Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

The TSC1 Tumor Suppressor Hamartin Interacts with Neurofilament-L and Possibly Functions as a Novel Integrator of the Neuronal Cytoskeleton

2002; Elsevier BV; Volume: 277; Issue: 46 Linguagem: Inglês

10.1074/jbc.m207211200

ISSN

1083-351X

Autores

Luciana A. Haddad, Nicole Smith, Mark Bowser, Yo Niida, Vanishree Murthy, Charo Gonzalez-Agosti, Vijaya Ramesh,

Tópico(s)

Histiocytic Disorders and Treatments

Resumo

Tuberous sclerosis complex, an autosomal dominant disease caused by mutations in either TSC1 orTSC2, is characterized by the development of hamartomas in a variety of organs. The proteins encoded by TSC1 andTSC2, hamartin and tuberin, respectively, associate with each other forming a tight complex. Here we show that hamartin binds the neurofilament light chain and it is possible to recover the hamartin-tuberin complex over the neurofilament light chain rod domain spanning amino acids 93–156 by affinity precipitation. Homologous rod domains in other intermediate filaments such as neurofilament medium chain, α-internexin, vimentin, and desmin are not able to bind hamartin. In cultured cortical neurons, hamartin and tuberin co-localize with neurofilament light chain preferentially in the proximal to central growth cone region. Interestingly, in the distal part of the growth cone hamartin overlaps with the ezrin-radixin-moesin family of actin binding proteins, and we have validated the interaction of hamartin with moesin. These results demonstrate that hamartin may anchor neuronal intermediate filaments to the actin cytoskeleton, which may be critical for some of the CNS functions of the hamartin-tuberin complex, and abolishing this through mutations in TSC1 orTSC2 may lead to certain neurological manifestations associated with the disease. Tuberous sclerosis complex, an autosomal dominant disease caused by mutations in either TSC1 orTSC2, is characterized by the development of hamartomas in a variety of organs. The proteins encoded by TSC1 andTSC2, hamartin and tuberin, respectively, associate with each other forming a tight complex. Here we show that hamartin binds the neurofilament light chain and it is possible to recover the hamartin-tuberin complex over the neurofilament light chain rod domain spanning amino acids 93–156 by affinity precipitation. Homologous rod domains in other intermediate filaments such as neurofilament medium chain, α-internexin, vimentin, and desmin are not able to bind hamartin. In cultured cortical neurons, hamartin and tuberin co-localize with neurofilament light chain preferentially in the proximal to central growth cone region. Interestingly, in the distal part of the growth cone hamartin overlaps with the ezrin-radixin-moesin family of actin binding proteins, and we have validated the interaction of hamartin with moesin. These results demonstrate that hamartin may anchor neuronal intermediate filaments to the actin cytoskeleton, which may be critical for some of the CNS functions of the hamartin-tuberin complex, and abolishing this through mutations in TSC1 orTSC2 may lead to certain neurological manifestations associated with the disease. Tuberous sclerosis complex (TSC), 1The abbreviations used are: TSC, tuberous sclerosis complex; ERM, ezrin, radixin, and moesin; NF-L, neurofilament light chain; NF-M, neurofilament medium chain; NF-H, neurofilament heavy chain; DIV, day in vitro ; aa, amino acid(s); GST, glutathione S-transferase a severe multisystem disorder with variable expression, is characterized by having misaligned dysplastic cells with normal growth (hamartias) and benign slow-growing lesions (hamartomas) commonly affecting brain, kidneys, heart, and skin (1Gomez M. Sampson J. Holtes-Whittemore V. Tuberous Sclerosis Complex. 3rd Ed. Oxford University Press, 1999Google Scholar). TSC is an autosomal dominant disease caused by mutations in tumor suppressor genes TSC1 or TSC2, which encode the proteins hamartin and tuberin, respectively (2Consortium E.C.T.S. Cell. 1993; 75: 1305-1315Abstract Full Text PDF PubMed Scopus (1517) Google Scholar, 3van Slegtenhorst M. deHoogst R. Hermans C. Nellist M. Janssen B. Verhoef S. Lindhout D. van den Ouweland A. Halley D. Young J. Burley M. Jeremiah S. Woodward K. Nahmias J. Fox M. Ekong R. Osborne J. Wolfe J. Povey S. Snell R.G. Cheadle J.P. Jones A.C. Tachataki M. Ravine D. Kwiatkowski D.J. Science. 1997; 277: 805-808Crossref PubMed Scopus (1407) Google Scholar). TSC lesions show abnormalities in cell proliferation, differentiation, and migration suggesting a role for hamartin and tuberin in these cellular processes. Hamartin contains a putative transmembrane domain at aa 127–144 and a coiled-coil domain spanning aa 719–998 (3van Slegtenhorst M. deHoogst R. Hermans C. Nellist M. Janssen B. Verhoef S. Lindhout D. van den Ouweland A. Halley D. Young J. Burley M. Jeremiah S. Woodward K. Nahmias J. Fox M. Ekong R. Osborne J. Wolfe J. Povey S. Snell R.G. Cheadle J.P. Jones A.C. Tachataki M. Ravine D. Kwiatkowski D.J. Science. 