|Home | About | Journals | Submit | Contact Us | Français|
The LDL receptor-related protein 1 (LRP1) mediates internalization of a large number of proteins and protein-lipid complexes and is widely implicated in Alzheimer's disease. The cytoplasmic domain of LRP1 (LRP1-CT) can be phosphorylated by activated protein-tyrosine kinases at two NPXY motifs in LRP1-CT; Tyr 4507 is readily phosphorylated and must be phosphorylated before phosphorylation of Tyr 4473 occurs. Pull-down experiments from brain lysate revealed numerous proteins binding to LRP1-CT, but the results were highly variable. To separate which proteins bind to each NPXY motif and their phosphorylation dependence, each NPXY motif microdomain was prepared in both phosphorylated and non-phosphorylated forms and used to probe rodent brain extracts for binding proteins. Proteins that bound specifically to the microdomains were identified by LC-MS/MS, and confirmed by western blot. Recombinant proteins were then tested for binding to each NPXY motif. The NPXY4507 (membrane distal) was found to interact with a large number of proteins, many of which only bound the tyrosine-phosphorylated form. This microdomain also bound a significant number of other proteins in the unphosphorylated state. Many of the interactions were later confirmed to be direct with recombinant proteins. The NPXY4473 (membrane proximal) bound many fewer proteins and only to the phosphorylated form.
The LDL receptor-related protein 1 (LRP1) is an integral membrane protein that carries out the internalization of a large number of proteins, protein complexes and lipoproteins including very low-density lipoproteins, chylomicron remnants, activated α2 macroglobulin, and the amyloid precursor protein [1, 2]. LRP1 plays a role in lipid transport, the uptake of protease-inhibitor complexes and has been implicated in Alzheimer’s disease .
LRP1 is composed of an entirely extracellular α-chain of 3924 amino acids that is non-covalently linked to a β-chain of 601 amino acids that contains an extracellular region, a single transmembrane region, and a highly conserved cytoplasmic region composed of 100 amino acids including four tyrosine residues. Two of these, Tyr 4473 and Tyr 4507, are present in the context of Asn-Pro-X-Tyr (NPXY) motifs. The NPXY motif was first identified in the LDL receptor where it is essential for clathrin-mediated internalization . The CED-6/GULP protein involved in receptor trafficking has also been shown to bind LRP1 and the membrane-distal NPXY4507 motif was implicated in binding . The NPXY4507 motif in LRP1 has been implicated in binding other PTB domain-containing intracellular signaling proteins such as ShcA, Fe65 and Disabled (Dab1) [6–8]. Yeast two hybrid analysis of the LRP1-CT found a number of proteins including several with PTB domains . This study also found that LRP1-CT interacted with a homolog of PIP 4, 5 kinase hinting of a broader role for LRP1, perhaps in assembling signaling complexes. Indeed, other members of the LRP family also play important roles as mediators of signal transduction .
LRP1 is tyrosine phosphorylated in v-Src transformed mouse fibroblasts, and Tyr 4507, one of two NPXY tyrosines, was identified as the principle v-Src phosphorylation site in LRP1 . Others have found that LRP1 is tyrosine phosphorylated in response to platelet-derived growth factor . The ShcA PTB domain is known to bind to phosphorylated tyrosine residues present in the context of NPXY motifs, and has been shown to bind phosphorylated LRP1 [6, 12]. We recently showed that phosphorylation of the two NPXY motifs in LRP1-CT is sequential. Phosphorylation of the membrane-distal NPXY4507 motif microdomain causes exposure of the membrane-proximal NPXY4473 motif microdomain which is then subsequently phosphorylated . In addition, tyrosine phosphorylation at NPXY4473 prevented binding of Snx17 and enhanced binding of Shp2, which has two SH2 domains that can each bind an NPXY motif . These observations led us to investigate whether other proteins might interact with the NPXY-containing microdomains of LRP1-CT and whether their binding might depend on tyrosine phosphorylation. Since experiments in which LRP1 is isolated from cells could not reveal which NPXY motif microdomain was the site of binding nor the phosphorylation dependence, we decided to prepare each NPXY motif microdomain separately in both phosphorylated and unphosphorylated forms to discover specific protein-protein interactions that were occurring at each site.
