Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

Coordinated Folding and Association of the LIN-2, -7 (L27) Domain

2002; Elsevier BV; Volume: 277; Issue: 38 Linguagem: Inglês

10.1074/jbc.m205856200

ISSN

1083-351X

Autores

Baruch Z. Harris, Shivkumar Venkatasubrahmanyam, Wendell A. Lim,

Tópico(s)

Polysaccharides and Plant Cell Walls

Resumo

LIN-2, -7 (L27) homology domains are putative protein-protein interaction modules found in several scaffold proteins involved in the assembly of polarized cell-signaling structures. These specific interaction pairs are well conserved across metazoan species, from worms to man. We have expressed and purified L27 domains from multiple species and find that certain domains from proteins such as Caenorhabditis elegans LIN-2 and LIN-7 can specifically heterodimerize. Biophysical analysis of interacting L27 domains demonstrates that the domains interact with a 1:1 stoichiometry. Circular dichroism studies reveal that the domains appear to function as an obligate heterodimer; individually the domains are largely unfolded, but when associated they show a significant increase in helicity, as well as a cooperative unfolding transition. These novel obligate interacting pairs are likely to play a key role in regulating the organization of signaling proteins at polarized cell structures. LIN-2, -7 (L27) homology domains are putative protein-protein interaction modules found in several scaffold proteins involved in the assembly of polarized cell-signaling structures. These specific interaction pairs are well conserved across metazoan species, from worms to man. We have expressed and purified L27 domains from multiple species and find that certain domains from proteins such as Caenorhabditis elegans LIN-2 and LIN-7 can specifically heterodimerize. Biophysical analysis of interacting L27 domains demonstrates that the domains interact with a 1:1 stoichiometry. Circular dichroism studies reveal that the domains appear to function as an obligate heterodimer; individually the domains are largely unfolded, but when associated they show a significant increase in helicity, as well as a cooperative unfolding transition. These novel obligate interacting pairs are likely to play a key role in regulating the organization of signaling proteins at polarized cell structures. In multicellular organisms, many cells are highly polarized to allow for specific cell-cell communication. An essential feature of polarized cells is the organization of signaling proteins into distinct membrane domains (1Bredt D.S. Cell. 1998; 94: 691-694Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar, 2Dotti C.G. Simons K. Cell. 1990; 62: 63-72Abstract Full Text PDF PubMed Scopus (358) Google Scholar). For example, both neurons and epithelial cells actively maintain regions of the cell surface to which receptors, downstream signaling molecules, and structural proteins are selectively targeted. To accomplish this function, cells often use multivalent scaffolding proteins that bring together diverse signaling molecules into supramolecular complexes. These scaffolding proteins generally have multiple protein-protein recognition modules that play an important role in the assembly and dynamic regulation of signaling complexes (3Fanning A.S. Anderson J.M. Curr. Opin. Cell Biol. 1999; 11: 432-439Crossref PubMed Scopus (275) Google Scholar). Multivalent scaffolding molecules assemble such structures by participating in two general classes of binding interactions. First, they bind to non-scaffolding molecules, usually signaling proteins such as receptors or downstream effectors. These interactions are often mediated by well characterized modular recognition domains, including PSD-95·Dlg·ZO-1 homology (PDZ) 1The abbreviations used are: PDZ, PSD-95·Dlg·ZO-1 homology; GST, glutathioneS-transferase; NTA, nitrilotriacetic acid; DTT, dithiothreitol; PBS, phosphate-buffered saline. domains (4Harris B.Z. Lim W.A. J. Cell Sci. 2001; 114: 3219-3231Crossref PubMed Google Scholar), which bind target proteins through short peptide motifs. Second, scaffold proteins can form higher order multiprotein networks through homo- or hetero-oligomerization with other scaffolding molecules (5Sheng M. Kim E. J. Cell Sci. 2000; 113: 1851-1856Crossref PubMed Google Scholar). Far less is understood about the mechanism of such oligomerization interactions, although they are also essential for proper complex assembly. A notable example of a multiprotein scaffold network is the LIN-2/LIN-7/LIN-10 heterotrimeric complex (6Rongo C. Cytokine Growth Factor Rev. 2001; 12: 349-359Crossref PubMed Scopus (13) Google Scholar), which is responsible for receptor targeting in polarized cells in a variety of metazoan organisms (7Craven S.E. Bredt D.S. Cell. 1998; 93: 495-498Abstract Full Text Full Text PDF PubMed Scopus (429) Google Scholar, 8Hoskins R. Hajnal A.F. Harp S.A. Kim S.K. Development. 