP fused to YFP served as a positive control for maximal protein-protein interaction (FR ϭ 9.0 Ϯ 1.3; Fig. 2a, first column). The lower limit of the FRET ratio that was indicative of a nonspecific interaction was established using a construct expressing only YFP together with a construct expressing p75 FL-CFP . In this condition, FRET could only occur through nonspecific interactions between the soluble acceptor fluorophore and the donor molecule tethered to p75 NTR . This resulted in a FRET ratio close to 1 (FR ϭ 1.2 Ϯ 0.2; Fig. 2a, second column), indicating that there was no energy transfer above background between the two fluorophores (37). In a second control experiment, p75 FL-CFP was co-transfected with p75 FL in which YFP was fused to the N terminus of the receptor. Again, FRET was at background levels, as these fluorophores were located on oppo-site sides of the membrane (FR ϭ 1.0 Ϯ 0.1; Fig. 2a, third column). By contrast, FRET between two p75 FL receptors containing donor/acceptor fluorophores fused to their C termini was significantly increased (FR ϭ 2.4 Ϯ 0.1; Fig. 2a, fourth column), suggesting a significant protein-protein interaction.

p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage
As seen by microscopy, significant FRET ratios, equivalent to that of the wild type p75 NTR receptors, were obtained between two p75 ECD proteins (Fig. 2b, first column). Two p75 NTR proteins lacking the extracellular domain (p75 CTF ) also generated a significant FRET ratio (Fig. 2b, second column). This indicated that neither the extracellular nor the intracellular domains were required for self-association, and implied that the transmembrane domain mediates p75 NTR dimerization. Consistent with this, the FRET ratio of two p75 NTR intracellular domain fragments (p75 ICD ) was not significantly different from that of random interactions (Fig. 2b, third column).
To determine whether the transmembrane domain is sufficient for p75 NTR homodimerization, we first tested whether the truncated p75 ECD-YFP protein could associate with wild type p75 FL-CFP . A significant FRET ratio was measured for interactions between p75 FL and p75 ECD (Fig. 2c, first column). This somewhat surprising result suggested that the CFP linked to the 138-amino acid intracellular domain of p75 FL was, nonetheless, within close proximity to the YFP located immediately adjacent to the plasma membrane due to its fusion to p75 ECD . Based on this, we postulated that the C terminus of the intracellular domain of p75 NTR was normally oriented toward the plasma membrane. To investigate this further, we generated a construct in which the structurally ordered amino acids Pro 295 -Pro 296 -Pro 297 within the intracellular juxtamembrane region were mutated to alanines (p75 AAA ), to introduce conformational flexibility and allow movement of the intracellular domain, e.g. away from the plasma membrane. This resulted in a significant decrease in p75 NTR FRET, supporting our hypothesis that the C terminus of the intracellular domain of p75 NTR is normally oriented toward the membrane (Fig. 2c, second column).
We next used a p75 NTR variant that contains the transmembrane sequence plus 14 amino acids of both the juxtamembrane extracellular and intracellular domains, which are 100% conserved in the vertebrate orthologs of p75 NTR (p75 TM ). p75 TM-YFP was appropriately trafficked to the plasma membrane of HEK293 and/or p75 FL as outlined in the figure (blue represents the CFP donor, yellow represents the YFP acceptor). FRET ratios for each p75 NTR pair were compared (ANOVA) to those of the negative control (second column). ***, p Ͻ 0.001 (mean Ϯ S.E., n ϭ 5 experiments). b, FRET ratio of interactions between truncated forms of p75 NTR (p75 ECD , p75 CTF , and p75 ICD ) with each other, as outlined in the figure. FRET ratios for each p75 NTR pair were compared (ANOVA) to those of the negative control (panel a, second column). ***, p Ͻ 0.001 (mean Ϯ S.E., n ϭ 4 experiments; ANOVA). The intracellular domain fragments of p75 NTR had a FRET ratio that was not significantly different from that of random interactions. c, the FRET ratio calculated for p75 NTR truncation and p75 AAA constructs fused to YFP when co-transfected with p75 FL-CFP . FRET ratios for p75 ECD and p75 TM paired with p75 FL were compared (ANOVA) to those of the negative control (panel a, second column). p75 AAA paired with p75 FL had a FRET ratio that was significantly lower than that of p75 FL :p75 FL (panel a, fourth column). ***, p Ͻ 0.001 (mean Ϯ S.E., n ϭ 4 experiments; ANOVA). d, confocal micrograph of a HEK293 cell expressing p75 TM-YFP , showing its location at the plasma membrane of the cell. e, Western blots of immunoprecipitates and lysates of HEK293 cells transfected with either p75 FL , p75 N-Gly , or p75 ICD constructs and co-transfected with p75 TM-YFP . p75 FL and p75 N-Gly but not p75 ICD was immunoprecipitated by the minimal p75 NTR transmembrane domain protein (representative of n ϭ 3 experiments). FL, full-length p75 NTR ; CTF, C-terminal fragment; IP, immunoprecipitation; WB, Western blot. p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 cells (Fig. 2d) and, when paired with p75 FL-CFP for FRET experiments, a significant FRET ratio was recorded, suggesting a specific protein interaction (Fig. 2c, third column). Consistent with this, full-length p75 NTR but not p75 ICD was co-immunoprecipitated with p75 TM-YFP (Fig. 2e, second and first lanes 2, respectively). These results indicate that the transmembrane domain of p75 NTR is sufficient for dimer formation.