The microdomains containing Tyr 4507 (Cys-amino aminohexanoic acid-TNFTNPVY4507ATLY) and Tyr 4473 (Cys-amino aminohexanoic acid-VEIGNPTY4473KMYEGGE) in both phosphorylated and unphosphorylated forms were synthesized and purified from AnaSpec, San Jose, CA. The peptides were immobilized using Sulfolink coupling gel according to the manufacturer’s directions (Pierce, Rockford, IL). Due to poor solubility of the unphosphorylated peptides all immobilization reactions were carried out in 3.5M Guanidine HCl, TBS pH 8.0. The amount of peptide immobilized was quantified by UV absorbance. Efficiency of immobilization was found to be 1.5–2.5mg peptide per milliliter of beads.
To verify that peptides were correctly immobilized and accessible, 10µL of peptide beads were added to a 25µg/mL chymotrypsin solution in TBS pH 7.4 and agitated at room temperature for 2 hours. Resulting peptides were mixed 1:1 with alpha-cyano-cinnamic acid (Agilent). Analysis by MALDI-TOF on an Applied Biosystems DE-STR yielded m/z 813.3 for NPXY4473 [(Y)KMYGEEG], 1063.5 for NPXY4507 [(F)TNPVYATLY], and 1143.4 for NPXpY4507 [(F)TNPVpYATLY], suggesting that peptides were susceptible to proteases, thus exposed on the bead surface.
As a further control for specificity of binding, the immobilized NPXpY4507 microdomain (20µL of beads) was treated with 10U shrimp alkaline phosphatase (SAP) (Promega) at 37°C for 12 hours in TBS pH 9.0, 1mM MgCl2. Non-phosphorylated NPXY4507 as well as NC beads were also treated with SAP the same way to rule out any other effects the phosphatase treatment may have on the beads.
Control and v-Src-transformed 14.30 and 14.30S3 mouse fibroblasts, and HEK293 cells were grown in Dulbecco’s modified Eagle’s medium containing 4 mM L-glutamine, 25 mM glucose and 10% calf serum at 37°C in 10% CO2 until confluent and then starved for 12 hours in serum free medium. MEF cells were obtained from ATCC (Manassas, VA) and grown in Dulbecco’s modified Eagle’s medium containing 4 mM L-glutamine, 25 mM glucose and 10% fetal bovine serum at 37°C in 10% CO2. The 11H4 hybridoma was obtained from ATTC (Manassas, VA) and grown in Iscoves-modified Dulbecco’s medium containing 25 mM L-glutamine and 10% FBS at 37°C in 10% CO2.
Brains of 3 month old Long Evans rats were resuspended in PLC lysis buffer (50 mM Hepes pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X100, 1.5 mM MgCl2 1 mM EGTA, 100 mM NaF, 10 mM Sodium Pyrophosphate, 500 µM Sodium Vanadate, 1 mM PMSF, 2 mM DTT, with protease inhibitor cocktail (P8340, Sigma-Aldrich) (1 brain in 10 ml) by Dounce homogenization. Lysates were cleared by high-speed centrifugation and incubated batch-wise with immobilized peptides (20µL) at 4°C on a rocker for 2 hrs. The beads were washed four times with PLC-lysis buffer, and bound proteins were analyzed by SDS-PAGE.