1996; 122: 97-111Crossref PubMed Google Scholar, 9Kaech S.M. Whitfield C.W. Kim S.K. Cell. 1998; 94: 761-771Abstract Full Text Full Text PDF PubMed Scopus (316) Google Scholar, 10Simske J.S. Kaech S.M. Harp S.A. Kim S.K. Cell. 1996; 85: 195-204Abstract Full Text Full Text PDF PubMed Scopus (234) Google Scholar, 11Whitfield C.W. Bénard C. Barnes T. Hekimi S. Kim S.K. Mol. Biol. Cell. 1999; 10: 2087-2100Crossref PubMed Scopus (91) Google Scholar, 12Rongo C. Whitfield C.W. Rodal A. Kim S.K. Kaplan J.M. Cell. 1998; 94: 751-759Abstract Full Text Full Text PDF PubMed Scopus (224) Google Scholar, 13Butz S. Okamoto M. Südhof T.C. Cell. 1998; 94: 773-782Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 14Cohen A.R. Woods D.F. Marfatia S.M. Walther Z. Chishti A.H. Anderson J.M. Wood D.F. J. Cell Biol. 1998; 142: 129-138Crossref PubMed Scopus (322) Google Scholar, 15Ide N. Hirao K. Hata Y. Takeuchi M. Irie M. Yao I. Deguchi M. Toyoda A. Nishioka H. Mizoguchi A. Takai Y. Biochem. Biophys. Res. Commun. 1998; 243: 634-638Crossref PubMed Scopus (7) Google Scholar, 16Irie M. Hata Y. Deguchi M. Ide N. Hirao K. Yao I. Nishioka H. Takai Y. Oncogene. 1999; 18: 2811-2817Crossref PubMed Scopus (43) Google Scholar, 17Jo K. Derin R., Li, M. Bredt D.S. J. Neurosci. 1999; 19: 4189-4199Crossref PubMed Google Scholar). In this heterotrimeric complex, each scaffold protein has one or more PDZ domains that specifically interact with other partner/signaling proteins (Fig.1). In addition, the three proteins interact with one another independently of their PDZ domains. LIN-2 homologs are members of a larger family of scaffold proteins referred to as membrane-associated guanylatekinases (MAGUKs), which contain PDZ, SH3, and guanylate kinase-like domains (3Fanning A.S. Anderson J.M. Curr. Opin. Cell Biol. 1999; 11: 432-439Crossref PubMed Scopus (275) Google Scholar, 7Craven S.E. Bredt D.S. Cell. 1998; 93: 495-498Abstract Full Text Full Text PDF PubMed Scopus (429) Google Scholar); LIN-7 and LIN-10 comprise distinct families of scaffolding molecules. The LIN-2·LIN-7·LIN-10 complex was first identified inCaenorhabditis elegans. Mutation of these proteins leads to aberrant localization of the epidermal growth factor receptor-like tyrosine kinase receptor LET-23 in vulval precursor epithelial cells; instead of being properly targeted to the basolateral surface where it is positioned to receive its extracellular signal, LET-23 is found at the apical membrane. Mislocalization of the receptor leads to a loss in LET-23-dependent signaling and defects in vulval induction (8Hoskins R. Hajnal A.F. Harp S.A. Kim S.K. Development. 1996; 122: 97-111Crossref PubMed Google Scholar, 9Kaech S.M. Whitfield C.W. Kim S.K. Cell. 1998; 94: 761-771Abstract Full Text Full Text PDF PubMed Scopus (316) Google Scholar, 10Simske J.S. Kaech S.M. Harp S.A. Kim S.K. Cell. 1996; 85: 195-204Abstract Full Text Full Text PDF PubMed Scopus (234) Google Scholar, 11Whitfield C.W. Bénard C. Barnes T. Hekimi S. Kim S.K. Mol. Biol. Cell. 1999; 10: 2087-2100Crossref PubMed Scopus (91) Google Scholar). The LIN-2, LIN-7, and LIN-10 proteins have been found to form a trimeric complex. In addition, the PDZ domain from LIN-7 interacts with the C terminus of LET-23 (9Kaech S.M. Whitfield C.W. Kim S.K. Cell. 1998; 94: 761-771Abstract Full Text Full Text PDF PubMed Scopus (316) Google Scholar). Although the LIN-2·LIN-7·LIN-10 complex is probably required simply for localization of LET-23 and therefore only indirectly involved in signaling, loss of any member of the complex has a similar phenotype with respect to vulval induction as loss of the receptor (10Simske J.S. Kaech S.M. Harp S.A. Kim S.K. Cell. 1996; 85: 195-204Abstract Full Text Full Text PDF PubMed Scopus (234) Google Scholar). LIN-10 has also been observed to play a crucial role in neuronal receptor targeting in C. elegans(12Rongo C. Whitfield C.W. Rodal A. Kim S.K. Kaplan J.M. Cell. 1998; 94: 751-759Abstract Full Text Full Text PDF PubMed Scopus (224) Google Scholar), suggesting that the LIN-2, LIN-7, and LIN-10 proteins function in polarized targeting in a variety of cell types. Homologs of LIN-2, LIN-7, and LIN-10 have also been identified in mammals; they are called CASK/PALS, VELI/MALS, and MINT/X11, respectively (13Butz S. Okamoto M. Südhof T.C. Cell. 1998; 94: 773-782Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 14Cohen A.R. Woods D.F. Marfatia S.M. Walther Z. Chishti A.H. Anderson J.M. Wood D.F. J. Cell Biol. 1998; 142: 129-138Crossref PubMed Scopus (322) Google Scholar, 15Ide N. Hirao K. Hata Y. Takeuchi M. Irie M. Yao I. Deguchi M. Toyoda A. Nishioka H. Mizoguchi A. Takai Y. Biochem. Biophys. Res. Commun. 