b -Amyloid Peptide-induced Apoptosis Regulated by a Novel Protein Containing a G Protein Activation Module*

Degeneration of neurons in Alzheimer’s disease is mediated by b -amyloid peptide by diverse mechanisms, which include a putative apoptotic component stimulated by unidentified signaling events. This report de-scribes a novel b -amyloid peptide-binding protein (de-noted BBP) containing a G protein-coupling module. BBP is one member of a family of three proteins containing this conserved structure. The BBP subtype bound human b -amyloid peptide in vitro with high affinity and specificity. Expression of BBP in cell culture induced caspase-dependent vulnerability to b -amyloid peptide toxicity. Expression of a signaling-deficient dominant negative BBP mutant suppressed sensitivity of human Ntera-2 neurons to b -amyloid peptide mediated toxicity. These findings suggest that BBP is a target of neuro-toxic b -amyloid peptide and provide new insight into the molecular pathophysiology of Alzheimer’s disease. Genetic and biochemical data have coalesced to establish that b -amyloid peptide (A b ) 1 is a causative factor in neuron death and the consequent dimunition of cognitive abilities observed in Alzheimer’s disease (1, 2). Plasma lipoproteins and their cell surface receptors influence sequestration and clear-ance of soluble A b , contributing to the

Genetic and biochemical data have coalesced to establish that ␤-amyloid peptide (A␤) 1 is a causative factor in neuron death and the consequent dimunition of cognitive abilities observed in Alzheimer's disease (1,2). Plasma lipoproteins and their cell surface receptors influence sequestration and clearance of soluble A␤, contributing to the etiology of the disease (3)(4)(5)(6). Inflammatory responses and oxidative damage also appear to contribute to the loss of neurons in Alzheimer's disease (7)(8)(9)(10). Although the earliest cellular perturbations remain unclear, recent findings indicate that A␤ may act as an initiating factor in the death of neurons by inducing signaling pathways leading to apoptosis (11)(12)(13)(14)(15)(16)(17). However, the specific molecular target(s) transducing these A␤ effects has not been identified. The intracellular protein ERAB can bind A␤ in vitro, and neuroblastoma cells expressing recombinant ERAB undergo apoptosis when treated with exogenously added A␤ (18), but the mechanism by which ERAB may affect apoptotic signaling remains obscure. We identified a novel human ␤-amyloid peptide binding protein (BBP) utilizing yeast 2-hybrid technology. Analysis of the BBP amino acid sequence revealed the presence of a structural module related to that of the 7 transmembrane domain G protein-coupled receptor superfamily and known to be important in heterotrimeric G protein activation. Data suggest that BBP mediates cellular vulnerability to A␤ toxicity through a G protein-regulated program of cell death. Two related proteins (BLP1, BLP2; BBP-like proteins) were identified by sequence and structural similarities to BBP, but only the BBP subtype regulates a response to A␤.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Systems-Yeast 2-hybrid (Y2H) expression plasmids were constructed in the vectors pAS2 and pACT2 (19). Strain CY770 (20) served as host for Y2H assays. Sequences encoding A␤ 42 were amplified by PCR using primers incorporating restriction sites for subsequent ligation into pAS2, using a human APP (amyloid precursor protein) cDNA clone as template. A Y2H plasmid library consisting of cDNA fragments isolated from human fetal brain cloned into the yeast 2-hybrid expression vector pACT2 (CLONTECH Laboratories) was screened with pAS2-A␤. Interactors were identified on synthetic complete yeast growth medium lacking histidine and containing 3-aminotriazole at a concentration of 25 mM. Y2H expression plasmids for BBP deletion analyses were generated by standard methods. Expression of all hybrid proteins was confirmed by immunoblot using antibodies (CLONTECH) directed to the Gal4 sequences. ␤-galactosidase assays to measure Y2H interactions were conducted in strain CY770 containing the indicated Y2H expression plasmids plus the GAL UAS -lacZ reporter plasmid pRY131 (21).