Dimeric P75 NTR Undergoes ␥-Secretase Cleavage but This Does Not Regulate Dimerization-Based on the above results, we next asked whether monomers or dimers of p75 NTR were being cleaved. This was achieved by determining whether the process of RIP, which occurs at low basal levels in HEK293 cells (Fig. 3), decreased the extent of full-length p75 NTR dimerization. We hypothesized that a fractional loss of receptor dimers due to ␥-cleavage would produce fragments that do not undergo FRET, and thus a change in the FRET ratio would occur under conditions of altered rates of RIP. First, ␣and ␥-cleavage of p75 NTR were induced by treatment with the phorbol ester PMA, which resulted in a significant decrease in p75 NTR FRET (Fig. 3a, second column). As the intracellular domain of p75 NTR is rapidly degraded, this was inhibited by treatment with the proteasome inhibitor ␤-clasto-lactacystin. This treatment did not rescue the extent of FRET change (Fig.   3a, third column), consistent with the soluble p75 ICD fragments not producing a significant FRET signal (Fig. 2b, third column). By contrast, in the absence of PMA treatment but the presence of the ␣-secretatse inhibitor TAPI-2, which prevents ␣and thus ␥-cleavage of p75 NTR , a significant increase in the FRET ratio was recorded as compared with that observed in wild type p75 NTR -expressing cells undergoing low basal rates of RIP (Fig.  3a, fourth column). To confirm that a basal level of p75 NTR proteolysis was responsible for reduced p75 NTR :p75 NTR FRET, we also used a noncleavable mutant of p75 NTR containing an N-glycosylation site (p75 N-Gly )(7), which co-immunoprecipitates with the p75 NTR transmembrane domain (Fig. 2e, fourth lane) but is not cleaved by ␣-secretase, thereby also preventing ␥-secretase cleavage (Fig. 3a, fifth lane). As with pharmacological ␣-cleavage inhibition, cells co-expressing full-length noncleavable p75 N-Gly-YFP together with p75 FL-CFP yielded a significantly higher FRET ratio (Fig. 3a, fifth column) than that recorded for cells expressing p75 FL FRET pairs.
To explicitly test whether ␥-secretase release of the intracellular domain of p75 NTR to the cytoplasm was responsible for the decrease in p75 NTR FRET during RIP, ␥-secretase activity was blocked by adding a specific ␥-secretase inhibitor, compound E. This treatment, which leads to an accumulation of the . ␣and ␥-cleavage were activated with PMA, intracellular domain degradation was inhibited with ␤-clasto-lactacystin (CL), ␣-cleavage was inhibited with TAPI-2 or the p75 N-Gly mutation, and ␥-secretase was inhibited with compound E (Cmp E). Inhibition of ␣and ␥-secretases by TAPI-2, the noncleavable N-Gly mutation or compound E, significantly increased the p75 NTR FRET ratio, whereas release of the p75 NTR intracellular domain promoted by PMA significantly reduced the FRET ratio. **, p Ͻ 0.01; ***, p Ͻ 0.001 (mean Ϯ S.E., n ϭ 4 -8 experiments). CTF, C-terminal fragment; ICD, intracellular domain fragment. b, Western blots of cell lysates from HEK293 transfected with p75 FL either alone or together with p75 TM or p75 N-Gly-TM (not detectable by the anti-p75 NTR antibody) and treated with PMA to induce ␣and ␥-cleavage and compound E to inhibit ␥-cleavage (as marked; all conditions contain the proteasome inhibitor, CL). Co-expression of these constructs had no appreciable effect on the rate at which full-length p75 NTR underwent RIP and there was no change in the CTF:ICD ratios. p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage p75 NTR C-terminal fragment, resulted in an increase in the FRET ratio equivalent to that resulting from ␣-cleavage inhibition (Fig. 3a, sixth column). These results indicate that dimeric p75 NTR is capable of being processed by both ␣and ␥-secretase.