1998; 243: 634-638Crossref PubMed Scopus (7) Google Scholar, 16Irie M. Hata Y. Deguchi M. Ide N. Hirao K. Yao I. Nishioka H. Takai Y. Oncogene. 1999; 18: 2811-2817Crossref PubMed Scopus (43) Google Scholar, 17Jo K. Derin R., Li, M. Bredt D.S. J. Neurosci. 1999; 19: 4189-4199Crossref PubMed Google Scholar, 18Tseng T.C. Marfatia S.M. Bryant P.J. Pack S. Zhuang Z. O'Brien J.E. Lin L. Hanada T. Chishti A.H. Biochim. Biophys. Acta. 2001; 1518: 249-259Crossref PubMed Scopus (33) Google Scholar, 19Hata Y. Butz S. Sudhof T.C. J. Neurosci. 1996; 16: 2488-2494Crossref PubMed Google Scholar, 20Kamberov E. Makarova O. Roh M. Liu A. Karnak D. Straight S. Margolis B. J. Biol. Chem. 2000; 275: 11425-11431Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar). These proteins have been observed in most neuronal and epithelial cell types and form a variety of dimeric or trimeric complexes (13Butz S. Okamoto M. Südhof T.C. Cell. 1998; 94: 773-782Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 21Biederer T. Sudhof T.C. J. Biol. Chem. 2000; 275: 39803-39806Abstract Full Text Full Text PDF PubMed Scopus (198) Google Scholar, 22Borg J.P. Straight S.W. Kaech S.M. de Taddéo-Borg M. Kroon D.E. Karnak D. Turner R.S. Kim S.K. Margolis B. J. Biol. Chem. 1998; 273: 31633-31636Abstract Full Text Full Text PDF PubMed Scopus (170) Google Scholar, 23Borg J.P. Lopez-Figueroa M.O. de Taddeo-Borg M. Kroon D.E. Turner R.S. Watson S.J. Margolis B. J. Neurosci. 1999; 19: 1307-1316Crossref PubMed Google Scholar, 24Setou M. Nakagawa T. Seog D.H. Hirokawa N. Science. 2000; 288: 1796-1802Crossref PubMed Scopus (607) Google Scholar, 25Straight S.W. Karnak D. Borg J.P. Kamberov E. Dare H. Margolis B. Wade J.B. Am. J. Physiol. 2000; 278: F464-F475Crossref PubMed Google Scholar). In these complexes, PDZ domains in each of the scaffold proteins are used to bind various signaling or transport proteins (9Kaech S.M. Whitfield C.W. Kim S.K. Cell. 1998; 94: 761-771Abstract Full Text Full Text PDF PubMed Scopus (316) Google Scholar, 12Rongo C. Whitfield C.W. Rodal A. Kim S.K. Kaplan J.M. Cell. 1998; 94: 751-759Abstract Full Text Full Text PDF PubMed Scopus (224) Google Scholar, 13Butz S. Okamoto M. Südhof T.C. Cell. 1998; 94: 773-782Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 21Biederer T. Sudhof T.C. J. Biol. Chem. 2000; 275: 39803-39806Abstract Full Text Full Text PDF PubMed Scopus (198) Google Scholar, 24Setou M. Nakagawa T. Seog D.H. Hirokawa N. Science. 2000; 288: 1796-1802Crossref PubMed Scopus (607) Google Scholar, 26Olsen O. Liu H. Wade J.B. Merot J. Welling P.A. Am. J. Physiol. 2002; 282: C183-C195Crossref PubMed Scopus (64) Google Scholar). In mammalian dendrites, the PDZ domain from VELI/MALS binds the C-terminal tail of theN-methyl-d-aspartate receptor subunit NR2B in a fashion analogous to the binding of the LIN-7 PDZ domain to the C-terminal tail of LET-23. In addition, the PDZ domain from MINT1/X11α binds to the kinesin superfamily motor protein KIF17. The CASK/PALS protein tethers the other two scaffold proteins together, just as LIN-2 tethers LIN-7 and LIN-10 (24Setou M. Nakagawa T. Seog D.H. Hirokawa N. Science. 2000; 288: 1796-1802Crossref PubMed Scopus (607) Google Scholar). It has been proposed that the entire heterotrimeric complex is involved in transport of vesicles containing N-methyl-d-aspartate receptors to the postsynaptic density (24Setou M. Nakagawa T. Seog D.H. Hirokawa N. Science. 2000; 288: 1796-1802Crossref PubMed Scopus (607) Google Scholar). More recent evidence indicates an important role for mammalian LIN-2·LIN-7·LIN-10 proteins in receptor localization and endosomal sorting in epithelial cells (25Straight S.W. Karnak D. Borg J.P. Kamberov E. Dare H. Margolis B. Wade J.B. Am. J. Physiol. 2000; 278: F464-F475Crossref PubMed Google Scholar, 27Straight S. Chen L. Karnak D. Margolis B. Mol. Biol. Cell. 2001; 12: 1329-1340Crossref PubMed Scopus (42) Google Scholar). Little is known about the interactions used to assemble the trimeric scaffold network. Recently, Doerks et al. (28Doerks T. Bork P. Kamberov E. Makarova O. Muecke S. Margolis B. Trends Biochem. Sci. 2000; 25: 317-318Abstract Full Text Full Text PDF PubMed Scopus (46) Google Scholar) identified regions of significant sequence homology within many members of the LIN-2 and LIN-7 family, which they termed "L27" domains, forLIN-2/LIN-7 dimerization. These putative domains are all within regions of LIN-2 and LIN-7 (or homologs) that have previously been implicated in association of the two scaffold proteins (13Butz S. Okamoto M. Südhof T.C. Cell. 1998; 94: 773-782Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 22Borg J.P. Straight S.W. Kaech S.M. de Taddéo-Borg M. Kroon D.E. Karnak D. Turner R.S. Kim S.K. Margolis B. J. Biol. Chem. 1998; 273: 31633-31636Abstract Full Text Full Text PDF PubMed Scopus (170) Google Scholar, 23Borg J.P. Lopez-Figueroa M.O. de Taddeo-Borg M. Kroon D.E. Turner R.S. Watson S.J. Margolis B. J. Neurosci. 