1997; 277: 805-808Crossref PubMed Scopus (1407) Google Scholar). Tuberin shows limited homology to the catalytic domain of Rap1 GTPase-activating protein at its C-terminal domain with reported weak activities for Rap1a and Rab5 GTPase-activating protein in vitro(4Wienecke R. Konig A. DeClue J.E. J. Biol. Chem. 1995; 270: 16409-16414Abstract Full Text Full Text PDF PubMed Scopus (350) Google Scholar, 5Xiao G.H. Shoarinejad F. Jin F. Golemis E.A. Yeung R.S. J. Biol. Chem. 1997; 272: 6097-6100Abstract Full Text Full Text PDF PubMed Scopus (321) Google Scholar). Hamartin and tuberin associate in vivo indicating that they participate in a common biochemical pathway (6Plank T.L. Yeung R.S. Henske E.P. Cancer Res. 1998; 58: 4766-4770PubMed Google Scholar, 7van Slegtenhorst M. Nellist M. Nagelkerken B. Cheadle J. Snell R. van den Ouweland A. Reuser A. Sampson J. Halley D. van der Sluijs P. Hum. Mol. Genet. 1998; 7: 1053-1057Crossref PubMed Scopus (496) Google Scholar, 8Murthy V. Haddad L.A. Smith N. Pinney D. Tyszkowski R. Brown D. Ramesh V. Am. J. Physiol. 2000; 278: F737-F746PubMed Google Scholar, 9Gutmann D.H. Zhang Y. Hasbani M.J. Goldberg M.P. Plank T.L. Henske E.P. Acta Neuropathol. (Berlin). 2000; 99: 223-230Crossref PubMed Scopus (49) Google Scholar, 10Catania M.G. Johnson M.W. Liau L.M. Kremen T.J. deVellis J.S. Vinters H.V. J. Neurosci. Res. 2001; 63: 276-283Crossref PubMed Scopus (20) Google Scholar). This association is mediated through aa 302–430 of hamartin and aa 1–418 of tuberin (11Hodges A.K., Li, S. Maynard J. Parry L. Braverman R. Cheadle J.P. DeClue J.E. Sampson J.R. Hum. Mol. Genet. 2001; 10: 2899-2905Crossref PubMed Scopus (100) Google Scholar), and formation of this complex prevents tuberin ubiquitination (12Benvenuto G., Li, S. Brown S.J. Braverman R. Vass W.C. Cheadle J.P. Halley D. Sampson J.R. Wienecke R. DeClue J.E. Oncogene. 2001; 19: 6306-6316Crossref Scopus (209) Google Scholar) and hamartin self-oligomerization (13Nellist M. van Slegtenhorst M. Goedbloed M. van den Ouweland A.M. Halley D.J. van der Sluijs P. J. Biol. Chem. 1999; 274: 35647-35652Abstract Full Text Full Text PDF PubMed Scopus (159) Google Scholar). Some pathological mutations in TSC1 and TSC2 abolish the interaction between these proteins (11Hodges A.K., Li, S. Maynard J. Parry L. Braverman R. Cheadle J.P. DeClue J.E. Sampson J.R. Hum. Mol. Genet. 2001; 10: 2899-2905Crossref PubMed Scopus (100) Google Scholar, 14Nellist M. Verhaaf B. Goedbloed M.A. Reuser A.J. van den Ouweland A.M. Halley D.J. Hum. Mol. Genet. 2001; 10: 2889-2898Crossref PubMed Scopus (88) Google Scholar). Rodent models of Tsc1 and Tsc2 develop predominantly renal cyst adenomas with liver hemangiomas observed only in mouse models (15Eker R. Diagn. Histopathol. 1981; 4: 99-110PubMed Google Scholar, 16Yeung R.S. Xiao G.H. Jin F. Lee W.C. Testa J.R. Knudson A.G. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 11413-11416Crossref PubMed Scopus (273) Google Scholar, 17Kobayashi T. Hirayama Y. Kobayashi E. Kubo Y. Hino O. Nat. Genet. 1995; 9: 218Crossref Scopus (293) Google Scholar, 18Kobayashi T. Minowa O. Kuno J. Mitani H. Hino O. Noda T. Cancer Res. 1999; 59: 1206-1211PubMed Google Scholar, 19Onda H. Lueck A. Marks P.W. Warren H.B. Kwiatkowski D.J. J. Clin. Invest. 1999; 104: 687-695Crossref PubMed Scopus (323) Google Scholar, 20Kobayashi T. Minowa O. Sugitani Y. Takai S. Mitani H. Kobayashi E. Noda T. Hino O. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 8762-8767Crossref PubMed Scopus (192) Google Scholar, 21Kwiatkowski D.J. Zhang H. Bandura J.L. Heiberger K.M. Glogauer M. el-Hashemite N. Onda H. Hum. Mol. Genet. 2002; 11: 525-534Crossref PubMed Scopus (541) Google Scholar). Mutations in either Drosophila Tsc1 or Tsc2 reveal identical phenotypes with increased cell and organ size (22Ito N. Rubin G. Cell. 1999; 96: 529-539Abstract Full Text Full Text PDF PubMed Scopus (226) Google Scholar, 23Tapon N. Ito N. Dickson B.J. Treisman J.E. Hariharan I.K. Cell. 2001; 105: 345-355Abstract Full Text Full Text PDF PubMed Scopus (454) Google Scholar, 24Potter C.J. Huang H. Xu T. Cell. 2001; 105: 357-368Abstract Full Text Full Text PDF PubMed Scopus (446) Google Scholar, 25Gao X. Pan D. Genes Dev. 2001; 15: 1383-1392Crossref PubMed Scopus (390) Google Scholar). In this system, the two proteins function together to restrict growth and proliferation and appear to interact with growth regulators in the insulin signaling pathway (23Tapon N. Ito N. Dickson B.J. Treisman J.E. Hariharan I.K. Cell. 2001; 105: 345-355Abstract Full Text Full Text PDF PubMed Scopus (454) Google Scholar, 24Potter C.J. Huang H. Xu T. Cell. 2001; 105: 357-368Abstract Full Text Full Text PDF PubMed Scopus (446) Google Scholar, 25Gao X. Pan D. Genes Dev. 2001; 15: 1383-1392Crossref PubMed Scopus (390) Google Scholar). Several experimental approaches clearly establish that hamartin and tuberin function together as a complex. It is likely