The proteins that bound to the NPXpY4507 microdomain were separated by 1-D SDS-PAGE using 10% Bis-Tris NuPAGE gels (Invitrogen). Gels were stained overnight with Gel Code Blue (Pierce) according to manufacturer’s suggested protocol. Bands unique to the NPXpY4507 –bound sample were excised, washed twice with 200 µL of 50% acetonitrile (ACN) and 50% 5 mM DTT / 25 mM NH4HCO3 (pH 7.4), reduced in 100mM DTT for 30 minutes at 50°C, washed again, and subsequently alkylated with 100mM iodoacetimide. The gel pieces were dehydrated in pure ACN, rehydrated by addition of 20 µL of ice-cold 10 ng/ µL trypsin (Promega) in 25 mM NH4HCO3 (pH 7.4), incubated on ice for 30 min and the remaining trypsin solution was removed and replaced with fresh 25 mM NH4HCO3 (pH 7.4). The digestion was allowed to continue at 37°C overnight. The peptide mixture was then acidified with 2 µL of 2% trifluoroacetic acid (TFA) (Fluka) and 2 µL of acetonitrile, vortexed for 30 min, and the supernatant extracted. Finally, 20 µL of 20% acetonitrile/ 0.1% TFA was added followed by vortexing to extract the remaining peptides and combined with the previous fraction. The combined extractions were concentrated in a speedvac prior to mass spectrometry.
Rat brain lysates (3 mls) were cleared by high-speed centrifugation and incubated batch-wise with immobilized peptides (35µL) at 4°C on a rocker for 2 hrs. After washing as before, the beads were washed twice with TBS 2mM EDTA, pH 8.0, then resuspended in the same buffer with 0.5% Rapigest® (Waters), and boiled for 10 min to elute bound proteins. The mixture was treated with 3mM TCEP (Sigma) for 30 min, then 6mM iodoacetamide (Sigma) for 30 min, and finally quenched with another 3mM TCEP. Freshly dissolved proteomics grade trypsin (10µg, Roche), was added to each sample, and digestion was carried out overnight at 37°C. The digestion was stopped with the addition of 100mM HCl, and incubated at 37°C for 1 hr to breakdown the Rapigest®. Samples were then centrifuged for 30 min at 4°C to remove beads and insoluble degraded lipid. Resulting peptides were extracted via C18 solid phase extraction (Varian, A57203) according to manufacturer’s suggested protocol, and concentrated by speedvac prior to MS analysis.
Digested samples were analyzed by electrospray ionization using a QSTAR-Elite hybrid mass spectrometer (Applied Biosystems) interfaced to a capillary column (180 ID, 365 OD) packed with C18 and capped with 0.2 mm filters on either end connected via 4 cm 25 ID (365 OD) capillary to the nanospray source. The spray needle was a pulled capillary 180~15 tip packed with C18. One dimensional chromatography was used for the samples from in-gel digests, whereas 2-dimensional “MUDPIT” chromatography was used for the in solution digests. For 2D chromatography, a strong cation exchange column was inserted in between the autosampler and injector valve. To accomplish the 2D chromatographic separation, the TEMPO LC high flow rate Channel 1 (10 µl/min) was used to load and separate on the SCX column with buffers A: 2.5%ACN, 0.2 % TFA 0.005%TFA; and Channel 1 B: 500mM ammonium acetate, 5%ACN, 0.2 % TFA. To separate on the C18 reverse phase, the low flow rate Channel 2 (400 nl/min flow) with buffer A: 2.5%ACN, 0.2 %FA 0.005%TFA and Channel 2 B: 100%ACN, 0.2 %FA 0.005%TFA was used. MUDPIT runs were carried out as described previously  with an 8 step ion exchange elution. Each was then separated on a 35-min linear gradient from 5 to 40% buffer B at a flow rate of 400 nL/min. LCMSMS data were acquired in a data-dependent fashion by selecting the 4 most intense peaks with charge state 2–4 that exceeded 10 counts with exclusion of former target ions set to always and the mass tolerance for exclusion set to 100 ppm. TOF MS were acquired at m/z 400–1600 Da for 1 sec with 20 time bins to sum. MSMS spectra were acquired from m/z 65 – 2000 Da using "enhance all" and 20 time bins to sum, Dynamic Background Subtract, Automatic Collision Energy, and Automatic MS/MS Accumulation with the Fragment Intensity Multiplier set to 4 and Maximum Accumulation set to 2 sec before exiting the scan.