1999; 19: 1307-1316Crossref PubMed Google Scholar, 27Straight S. Chen L. Karnak D. Margolis B. Mol. Biol. Cell. 2001; 12: 1329-1340Crossref PubMed Scopus (42) Google Scholar). Moreover, the L27 homology region appears to be the minimal region required for LIN-2/LIN-7 association. Therefore, the L27 domain is postulated to be a protein-protein interaction module critical for the assembly of this higher order scaffolding structure (29Lee S. Fan S. Makarova O. Straight S. Margolis B. Mol. Cell. Biol. 2002; 22: 1778-1791Crossref PubMed Scopus (123) Google Scholar, 30Roh M.H. Makarova O. Liu C.J. Shin Lee S. Laurinec S. Goyal M. Wiggins R. Margolis B. J. Cell Biol. 2002; 157: 161-172Crossref PubMed Scopus (299) Google Scholar). Here we report the expression, purification, and biochemical characterization of several L27 domains from both C. elegansand mammals. We observe a specific interaction between domains from the LIN-2/CASK family and domains from the LIN-7/VELI/MALS family. This interaction functions both within species and across species. We also observe that L27 domains assemble into a 1:1 heterodimer with binding affinities in the low micromolar range. Finally, we find that isolated L27 domains appear to be relatively unstructured but adopt a stable folded structure upon heterodimeric association. Thus, the domains act as obligate heterodimers. Using the L27 sequence profile of Doerks et al. (28Doerks T. Bork P. Kamberov E. Makarova O. Muecke S. Margolis B. Trends Biochem. Sci. 2000; 25: 317-318Abstract Full Text Full Text PDF PubMed Scopus (46) Google Scholar), we identified putative L27 domains in the following proteins; from C. elegans, LIN-2a and LIN-7 (8Hoskins R. Hajnal A.F. Harp S.A. Kim S.K. Development. 1996; 122: 97-111Crossref PubMed Google Scholar, 10Simske J.S. Kaech S.M. Harp S.A. Kim S.K. Cell. 1996; 85: 195-204Abstract Full Text Full Text PDF PubMed Scopus (234) Google Scholar); from Rattus norvegicus, the LIN-7 homolog LIN-7-BA (16Irie M. Hata Y. Deguchi M. Ide N. Hirao K. Yao I. Nishioka H. Takai Y. Oncogene. 1999; 18: 2811-2817Crossref PubMed Scopus (43) Google Scholar); and from Homo sapiens, the LIN-2 homolog CASK (14Cohen A.R. Woods D.F. Marfatia S.M. Walther Z. Chishti A.H. Anderson J.M. Wood D.F. J. Cell Biol. 1998; 142: 129-138Crossref PubMed Scopus (322) Google Scholar). Here we employ the following nomenclature to describe each L27 domain: the abbreviated name of the species of origin (C.e., R.n., or H.s.), followed by the name of the orthologous protein from C. elegans (LIN-7, LIN-2) and N or C denoting N- or C- terminal, for cases in which a protein has two domains. For example, the H. sapiens C-terminal CASK L27 domain is denoted by "H.s. LIN-2C." DNA regions encoding the putative L27 domains were cloned by PCR from cDNA libraries and expressed as fusion proteins to both N-terminal hexahistidine (His6) and glutathioneS-transferase (GST) tags for purification and biochemical characterization. Regions cloned comprised the following amino acids:C. elegans LIN-7 115–180 (C.e. LIN-7), C. elegans LIN-2a 363–430 (C.e. LIN-2N) and 423–487 (C.e. LIN-2C),R. norvegicus LIN-7-BA 10–75 (R.n. LIN-7), and H. sapiens CASK 335–401 (H.s. LIN-2N) and 394–460 (H.s. LIN-2C). The cloned regions include five extra amino acids added to both the N and C termini of the putative homology regions as a safety margin to allow for expression of stable and soluble domains. PCR products were ligated into expression vectors pBH4 (31Hillier B.J. Christopherson K.S. Prehoda K.E. Bredt D.S. Lim W.A. Science. 