that, depending on the cell or tissue type, additional proteins exist in this complex, and, therefore, identification of binding partners for hamartin and tuberin is essential. Recently, hamartin was shown to interact with the ERM (ezrin, radixin, and moesin) family of actin-binding proteins and is thought to have a role in cell adhesion and activation of the small GTP-binding protein Rho (26Lamb R.F. Roy C. Diefenbach T.J. Vinters H.V. Johnson M.W. Jay D.G. Hall A. Nat. Cell Biol. 2000; 2: 281-287Crossref PubMed Scopus (278) Google Scholar). Using the yeast two-hybrid system, we have isolated the neurofilament-light chain (NF-L) and have confirmed that NF-L is a physiologically relevant interactor for hamartin. Studies performedin vitro indicate that NF-L is part of the hamartin-tuberin complex. Hamartin and tuberin co-localize with NF-L and the ERM proteins in cultured primary cortical neurons with enrichment in neuronal growth cones. The association of hamartin with NF-L and ERM proteins in the proximal and distal part of the growth cone, respectively, suggests that this tumor suppressor may function as an integrator of the neuronal intermediate filament and the actin cytoskeletal network. cDNA was synthesized from human fetal cerebellum mRNA using the Superscript preamplification system for first-strand cDNA synthesis (Life Technologies, Inc., Rockville, MD) followed by reverse transcription-PCR using specific primers and Pfu DNA polymerase (Stratagene, La Jolla, CA). The bait used in the yeast two-hybrid system encoded hamartin amino acids 674 through the stop codon and was amplified with primers 5HAMC (5′-CGGCGTCGACCTTGGACCCACTTTGGAGGC-3′) and 3HAMC (5′-CGGCGTCGACCTTTAGCTCTCTTCATGATGAGT-3′) and cloned into the vector pGBT9 (Clontech, Palo Alto, CA). TheNF-L insert, isolated from the cDNA library, was subcloned into the EcoRI site of the pGEX4T-1 and pGAD424 vectors. Full-length TSC1 in pcDNA3 has been described previously (8Murthy V. Haddad L.A. Smith N. Pinney D. Tyszkowski R. Brown D. Ramesh V. Am. J. Physiol. 2000; 278: F737-F746PubMed Google Scholar). Full-length NF-L was cloned into pcDNA3 by PCR amplification of an Image (National Institute of Health) clone obtained from Research Genetics (#631980) using primers NFL1S (5′-GCGGGATCCATGAGTTCCTTCAGCTACGAGCCG-3′) and NFL3A (5′-CAGGCGGCCGCTCAATCTTTCTTCTTAGCTGCTTG-3′). NF-L domain constructs were obtained by reverse transcription-PCR from fetal cerebellum cDNA and cloned directionally into the pGEX4T-1 vector. The amino acid numbers defining head, rod, and tail domains were taken from human NF-L Genpept sequence accession number P07196. Primers used to amplify the various domains of NF-L were as follows: the head domain (aa 1–92) using NFL1S and NFL5A (5′-GCGCTCGAGCTGCGCCTTCTCCTGCGTGCG-3′), the first 64 residues of the rod domain (aa 93–156) using NFL2S (5′-GCGGGATCCCTCCAGGACCTCAATGACCGC-3′) and NFL2A (5′-ATAGCGGCCGCCTACTCGTTGGTGGTGGCATCTTC-3′), the last 241 residues of the rod domain (aa 157–397) using NFL3S (5′-GCGGGATCCAAGCAGGCGCTCCAGGGCGAG-3′) and NFL7A (5′-GCGGTCGACGGTCTCCTCGCCTTCCAAGAG-3′). NF-M construct (aa 1–165) was amplified with primers NFM1S (5′-GGAATTCATGAGCTACACGTTGGACTCG-3′) and NFM1A (5′-CGTGTCGACCTGAGCCTTCTCGTGGTTCACG-3′), and an α-internexin construct (aa 1–165) was made with primers AINT1S (5′-GGAATTCATGAGCTTCGGCTCGGAGCAC-3′) and AINT1A (5′-CGTGTCGACCTGCGAGCCGAGCTGGCG-3′). NF-M and α-internexin constructs were directionally cloned into the vector pGAD424 (Clontech, Palo Alto, CA). A desmin construct (aa 22–177) was obtained by sub-cloning full-length mouse desmin cDNA, kindly provided by Dr. R. Kothary from Ottawa General Hospital Research Institute, into SmaI and SalI sites of pGAD424 vector. Full-length 3′ myc-tagged vimentin in pcDNA3 was a kind gift from Dr. R. D. Goldman, Northwestern University Medical School. All clones obtained by PCR were completely sequenced. Hamartin polyclonal antibodies HF3 and HF6 have been described earlier (8Murthy V. Haddad L.A. Smith N. Pinney D. Tyszkowski R. Brown D. Ramesh V. Am. J. Physiol. 2000; 278: F737-F746PubMed Google Scholar, 27Murthy V. Stemmer-Rachamimov A.O. Haddad L.A. Roy J.E. Cutone A.N. Beauchamp R.L. Smith N. Louis D.N. Ramesh V. Acta Neuropathol. (Berlin). 