Data were analyzed by Analyst 2.0 (Applied Biosystems) and subjected to two different database search protocols. First, spectra were searched with Protein Pilot using the paragon algorithm  against both the rat database alone (Rattus norvegicus, ncbi 07/06/2006, ftp://ftp.ncbi.nih.gov/genomes/R_norvegicus/protein/) as well as the non-redundant swissprot database (swissprot 10/29/2008, ftp://ftp.ncbi.nih.gov/blast/db/FASTA/). Search parameters included fixed carboxymethyl modification of cysteines and included common forms of variable modifications. Only protein hits with at least 2 strong unique peptide matches (>95) were kept. Any hits in the negative control were subtracted from the other samples. The spectra were also searched using Mascot 2.2.1 (Matrix Science) with Mascot Daemon 2.2 (Matrix Science) data import filter parameters set as follows: default precursor charge state 2–4; precursor and MSMS data centroiding using 50% height and 0.05 amu merge distances. MSMS peaks with intensity less than 1% of the base peak were discarded, as were MSMS spectra with less than 22 peaks remaining. Data were searched against the Swissprot database obtained at ftp://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/ containing 237,168 sequences. The search identified tryptic peptides with up to 2 missed cleavages and used mass tolerances of 100 ppm (MS) and 0.10 Da (MSMS), with variable modifications as follows: deamidation (NQ), oxidation (M), pyro-Glu (N-term E). The search results indicated that individual ion scores > 42 indicate identity or extensive homology (P < 0.05). Numbers of peptides reported are for non-redundant peptides. MASCOT and Protein Pilot both identified the same proteins, so the data from Protein Pilot are presented.
Common contaminants including keratin, trypsin, and tubulin were omitted from final table. Results for each microdomain were compared to the results without immobilized peptide and only those proteins that were not observed in the negative control were considered “real”. This criteria resulted in 14/39 “real” hits for the NPXpY4507 microdomain, 31/80 “real” hits for the NPXY4507 microdomain, 3/23 “real” hits for the NPXpY4473 microdomain, and no “real” hits for the NPXY4473 microdomain.
A polyclonal anti-Shp-2 serum was raised against a GST fusion protein containing both SH2 domains. The same technique was used to generate polyclonal anti-Grb2, anti-ShcA, and anti-CSK. a polyclonal anti-LRP1 serum was raised against a GST-fusion protein containing the cytoplasmic domain. Anti-phosphotyrosine monoclonal antibody 4G10, polyclonal anti-p85 PI3K serum, anti-HA (12CA5) and anti-Fe65 (3H6) were purchased from Upstate (Charlottesville, VA). Anti-PLCγ and anti-GAPDH (14C10) were purchased from Cell Signaling Technologies (Danvers, MA). Anti-Crkl (SC-31694) was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-Src monoclonal antibody 327 was a gift from Dr. T. Hunter (La Jolla, CA). Anti-14-3-3 (57-0700), anti-Ubiquitin (13–1600) and anti-CamK2B (13–9800) were purchased from Zymed (Carlsbad, CA). Anti-GST antibody (27457701) was from GE Healthcare. Immunoprecipitation and immunoblotting were carried out exactly as described previously , with the exception of Ubiquitin blots which were autoclaved for 20 minutes after transfer for increased sensitivity .