1999; 284: 812-815Crossref PubMed Scopus (470) Google Scholar) for His6 fusions or pGEX4T1 for glutathione S-transferase fusions. All constructs were verified by DNA sequencing on both strands. Protein constructs were expressed in the E. coli strain BL21 pLYS-S (Invitrogen). Cultures were grown at 25 °C because of increased protein solubility at this temperature and induced with 1 mm isopropyl-1-thio-β-d-galactopyranoside. After induction for 6 h, cultures were harvested by centrifugation, resuspended in ∼25 ml of 50 mm Tris-Cl, pH 7.5, 300 mm NaCl per original liter of culture, and subjected to freeze-thaw treatment. Cell suspensions were lysed by sonication and cleared by centrifugation at 20,000 ×g. Three forms of recombinant proteins were used in this study. First, crude fractionated lysates containing His6 fusions for use as "prey" in GST-pulldown assays were prepared by expression as described above, followed by incubation with Ni2+-NTA Superflow™ resin (Qiagen) for ∼30 min. The resin was extensively washed with 50 mm Tris-Cl, pH 7.5, 300 mm NaCl, and bound protein eluted with 250 mm imidazole in 50 mm sodium phosphate buffer, 300 mm NaCl, pH 8.0 to >50% purity. His6 fusion constructs purified in this manner were then dialyzed extensively against 20 mm HEPES pH 7.0, 1 mm DTT and used immediately for pulldown assays. Second, GST fusions for use as "bait" in GST-pulldown assays were incubated with glutathione-agarose resin (Sigma) for 30 min at 4 °C followed by extensive washing in 1× phosphate-buffered saline (PBS), 2 mm DTT. Third, untagged proteins used in quantitative biochemical experiments were generated by thrombin cleavage of the GST fusion proteins. Cleavage was performed by adding thrombin (Hematologic Technologies, Inc.) at a concentration of ∼6.5 μg/ml to a slurry of fusion protein bound to glutathione-agarose. Cleavage reactions were rotated overnight at 4 °C, and cleaved protein was separated from the resin by filtration. Cleaved L27 domains were further purified by ion exchange chromatography in 20 mm Tris-Cl, 1 mm DTT, 0–500 mm NaCl, pH 7.0 on an HR10/10 Source Q Sepharose column (Amersham Biosciences) to >99% purity and stored in 20 mm HEPES pH 7.0, 1 mmDTT. GST fusions of L27 domains for use as bait were expressed and purified as described above. 25 μl of GST-L27 domain slurry was incubated with ∼150 μl of eluted His6 L27 domain in a final buffer of 20 mmHEPES pH 7.0, 10 mm NaCl, 2 mm DTT for 30 min, followed by three washes with 1× PBS containing 0.1% Triton X-100, then two washes in 1× PBS with no detergent. Bound proteins were eluted from the resin in 2× SDS sample buffer and subsequently visualized with SDS-PAGE followed by Western blotting with anti-hexahistidine antibodies. Purified H.s. LIN-2C and C.e. LIN-2C domains were conjugated to a fluorescent dansyl moiety by covalent linkage to endogenous cysteines (corresponding to residues 464 and 473 in C. elegans LIN-2a and residue 412 inH. sapiens CASK). 100 μm protein was incubated with 2 mm dansyl iodoacetamide (Molecular Probes) in 20 mm HEPES pH 7.0 overnight at 4 °C. Labeled proteins were purified over a PD-10 gel filtration column (Amersham Biosciences) and redialyzed into storage buffer (20 mm HEPES pH 7.0, 1 mm DTT). Fluorescence was assayed in a 1 × 1 cm stirred cell cuvette in a buffer containing 50 mm HEPES pH 7.5, 50 mm NaCl at 20 °C using a Photon Technology Inc. Quantum Master fluorometer. Changes in fluorescence intensity were recorded, and data were fit to a bimolecular binding equation using Profit 5.1.0 (Quantum Soft). For conditions in which [L]≫[P], we fit the data as described previously (32Harris B.Z. Hillier B.J. Lim W.A. Biochemistry. 2001; 40: 5921-5930Crossref PubMed Scopus (107) Google Scholar). For conditions in which [L] approached [P], we used the analysis described by Nguyenet al. (33Nguyen J.T. Turck C.W. Cohen F.E. Zuckermann R.N. Lim W.A. Science. 1998; 282: 2088-2092Crossref PubMed Scopus (277) Google Scholar). All binding experiments were repeated 3–4 times. Binding of dansylated 2C domains to C.e. LIN-7 was verified independently by GST pulldown assay and showed no detectable difference from unlabeled 2C domains. The R.n. LIN-7 L27 domain was too insoluble to purify in quantities sufficient to perform binding assays. To confirm that the dansyl modifications did not alter binding affinity, we also performed competition binding assays (32Harris B.Z. Hillier B.J. Lim W.A. Biochemistry. 2001; 40: 5921-5930Crossref PubMed Scopus (107) Google Scholar) in which the unlabeled proteins were used to compete off identical labeled proteins. Quantitative analysis of these data revealed no effect of the label (data not shown). Analytical ultracentrifugation experiments were performed on a Beckman Optima XL-A analytical ultracentrifuge with dual beam detection at 280 or 504 nm and with an An-60Ti rotor having sectored cells composed of sapphire- and carbon-impregnated Epon. Data were parsed with Win ReEdit version 0.999.0023 and analyzed with WinNonLin version 1.060 (David Yphantis, University of Connecticut) and Profit 5.1.0 (Quantum Soft). Single species experiments were fit to Equation 1. ATOT=A0eςr22−r022Equation 1 In Equation 1 A TOT is the total absorbance measured by the spectrophotometer, A 0is the absorbance constant, r is