As the majority of known ␥-secretase substrates are dimers (16) and the enzymatic component, presenilin, is also a dimer (46), we next tested whether ␥-secretase requires a dimeric receptor substrate or is capable of cleaving only one component of a dimeric receptor complex at a time. To do this we took advantage of the requirement of ␣-cleavage for subsequent ␥-cleavage (12,16) and the resistance to ␣-secretase cleavage of p75 NTR containing the N-Gly mutation; this modification did not affect the ability of p75 NTR to form dimers when paired with wild type p75 NTR (Fig. 3a, fifth column, and Fig. 2e, fourth column). We therefore substituted the N-Gly mutation into our minimal transmembrane protein (p75 N-Gly-TM ).
The transmembrane domain construct, either with or without the N-Gly substitution, was overexpressed at a 2:1 ratio with the p75 FL construct to enhance formation of p75 N-Gly-TM : p75 FL heterodimers. Use of this truncated construct ensured that any resulting cleavage fragments were derived from p75 FL . No effect on the rate of intracellular domain fragment production from the full-length p75 NTR protein was observed under either basal conditions or following stimulation with PMA (Fig.  3b). This indicated that dimerization of p75 NTR with a noncleavable substrate had no discernable effect on its ability to undergo RIP. Therefore, either only monomers of p75 NTR were being cleaved or one-half of the mutant:wild type p75 NTR heterodimer was able to be cleaved. As the previous experiment had demonstrated that p75 NTR dimers undergo RIP, these results suggest that a dimeric substrate (a dimer of p75 NTR C-terminal fragments) is not a requirement for the ␥-secretase complex to interact with and cleave p75 NTR .
TrkA Signaling-induced ␥-Secretase Cleavage Reduces p75 NTR C-terminal Fragment Homo-oligomerization-We next investigated the influence of ligand and TrkA on p75 NTR dimerization and RIP. The addition of 100 ng/ml of NGF had no effect on the extent of FRET between two p75 FL or two p75 CTF proteins (Fig. 4, a and b). Similarly, sortilin co-expression with or without ligand had no effect on FRET levels (Fig. 4b). However, co-expression of TrkA either with or without NGF treatment resulted in a significant decrease in the FRET ratio compared with that obtained with only p75 NTR -expressing cells (Fig. 4b).
As overexpression of TrkA can result in autophosphorylation (7), and Trk signaling induces p75 NTR RIP (9, 11), we asked whether TrkA signaling was responsible for the reduction in p75 NTR oligomerization. In contrast to the kinase-active TrkA, a kinase-inactive mutant (TrkA K538R ) produced a significant increase in the FRET ratio compared with that observed in the absence of co-receptor (Fig. 4b).
To determine whether the TrkA-induced decrease in FRET was due to conformational changes in the p75 NTR intracellular domain, we used the truncated form of p75 NTR in FRET experiments performed on a per cell basis. Using this method we replicated the earlier finding that the noncleavable form of p75 NTR (p75 ECD-N-Gly ) had a significantly increased homo-in-teraction (E max ϭ 25.0 Ϯ 0.6), and that TrkA K538R expression also significantly increased the FRET between p75 ECD proteins (E max ϭ 25.3 Ϯ 1.0) to a level equivalent to the FRET measured between two noncleavable p75 ECD-N-Gly proteins (Fig. 4c). By contrast, TrkA again significantly decreased the FRET level of p75 ECD homodimers (E max ϭ 19.1 Ϯ 0.3; Fig. 4b). No significant change was detected between p75 ECD-N-Gly oligomers when TrkA was co-expressed (E max ϭ 24.9 Ϯ 0.8; Fig. 4c). Importantly, neither TrkA dimerization induced by the addition of ligand, nor the ability of p75 NTR to be cleaved, affected the extent of interaction between p75 NTR and TrkA, as measured by p75 ECD-CFP to TrkA ECD-YFP FRET (Fig. 4d).
These results indicated that TrkA-mediated signals were affecting the self-association of p75 NTR , specifically C-terminal fragment dimerization, due to ␥-secretase cleavage. We therefore next measured self-association of p75 CTF proteins in the presence of TrkA. Analogous to cleavable p75 FL this caused a significant decrease in the p75 CTF :p75 CTF FRET value, whereas TrkA K538R had no effect on this FRET ratio (Fig. 4e). Furthermore, inhibition of ␥-secretase cleavage prevented the significant decrease in p75 NTR FRET observed due to TrkA co-expression (Fig. 4f).
