Huntington's disease (HD) occurs when the widely expressed protein huntingtin contains an expanded glutamine repeat. The selective degeneration and neuronal morphologic abnormalities of HD may involve interactions with proteins that bind to huntingtin, such as HAP1. The biological significance of this interaction is unclear because neither HAP1 nor huntingtin have significant homology to known proteins. Therefore, we sought to identify HAP1-binding proteins. Using the yeast two-hybrid system, we isolated a rat cDNA encoding part of a protein that interacts with HAP1, and we confirmed the specificity of this interaction using an in vitro protein-binding assay. We called the protein Duo because it is closely related to the human protein Trio but is shorter. Northern blot analysis indicates brain-specific expression of Duo. Human Duo contains a guanine nucleotide exchange factor (GEF) domain that is likely to be rac1-specific, a pleckstrin homology (PH) domain and spectrin-like repeat units. These data support the hypothesis that huntingtin is involved in vesicle trafficking and cytoskeletal functions, and raise the possibility of a role for huntingtin in the regulation of a ras-related signaling pathway.
Huntington's disease (HD) is an autosomal dominant neurodegenerative disorder characterized by motor impairments and dementia. The pathology of HD is brain specific, with the caudate and putamen showing the most dramatic neuronal morphological abnormalities and degeneration (1 ). HD is caused when a 350 kDa protein called huntingtin contains an expanded glutamine repeat (2 ). While the neuronal abnormalities in HD are selective, huntingtin is widely expressed (3 ). The mutation is believed to cause a `gain of function', possibly by altering interactions of huntingtin with other brain-specific proteins (1 ,4 ). Several huntingtin-interacting proteins have been identified, but so far only huntingtin-associated protein 1 (HAP1) and the recently described huntingtin-interacting-protein 1 (HIP1) bind to huntingtin in a repeat length-dependent manner (4 -7 ). Since both huntingtin and HAP1 have no homology to other proteins in the database, the significance of this interaction is unclear. Therefore, we sought to find HAP1-binding proteins using the yeast two-hybrid system.
A fusion protein consisting of the GAL4 DNA-binding domain and the N-terminus of HAP1 (amino acids 1-313) (pPC97 N-term/HAP1) was used as bait to screen a random primed adult rat hippocampus/cerebral cortex cDNA library prepared in the GAL4 activation domain-encoding vector, pPC86 (8 ,9 ). Among the 100 000 colonies screened, 45 colonies were positive in a [beta]-galactosidase assay and 21 were similar to clone N9-3 that encodes 587 amino acids of rat Duo. Duo N9-3 is positive in a [beta]-galactosidase assay when co-expressed with HAP1 and negative when co-expressed with the vector itself or with other proteins such as huntingtin (Fig. 1 ). The specificity of the interaction was confirmed using an in vitro protein-binding assay (Fig. 2 ). HAP1 expressed in HEK293 cells binds to a glutathione-S-transferase (GST) Duo N9-3 fusion protein and not to GST by itself or GST attached to another protein nNOS (amino acids 201-406 of neuronal nitric oxide synthase). Thus, the interaction between HAP1 and Duo N9-3 is specific.
The full-length Duo cDNA contig was generated mostly by screening human brain cDNA libraries (Fig. 3 A). The contig is 6469 bp long and encodes a 1663 amino acid protein with no signal sequence or hydrophobic sequence that could span the lipid bilayer, suggesting that Duo is neither secreted nor a classical membrane protein (Fig. 3 A and B). The human Duo and rat Duo N9-3 amino acid sequences show 98.5% identity (Fig. 3 B); therefore, rat HAP1 is likely to bind to human Duo. BLAST analysis indicated that Duo shows a high degree of sequence similarity with Trio (10 ) and PCIP-10 (11 ). Trio interacts with the leukocyte common antigen-related (LAR) transmembrane protein-tyrosine phosphatase (10 ). PCIP-10 is a peptidylglycine [alpha]-amidating monooxygenase (PAM) COOH-terminal interacting protein (11 ). Human Duo has 80.6% identity with the human protein Trio, and rat Duo N9-3 is identical (except for a nine amino acid insert) to rat PCIP-10 ( (Fig. 3 A). Thus, Trio and Duo/PCIP-10 are two different proteins. Duo contains a Dbl-like guanine nucleotide exchange factor pleckstrin homology (GEF-PH) domain and spectrin-like repeats (Fig. 3 C). The GEF domain in Duo is 98.7% similar to the rac1-specific GEF-D1 domain in Trio (Fig. 3 C) (10 ).