2001; 101: 202-210Crossref PubMed Scopus (26) Google Scholar). A polyclonal antibody C20 (Santa Cruz Biotechnology, Santa Cruz, CA) was employed for tuberin. For NF-L either monoclonal antibody (clone DA2, Zymed Laboratories Inc., San Francisco, CA) or a polyclonal antibody (Chemicon, Temecula, CA) was used. Anti-vimentin monoclonal antibody was obtained from Dako (clone V9, Glostrup, Denmark), and the anti-ERM 13H9 monoclonal supernatant was a generous gift of Dr. F. Solomon at the Massachusetts Institute of Technology. Donkey anti-mouse conjugated to Cy3, donkey anti-rabbit conjugated to Cy2 (Jackson ImmunoResearch Laboratories, West Grove, PA), and donkey anti-rabbit Alexa488 (Molecular Probes, Eugene, OR) were used as secondary antibodies. Hamartin C-terminal bait (aa 674–1164) was used to screen a human fetal brain cDNA library made in fusion with the GAL4 activation domain in the vector pGAD10 (Clontech, Palo Alto, CA) using the yeast strain Hf7c containing the reporter genes his3 and lacZ. One million co-transformants were screened for the two reporter genes on plates containing media with amino acid selection. Forhis3 gene transcription activation, 20 mm3-amino-1,2,4-triazole (Sigma, St. Louis, MO) was added to decrease the possibility of false-positive results. Yeast two-hybrid clones pVA3 (murine p53), pTD1 (SV40 large T-antigen), and pLAM5 (5′-end of lamin C) were used as controls (Clontech, Palo Alto, CA). Interactor controls WASP and WIP were kindly provided by Dr. N. Ramesh (Children's Hospital, Boston, MA). Dystrobrevin-α exons 14 through 16 were amplified by PCR from a full-length clone in pCR2.1, kindly provided by Dr. L. M. Kunkel (Children's Hospital, Boston, MA), and were subcloned into either pGBT9 or pGAD424 vectors. Hamartin bait deletion constructs were obtained by digestion of the bait with PstI (Fig. 3 A, #2) orEcoRI (Fig. 3 A, #5), followed by vector purification, ligation, and transformation. Other constructs were generated by restriction digestion (Fig. 3 A,#3 with EcoRI, #4 withSmaI, and #6 with BamHI andBglII), followed by gel purification of respective bands and subcloning into pGBT9 vector. Cos-7 cells, co-transfected with full-length hamartin and either full-length NF-L or vimentin, were lysed in 150 mm NaCl, 50 mmTris-HCl (pH 8.0), 1% Triton-X-100, complete protease inhibitor mixture (Roche Molecular Biochemicals, Indianapolis, IN), 2 mm EDTA, and phosphatase inhibitors (1 mmsodium orthovanadate, 50 mm sodium fluoride). Lysates were precleared with normal rabbit serum and a mixture of protein A- and protein G-agarose beads (Roche Molecular Biochemicals). Precleared lysates were subjected to immunoprecipitation with anti-hamartin antibody HF6, and beads were washed extensively with lysis buffer containing decreasing concentrations of Triton X-100 (from 1% to 0.25%). After the final wash, beads were resuspended in 1× sample buffer (33% glycerol, 6.7% SDS, 330 mm dithiothreitol), resolved by 6% SDS-PAGE, transferred to nitrocellulose membrane, and subjected to immunoblot analysis using anti-NF-L antibody or anti-vimentin antibody. Immunoreactive bands were visualized by ECL (Amersham Biosciences, Piscataway, NJ). HeLa cells were used as an endogenous source of hamartin. Cells were lysed as described above, and the lysate was incubated with 600 pmol of GST-NF-L or GST immobilized on glutathione-Sepharose 4B beads. The beads were washed extensively with phosphate-buffered saline containing Pefabloc, resuspended in 1× sample buffer, subjected to 6% SDS-PAGE, and immunoblotted with anti-hamartin antibody HF6. Neuron cultures were established using published protocols (28Lin J.W., Ju, W. Foster K. Lee S.H. Ahmadian G. Wyszynski M. Wang Y.T. Sheng M. Nat. Neurosci. 2000; 3: 1282-1290Crossref PubMed Scopus (408) Google Scholar). Briefly, brain cortexes of Sprague-Dawley rat embryos (E20) were dissected, submitted to enzymatic dissociation of cells with trypsin (Invitrogen, Rockville, MD) at room temperature, followed by mechanical dissociation in Hanks' balanced salt solution (Invitrogen). 5 × 104 cells were plated in each well of a six-well plate containing poly-d-lysine/laminin-coated coverslips (BD-Bioscience, Bedford, MA) and 2 ml of media. Neurobasal media (Invitrogen) supplemented with 2% B-27 (Invitrogen) was used. Media was initially changed on the second day in vitro (DIV) and subsequently once every week. Cells were fixed in 4% paraformaldehyde at 37 °C for 30 min on the 10th or 16th DIV. Fixed neurons were permeabilized with 0.1% Nonidet P-40 in phosphate-buffered saline for 15 min, blocked for 1 h in 10% normal goat serum, incubated with anti-hamartin polyclonal antibody HF6 (1:10) and either anti-NF-L monoclonal antibody (1:100) or anti-ERM monoclonal antibody (1:100) for 1 h, then incubated again with Alexa488-conjugated goat anti-rabbit antibody (1:1500) and Cy3-conjugated donkey anti-mouse (1:1000) for 30 min. For tuberin staining, C20 was used at a dilution of 1:100 followed by Cy2-conjugated donkey anti-rabbit (1:1000) antibody. Coverslips were mounted on Gelvatol mounting media (polyvinyl alcohol resin grade 205, Air Products and Chemicals, Lehigh Valley, PA). Images were captured using LSM 5 Pascal software coupled to a Zeiss LSM Pascal Vario 2 RGB confocal system. To identify proteins that associate with hamartin, the C-terminal half of the protein (aa 674–1164) containing the large coiled-coil domain was used as a bait to screen a human fetal brain cDNA library using the GAL4 yeast two-hybrid system. Screening was performed on a total of 1 × 106 clones with 15 positive clones identified representing three groups. One of the cDNA groups, which comprised 13 positive clones, encoded the head and part of