The Motor Protein Kinesin-1 Links Neurofibromin and Merlin in a Common Cellular Pathway of Neurofibromatosis

Mutations in either of the two tumor suppressor genes NF1 (neurofibromin) and NF2 (merlin) result in Neurofibromatosis, a condition predisposing individuals to developing a variety of benign and malignant tumors of the central and peripheral nervous systems. Here we report the identification of two distinct NF1-containing complexes one in the soluble and the other in the particulate fraction of Hela extract. We show that the soluble NF1 complex delineates a large holo-NF1 complex (2 MDa) encompassing the components of a smaller particulate core-NF1 complex (400 kDa). Purification of the core-NF1 complex followed by mass spectrometric analysis revealed the motor protein, kinesin-1 heavy chain (HsuKHC/KIF5B), as a catalytic subunit of both NF-1-containing complexes. Importantly, although NF1 and NF2 are not in a stable association, NF2 is also a component of a distinct kinesin 1-containing complex. These results point to kinesin-1 as a common denominator between NF1 and NF2. We have initiated the biochemical characterization of NF1- and NF2-containing complexes from mammalian cells. These experiments led to the identification of two distinct NF1-containing complexes. We show that while NF1 purified from the soluble fraction reside in a large complex of approximately 2MDa, NF1 in the particulate fraction is a component of a smaller complex of 400 kDa. To gain insights into the functions of these complexes, we used a combination of conventional and affinity chromatography to purify the smaller core-NF1 complex from the particulate fraction. We have identified the catalytic subunit of this complex as the motor protein kinesin-1. Importantly, we show that although NF1 and NF2 proteins are not stably associated, NF2 is also a component of a distinct kinesin-1-containing complex.


Introduction
Neurofibromatosis type 1 (NF1) or von Recklinghausen disease is a common neurological genetic disease that affects 1 in 3500 individuals world wide (1,2).
Mutations in the human NF1 gene lead to a common neurocutaneous disorder characterized by begnin tumors (neurofibromas and giomas), abnormal distribution of melanocytes (cafe-au-lait spots), and malignant tumors, including neurofibrosarcomas, pheochromocytomas, rhabdomyosarcomas, astrocytomas, and juvenil myeloid leukemias. NF1 patients also exhibit cognitive deficits and other symptoms unrelated to cancer, affecting neural crest-derived tissues outside of the nervous system reflective of a role for NF1 in developmental control (2,3).
NF1 encodes a large protein of 2818 amino acids designated neurofibromin (4)(5)(6). The protein is highly conserved from yeast to human. Neurofibromin is expressed ubiquitously in human, with the highest expression in adult peripheral and central nervous systems (7). The protein contains a GAP-related domain (GRD) that shares homology to known GTPase-activating proteins (GAPs). NF1-GRD has been shown to act as a GAP for the Ras family of small GTPases (8)(9)(10). Thus, several studies suggest that the tumor-suppressor activity of neurofibromin depend on its ability to negatively regulate the ras-mediated signalling pathway that regulate cell growth and differenciation in a variety of cell types (11). Neurofibromatosis type 2 (NF2) is an autosomal dominant disorder implicated in the development of sporadic schwannomas, meningiomas, ependymonas, and astrocytomas (12)(13)(14). NF2 gene encodes a 595-amino acid protein termed merlin belonging to the ERM (Ezrin, Radixin and Moesin) family that link the actin cytoskeleton to cell surface glycoproteins (15 4 We have initiated the biochemical characterization of NF1-and NF2-containing complexes from mammalian cells. These experiments led to the identification of two distinct NF1-containing complexes. We show that while NF1 purified from the soluble fraction reside in a large complex of approximately 2MDa, NF1 in the particulate fraction is a component of a smaller complex of 400 kDa. To gain insights into the functions of these complexes, we used a combination of conventional and affinity chromatography to purify the smaller core-NF1 complex from the particulate fraction. We have identified the catalytic subunit of this complex as the motor protein kinesin-1. Importantly, we show that although NF1 and NF2 proteins are not stably associated, NF2 is also a component of a distinct kinesin-1-containing complex.