By visual inspection, we identified four spectrin-like repeats that fit 60-80% the nomenclature introduced by Speicher and Marchesi (12 ) (Fig. 3 A). Position #1 in each unit corresponds to amino acid #280 in SP-D1, #507 in SP-D2, #861 in SP-D3 and #1101 in SP-D4. The residues found that fit the consensus are: (i) tryptophan at position #45 and leucine at position #26; (ii) hydrophobic amino acids at positions #1 (I, V or L), #15 (L, V or F), #35 (F or L) and #46 (I); and (iii) charged residues at positions #22 (R, T or K), #38 (D, E or S), #41 (K, Q or E) and #72 (T, K or E). A fifth spectrin-like repeat is found between SP-D2 and SP-D3 (amino acids 744-849), which has all the above characteristics except for T at position #35, N at position #46 and hydrophobic amino acids at positions #22 (L) and #41 (V). Less conserved spectrin-like repetitive units exist such as one beginning at position #744 (Fig. 3 A) and #612 (not shown). Using other criteria (11 ,13 ), as many as nine spectrin-like repeats can be described, but we have enumerated the four which are described in Trio (10 ) and fit the criteria of Speicher and Marchesi (12 ). A scan of the Duo N9-3 sequence by the paired-coils program revealed one coiled-coil within amino acids 914-943 with a particularly high probability (0.769; significance of 0.5) (Fig. 3 A).
Northern blot analysis revealed two brain-specific mRNAs of ~7.0 and 4.7 kb (Fig. 4 ). The mRNA of ~7 kb corresponds to the length of the cDNA isolated (6469 bp) (Fig. 3 A). The 4.7 kb mRNA could be an alternatively polyadenylated form. None of the peripheral tissues analyzed contained significant levels of Duo mRNA (Fig. 4 a). Brain-selective expression of Duo occurred in both rat and human, with high levels in human cerebral cortex, putamen, amygdala, hippocampus and caudate nucleus; lower levels were present in brainstem and cerebellum (Fig. 4 b). These results indicate that Duo is expressed predominantly in the brain. ab
Figure
We have identified a protein we term Duo that interacts specifically with HAP1 (Figs 1 and 2 ). Like HAP1 (14 ), it is expressed predominantly in the brain (Fig. 4 a). Human Duo is 80.6% identical to human Trio (10 ) and contains at least four or five spectrin-like repeats and a Dbl-like GEF-PH domain; the GEF domain is 98.7% similar to the rac1-specific GEF-D1 domain in Trio (Fig. 3 ). The fragments of rat Duo and rat PCIP-10 (11 ) cDNA isolated in the yeast two-hybrid system overlap considerably, show nearly complete identity and contain spectrin-like repeats and a stringently defined coiled-coil (Fig. 3 ). Since the strength of the interaction between huntingtin and HAP1 is dependent on the glutamine repeat length in huntingtin, and since HAP1 has brain-specific expression, HAP1, and thus Duo, may be involved in the pathogenesis of HD. The pattern of expression of Duo in the brain (Fig. 4 b) includes regions affected in HD but does not perfectly match the pattern of pathology of HD. The pattern of degeneration in HD may also relate to intrinsic cellular vulnerability (15 ,16 ).
Duo has characteristics of membrane cytoskeletal proteins (17 ); it contains the repeats typically found in all members of the spectrin supergene family. All members of this family bind actin. Although Duo lacks a consensus actin-binding site (17 ,18 ), it may interact with actin indirectly, as observed for [alpha]-fodrin and [alpha]-spectrin. This possibility is supported by the results of BLAST analysis which indicate that Duo is more similar to the [alpha] rather than the [beta] chain of fodrin/spectrin. The association of HAP1 with a cytoskeletal protein, Duo, is consistent with the findings that HAP1 is enriched in subcellular fractions containing cytoskeletal elements (14 ) and huntingtin binds with a cytoskeletal protein called HIP1 (6 ,7 ).
An extensive overlap exists between the sequences of Duo (PCIP-10) that contain the HAP1-binding site and the PAM-binding region (Fig. 3 ). This region has a particularly stringently defined coiled-coil (0.769), which are structures involved in protein-protein interactions at the cytoskeleton (20 ). Whether HAP1 and PAM bind Duo (PCIP-10) simultaneously or not remains to be determined. The luminal domain of PAM is critical to the production of bioactive peptides stored in large dense core vesicles. The cytoplasmic domain, which binds PCIP-10/Duo (11 ), is believed to be involved in the biogenesis of secretory granules and has a sorting signal for internalization from the cell surface (21 ). PAM, PCIP-10, HAP1 and huntingtin are expressed in neurons, including cortical neurons (11 ,14 ,22 -25 ). Huntingtin has been associated with exocytic and endocytic vesicles (14 ,22 -25 ). Thus, we speculate that interactions between these proteins may take place during biogenesis of secretory granules and/or internalization from the cell surface.