the radius, andr 0 is the minimum radius (i.e. radius of meniscus). ς denotes the following expression shown as Equation2. ς=Mω2(1−v̄ρ)RTEquation 2 In Equation 2 M is the molecular mass (g/mol), ω is the angular velocity of the rotor (rads/s), v̄ is the partial specific volume (cm3/g), ρ is the solvent density, R is the gas constant, and T is the absolute temperature in degrees Kelvin. Two-species data were fit using Equation 3. ATOT=A1eς1r22−r022+A2eς2r22−r022Equation 3 In Equation 3 variables are defined as above, andA TOT is treated as the sum of two distributions;A 1 and ς1 correspond to monomer values, and A 2 and ς2 correspond to heterodimer values. Experiments with L27 monomers alone were checked using WinNonLin; data fit in all cases to a single-species model with monomeric molecular weight, and no higher aggregation states could be fit. Analysis of heterodimers with one labeled species was performed as described (34Mullins R.D. Kelleher J.F., Xu, J. Pollard T.D. Mol. Biol. Cell. 1998; 9: 841-852Crossref PubMed Scopus (75) Google Scholar, 35Bubb M.R. Lewis M.S. Korn E.D. J. Biol. Chem. 1994; 269: 25587-25591Abstract Full Text PDF PubMed Google Scholar). Proteins used for analytical ultracentrifugation experiments were purified as described above or specifically labeled as follows: purified H.s. LIN-2C and C.e. LIN-2C domains were conjugated to a fluorescent BODIPY moiety by covalent linkage to endogenous cysteines (corresponding to residues 464 and 473 in C. elegans Lin-2a and residue 412 in H. sapiens CASK). Labeling was performed by incubating 100 μm protein with 2 mm BODIPY maleimide (Molecular Probes) in 20 mm HEPES pH 7.0 overnight at 4 °C. Labeled proteins were purified over a PD-10 gel filtration column (Amersham Biosciences) and redialyzed into a storage buffer of 20 mm HEPES, 1 mm DTT, pH 7.0. Binding of BODIPY-labeled 2C domains to C.e. LIN-7 was verified independently by GST pulldown assay and showed no detectable difference from unlabeled 2C domains. We were not able to exceed a C.e. LIN-7:LIN-2C ratio of 25:1 as the C.e. LIN-7 monomer shows significant non-ideal behavior at concentrations ≥25 μm, which leads to a decrease in the apparent molecular weight of either heterodimer. Circular dichroism spectroscopy was performed using either an AVIV model 62DS or JASCO 715 spectrapolarimeter. All scans were performed with untagged, purified proteins in a 0.1-cm quartz cuvette in 20 mm HEPES pH 7.5; data for scans were collected at 1 nm intervals, with 1 s averaging time per data point and a total of 10 repeats per scan. Temperature denaturation studies were performed in 0.1-, 0.5-, or 1.0-cm path length quartz cuvettes with 50 mm sodium phosphate buffer, pH 7.5, 10 mm NaCl. Data for melts was collected at 1 °C intervals, with 60 s averaging time and 60 s of equilibration time per data point. Proteins subjected to temperature melts were scanned in every case (200–300 nm) before and after each melt to verify reversibility. To assess the binding capacity of putative L27 domains, we performed GST-pulldown assays, cross-testing all possible combinations of the six cloned domains (C.e. and R.n. LIN-7, C.e. and H.s. LIN-2N, C.e. and H.s. LIN-2C). In each case, a GST-tagged bait domain was used to pull down a His6-tagged "prey" domain. As shown in Fig. 2, we identified a specific interaction between LIN-2C and LIN-7 L27 domains. This interaction is conserved across species; both mammalian and C. elegans LIN-7 L27 domains are able to bind LIN-2C L27 domains from either species. These data are consistent with previous findings that mammalian LIN-2C domains bind mammalian LIN-7 domains (13Butz S. Okamoto M. Südhof T.C. Cell. 1998; 94: 773-782Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 22Borg J.P. Straight S.W. Kaech S.M. de Taddéo-Borg M. Kroon D.E. Karnak D. Turner R.S. Kim S.K. Margolis B. J. Biol. Chem. 1998; 273: 31633-31636Abstract Full Text Full Text PDF PubMed Scopus (170) Google Scholar, 23Borg J.P. Lopez-Figueroa M.O. de Taddeo-Borg M. Kroon D.E. Turner R.S. Watson S.J. Margolis B. J. Neurosci. 