RACE and Genomic Cloning-A modified rapid amplification of cDNA ends (RACE) protocol was utilized to obtain 5Ј-BBP sequences absent in the original Y2H clone. First strand DNA synthesis was performed using the rTth thermal-stable polymerase system (Perkin Elmer Life Sciences) with human hippocampus mRNA as template. The Marathon™ cDNA synthesis kit (CLONTECH) was used for subsequent second strand cDNA generation and amplification. In addition, a human genomic lambda library (Stratagene) was screened with a randomly primed BBP protein coding region probe. Positive clones were purified and subjected to further analysis by hybridization to a 45nucleotide probe directed to the most 5Ј sequences known from the original cDNA clone. The nucleotide sequence of the upstream BBP region from a positive clone was identical to cDNA sequences identified by RACE. Mouse BBP cDNA and human BLP1 and BLP2 cDNA sequences were initially assembled from expressed sequence tag (EST) data extracted from Genbank TM /EBI. These segments, which did not include potential translation start sequences but extended 3Ј to poly(A) sequences, were amplified by RT-PCR from mouse or human brain mRNA. Subsequently, RACE was used to isolate 5Ј sequences for human BLP1 and BLP2. RACE failed for mouse BBP, but 5Ј sequences were obtained to a potential initiating ATG from genomic cloning (data not described). The three human cDNAs were assigned accession numbers AF353990, AF353991, and AF353992; mouse BBP AF353993.
Antibody Generation and Immunoblots-Predicted BBP ectodomain sequences were synthesized as five nonoverlapping peptides. The peptides were pooled and conjugated to activated KLH carrier protein per vendor's instructions (Pierce). Chickens were injected intramuscularly with 0.1 mg of peptides/KLH each week for 4 weeks. Eggs were collected and tested for IgY titer to each BBP peptide by ELISA. IgY was partially purified from egg yolk by dilution and ammonium sulfate precipitation (22). This sample was further purified by solid phase affinity binding to BBP peptide composed of residues 42-81 (Fig. 1A). Expression of recombinant BBP protein was evaluated in Chinese Hamster Ovary cell lysates. Cells were transfected with pBBP by Lipo-fectAMINE-PLUS per manufacturer's (Life Technologies, Inc.) instructions. Cells were suspended in hypotonic buffer (50 mM Tris, pH 7.2; 1 mM EDTA) plus proteinase inhibitors and were maintained on ice. Cells were disrupted using a polytron and debris was removed by centrifugation at 2,000 rpm in a microcentrifuge. Soluble and membrane fractions were separated by centrifugation at ϳ200,000 ϫ g using a 45Ti rotor in a TL100 centrifuge (Beckman Instruments). The membrane pellet was resolubilized in phosphate-buffered saline (PBS) with 1% Triton X-100 plus proteinase inhibitors. Laemmli's buffer with detergent and 2-mercaptoethanol were added to aliquots containing 50 g of protein, and samples were boiled for 5 min prior to electrophoresis in a 4 -10% Tris-glycine NuPage gel (NOVEX). Samples were transferred to a polyvinylidene difluoride membrane by the semi-dry method (BIO-RAD). Blots were probed with the chicken anti-BBP antibody described above, using rabbit anti-IgY conjugated to horseradish peroxide (Promega) as a secondary detection reagent. Proteins were visualized by development with the ECL-Plus reagent and exposure to Hyperfilm (Amersham Pharmacia Biotech). Deglycosylation of proteins was achieved using the enzymes PNGase-F, NANase II, and O-glycosidase DS per manufacturer's instructions (Bio-Rad).
Northern and RT-PCR Analyses-Human multiple tissue mRNA Northern blots were obtained from CLONTECH. BBP sequences extending from the original Y2H fusion junction to the poly(A) region were used to generate a radiolabeled probe. Similar BLP1 and BLP2 cDNAs were used to generate probes. Control ␤-actin DNA was provided by the manufacturer. Hybridizations were performed per manufacturer's (CLONTECH) instructions. Quantitative RT-PCR was performed utilizing a PE Applied Biosystems 7700 instrument per manufacturer's instructions, probing with sequences contained within the BBP protein coding region. Detection reagents for glyceraldehyde phosphate dehydrogenase mRNA were used to qualitatively assess RNA samples and to standardize reactions.
In Situ Hybridization-DNA templates for riboprobe synthesis were prepared by PCR using each human BBP or BLP cDNA. Subsequent riboprobes were generated to 3Ј-untranslated regions using the Riboprobe Gemini TM System (Promega). In situ hybridization histochemistry using sections of cynomolgus monkey (Macaca fascicularis) forebrain was performed as described previously (23). Emulsion autoradiograms were developed and fixed according to the manufacturer's (Kodak) instructions, and the underlying tissue sections were stained with hematoxylin. To assess nonspecific labeling, a control probe was generated from a template provided in the Riboprobe Gemini TM System kit. No specific hybridization was observed in control sections.