The GEF domain in Duo is nearly identical to the rac1-specific GEF-D1 domain in Trio (10 ) and, after submission of this paper, an article was published on PCIP-10 (Duo) reporting its rac1-specific binding (26 ). Functional GEF domains activate small GTP-binding proteins like rac and may have adjacent PH domains regulating their activity (27 ). The GEF domain in Duo has a juxtaposed PH domain and therefore is likely to activate rac1. Rac1 is involved in regulating cytoskeleton (actin) organization, endocytosis, exocytosis and free radical production (28 ,29 ). Rac1 and the other members of the Rho family of GTPases have a role in neuritogenesis (30 ). Corticotrope tumor cells stably expressing PCIP-10 (Duo) produce longer and more highly branched neuritic processes than non-transfected cells (26 ). Thus, the results in this study raise the possibility that huntingtin may be involved indirectly in cellular functions regulated by ras-related signaling pathways, consistent with the altered morphology seen in some surviving neurons in HD (31 ,32 ).
The HAP1 cDNA fragment corresponding to bp 40-939 was isolated by PCR and subcloned into the SalI and NotI sites of the pPC97 vector in-frame with the GAL4 DNA-binding domain (8 ,9 ). Y190 yeast cells were transformed with pPC97-Nterm HAP1 and subsequently with an adult rat hippocampus random primed cDNA library cloned into pPC86. The protocols used in this yeast two-hybrid system were performed as described previously (8 , 9 ; Clontech).
Escherichia coli strain BL21 transformed with pGEX-4T2-N9-3, pGEX-4T2-amino acids 201-406 of nNOS, or pGEX-4T2 alone were grown in LB induced with isopropyl-[beta]-d-thiogalactopyranoside (IPTG), and the GST fusion proteins synthesized were coupled to glutathione-Sepharose 4B beads as described by the manufacturers (Pharmacia). BL21 pGEX-4T2-N9-3 was grown at 30oC instead of 37oC , and induced with 30 [mu]M IPTG instead of 100 [mu]M, and all samples were freeze-thawed three times before sonication. A construct containing the expression vector PRK and the HAP1 cDNA under the control of a cytomegalovirus (CMV) promoter was transiently transfected into HEK293 cells using the calcium-mediated DNA transfection (BBS) method (33 ). The cells were collected by trypsinization, washed once in medium containing fetal calf serum, twice in phosphate-buffered saline (PBS), once in 50 mM Tris pH 7.4 10 mM EDTA, and stored at -80oC in aliquots. Just before use, the cells were sonicated at 1-3 mg/ml for 10 s in sonication buffer [50 mM Tris pH 7.4, 160 mM NaCl, 2.5 mM MgCl2, 1.5 mM CaCl2, 2.5 mM KCl, 1% Triton, 1 tablet/50 ml of TM complete protease inhibitors (Boehringer Mannheim)], centrifuged at 25 000 g for 10 min at 4oC, and the supernatant was used in the binding assay. The conditions used are based on previously described procedures (5 ,11 ): 5 [mu]g of sonicate and 20 [mu]l of 50% GST, GST-NOS amino acids 201-406 or GST-Duo-N9-3 Sepharose 4B- glutathione beads (pre-washed in the same buffer) were added to 1.4 ml of binding buffer (10 mM Tris-HCl pH 7.5, 150 mM potassium acetate, 1 mM MgCl2 and 0.5 mM CaCl2 pH 7.5), incubated overnight at 4oC mixing gently, washed at 4oC four times for 10 min each using binding buffer, and subjected to SDS-PAGE and Western blot analysis.
A rat Duo PCR product (bp 40-230) was generated by standard procedures. This PCR product, a human Duo cDNA clone containing bp 4982-5537, and actin cDNA were radiolabeled using the random primed kit (Boehringer Mannheim). Northern blots (Clontech) were pre-hybridized using the express hybridization buffer at 42oC for 2-3 h, hybridized with 1 * 106 c.p.m./ml overnight at 68oC, and the blots were washed 2-3 times for 30 min each at 50oC, and exposed for 12-48 h.