Endogenous TrkA activation by NGF similarly affected p75 NTR dimerization. Activation of TrkA has previously been shown to induce p75 NTR RIP within minutes (9). Therefore p75 NTR constructs were transfected into dorsal root ganglia neurons and FRET was measured. Following the addition of NGF, a 25% reduction in the basal p75 NTR FRET was observed within 5 min, which was prevented by the treatment of neurons with either TAPI-2 or compound E (Fig. 4g). Indeed, when RIP was prevented, the resulting increased p75 NTR FRET ratio was suggestive of an increase in the number of homodimers.
Together, these data suggest that the physical presence of TrkA promotes full-length p75 NTR :p75 NTR oligomerization, whereas TrkA activation not only activates ␣-cleavage (9, 11), but also promotes the dissociation of p75 NTR C-terminal fragment dimers through facilitation of ␥-secretase cleavage.
Residues within the Transmembrane Domain of p75 NTR Are Required for ␥-Secretase Cleavage but Not Dimerization-Given that many ␥-secretase substrates contain transmembrane GXXXG or related dimerization motifs, we investigated whether the AXXXG motif of two p75 NTR transmembrane sequences could self-associate by modeling the structure of the transmembrane domain. This was performed based on the NMR structure of a GXXXG-containing protein, glycophorin A (GpA) (42). The GpA dimer interface utilizes the sequence 79 GVXXGV 84 (Fig. 5a). Therefore the model was based on p75 NTR dimerizing through 262 AVXXGL 267 . Our model revealed that residues in the putative dimer interface were comparable in both the GpA and p75 NTR structures (Fig. 5). Specifically, the side chains of Ala 262 and Gly 266 of one p75 NTR monomer were able to pack against Ile 259 , Val 263 , and Leu 267 in the adjacent monomer (Fig. 5b); the presence of small glycine and alanine residues in the AXXXG motif allows the helix backbones to reach close proximity and the larger side chains to pack in a "ridges into grooves" fashion (42).
To interfere with this interface, we substituted the Ala 262 and Gly 266 residues (residues 12 and 16 of the transmembrane p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 43817 p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage domain) with leucine, generating a p75 LXXXL expression construct. Modeling of the p75 LXXXL mutant transmembrane revealed that the introduction of bulkier leucine residues at positions 262 and 266 would result in steric clashes between these residues and the three branched hydrophobic residues of the opposite monomer (Fig. 5c). These mutations would therefore alter the dimerization face if it is mediated by the AXXXG domain.
Another transmembrane domain residue implicated in modulating dimerization of p75 NTR is Cys 257 , which can mediate a disulfide bond between the p75 NTR monomers (24). Surprisingly, our modeling highlighted that Cys 257 was on the opposite face of the transmembrane domain to the AXXXG motif, indicating that the two possible modes of dimerization were mutually exclusive for a single pair of p75 NTR molecules (Fig. 5b). We therefore generated a construct in which the single transmembrane cysteine residue was changed to alanine (p75 C257A ), as well as a construct replacing the transmembrane domain with that of the tumor necrosis factor receptor transmembrane domain (p75 TNFR ), which contains cysteine residues but otherwise is not conserved compared with p75 NTR in terms of amino acid sequence (Fig. 5d).
We then measured the ability of these mutant p75 NTR variants to form covalent dimers. As previously reported, under nonreducing conditions (i.e. in the absence of DTT), a p75 NTRimmunopositive band was observed at ϳ160 kDa (Fig. 6a, asterisk). This complex is likely to be mediated by the disulfide bond via the transmembrane Cys 257 as it was not detected in lysates of HEK293 cells expressing p75 C257A (Fig. 6a, arrowhead). However, p75 LXXXL and p75 TNFR proteins formed covalent complexes to a similar extent and size as wild type p75 NTR (Fig.  6a, asterisk). By contrast, although p75 C257A retained the ability to FRET with p75 FL (Fig. 6b, second column), the ability of the p75 LXXXL and p75 TNFR proteins to undergo FRET with wild type p75 NTR was significantly reduced (Fig. 6b, third and fourth columns; FRET ratios: 1.08 Ϯ 0.27 and 1.52 Ϯ 0.03, respectively). p75 LXXXL was nonetheless able to undergo significant FRET when paired with another p75 LXXXL protein (FRET ratio: 1.97 Ϯ.0.04; n ϭ 3, p Ͻ 0.001, cf. fourth column), indicating that the reduction in FRET between p75 FL and p75 LXXXL proteins may be due to a change in the relative position of the two C-terminal fluorophores rather than a disruption to dimerization itself.