1999; 19: 1307-1316Crossref PubMed Google Scholar, 27Straight S. Chen L. Karnak D. Margolis B. Mol. Biol. Cell. 2001; 12: 1329-1340Crossref PubMed Scopus (42) Google Scholar,29Lee S. Fan S. Makarova O. Straight S. Margolis B. Mol. Cell. Biol. 2002; 22: 1778-1791Crossref PubMed Scopus (123) Google Scholar). Within this set, we did not identify any binding partners for the LIN-2N L27 domain, although recent studies suggest that this domain may be responsible for interaction with another multidomain scaffolding protein, SAP97 (29Lee S. Fan S. Makarova O. Straight S. Margolis B. Mol. Cell. Biol. 2002; 22: 1778-1791Crossref PubMed Scopus (123) Google Scholar). Deletion analysis (data not shown) indicates that the boundaries identified by sequence homology (28Doerks T. Bork P. Kamberov E. Makarova O. Muecke S. Margolis B. Trends Biochem. Sci. 2000; 25: 317-318Abstract Full Text Full Text PDF PubMed Scopus (46) Google Scholar) correspond to those of the functional domain, deletion of 5–10 additional amino acids results in loss of binding, whereas the addition of 5–10 residues has no noticeable effect on binding. To quantitatively measure the binding affinity of LIN-2C and LIN-7 L27 domains, we took advantage of the fact that both the C.e. LIN-2C and H.s. LIN-2C domains have cysteine residues, to which we covalently attached a fluorescent dansyl group. Dansyl fluorescence is very sensitive to local environment and often changes significantly in response to protein binding. Binding was assayed by measuring the change in fluorescence with increasing amounts of the C.e. LIN-7 domain. Results of the fluorescence perturbation binding assay for the C.e. LIN-7:C.e. LIN-2C(dansyl) pair are shown in Fig.3 A (filled circles). The data could be fit well to a bimolecular binding reaction, with a K d = 14 ± 4 μm(Fig. 3 A). Scatchard analysis (Fig. 3 B) is consistent with a 1:1 stoichiometric ratio. For all cases, we confirmed that the dansyl modification did not perturb the affinity by performing a competition binding assay in which the dansylated probe was competed off by addition of undansylated domain. Fits to these data (not shown) indicate no significant difference in affinity between the modified and unmodified domains. Addition of up to 50–100 μm of most other purified L27 domains (either 2C or 2N) resulted in no significant change in fluorescence (Fig. 3 A, open squares), confirming the previous GST-pulldown result (Fig. 2) that this hetero-oligomer interaction is highly specific. The only other L27 domain pair that showed significant binding was the homologous interspecies pair of the H.s. LIN-2C and C.e. LIN-7 domains. Interestingly, this interspecific pair was found to have aK d = 88 ± 6 nm (Fig.3 C) and is therefore ∼100-fold higher in affinity than the corresponding intraspecies pair. It is possible that the C.e. LIN-2C domain requires additional regions outside of the domain for high affinity interaction. However, given the high homology between the domains (28% identity, 46% similarity), this explanation seems unlikely. It is more likely that this result represents a real difference in affinity. This finding raises the interesting possibility that the biologically relevant intraspecies pair has evolved to have a suboptimal affinity, which may be more amenable to regulated multiprotein complex assembly. Although LIN-7 and LIN-2 homologs have been shown to bind one another in a wide variety of systems, the precise stoichiometry of the complex is not known. Therefore, the possibility exists that L27 domains form higher order oligomers rather than a simple heterodimer. To further characterize the stoichiometry of the complex and obtain a molecular weight for the interacting pair, we performed equilibrium analytical ultracentrifugation on the tight binding H.s. LIN-2C:C.e. LIN7 L27 pair. H.s. LIN-2C monomer was selectively labeled with a BODIPY fluorophore exhibiting a peak absorbance at 504 nm (at this wavelength nonlabeled monomers have no detectable absorbance). Upon mixing 1 μm labeled H.s. LIN-2C monomer with an excess amount of unlabeled C.e. LIN-7, we were able to follow the equilibrium distribution of the labeled species in an analytical ultracentrifugation experiment. For the C.e. LIN-7:H.s. LIN-2C pair, the fraction bound at these concentrations is very high, and the data are in agreement with a 1:1 heterodimer and inconsistent with alternative higher order oligomers (Fig.4 A). Analytical ultracentrifugation of H.s. LIN-2C, C.e. LIN-2C, or C.e. LIN-7 monomers alone gave good fits to a monomer molecular weights over a range of concentrations of ∼50–200 μm with no evidence of homodimerization. Thus, we conclude that the LIN-7:LIN-2C L27 domain pair forms a simple heterodimer. We assessed the stability and secondary structure of L27 domains, both as individual proteins and in bound complexes, using circular dichroism spectroscopy. Spectra for L27 monomers show little ellipticity at 222 nm (Fig. 5 A), characteristic of proteins that have little helical structure. However, we observed a striking increase in ellipticity at 222 nm upon mixing of the monomers; the ellipticity observed is significantly greater than the arithmetic sum of the individual monomer spectra. Specifically, mixed 7 and 2C domains, not other domains