A PCR product (bp 40-230) derived from rat Duo N9-3 was radiolabeled using the random primed kit (Boehringer Mannheim) and used as a probe to screen a human frontal cortex cDNA [lambda]ZAP library (Stratagene). The 3' region of the human Duo cDNA was obtained in a second screening using an oligonucleotide probe derived from a sequence encoding part of the fourth spectrin-like domain (5'-CAGCAGTTCCTGAGTCTCCTCTGTGGTCTCTCCAGTGGAGGTATG-3'). The cDNA inserts from positive phage clones were isolated using the in vivo excision method (Stratagene) and four clones (from a total of 12) were fully sequenced on both strands using an ABI automated sequencing apparatus (Johns Hopkins University Core Genetics Facility). 5' RACE of human whole brain mRNA was performed according to the instructions of the manufacturer (Life Technologies). A homology search was performed against GenBank, EST database (34 ) and ProfileScan (ISREC bioinformatics group). EST46985 was purchased from Research Genetics and sequenced on both strands.
A peptide (RQRSSMPAGSVTHC) corresponding to amino acids 410-414 of rat HAP1 plus C-terminal cysteine (added for coupling purposes) was synthesized (Research Genetics) and conjugated to maleimide-activated keyhole limpet hemocyanin (KLH) and bovine serum albumin (BSA) (Pierce) essentially as described in the manufacturer's directions. New Zealand White rabbits were immunized with the conjugates (Cocalico Biologicals, Inc., Reamstown, PA) using the KLH conjugates for the first two injections (days 0 and 14) and the BSA conjugates for subsequent boosts (days 21 and 51 and each 1-2 months thereafter). Antibodies were affinity purified from antisera prepared from samples collected after day 90. Antiserum (15-20 ml) was first treated by batch incubation overnight at 4oC with a Sulfolink column (Pierce) which had been reacted previously with 2-mercaptoethanol. Antibodies were then affinity purified using the peptide immunogen immobilized on a Sulfolink column essentially according to the directions of the manufacturer. The antiserum was incubated batchwise with 2 ml of the affinity matrix overnight at 4oC in the presence of 2 mM EDTA and a cocktail of protease inhibitors. The affinity resin was then poured into a column, washed extensively as described (23 ) and eluted with 90 mM glycine-HCl (pH 2.5) containing 10% ethylene glycol. The fractions collected were neutralized by the addition of 100 [mu]l of 2 M Tris-HCl (pH 7.4). The fractions containing the highest absorbance at 280 nm were pooled and dialyzed extensively against 20 mM HEPES (pH 7.4), 0.9% NaCl followed by dialysis into 20 mM HEPES (pH 7.4), 100 mM NaCl, 50% glycerol. Antibodies were stored at -20oC. The anti-peptide antibody recognized a band at the same location as the anti-fusion protein antibody (5 ) in a Western blot of protein extracts from transfected cells expressing HAP1a.
The sequences of the partial rat Duo cDNA and full-length human Duo cDNA were submitted to GenBank and the corresponding accession numbers are U94189 and U94190, respectively.
We thank Betty Eipper and Dick Mains for helpful discussions and for sharing cDNA sequence and other information prior to publication. We thank Samie Jaffrey for the pGEX-4T2-nNOS construct and the Genetics Core Facility, especially Roxann Ashworth, for DNA sequencing and analysis. We thank Jon Wood for comments on the manuscript. This work was supported by a MSD award from the American Heart Association to V.C., a Pew Fellowship award to S.E., an NIMH grant MH01152 to P.W. and an NINDS grant NS16375 to C.A.R. This work was also supported by a bequest from Sar and Brita Ann Levitan and by the DeVelbiss Fund.
Under an agreement between Guilford Pharmaceuticals, Inc. and the Johns Hopkins University, the author is entitled to a share of sales royalty received by the University from Guilford Pharmaceuticals, Inc. The terms of this arrangement are being managed by the University in accordance with its conflict of interest policies.
We have recently found that HAP1 binds to other cytoskeletal and motor-related proteins, including p150Glued of the dynactin complex [Engelender, S., Sharp, A.H., Colomer, V., Tokito, M.K., Lanahan, A., Worley, P., Holzbaur, E.L.F., and Ross, C.A. (Hum. Mol. Genet., submitted) Huntingtin associated protein 1 (
Human Molecular Genetics
Pages
Introduction
Results
Identification of Duo as a HAP1-specific interactor
Isolation of full-length human Duo cDNA
Expression of Duo mRNA
Discussion
Materials And Methods
Yeast two-hybrid screening
Protein-binding assay
Northern blot analysis
cDNA library screening and RACE
HAP1 antibody preparation
GenBank accession number
Acknowledgements
Disclosure Statement
References
Note Added In Proof
REFERENCES
This page is maintained by OUP admin. Last updated Wed Aug 13 15:52:16 BST 1997. Part of the OUP Journals World Wide Web service. Copyright Oxford University Press, 1996