To test this, variant p75 NTR proteins (lacking YFP) were cross-linked. All p75 NTR variants were found within a complex with combined molecular mass of ϳ140 kDa (Fig. 6a, hash symbol), twice the molecular mass of monomeric p75 NTR , but less than the complex observed in nonreducing conditions (Fig. 6a,  asterisk). Similarly, despite their altered sequences, proteins from all three constructs (lacking YFP), and their C-terminal fragments, were able to be pulled down by p75 TM-YFP (Fig. 6c), indicating that they retained the ability to associate and form dimers with the wild type p75 NTR transmembrane domain.
Finally, the rate of ␥-secretase cleavage of the p75 NTR proteins that contained transmembrane domain mutations was examined. In HEK293 cells, ␥-secretase cleavage of p75 NTR proteins in which the transmembrane domain cysteine (p75 C257A ) had been substituted was at a significantly reduced rate under both basal conditions and, more obviously, following stimulation with PMA (Fig. 6d, p Ͻ 0.01, n ϭ 4). This indicated that, although covalent dimerization is not a requirement for cleavage to occur, Cys 257 facilitates this cleavage event. The level of the intracellular domain fragment in cells expressing either p75 TNFR or p75 LXXXL was negligible in both basal and stimulated conditions, indicating that ␥-secretase cleavage (but FIGURE 4. TrkA regulates p75 NTR dimerization and RIP. a, the FRET ratios measured live between p75 CTF and p75 FL pairs in transfected HEK293 cells following the addition of 100 ng/ml of NGF. b, the FRET ratio of p75 FL constructs in transfected HEK293 cells 24 h after the addition of 100 ng/ml of NGF or 40 ng/ml of pro-NGF. Co-transfection with sortilin, TrkA, or kinase-inactive TrkA K538R is as indicated. c, FRET analysis as determined by the E max of cleavable p75 NTR extracellular domain constructs (p75 ECD ) and noncleavable extracellular domain constructs (p75 ECD-N-Gly ) fused to CFP or YFP. Nonspecific interactions were standardized to the membrane marker CTH-7A-L (first column). By comparison, there was significant p75 ECD self-association (#, p Ͻ 0.001). TrkA co-expression significantly decreased, whereas TrkA K538R significantly increased p75 ECD FRET. TrkA co-expression had no significant effect on p75 ECD-N-Gly FRET. p75 ECD FRET E max values were compared with the E max values of p75 ECD without co-receptor (mean Ϯ S.E., n ϭ 7-8 experiments). d, levels of FRET between TrkA ECD-YFP and p75 ECD-CFP or p75 ECD-N-Gly-CFP constructs. E max values were compared with the E max values of p75 ECD-YFP co-transfected with control CTH-7A-L. p75 NTR -TrkA FRET was not significantly altered after the application of NGF or introduction of the uncleavable N-glycosylation mutation (mean Ϯ S.E., n ϭ 4 experiments; significance is relative to the second column, except where indicated by #, which is nonsignificant compared with the third column). e, co-expression of TrkA, but not TrkA K538R , significantly reduced the FRET ratio of two p75 CTF proteins (mean Ϯ S.E., n ϭ 3 experiments). f, the TrkA activity-induced decrease in p75 NTR FRET was prevented by treatment of cells with the ␥-secretase inhibitor compound E (Cmp E). g, FRET ratios were calculated for dorsal root ganglia neurons isolated from p75 NTRϪ/Ϫ animals expressing p75 FL FRET constructs in the presence or absence of TAPI-2 or compound E as measured live over a period of 9 min following the administration of 5 ng/ml of NGF (mean Ϯ S.E.; n Ͼ5 neurons per condition from 3 experiments). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. Cys 257 proposed to form a disulfide bond is shown in red. The introduction of leucine residues at positions 262 and 266 of the p75 LXXXL mutant would result in steric clashes (indicated by circles) that would alter helix packing. d, sequence alignment of the transmembrane domains of GpA, p75 NTR , and TNFR, with glycine and alanine residues that putatively facilitate helix-helix interaction highlighted in orange, and cysteine residues in red. p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 not ␣-cleavage) of these proteins was also significantly impaired (Fig. 6d). However, when transfected into TrkA-expressing PC12 cells, intracellular domain fragments were produced from each of the p75 NTR mutant constructs except p75 N-Gly (Fig. 6e). For all constructs, on average, PMA induced a 2-fold increase, and NGF a 1.2-1.5-fold increase, in the rate of cleavage. Although the rate of ␥-cleavage remained significantly impaired for p75 C257A , p75 LXXXL , and p75 TNFR , accumulating up to 4-fold higher levels of C-terminal fragment (Fig. 6e), the amount of intracellular domain fragment generated in the mutants mirrored the level of the C-terminal fragment. There-fore, the ability of p75 NTR to heterodimerize with TrkA, in addition to the p75 NTR transmembrane sequence and/or structure, appears to enhance the ability of ␥-secretase to dock with and/or cleave the p75 NTR C-terminal fragment.