In Vitro Binding Assays-The tagged BBP segment used in in vitro binding assays was produced by amplifying cDNA encoding the BBP protein from the clone 14 junction to 1 ⁄2tm, incorporating DNA encoding the c-Myc epitope EQKLISEEDL in the 5Ј-PCR primer. This DNA was cloned into the pET28a vector and transformed into E. coli strain BL21/DE3 (Novagen). Recombinant BBP protein was isolated in inclusion bodies extracted by B-PER reagent (Pierce). Samples were resuspended in 20 ml of 1ϫ binding buffer (Novagen) containing 6 M urea and purified using His-Bind™ chromatography (Novagen) in the presence of 6 M urea. Following elution, urea was removed by dialysis against binding buffer. To prepare Sepharose-antibody complex, 1.5 mg of swelled protein A-Sepharose (Sigma) was mixed with 3.3 ml of rabbit anti-mouse IgG (Jackson Immunoresearch Laboratories, Inc.) in cold immunoprecipitation buffer (IPB; Ref. 24) plus 0.02% NaN 3 with rotation overnight at 4°C. After washes, the 6E10 anti-A␤ antibody (Senetek PLC) was added, and the mixture was rotated overnight at 4°C. Purified BBP was combined with 500 ng of A␤ 40 at a 1:1 molar ratio in 200 l of 50 mM Tris pH 7.6, 150 mM NaCl, 1% Nonidet P-40. For A␤40 -1 competition experiments, 25 g (50ϫ molar excess) of A␤40 -1 was combined with BBP and A␤1-40. Samples were agitated at room temperature for 1 h, washed three times with 200 l of IPB, resuspended in 20 l of 2ϫ Laemmli sample buffer containing ␤-mercaptoethanol, boiled for 5 min, and loaded on a 10 -20% Tris-HCl polyacrylamide gel. Standard Western blot procedures utilizing Ni 2ϩhorseradish peroxidase were used to detect BBP. ELISA plates for solid phase binding measurements were coated with A␤ 40 peptide and blocked with 5% BSA as described elsewhere (25). Forty-residue human or rodent ␤-amyloid peptide and reverse human peptide were obtained from AnaSpec, Inc. Peptides were dissolved and stored in hexafluoroisopropyl alcohol at 1 mg/ml. Samples were lyophilized by pervasion with nitrogen and then resuspended in PBS and immediately frozen as disaggregated peptide. To achieve consistent aggregation, the peptides were resuspended in cell growth medium and divided into 0.13-ml aliquots in a 96-well plate. The plate was shaken at 500 rpm for 5 h. Samples were then combined and normalized to a final A␤ concentration of 50 M and immediately frozen. The human A␤ peptide was evaluated for circular dichroism and thioflavin-S fluorescence and shown to be predominantly aggregated, but not fibrillar. Purified tagged BBP protein was prepared in PBS plus 5% BSA. Protein was applied to A␤-coated plates for 1 h at room temperature, washed 5ϫ with PBS plus 0.1% Tween 20. Bound BBP was detected by application of mouse anti-Myc antibody (CalBiochem) followed by five washes, with subsequent application of donkey anti-mouse IgG conjugated to alkaline phosphatase (Jackson Immunoresearch Laboratories, Inc.). Following five washes and development with an alkaline phosphatase substrate kit (BIORAD), A 405 values were obtained. Values obtained from wells lacking A␤ peptide were subtracted from total bound BBP1 to calculate specific binding. Competition assays were performed by adding peptides to BBP (50 nM) in PBS, 5% BSA and incubating at 4 o overnight prior to application to A␤-coated plates. Approximate binding coefficients were calculated by nonlinear regression analysis using PRISM software (Graphpad, Inc.).