GST tagged SH2 domains from human Grb2 (58–159), PLCγ-1 (“N” 544–659 & “C” 663–759), and PI3K (“N” 314–446 & “C” 614–724) and full length 14-3-3γ (genbank ID 21464100) were cloned into pGEX2T (GE Healthcare) using BamH1 and EcoR1 restriction sites. BL21-DE3 cells were grown to OD600 0.5 at 37°C and protein expression induced with 0.3mM IPTG at 18°C overnight. Cells were harvested by centrifugation, resuspended in 50mM Tris pH 8.0, 500mM NaCl, 1mM DTT, and 1mM PMSF (Sigma) and lysed by sonication. After clearing debris by centrifugation at 12krpm for 40 min, cell extract was loaded over Glutathione-sepharose (GE Biosciences) according to manufacturer’s instructions. Eluted protein was buffer exchanged and concentrated over Y10K centricon (Millipore). 14-3-3γ was also cloned into pET15 with an N-terminal His tag, and purified by Nickel affinity chromatography followed by gel filtration. Recombinant purified His tagged CSK was a gift from Patricia Jennings. His-tagged CamKIIβ was purchased from Sigma. Purified calmodulin was purchased from Millipore. SH2 domains of Shp-2 and HA tagged Snx17 were expressed and purified as previously described . For recombinant pulldown reactions 10µL of NPXY beads were incubated in a 100nM solution of protein in the same PLC-lysis buffer used for previous pulldowns except for CamKIIβ which contained 50nM CamKIIβ and 200nM Calmodulin. Reactions were washed and blotted as described above.
Several groups have shown that LRP1 can become phosphorylated on Tyr in the distal NPXY4507 motif by activated protein-tyrosine kinases [6, Barnes, 2003 #17, 17]. Association of LRP1-CT with “signaling” proteins such as Shp2 and Src has also been demonstrated . These observations led us to hypothesize that other proteins may also bind to LRP1 in a phosphorylation dependent manner.
To find out whether LRP1 is involved in multiple phosphotyrosine-dependent protein-protein interactions, LRP1 was isolated by immunoprecipitation from control and v-Src transformed fibroblasts and the immunoprecipitate was analyzed by anti-P.Tyr immunoblotting. The results show LRP1 is associated with several P.Tyr-containing proteins in v-Src transformed cells (Figure 1A). A significant increase in these interactions was observed when cells were treated with protein tyrosine phosphatase inhibitors orthovanadate and hydrogen peroxide (Figure 1A).
Because tyrosine phosphorylation sites can act as docking sites for other proteins we asked whether any of the proteins that coimmunoprecipitate with LRP1 require Tyr 4507 phosphorylation for binding. The NPXpY4507 microdomain was prepared and immobilized on agarose beads (TNFTNPV(pY4507)ATLY). This immobilized microdomain was first used to probe for binding proteins v-Src transformed fibroblasts (Figure 1B). The results again showed a number of tyrosine-phoshphorylated proteins were specifically binding to the microdomain.
Because LRP1 is highly expressed in the mammalian brain, we sought to find out whether these interactions were only found in v-Src transformed fibroblasts, or whether they were also present in brain tissue. Figure 2 shows that a similarly large number of proteins were found to bind specifically to the phosphorylated microdomain when rat brain lysates were probed. The bands corresponding to proteins binding specifically to the NPXpY4507 microdomain were excised and the proteins were identified by LC-MSMS sequencing (Table 1). A large number of signaling proteins were found to bind specifically to this microdomain in its phosphorylated form.
To better characterize the phosphorylation dependence of the NPXY microdomains, the four microdomains were prepared and analyzed simultaneously; the NPXY4473 microdomain, NPXpY4473 microdomain, NPXY4507 microdomain, and NPXpY4507 microdomain. Anti-P.Tyr blotting of the bound proteins showed a smear of tyrosine phosphorylated proteins binding to the phosphorylated NPXY4507 and NPXY4473, but not to the unphosphorylated forms (Figure 3A). To detect all proteins interacting with these peptides the bound proteins were analyzed by silver staining (Figure 3B). It was evident that while the NPXY4473 microdomain did not seem to have many specifically bound proteins, the NPXY4507 microdomain had a large number of specific interactions. To ensure the differences observed between NPXpY4507 and NPXY4507 were not the result of anything other than phosphorylation, NPXpY4507 beads were treated with shrimp alkaline phosphatase. The resulting silver stained gels showed that removal of the phosphate from NPXY4507 yields a binding profile identical to unphosphorylated NPXY4507 (Figure 4).