DISCUSSION
Here we show that p75 NTR forms dimers via its transmembrane domain. Neither the extracellular domain (27,36) or ligand, nor the intracellular domain fragment of p75 NTR released following ␥-secretase cleavage, contributed to self-association to a substantial degree. Although p75 NTR dimers were p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage capable of being processed by both ␣and ␥-secretases, a dimeric p75 NTR substrate was not required for RIP. Mutation of amino acids within the transmembrane sequence that have been implicated in p75 NTR dimerization had only subtle effects on homodimerization but significantly affected normal rates of ␥-secretase cleavage. Although ligand activation of p75 NTR did not significantly affect rates of RIP in the cells examined here, signals mediated by TrkA activation induced the cleavage of p75 NTR , thereby reducing the self-association of the intracellular domain fragments. Furthermore, TrkA expression alone increased full-length p75 NTR homodimerization, which appeared to facilitate subsequent ␥-secretase cleavage.
The Transmembrane Domain of p75 NTR Mediates Selfassociation-A number of lines of evidence now suggest that both exogenously and endogenously expressed p75 NTR is capable of self-associating in the absence of a ligand (24,25,33). Such interactions were measured here through cross-linking, immunoprecipitation, and FRET analyses in live cells. The extracellular domains of p75 NTR do not dimerize in the absence of a ligand (25,33), and our studies revealed that, once cleaved by ␥-secretase, the intracellular domain of p75 NTR also does not self-associate. In addition, the transmembrane domain alone is sufficient for self-association, consistent with reports that p75 NTR dimers are stabilized by an intermolecular disulfide bond mediated by Cys 257 within the transmembrane domain (35). Although covalent p75 NTR -immunoreactive complexes required Cys 257 (Fig. 5a), we found that self-association, close enough for FRET, cross-linking and co-immunoprecipitation to occur (Fig. 5, a-d), remains possible when the cysteine is mutated to an alanine, indicating that other residues in the transmembrane domain also or alternatively mediate dimerization.
The AXXXG motif within the transmembrane domain has also been implicated in mediating dimerization of p75 NTR (24). However, substitution of AXXXG with LXXXL did not prevent p75 NTR transmembrane co-immunoprecipitation or the formation of higher molecular weight covalent and chemically cross-linked complexes. Nonetheless, we and others (24) found that mutation of this motif reduces the FRET signal between the mutant p75 NTR and a wild-type p75 NTR molecule. Although such a finding can be indicative of loss of dimer association, FRET signals can also be reduced if the fluorophores are no longer in alignment, i.e. if their orientation is rotated such that the emission dipole of the donor forms a 90°angle with the absorption dipole of the acceptor, preventing energy transfer (40). As our modeling predicted that the LXXXL mutations would change the crossing angle of the helices (Fig. 4c), this would be likely alter the relative orientation of both the transmembrane and intracellular domains, an explanation for the change in FRET (Fig. 6b). We therefore interpret the reduced FRET signal in this scenario as indicating a change in the orientation of the transmembrane structure of p75-LXXXL relative to that of wild type p75 NTR . Although not necessarily preventing receptor dimerization, it is likely that these mutations change the interactions that make up the dimer interface.