alone or in combination, show CD spectra that are characteristic of highly α-helical folded proteins (Fig.5 A). We observed this for both interacting pairs (C.e. LIN-7:C.e. LIN-2C and C.e. LIN-7:H.s. LIN-2C). The difference spectra indicate the formation of a highly helical structure only upon interaction. However, we do not observe significant changes for non-binding L27 pairs (Fig. 5 B). To further characterize this conformational change upon L27 heterodimerization, we performed temperature-melting experiments on both monomers and heterodimers at 222 nm, a wavelength diagnostic of α-helical secondary structure. We performed these experiments with the high affinity C.e. LIN-7:H.s. LIN-2C pair. As shown in Fig.5 C, neither of the individual domains shows a cooperative transition with increasing temperature. The H.s. LIN-2C:C.e. LIN-7 heterodimer, however, displays a clear cooperative unfolding transition. This supports a model in which binding and folding events are linked for the LIN-2C:LIN-7 L27 pair; the individual monomers are largely unfolded but become a stable, helical, folded unit upon association. If binding and folding events for the LIN-2C:LIN-7 pair are linked, apparent stability of the heterodimer would show a dependence on concentration. To test this, we measured the melting temperature of varying equimolar mixtures of H.s. LIN-2C-C.e. LIN-7, from 1 to 10 μm each, by CD. As shown in Fig. 5 D, we observe a trend of increasing stability with increasing concentration, with an overall change in T m from ∼30 to ∼37 °C over the range we measured. In further support of this model, we find that a construct of artificially tethered L27 domains, in which the effective concentration of each L27 monomer for its intramolecular binding partner should be very high, also shows cooperative unfolding behavior with a higher T m of ∼65 °C (data not shown). Together these data indicate that individual L27 domains are largely unfolded but adopt a helical structure when they interact to form a 1:1 simple heterodimer. L27 domains are therefore very different from most other modular protein-protein recognition domains used in signal transduction, as most such domains are independently folding units that can be transferred both structurally and functionally by recombination. This unusual property of L27 domains makes them similar to coiled-coil domains, which can also serve as specific homo- or hetero-oligomerization modules. Coiled-coils are largely unstructured independently but form a helical bundle when associated (36Hurst H.C. Protein Profile. 1995; 2: 101-168PubMed Google Scholar, 37Thompson K.S. Vinson C.R. Freire E. Biochemistry. 1993; 32: 5491-5496Crossref PubMed Scopus (188) Google Scholar, 38Alber T. Curr. Opin. Genet. Dev. 1992; 2: 205-210Crossref PubMed Scopus (250) Google Scholar). Is it possible that L27 domains are related to the coiled-coil family? A region of certain LIN-7 homologs corresponding to the L27 domain has in fact been referred to as a potential coiled-coil domain (26Olsen O. Liu H. Wade J.B. Merot J. Welling P.A. Am. J. Physiol. 2002; 282: C183-C195Crossref PubMed Scopus (64) Google Scholar, 27Straight S. Chen L. Karnak D. Margolis B. Mol. Biol. Cell. 2001; 12: 1329-1340Crossref PubMed Scopus (42) Google Scholar). However, we believe that it is unlikely that the L27 domain is a canonical coiled-coil. The region of LIN-7 that may be a putative coiled-coil is extremely short (15 amino acids), especially compared with the length of L27 domains (∼65 residues) required for association. Further, analysis of multiple aligned L27 domain sequences shows no consistent propensity to form coiled-coil domains (according to either the COILS program or MULTICOIL analysis) (39Schultz J. Copley R.R. Doerks T. Ponting C.P. Bork P. Nucleic Acids Res. 2000; 28: 231-234Crossref PubMed Scopus (1054) Google Scholar, 40Lupas A. Van Dyke M. Stock J. Science. 1991; 252: 1162-1164Crossref PubMed Scopus (3482) Google Scholar, 41Wolf E. Kim P.S. Berger B. Protein Sci. 1997; 6: 1179-1189Crossref PubMed Scopus (649) Google Scholar). Even within the LIN-7 family, the putative coiled-coil motif is not conserved (26Olsen O. Liu H. Wade J.B. Merot J. Welling P.A. Am. J. Physiol. 2002; 282: C183-C195Crossref PubMed Scopus (64) Google Scholar). Although it may be possible that some element of the L27 heterodimer adopts a coiled-coil-like structure, the L27 domain is unlikely to be a canonical coiled-coil. A definitive answer to this question awaits high resolution structural information for the complex. Nonetheless, it appears likely that these domains represent a novel class of obligate heterodimerization modules.

Referência(s)