Cellular Assays-Human Ntera-2 (Nt2) stem cells (26) were maintained in Dulbecco's Modified Eagle's medium supplemented with 10% fetal bovine serum. Human SH-SY5Y cells were maintained in 1:1 Dulbecco's Modified Eagle's medium/F-12 Ham nutrients plus 15% fetal bovine serum. BBP cDNAs were modified by PCR for expression from the vector pcDNA3.1 (Invitrogen). Mutation of the arginine codon to glutamate within the DRF motif of BBP cDNA was performed using the QuickChange™ system (Stratagene). Cell viability determinations were performed with a CellTiter kit (Promega) measuring conversion of MTS tetrazolium to formazan. For nuclear morphology assays, expression constructs were introduced with pEGFP-N1 (CLONTECH) into cells by electroporation. DNA amounts were 7.5 g of subject DNA plus 2.5 g of pEGFP-N1. Approximately 24 h after transfection, growth medium was replaced with medium containing 10 M A␤ peptide. Forty-eight hours after A␤ addition, the chromatin-specific dye Hoechst 33342 (Molecular Probes, Inc.) was added to a concentration of 10 ng/ml. Medium was removed after 10 min, and cells were washed with PBS. Cells were then fixed by immersion in PBS containing 4% paraformaldehyde. A minimum of 200 transfected (EGFP ϩ ) cells per sample were scored manually by fluorescence microscopy. Statistical comparisons of apoptotic nuclei counts were conducted using Yates G-test of probability for categorical data. All experiments were repeated greater than three times with the same results. Human Nt2 neurons were derived by treatment of stem cells with retinoic acid (26). Neurons were transfected by the Profection TM system (Promega) using 4.5 g of test plasmid plus 1.5 g of pEGFP-N1 per sample. Similar to the other cellular assays, growth medium was replaced with medium containing 10 M aggregated A␤ 24 h after transfection. Following incubation for 48 h, cells were fixed and nuclear morphologies of transfected (i.e. EGFP ϩ ) cells were scored as described above. Transfection efficiencies, determined as the fraction of green fluorescent cells, for Nt2 stem cells and neurons were routinely 5-6% and 1-2%, respectively.

Identification of a Novel ␤-Amyloid
Peptide-binding Protein-A Y2H genetic screen was developed to identify proteins that interact with human A␤. The Y2H assay strain CY770, expressing A␤ 42 fused to the yeast Gal4 DNA-binding domain, was transformed with a human fetal brain cDNA Y2H library. A single clone, denoted no. 14, was selected for further characterization as it produced consistent reporter gene activation and contained a substantial open reading frame continuous with that of the Gal4 transcriptional activation domain. The cDNA insert of clone 14 comprised 984 base pairs, terminating in a poly(A) tract. This sequence encoded 201 amino acids with two regions of sufficient hydrophobicity and length to transverse a cellular membrane, indicated in Fig. 1A as tm1 and tm2. Examination of available databases revealed numerous ESTs, but no entry contained a complete open reading frame or attributed a potential function to the protein, which we designate BBP.
BBP Protein Expression-The BBP cDNA contained in clone 14 lacked a potential native translation initiation site. BBP 5Ј-DNA sequences were identified from both cDNA and genomic DNA sources. The additional cDNA sequence contained three possible start codons preceded by an inframe stop codon. The deduced sequence of the translation product shown in Fig. 1A begins at the first potential start site possessing suitable context for efficient translation initiation (27), and as suggested experimentally by recombinantly expressed BBP (data not shown). Peptides contained in the predicted ectodomains of BBP were used to immunize chickens. The chicken IgY was purified and utilized to probe partitioned soluble and membrane fractions produced from cells transfected with a BBP cDNA expression plasmid. A broad specific band of 36 -42 kDa was observed in the membrane-enriched fraction (Fig. 1B); no protein was detected in the soluble fraction. Enzymatic removal of carbohydrates resulted in a shift to 19 kDa, demonstrating that the BBP protein is extensively glycosylated. These data suggest that mature BBP is an integral membrane glycoprotein likely derived from a precursor containing a cleavable secretory signal leader. To investigate this prediction, recombinant human BBP protein lacking putative transmembrane (tm) domains was produced as a fusion to immunoglobulin Fc domain to facilitate protein stability and detection. This fusion protein was found to be secreted from cultured cells. Amino terminal sequencing of the protein revealed a single product processed to the position indicated in Fig. 1A. The site of cleavage is canonical to the signal peptidase site consensus (28). The cleaved protein contains 170 amino acids, consistent with the migration of deglycosylated protein at ϳ19 kDa (Fig. 1B).