Proteins bound to each NPXY motif were eluted and analyzed by in solution trypsin digest followed by 2D LC-MSMS. Proteins with at least two strong (>95 confidence) unique peptide hits are listed in Table 2–Table 4. The phosphorylated membrane distal NPXY4507 microdomain showed dozens of strong protein interactions (Table 2) whereas the phosphorylated membrane proximal NPXY4473 microdomain had very few (Table 3). Besides Grb-2 and Shc-3, both of which were also found in the NPXpY4507 sample, the only protein identified as interacting specifically with NPXpY4473 was peptidyl-prolyl cis-trans isomerase A (Table 3). The unphosphorylated NPXY4507 microdomain also bound many more proteins (Table 4) than the unphosphorylated NPXY4473, which had no protein hits that were not also found in the negative control. A few of the proteins that were observed in the in gel digest analysis of the phosphorylated NPXY4507 interacted more strongly with the unphosphorylated NPXY4507 microdomain. The most likely explanation of these results is that some dephosphorylation of the microdomain occurred during the experiment prior to in gel digestion.
To confirm the binding of several of the identified proteins to the microdomains of LRP1, bound proteins were resolved by SDS-PAGE, transferred to PVDF membranes and analyzed by immunoblotting. This experiment confirmed that PI3-kinase, Shp-2, PLC-γ, Src, CSK, Shc-3, and Grb-2, present in lysates of rat brains are able to bind to the NPXpY4507 microdomain (Figure 5). As expected from the mass spectrometry data, Shc-3 and Grb2 were found to interact with the NPXpY4473 microdomain also. GAPDH and 14-3-3γ can bind to the non phosphorylated NPXY4507. In agreement with our mass spectrometry data, CamKII seems to bind preferentially to unphosphorylated NPXY4507 microdomain with a weaker signal in the NPXpY4507 microdomain lane.
A few expected binding interactions were not observed in our mass spectrometry experiments, and these were also probed by Western blotting. Fe65 had previously been reported to interact with LRP at the NPXY motifs , but was not identified in our study. Blotting of the NPXY-bound proteins for Fe65 revealed that indeed, it could bind to both the phosphorylated and unphosphorylated NPXY4507 microdomains. We and others had previously shown that Sorting nexin 17 (Snx17) bound LRP1-CT and that this binding depended on the presence of the NPXY4473 motif [13, 19]. Figure 6 shows that recombinant HA tagged Snx17 binds both the unphosphorylated NPXY4473 and the unphosphorylated NPXY4507 motif. These results recapitulate our previous finding in v-Src transformed fibroblasts in which phosphorylation inhibited Snx17 binding, but add an additional finding that Snx17 binds to both of the NPXY motifs in LRP1-CT. It is not too surprising that Snx17 and FE65 were not observed in the proteomics analysis as we have observed low abundance of these proteins in our rodent lysates .
Given the large number of proteins that were observed to bind to the NPXY4507 microdomain, it was possible that most of the interactions were indirect and that the proteins were bound in some sort of large signaling complex. To test for direct interactions, recombinant proteins were prepared and probed for binding to the NPXY microdomains. These experiments revealed that 14-3-3γ reproducibly interacts with only the NPXY4507 microdomain in either a GST tagged or a His-tagged construct (Figure 6). All SH2 domains from PLCγ, Grb-2, PI3K, Shp-2, and full length CSK were also tested either with a GST or His8-Ubiquitin tag. These experiments revealed that certain SH2 domains can bind both phosphorylated NPXY4473 and NPXY4507 motifs, and others only bind avidly to NPXpY4507 (Figure 6). Recombinant CamKIIβ was able to bind to both phosphorylated and non-phosphorylated NPXY4507.
One of the most abundant proteins that bound to the NPXpY4507 motif microdomain was identified as Shp-2. Since previous studies have shown that Shp-2 can interact with both phosphorylated NPXY4473 and NPXY4507 of LRP , we used recombinant SH2(N) or SH2(C) domains of Shp-2 fused to ubiquitin to test for direct binding to each NPXY motif. Consistent with our previous results, the NPXpY4507 is the high-affinity binding site (Figure 6). Only on much longer exposures could a band for SH2(C) be detected in the NPXpY4473 lane (data not shown).