Western blot analysis
For detection of the NF1 protein, affinity-purified polyclonal antibodies sc-68 (NF1GRP-D) raised against synthetic peptides corresponding to the carboxy terminal domain of the human NF1 gene product were used (Santa cruz Biotechnology). For detection of the NF2 protein, affinity-purified polyclonal antibodies sc331 (A-19) and sc332 (C-18) raised against synthetic peptides corresponding to the N-terminus and the C-terminus of the NF2 protein were used (Santa cruz Biotechnology). For detection of the kinesin-1 protein, one polyclonal antibody raised against the insert 1 region of the head of human uKHC (KIF5B) (gift from Vale R.'s lab) and two monoclonal antibodies H1 and H2 raised against bovine brain kinesin (Chemicon International, Inc) were used.

Protein Identification using LC-MS/MS
Gel bands were excised from Colloidal Coomassie stained gels, bands were destained, alkylated with iodoacetamide, and digested using modified Trypsin (Promega) for 16 hr at 37°C essentially. A portion of the extracted peptides were loaded to a nanocapillary reverse-phase 75••m column terminating in a nanospray 15••m tip (New Objective) packed with Porous R2 resin (Applied Biosystems). The nanocolumn was directly coupled to a ThermoFinnigan LCQ quadrupole ion trap mass spectrometer and peptides were eluted into the mass spectrometer using an acetic acid -acetonitrile gradient. Data was acquired using triple play mode to automatically obtain peptide masses, peptide charge states, and MS/MS spectra. The resulting data were searched against the non-redundant NCBI using TurboSEQUEST Browser to identify proteins.

Immunoaffinity-Purification of the NF1-containing Complex
Anti-NF1 antibodies (500 µg, C-terminal, Santa cruz Biotechnology sc-68) were cross-linked to Protein A-Sepharose (1 mL, Repligen) using standard techniques for affinity purification. The heparin fractions from HeLa cell and Calf brain were incubated

Identification of two distinct NF1-containing complexes : a soluble holo-NF1 and a particulate core-NF1
To gain insight into the biochemical properties of NF1 we fractionated HeLa soluble and particulate extracts (Fig.1a, (Fig.1a). This difference was further demonstrated once the two fractions were analyzed by gel filtration. The soluble NF1 complex eluted at an apparent molecular mass of 2 MDa (Fig. 1b, fraction 18 was the peak). In contrast, the particulate NF1 derived from either the 0.3 or 0.5 M KCl elution of P11 chromatography exhibited an elution profile consistent with a complex of 400 kDa (Fig. 1c). These sizes are estimated relative to globular protein standards, and assume that the complexes are themselves globular. If the complexes are elongated they would have a smaller mass. These results suggest the existence of two distinct NF1containing complexes, a large NF1-containing complex enriched in the soluble fraction and a smaller complex in the particulate fraction. It is possible that the NF1 complex in the particulate fraction is in association with the outer nuclear membrane. The association of NF1 with the particulate fraction was previously detected by immunofluorescence staining in neurons following detergent extraction (17).

Purification of the core-NF1 complex from HeLa and Calf brain particulate fraction revealed the presence of Kinesin-1
To define the polypeptide composition of NF1-containing complexes, we isolated the smaller NF1 complex. NF1 was purified from both HeLa cells and calf brain particulate extract using a combination of conventional and affinity chromatography following the scheme presented in Fig. 2a and b. Analysis of α-NF1 affinity eluate by SDS-PAGE and colloidal blue staining revealed the association of NF1 with three polypeptides of 150, 9 band as NF1 and identified the 110 kDa band as the motor protein kinesin-1 heavy chain (HsuKHC/KIF5B) (18). Western blot analysis confirmed the association of kinesin-1 and NF1 in both HeLa cells and calf brain ( Fig. 2a and b). Furthermore, immunoprecipitation experiments using three different kinesin-1 antibodies demonstrated a stable association of NF1 and kinesin-1 from particulate extract of both calf brain and HeLa cells (Fig. 3a and data not shown). Interestingly, NF1 derived from the soluble fraction is also in a stable association with kinesin-1 (Fig. 3b). We further confirm this association by immunoprecipitating endogenous NF1 with ectopically expressed FLAG-KIF5B using anti-FLAG antibodies followed by elution of bound material with Flag peptide (Fig. 3c).
Together, these results demonstrate the stable association of NF1 and kinesin-1 in both soluble and particulate fractions.