Two Modes of p75 NTR Dimerization?-Our modeling of the transmembrane domain revealed that the AXXXG motif and the Cys 257 residues are on opposite faces of the transmembrane domain helix. However, it is possible that the disulfide bond and the AXXXG motif can mediate self-association independently. Rather than participating in the formation of a single dimer, p75 NTR might form a tetramer or larger complex through an alternating series of noncovalent AXXXG interactions and disulfide bonds (e.g. see Ref. 47). Indeed, in nonreducing conditions, minor p75 NTR -immunopositive complexes were observed at sizes greater than 200 kDa (Fig. 6a). Similarly, TrkA co-expression increased p75 NTR homodimerization, but this did not alter Trk-p75 NTR FRET, indicating that larger heteromer complexes are possible. Furthermore, the size difference in the cross-linked and the disulfide bond-mediated p75 NTR immunoreactive complexes (see also Ref. 24) indicates that they may contain different constituents, i.e. that either the AXXXG motif or the transmembrane cysteine of p75 NTR facilitates interactions with a protein other than a second p75 NTR . The larger size of the DTT-reducible complex (greater than twice 75 kDa) suggests that p75 NTR dimerizes through the AXXXG domain and covalently bonds to another larger protein (neither Trk nor sortilin have transmembrane cysteines, making them unlikely candidates). Alternatively, both the AXXXG domain and Cys 257 mediate structurally different homodimers, with formation of the Cys 257 covalent p75 NTR homodimer being permissive for its interaction with a third, smaller protein. FIGURE 6. Residues within the transmembrane domain of p75 NTR are required for ␥-secretase cleavage but not dimerization. a, Western blots of lysates of HEK293 cells expressing either p75 FL , p75 C257A , p75 TNFR , or p75 LXXXL . Where indicated, lysates did not contain DTT (ϪDTT) or, prior to lysis, cells were treated with disuccinimidyl suberate (DSS) to cross-link proteins (ϩDSS). p75 NTR -immunopositive DTT-reducible (*) or cross-linked (#) complexes are indicated. p75 C257A did not form a covalent complex (Ͼ). b, the FRET ratio calculated for p75 NTR mutant constructs fused to YFP when co-transfected with p75 FL-CFP . The FRET ratio for each set of p75 NTR mutant constructs was compared with that for wild type p75 FL homo-oligomers (first column). Equivalent FRET ratios were observed between two full-length p75 NTR proteins and p75 FL :p75 C257A , whereas the p75 FL :p75 TNFR and p75 FL :p75 LXXXL FRET ratios were significantly reduced. ***, p Ͻ 0.001 (mean Ϯ S.E., n ϭ 4 -7 experiments). c, Western blots of immunoprecipitates and lysates of HEK293 cells transfected with p75 NTR mutant constructs and co-transfected with p75 TM-YFP . p75 C257A , p75 TNFR , and p75 LXXXL (and their C-terminal fragments) were co-immunoprecipitated by the minimal p75 NTR transmembrane domain protein. d, Western blot of HEK293 cells transfected with p75 FL , p75 C257A , p75 TNFR , or p75 LXXXL undergoing basal rates of cleavage or enhanced rates of cleavage following stimulation with PMA (ϩ) for 3 h. All conditions contain a proteasome inhibitor so as to visualize the intracellular domain fragment. p75 C257A had a significant reduction in the rate of ␥-cleavage following PMA stimulation compared with the rate of p75 NTR cleavage. A comparison of the ratio of densitometry measurements of C-terminal fragments to intracellular domain bands in lysates of cells transfected with p75 FL-YFP and p75 C257A-YFP following PMA treatment (mean Ϯ S.E.; n ϭ 4 experiments) revealed that p75 C257A produces significantly less intracellular domain fragment relative to wild type p75 NTR (p Ͻ 0.01, t test). Increased levels of the C-terminal fragment were detected in cells transfected with p75 TNFR or p75 LXXXL . However, even following PMA stimulation no intracellular domain fragment was detected. Blots are representative of 4 experiments. e, Western blot of p75 NTR -deficient PC12 cells transfected with p75 FL , p75 C257A , p75 TNFR , p75 LXXXL , or p75 N-Gly undergoing basal rates of cleavage or cleavage stimulated by treatment with PMA or NGF for 4 h. All conditions contain a proteasome inhibitor. Although all mutant constructs except p75 N-Gly produced intracellular domain fragments, the rates of ␥-secretase cleavage were significantly reduced, resulting in significant accumulation of the C-terminal fragments (C-terminal fragment intensity was normalized to the intensity of full-length p75 NTR in each lane and averaged for 3 experiments before being compared with the level of C-terminal fragment in unstimulated p75 FL -transfected cells). Blots are representative of 3 experiments. p75 NTR Transmembrane Sequence Regulates ␥-Secretase Cleavage DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 ␥-Secretase Recognizes Only Half of a Dimeric Substrate at a Time-Our experiments revealed that when p75 NTR could not be linked to a partner protein by a disulfide bond, the rate of ␥-secretase cleavage was significantly reduced, suggesting the residue normally facilitates this cleavage event. There is emerging evidence that components of the ␥-secretase complex (i.e. presenilin and APH-1) form dimers (16,46,48), and that it can recognize and cleave dimeric substrates (27,35), suggesting that the entire enzyme-substrate complex may be dimeric. As recognition of the N-terminal stump of the C-terminal fragment is a requirement for nicastrin to dock the protein substrate (12), we tested whether preventing ␣-cleavage and thus ␥-cleavage of one p75 NTR monomer within the dimer substrate had an impact on ␥-secretase processing of the other p75 NTR monomer. Little if any effect on total levels of intracellular domain production was observed in this experiment (Fig. 3b), demonstrating that ␥-secretase can recognize p75 NTR monomers within homodimers. Interestingly, forced dimerization of p75 NTR (through the addition of cysteines within its transmembrane domain) also has no discernable influence on rates of ␥-secretase cleavage (35), and dimerization in its self is insufficient to promote ␥-secretase cleavage of a widely studied ␥-secretase substrate, amyloid protein precursor (APP) (31). This highlights an assumption in the field that when a dimeric receptor undergoes RIP, the entire dimer is processed by metalloprotease to generate a dimeric C-terminal fragment, with both fragments subsequently being cleaved by ␥-secretases to release two intracellular domain fragments. Our work raises the possibility that only one-half of a dimer receptor complex is processed by RIP at a time. Indeed, as inactive TrkA co-expression increased p75 NTR FRET to a level equivalent to that between p75 N-Gly receptors, inactive TrkA:p75 NTR heterodimers might physically inhibit p75 NTR RIP.