the rod domain of NF-L spanning amino acids 8–156. The hamartin bait as well as the NF-L clone did not independently activate transcription of the reporter genes his3 andlacZ. The specificity of this interaction was tested using various unrelated baits, which did not activate transcription after co-transformation with either the TSC1 bait or NF-L (Fig.1). Among these controls, dystrobrevin-α in particular represents an unrelated protein containing a coiled-coil domain and did not show an interaction with either hamartin or NF-L, excluding the possibility that the interaction observed between NF-L and hamartin is nonspecifically mediated through the coiled-coil domains of these proteins. To further characterize the interaction between hamartin and NF-L, we tested the ability of hamartin to bind GST-NF-L in affinity precipitation assays. As shown in Fig.2 A, endogenous hamartin from HeLa cells was able to bind to a GST-NF-L fusion protein encoding aa 8–156 and did not bind to GST alone employed as a negative control. In addition, immunoprecipitation assays performed on Cos-7 cells, which had been co-transfected with full-length TSC1 and full-length NF-L cDNA clones, revealed that NF-L co-immunoprecipitates with hamartin (Fig. 2 B), confirming that NF-L is a physiologically relevant binding partner for the tumor suppressor hamartin. Coiled-coil domains are known to mediate protein-protein interactions (29Burkhard P. Stetefeld J. Strelkov S. Trends Cell Biol. 2001; 11: 82-88Abstract Full Text Full Text PDF PubMed Scopus (850) Google Scholar). To define the domain(s) responsible for the interaction of hamartin with NF-L, further deletion constructs within the TSC1 bait region were generated and employed in yeast two-hybrid assays to verify the activation of the reporter genes his3 andlacZ after co-transformation with NF-L. Deletions either within, upstream, or downstream of the coiled-coil domain completely abolished the activation of transcription (Fig.3 A). In addition, the coiled-coil domain alone could not activate transcription of the reporter genes. Thus we could not further define the domain in hamartin that is necessary for interacting with NF-L. It is possible that conformational changes caused by deletions mask the NF-L binding site in hamartin in this assay. Similar to other intermediate filaments, NF-L is composed of head (aa 1–92), rod (aa 93–397), and tail (aa 398–544) domains. The region of NF-L that was isolated in the yeast two-hybrid screen (aa 8–156) comprised most of the head and a part of the rod domains. To further analyze hamartin binding to NF-L, we expressed the head and rod regions of NF-L as GST fusion proteins and performed affinity precipitation assays as described above. A majority of the interaction between hamartin and NF-L was seen with the rod domain spanning aa 93–156 (Fig. 3 B). We also observed that the rest of the NF-L rod domain (aa 157–397) was not necessary for this interaction (data not shown). The NF-L rod domain is composed of heptad repeats forming four regions of coiled-coil domains (coils 1A, 1B, 2A, and 2B) connected by three linker regions. The NF-L rod domain segment that binds hamartin contains coil 1A, the first linker, and the first 19 residues of coil 1B. Hamartin and tuberin strongly associate with each other and may exist as part of a complex that contains additional components (13Nellist M. van Slegtenhorst M. Goedbloed M. van den Ouweland A.M. Halley D.J. van der Sluijs P. J. Biol. Chem. 1999; 274: 35647-35652Abstract Full Text Full Text PDF PubMed Scopus (159) Google Scholar). Hamartin and tuberin exhibit strong similarities in their expression pattern and subcellular distribution, although some differences do exist (8Murthy V. Haddad L.A. Smith N. Pinney D. Tyszkowski R. Brown D. Ramesh V. Am. J. Physiol. 2000; 278: F737-F746PubMed Google Scholar, 9Gutmann D.H. Zhang Y. Hasbani M.J. Goldberg M.P. Plank T.L. Henske E.P. Acta Neuropathol. (Berlin). 2000; 99: 223-230Crossref PubMed Scopus (49) Google Scholar) raising the possibility that protein interactors identified for each protein may or may not be part of the hamartin-tuberin complex. We raised the question whether tuberin along with hamartin could be retained in affinity precipitation assays performed with GST-NF-L. The NF-L domains spanning amino acids 8–156, as well as amino acids 93–156, are capable of retaining tuberin presumably through hamartin (Fig. 3 B) suggesting that NF-L may exist as part of the hamartin-tuberin complex in neurons. The rod domain of NF-L is about 50% homologous to rod domains of other intermediate filaments. Because the association of hamartin with NF-L is mostly mediated through the rod domain, we examined whether additional intermediate filaments with homologous rod domains are capable of interacting with hamartin. The intermediate filaments with the highest homology to NF-L include vimentin, desmin, glial fibrillary acidic protein, neurofilaments-medium (NF-M) and -heavy (NF-H) chains, α-internexin, peripherin, and cytokeratin 8. Vimentin is present in cells from mesenchymal origin and is highly expressed in fibroblasts, which are a major cellular component of TSC-associated skin lesions such as facial angiofibromas, shagreen patch, and subungual fibromas. Similarly, desmin, a muscle-specific intermediate filament, is of relevance for TSC cardiac rhabdomyoma. We therefore initially examined whether vimentin and desmin were capable of binding hamartin. Co-transfection of full-length TSC1 with full-length VIM, followed by immunoprecipitation with anti-hamartin antibody showed that vimentin does not co-immunoprecipitate with hamartin (Fig.4). Co-immunoprecipitations could not be carried out successfully on cells co-transfected with TSC1and DES, because exogenously expressed desmin consistently precipitated and bound nonspecifically to Sepharose beads employed in the assays. We therefore used the yeast two-hybrid system to examine the interaction of hamartin with desmin as well as two members of the class IV intermediate filaments, NF-M and α-internexin, in which NF-L is a member. With the exception of NF-L, the transcriptional activation of the reporter genes was not observed when TSC1 was co-transformed with any of the other intermediate filaments tested (Table I). These findings indicate that, among the intermediate filaments, NF-L is a distinct binding partner of hamartin with the interaction likely to be restricted to neurons.Table IInteraction of hamartin with other intermediate filaments assessed by yeast-two hybrid assayBinding domainActivation domainReporter gene activationhis3lacZBaitNF-L++BaitNF-M−−Baitα-Internexin−−BaitDesmin−−The respective rod domains of intermediate filament family members NF-M, α-internexin, or desmin did not activate the transcription of the reporter genes his3 and lacZ when co-transformed with TSC1 bait. Open table in a new tab The respective rod domains of intermediate filament family members NF-M, α-internexin, or desmin did not activate the transcription of the reporter genes his3 and lacZ when co-transformed with TSC1 bait. To examine whether hamartin and NF-L are present in similar subcellular locations within neurons, we performed indirect immunofluorescence microscopy on cultured cortical neurons obtained from rats. We observed an enrichment of hamartin staining in many neuronal growth cones with a weaker punctate pattern along the neurites. As expected, NF-L was present in neurites and growth cones, and the two proteins co-localized primarily in the proximal to central regions of growth cones (Fig. 5,A–C). We also examined the co-localization of tuberin with NF-L and observed that tuberin in fact overlaps with NF-L in the proximal to central regions of growth cones (Fig. 5,D–F) further supporting that hamartin together with tuberin exist in a complex with NF-L in neuronal growth cones. A recent study reported hamartin as a physiological binding partner for the ERM family of proteins. In this study, co-immunoprecipitation of endogenous ezrin with hamartin as well as co-localization of ERM proteins with hamartin were documented in primary human umbilical vein endothelial cells (26Lamb R.F. Roy C. Diefenbach T.J. Vinters H.V. Johnson M.W. Jay D.G. Hall A. Nat. Cell Biol. 2000; 2: 281-287Crossref PubMed Scopus (278) Google Scholar). However, it remained unclear whether the interactions between these proteins occurred in other cell types. Interestingly, several studies have demonstrated that the ERM proteins are enriched in growth cones of various neurons (30Goslin K. Birgbauer E. Banker G. Solomon F. J. Cell Biol. 1989; 109: 1621-1623Crossref PubMed Scopus (87) Google Scholar, 31Birgbauer E. Dinsmore J. Winckler B. Lander A.a. Solomon F. J. Neurosci. Res. 1991; 30: 232-241Crossref PubMed Scopus (37) Google Scholar, 32Gonzalez-Agosti C. Solomon F. Cell Motil. Cytoskeleton. 1996; 34: 122-136Crossref PubMed Scopus (28) Google Scholar). Radixin and moesin are implicated to play a key role in generating and maintaining the normal structural and functional organization of growth cones (32Gonzalez-Agosti C. Solomon F. Cell Motil. Cytoskeleton. 1996; 34: 122-136Crossref PubMed Scopus (28) Google Scholar, 33Paglini G. Kunda P. Quiroga S. Kosik K. Caceres A. J. Cell Biol. 1998; 143: 443-455Crossref PubMed Scopus (141) Google Scholar, 34Castelo L. Jay D. Mol. Biol. Cell. 1999; 10: 1511-1520Crossref PubMed Scopus (58) Google Scholar). We therefore examined whether hamartin and ERM proteins co-localize in neurons and observed that these two proteins indeed exist together in neuronal growth cones. Hamartin co-localizes with the ERM proteins in the central to distal part of growth cones (Fig.6 A, panels A–D). Similarly, tuberin also exhibits a strong co-localization with the ERM proteins in the distal region of growth cones (data not shown). As described previously ERM proteins are seen in the actin-rich peripheral regions of growth cones where hamartin is not enriched (panels C and D). Furthermore, we confirmed that interaction occurs between hamartin and ERM proteins by affinity precipitation assays employing GST fusion proteins expressing the N terminus of either moesin or the related NF2 tumor suppressor protein merlin and endogenous hamartin from HeLa cells. Hamartin was able to bind GST-moesin and, to a lesser extent, GST-merlin but not GST alone, in affinity precipitation assays (Fig. 6 B), which is in agreement with an earlier study (26Lamb R.F. Roy C. Diefenbach T.J. Vinters H.V. Johnson M.W. Jay D.G. Hall A. Nat. Cell Biol. 2000; 2: 281-287Crossref PubMed Scopus (278) Google Scholar). Taken together, these data suggest that in neurons, hamartin may integrate actin and NF-L via its direct interaction with the ERM proteins and NF-L. TSC is a developmental disorder characterized by abnormalities in cell migration, differentiation, and proliferation, and it is manifested by the presence of hamartomas and hamartias in various organs. Central nervous system lesions include cortical tubers, subependymal giant cell astrocytomas, subependymal nodules, and neuroglial heterotopias, which result in a variety of neurological manifestations, including mental retardation and seizures. The CNS lesions contain cells that can be classified as neurons and astrocytes as well as cells that are not easily classified. The neurons in these lesions appear to be primitive or aberrant (1Gomez M. Sampson J. Holtes-Whittemore V. Tuberous Sclerosis Complex. 3rd Ed. Oxford University Press, 1999Google Scholar). We have observed the expression of tuberin and hamartin to decline progressively after birth in many tissues; however, in the CNS their expression levels did not decrease with age. Throughout development and in adult CNS strong expression of both proteins was seen in neurons (27Murthy V. Stemmer-Rachamimov A.O. Haddad L.A. Roy J.E. Cutone A.N. Beauchamp R.L. Smith N. Louis D.N. Ramesh V. Acta Neuropathol. (Berlin). 2001; 101: 202-210Crossref PubMed Scopus (26) Google Scholar). Genetic analyses of TSC lesions demonstrate that a majority of renal angiomyolipomas seen in TSC are associated with loss of the wild-type allele as expected, whereas less frequent loss of heterozygosity is seen in cortical tubers (35Henske E.P. Scheithauer B.W. Short M.P. Wollmann R. Nahmias J. Hornigold N. van Slegtenhorst M. Welsh C.T. Kwiatkowski D.J. Am. J. Hum. Genet. 1996; 59: 400-406PubMed Google Scholar, 36Au K.S. Hebert A.A. Roach E.S. Northrup H. Am. J. Hum. Genet. 1999; 65: 1790-1795Abstract Full Text Full Text PDF PubMed Scopus (64) Google Scholar, 37Niida Y. Stemmer-Rachamimov A.O. Logrip M. Tapon D. Perez R. Kwiatkowski D.J. Sims K. MacCollin M. Louis D.N. Ramesh V. Am. J. Hum. Genet. 2001; 69: 493-503Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar). Careful analysis for subtle variations in TSC genes, molecular analysis of micro-dissected balloon cells from cortical tubers, as well promoter methylation studies performed in our laboratory failed to reveal the second somatic mutation in some of the clonally derived brain lesions. This suggests the possibility that somatic mutations may not be necessary and that other mechanisms may contribute to abnormal growth particularly in the CNS lesions (37Niida Y. Stemmer-Rachamimov A.O. Logrip M. Tapon D. Perez R. Kwiatkowski D.J. Sims K. MacCollin M. Louis D.N. Ramesh V. Am. J. Hum. Genet. 2001; 69: 493-503Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar). All these observations suggest that the hamartin-tuberin complex is likely to have distinct functions in the CNS. Neurofilaments are the major intermediate filaments in mature neurons and consist of three polypeptides, the light (NF-L), medium (NF-M), and heavy (NF-H) chains. NF-L homodimerizes and interacts with NF-M and NF-H, which extend side projections in the neurofilament meshwork. In this study we have shown that NF-L associates with hamartin through the first half of its rod domain. Other intermediate filaments such as vimentin, desmin, NF-M, and α-internexin, which exhibit about 50% homology in the first two coil regions of their respective rod domains with NF-L, do not show association with hamartin. Therefore, hamartin association with intermediate filaments is probably restricted to neurons and is specific to NF-L. Furthermore, in affinity precipitation assays using the first half of the NF-L rod domain, it is possible to recover tuberin bound to hamartin, suggesting that the three proteins are probably part of the same molecular complex in neurons. Co-localization of hamartin and tuberin with NF-L is enriched particularly in the proximal and central regions of neuronal growth cones implicating this as a potential site of interaction for these proteins. An earlier study found hamartin to be associated with the ERM family of actin-binding proteins in vitro and in vivo in human umbilical vein endothelial cells, and this interaction was shown to be required for activation of Rho by serum or lysophosphatidic acid (26Lamb R.F. Roy C. Diefenbach T.J. Vinters H.V. Johnson M.W. Jay D.G. Hall A. Nat. Cell Biol. 2000; 2: 281-287Crossref PubMed Scopus (278) Google Scholar). We have confirmed that hamartin interacts with moesin, and in cultured primary neurons hamartin and tuberin reveal strong co-localization with ERMs in the distal region of growth cones. Interestingly, among the ERM family members, only radixin and moesin are concentrated in neuronal growth cones, and ezrin is restricted mainly to the cell body (32Gonzalez-Agosti C. Solomon F. Cell Motil. Cytoskeleton. 