Kinesin-1 is also associated with a distinct soluble NF2-containing complex
Mutations in the NF2 gene also causes a similar disease manifestation as that of NF1 (12)(13)(14). We therefore asked whether NF1 and NF2 are stably associated.
Immunoprecipitation experiments using either anti-NF1 or anti-NF2 antibodies did not support an association between NF1 and NF2 proteins from the soluble fraction of HeLa cells (NF2 was not detected in particulate fraction, Fig. 4a). We then asked whether NF2 is also a stable component of kinesin-1 containing complexes. Immunoprecipitation experiments using both the N-and the C-terminal anti-NF2 antibodies revealed the stable association of NF2 and kinesin-1 (Fig. 4b). We confirmed the association by immunoprecipitating endogenous NF2 with ectopically expressed FLAG-KIF5B using anti-FLAG antibodies followed by elution of bound material with Flag peptide (Fig.4c).
Indeed, fractionation of soluble HeLa extract by gel filtration revealed the coelution of 10 NF2 and kinesin-1 in a large complex (Fig. 4d, fractions 16 through 20), although a fraction of NF2 was also detected at a smaller molecular mass (fractions 30-36). These results indicate that although NF2 is a component of a large kinesin-1-containing complex, this complex seems distinct from the NF1-containing complex. However, since both NF1 and NF2 are components of a large kinesin-1-containing complex, it is possible that a fraction of NF2 and NF1 are stably associated but that antibodies to each protein disrupts this association.

Discusssion
Kinesin-1 is a tetramer consisting of two 120-kDa heavy chains (KHC) and two 64-kDa light chains (KLC). Kinesin-1 heavy chain HsuKHC/KIF5B belongs to the kinesin protein superfamily (KIF) (19). This family has been shown to transport protein complexes, organelles and mRNA to specific destinations in an ATP-and microtubuledependent manner (20,21). Furthermore, some members of this family are also involved in chromosomal and spindle movements during mitosis and meiosis (22,23). Although, a stable association of kinesin-1 and NF1 or NF2 was an unexpected finding, it is consistent with previous microscopy studies indicating the sub-cellular localization of NF1 and NF2 with the cytoskeleton (17,(24)(25)(26)(27). Taken together, the association of NF1 and NF2 with the motor protein kinesin-1 suggests a role for these proteins in microtubule-mediated intracellular signal transduction pathways.
Recent studies have shown that the axonal transport of amyloid precursor protein (APP) in neurons are mediated by the direct biochemical interaction between APP and KLC, the light chain subunit of kinesin-1 (28,29). Considering that microtubule-11 dependent trafficking requires at least two entities -a cargo-bound receptor and the motor proteins -the authors proposed that APP may be a membrane cargo receptor for a kinesin-mediated axonal transport of β-secretase and presenilin-1 (29). In analogy with this model, the association between kinesin-1 and NF1 or NF2 might reflect a new function for these proteins in transport of vesicular cargoes within cells. Although NF1 has several known functions, including Ras GTPase-activating protein activity (8)(9)(10)      and α-TRAP220 (control) followed by Western analysis using α-NF1(sc-68) and α-KIF5B antibodies. Calf brain particulate extract was used as the input. (b) Western analysis using α-KIF5B antibodies following immunoprecipitation using the affinity-