By contrast, a decrease in p75 NTR FRET in neurons occurred within minutes of TrkA activation following the addition of NGF. This corresponds to the increased activity of the ␣-cleavage enzyme ADAM17 and production of the intracellular domain fragment (9), supporting our conclusion that ␥-secretase is primarily responsible for the observation. However, it remains possible that TrkA dimerization and activation causes a conformational change in the bound homomonomer p75 CTF fragments, facilitating cleavage by ␥-secretase.
Mutations That Alter Transmembrane Structure Prevent ␥-Secretase Cleavage of p75 NTR -Our work indicates that transmembrane conformation, in particular changes induced in p75 LXXXL and p75 TNFR proteins, is important for cleavage of p75 NTR by ␥-secretase. The molecular mechanism of RIP is not well understood; however, unraveling the secondary structure of the helical transmembrane domain to expose the backbone carbonyl for ␥-cleavage, rather than sequence conservation (49) or dimerization (31,35), is emerging as a key event in APP cleavage. It has been proposed that transmembrane glycine residues may enable destabilization of the ␣-helix to promote unraveling in the presence of the ␥-secretase complex (50). Extending the length of the ␣-helical structure of the transmembrane domain, in exchange for the wild type structure that becomes a random coil at the cleavage site, leads to a decrease in the rate of ␥-secretase cleavage of APP (30), supporting the idea that the ability of the helices to uncoil is important. The LXXXL mutation in p75 NTR may similarly stabilize the ␣-helix and thus prevent the transmembrane domain from unraveling to allow ␥-secretase access to its substrate.
Alternatively, alterations to the transmembrane helical structure through mutation have the potential to shift the site of enzyme docking and thus introduce mismatched interactions between the ␥-secretase site and the ␥-secretase enzyme, significantly reducing its ability to catalyze the intracellular domain release (32). It is also theoretically possible that GXXXG motifs within ␥-secretase components (48) facilitate docking between the enzyme and its substrate. The AXXXG motif within p75 NTR directly flanks the ␥-secretase cleavage site (AV/VVG), and disruption of this domain may therefore prevent its heterodimerization with enzyme components. One or a combination of these events would explain the reduced ability of ␥-secretase to cleave the mutant proteins.
Functional Significance of p75 NTR Cleavage-p75 NTR has been identified as an important regulator of cell death and survival in the nervous system, including during neurodegeneration, with processing by RIP and disulfide bond formation a crucial step in the activation of cell death signaling (7,36,51,52). Here we show that modification of the transmembrane structure of p75 NTR affects its cleavage by ␥-secretase, in agreement with findings reported for APP (27)(28)(29)(30)(31)(32). Our results indicate that merely inhibiting dimerization would not directly prevent RIP of p75 NTR because each substrate receptor may be processed as a monomer. However, receptor dimerization or structural changes caused or permitted by the transmembrane disulfide bond, which is necessary for cell death signaling, was shown here to facilitate ␥-secretase cleavage, consistent with the generation of the intracellular domain fragment being required for transcription-dependent cell death (8). TrkA activation also promoted ␥-secretase cleavage of both wild type and mutant p75 NTR , subsequent to increasing p75 NTR self-association. Sortilin can also regulate p75 NTR RIP (54). Whether these influences are primarily due to physical interaction (herein), activation of secretases (9,11), and/or movements within membrane compartments (7) remains to be determined. Other heterodimer interactions might also alter the rate of ␥-secretase cleavage, and impact on the ability of p75 NTR to signal (36,53), because they modulate the p75 NTR transmembrane structure. Further investigation of the effect of proteins that can associate with p75 NTR and modulate RIP are therefore warranted.