The LRP1 cytoplasmic domain has been implicated by association in cell signaling. We originally identified LRP1 as a Shc-binding protein , and later it was also shown to bind several PTB domain-containing proteins including disabled (Dab1) , and by yeast-two-hybrid studies JIP-1 and -2, CAPON, and a PIP4 kinase homolog among others . In addition, it interacts with the PDGF receptor-β , is phosphorylated in response to PDGFR ligand binding , and is at least sometimes localized to lipid rafts . It also binds Shp2 only in the phosphorylated form, and Snx17 only in the unphosphorylated form . These results were the impetus for the studies presented here, in which the two NPXY microdomains and their phosphorylated counterparts were separately analyzed for protein binding interactions in rodent brain lysates.
The only way to separately isolate such microdomains and to obtain them in pure form (either phosphorylated or unphosphorylated) is to prepare them synthetically. Previous work has shown that this approach is effective for identifying proteins that bind to NPXY motifs and in particular to probe the dependence of binding on tyrosine phosphorylation [22, 23]. This approach led to the identification of a variety of proteins that bind specifically to the phosphorylated or unphosphorylated NPXY microdomains of LRP1-CT. In some cases the binding depended strongly on phosphorylation state, as previously seen with peptide array studies with NPXY motifs . Based on our findings the NPXY4473 has very few detectable binding partners, while the NPXY4507 had many direct interactions both in the phosphorylated and unphosphorylated states. It should be noted that the NPXpY4473 microdomain serves as an important specificity control. The proteins observed to bind to the NPXpY4507 microdomain must be recognizing more than just the NPXpY sequence, or they would have also been observed to bind to the NPXpY4473 microdomain. Of all the signaling proteins that bound to the distal NPXpY4507 microdomain, only Shc3 and Grb2 bound also to the NPXpY4473 microdomain.
Many of the signaling proteins bound to the phosphorylated form of the LRP1 NPXY4507 microdomain including PLCγ, PI 3-kinase, Shp-1, Shp-2, Src, Fyn, ShcC and CrkL, are proteins that are recruited to the plasma membrane and are tyrosine phosphorylated in response to extracellular signals [24–31]. Aside from Shp-2, these are all previously unreported interactions, and identification of these proteins as potential LRP1 binding proteins further substantiates the model in which LRP1 functions as a signaling receptor. The finding that CamKII binds both NPXY4507 and NPXpY4507 microdomains gives yet more support for the role of LRP1 in signaling. Meaningful co-localization experiments will require introduction of mutant forms of the LRP1-CT at endogenous levels under controlled phosphorylation conditions. These are quite challenging experiments and were outside the scope of the current project.
Several independent studies have found that the adaptor protein ShcA binds to tyrosine phosphorylated LRP1. In the current study we have identified Shc-3 (ShcC) as an LRP1 binding protein in rat brain lysates. ShcC is a Shc family member that is specifically expressed in the brain [30, 32, 33]. It is well established that all Shc family members contain an amino-terminal PTB domain that binds to phosphorylated NPXY motifs. In the case of LRP1 it seems that both NPXY motifs in their phosphorylated forms can interact with ShcC.
Shp-2 is a protein-tyrosine phosphatase that contains two SH2 domains followed by a phosphatase domain. We have found that Shp-2 strongly binds to the NPXpY4507 motif and have confirmed that it also binds the phosphorylated LRP1 cytoplasmic domain . This is consistent with the fact that the amino-terminal SH2 domain of Shp-2 binds to phosphorylated tyrosine residues followed by non-polar residues at +1, +3 and +5 . This consensus is matched exactly by the sequence carboxy-terminal to NPXY4507, but not in NPXY4473 which has a much weaker apparent affinity for Shp2. The result that phosphorylated NPXY4473 can bind Shp-2 with weaker affinity agrees with previous results . The observation that Shp2 bound so strongly to the immobilized NPXpY4507 microdomain is most likely due to the fact that Shp2 has two SH2 domains, which can simultaneously engage the multiple copies of the microdomain that are available when it is immobilized. This phenomenon could also be occurring for the other proteins with multiple binding sites.