1996; 34: 122-136Crossref PubMed Scopus (28) Google Scholar, 33Paglini G. Kunda P. Quiroga S. Kosik K. Caceres A. J. Cell Biol. 1998; 143: 443-455Crossref PubMed Scopus (141) Google Scholar). Moesin and radixin have been clearly shown to regulate key aspects of growth cone development and maintenance modulating neurite formation and polarity (32Gonzalez-Agosti C. Solomon F. Cell Motil. Cytoskeleton. 1996; 34: 122-136Crossref PubMed Scopus (28) Google Scholar, 33Paglini G. Kunda P. Quiroga S. Kosik K. Caceres A. J. Cell Biol. 1998; 143: 443-455Crossref PubMed Scopus (141) Google Scholar, 34Castelo L. Jay D. Mol. Biol. Cell. 1999; 10: 1511-1520Crossref PubMed Scopus (58) Google Scholar). In neurons expressing reduced levels of radixin and moesin, growth cone alterations are accompanied by a dramatic disorganization of F-actin (33Paglini G. Kunda P. Quiroga S. Kosik K. Caceres A. J. Cell Biol. 1998; 143: 443-455Crossref PubMed Scopus (141) Google Scholar). It is likely that the association of hamartin-tuberin complex with ERM proteins is essential for growth cone maintenance and motility. The ERM proteins are highly regulated, actin-binding proteins and have emerged as key molecules in linking F-actin to specific membrane proteins, especially in cell surface structures (38Bretscher A. Chambers D. Nguyen R. Reczek D. Annu. Rev. Cell Dev. Biol. 2000; 16: 113-143Crossref PubMed Scopus (327) Google Scholar). The interaction of hamartin with ERM proteins is reported to be required upstream of Rho for lysophosphatidic acid-induced assembly of focal adhesions and actin stress fibers (26Lamb R.F. Roy C. Diefenbach T.J. Vinters H.V. Johnson M.W. Jay D.G. Hall A. Nat. Cell Biol. 2000; 2: 281-287Crossref PubMed Scopus (278) Google Scholar). Mouse embryonic fibroblast lines derived from Tsc1 null mice display a difference in actin and focal adhesion dynamics supporting a function for hamartin in the actin-based cytoskeleton (21Kwiatkowski D.J. Zhang H. Bandura J.L. Heiberger K.M. Glogauer M. el-Hashemite N. Onda H. Hum. Mol. Genet. 2002; 11: 525-534Crossref PubMed Scopus (541) Google Scholar). Our results documenting the interaction of hamartin with NF-L imply that hamartin could function as a novel integrator of the neuronal cytoskeleton (Fig.7). In this context it is worthwhile to note that several members of the plakin protein family serve as anchors that connect cytoskeleton filaments to junctional complexes and as linker proteins that cross-link various cytoskeletal elements. Interestingly, plakins are involved in both inherited and autoimmune diseases that affect the skin, heart, neuronal tissue, and skeletal muscle (39Leung C.L. Liem R.K. Parry D.A. Green K.J. J. Cell Sci. 2001; 114: 3409-3410PubMed Google Scholar). Thus the present study potentially adds hamartin to the growing list of versatile cytoskeletal linker proteins that are implicated in human disease. The CNS lesions of TSC are thought to arise from a defect in glial-neuronal differentiation and migration during embryonic development. It is believed that the least differentiated cells found in TSC brains fail to reach the post-mitotic stage and thus fail to migrate from the periventricular germinal matrix, developing into subependymal giant cell astrocytomas and subependymal nodules. More differentiated, but still dysplastic cells succeed in migrating to the cortical plate but fail to become incorporated into the normal cortical cytoarchitecture, forming cortical tubers or neuroglial heterotopias (40Caviness Jr., V.S. Takahashi T. Ann. N. Y. Acad. Sci. 1991; 615: 187-195Crossref PubMed Scopus (12) Google Scholar, 41Huttenlocher P.R. Wollmann R.L. Ann. N. Y. Acad. Sci. 1991; 615: 140-148Crossref PubMed Scopus (33) Google Scholar, 42Hirose T. Scheithauer B.W. Lopes M.B.S. Gerber H.A. Altermatt H.J. Hukee M.J. van den Berg S.R. Charlesworth J.C. Acta Neuropathol. 1995; 90: 387-399Crossref PubMed Scopus (132) Google Scholar). Therefore, it is tempting to speculate that the association of hamartin with NF-L and ERM proteins might play an essential role in neuronal migration and that abolishing this interaction by mutations in the TSC genes could explain certain neuronal differentiation and migration defects observed in TSC brains. We thank Roberta L. Beauchamp for her help in preparation of this manuscript and members of our laboratory for helpful comments on the manuscript.

Referência(s)