The unphosphorylated NPXY4507 microdomain specifically bound AP-2 adaptors and clathrin as well as many “house-keeping” and metabolic proteins (Table 3). These interactions were specific for the unphosphorylated microdomain and were not observed with the phosphorylated microdomain. These results show how phosphorylation is likely to affect receptor internalization and sorting. Indeed, such an effect has been previously observed in cells in which stimulation of PDGFR causes changes in LRP1 phosphorylation and subsequent internalization of both proteins .
In the course of this study nothing was found specifically interacting with NPXY4473 which was surprising. The peptide was clearly immobilized as the others based on UV follow up of the immobilization and chymotryptic digestion of the beads. Since Snx17 was previously shown to interact with this region of the LRP1-CT [13, 19], recombinant Snx17 was tested for binding with all four of the microdomains. In these experiments, specific binding to both NPXY4473 and NXPY4507 was observed. It is possible that in previous studies the NPXY4507 region was in a phosphorylated form thus unable to interact with Snx17, as we reported previously and observed again here . At the very least this demonstrates that NPXY4473 could have specific interactions in the brain, just below our detection limit in this study.
Several lines of evidence have implicated LRP1 in the development of Alzheimer’s disease. APP, the amyloid precursor protein, binds directly to LRP1 . In addition, it has been shown that Aβ constituent of amyloid plaques, can bind to α2M and ApoE; both of which are also ligands for LRP1 . Genetic evidence shows increased presence of certain alleles of LRP1, α2M and ApoE in Alzheimer’s disease patients [38–40]. Thus it is generally thought that LRP1’s involvement in Alzheimer’s disease is related to its ability to interact directly or indirectly with APP or Aβ. The proteins newly identified to bind to LRP1-CT microdomains in this study are peculiarly interesting when combined with recent findings that residues 4504–4510 of LRP affect APP processing and Aβ generation [41, 42]. Therefore these interactions at the NPXY4507 site could play a role in Aβ generation and thus the onset of Alzheimer’s disease. Recent RNAi experiments provide a functional link between Shc and APP processing . It has also been shown that APP can become phosphorylated on Y682 which can then interact with ShcA and ShcC.  although it was later found that phosphorylation of APP was not the underlying cause of its processing . The same study also showed that APP processing is affected by several tyrosine kinases including Src, and that the effects are dependent on the presence of LRP . Fluorescence lifetime imaging microscopy showed that APP and LRP are co-localized regardless of tyrosine phosphorylation of LRP . Thus, LRP’s affect on APP trafficking is probably best explained by LRP recruiting these signaling molecules to APP.
The specific interaction of NXPY4507 with 14-3-3 proteins provides another link between LRP1 and APP in the light of recent evidence that APP interacts with 14-3-3γ . Fe65 is another adaptor protein that has been shown to interact with both APP and LRP, and modulate Aβ generation . Fe65 from rat brains interacted only at the NPXY4507 in both phosphorylation states. This is in clear agreement that only the distal NPXY motif was shown to have effects on APP trafficking and Aβ generation .
The results described here identify several kinases that can bind to LRP1 including; CamKII, Src, Fyn, and CSK. Interestingly, CaM kinase II [48, 49] and Src family kinases [50–52] have been implicated previously in Tau phosphorylation. These observations suggest that LRP1 could also be involved in linking APP and Aβ to Tau hyperphosphorylation, although this conjecture will need to be studied in much more detail.
This work was supported by NIH grants RO1 AG025343 and R01 CA78629, GB was supported by the Heme and Blood Proteins Training Grant, T32-DK007233. We gratefully acknowledge Patricia Jennings for recombinant CSK and John Kuriyan for Cam kinase II.
The authors have no conflicts of interest.