BBP Relationship to the G Protein-coupled Receptor Superfamily-The BBP protein and translations of available ESTs were assembled, aligned, searched for conserved segments, and evaluated by the MoST protein motif search algorithm (29). First, these analyses revealed three distinct sets of ESTs in both the human and mouse datasets, indicating that BBP is one member of a structurally related protein family. Subsequently, orthologous sequences to mammalian BBP and the BBP-like proteins were also identified in the Drosophila melanogaster genome. No additional subtypes were detected in available databases. Human BLP1 and BLP2, and mouse and fly BBP cDNAs were isolated by RT-PCR methodologies using EST and genomic DNA information to guide primer design. The cDNA sequences encoding the mouse and fly BLP1 and BLP2 proteins were derived from EST and genomic DNA consensus determinations. Alignment of the three translation products from human, mouse, and fly revealed several key features ( Fig. 2A). Most striking is a segment exhibiting a significant relationship to the G protein-coupled receptor (GPCR) superfamily. Specifically, these proteins contain two potential tm domains with a high degree of similarity to the third and fourth tm domains of GPCRs ( Fig. 2A and B). The similarity includes the intervening hydrophilic loop, which contains the well characterized three amino acid motif, aspartate (D), arginine (R), and an aromatic residue (Y or F) (commonly referred to as the DRY sequence), that is conserved in most members of this receptor superfamily and has been shown to serve as the molecular trigger for G protein activation (30). In addition to a general similarity, Ͼ25% identity to the tm3 through tm4 segment of some GPCR members, other very highly conserved amino acids include a cysteine immediately preceding tm3 (BBP tm1) and a lysine marking the beginning of tm4 (BBP tm2). A tryptophan found in tm4 of ϳ95% of GPCRs is present at the equivalent position in the BLP1 and BLP2 subtypes. Preceding the tm domains, there is little homology between BBP/BLP subtypes, a common feature of receptor families sharing a conserved signal coupling domain, with unique activities determined by less conserved ectodomains. Each protein possesses a region of strong hydrophobicity near the amino terminus, indicative of an amino-terminal secretory signal. With the demonstrated functionality of the amino-terminal signal sequence in BBP, and in conjunction with the homologies to GPCR topology, it is predicted that the proteins transverse cellular membranes twice, with both termini luminal or extracellular as depicted in Fig. 2B. As with prototypic 7-tm domain G protein-coupled receptors, the BBP/ BLP proteins contain the important DRF motif appropriately positioned between two tm domains, juxtaposed to the first tm domain. This suggests that the proteins could modulate a heterotrimeric G protein regulatory pathway.
BBP Gene Expression-Tissue expression of BBP and BLP mRNAs was evaluated, revealing major transcripts of 1.25 kb, 1.35 kb, or 1.40 kb for BBP, BLP1, and BLP2, respectively, in all samples (Fig. 3A). Analysis of in situ hybridization autoradiograms of non-human primate forebrain obtained using riboprobes directed to each BBP mRNA demonstrated that all three genes are prominently expressed in medium to large cells in a pattern indicative of elevated expression in neurons as opposed to glial cells. BBP/BLP transcripts were observed in largely overlapping regional patterns, with greatest expression in the hippocampus and neocortex (Fig. 3B). The BBP riboprobe was also used to evaluate a human hippocampus/entorhinal cortex sample, revealing expression in virtually all neurons (data not shown). In summary, BBP mRNAs were observed in all tissues examined, and in situ analyses of brain samples revealed extensive expression in neurons of the hippocampus/entorhinal cortex and neocortical regions. This pattern is similar to the regionally-restricted neurodegeneration observed in Alzheimer's disease (31,32). More rigorous histopathological studies will be necessary to fully assess correlations between BBP expression and disease pathology.
BBP Interactions with A␤-Deletion variants of BBP were examined in Y2H assays to delineate the A␤-binding domain. Results are shown in Fig. 4A. The original clone 14 produced only a weak response with the A␤ fusion protein. Protein transmembrane domains have been observed to substantially attenuate Y2H responses, 2 as the reconstitution of transcription activator function must occur in the cell nucleus. Deletion of the second tm domain and carboxyl-terminal sequences of BBP resulted in much more robust Y2H activity, measured by activation of either HIS3 or lacZ reporter genes (clone14-⌬tm2). Moving in from the amino end, protein deleted for approximately half of the region between the clone 14 junction and tm1 maintained a positive interaction with A␤ (⌬60-⌬tm2). The deletion of 26 additional amino acids resulted in the loss of a 2 B. A. Ozenberger, unpublished data.

FIG. 3. The BBP genes are widely expressed, with prominent expression in neurons of the hippocampus/entorhinal cortex and neocortex.
A, nylon membranes blotted with 2-g size fractionated poly(A) RNA isolated from the indicated tissues were sequentially hybridized with the indicated radiolabeled cDNA probe. A single predominant band corresponding to the indicated length was observed in all lanes for each probe. Blots were also probed with ␤-actin as a loading and RNA integrity control. B, autoradiograms of coronal sections of cynomolgus monkey forebrain, taken at mid to caudal levels, processed to visualize the distribution of BBP, BLP1, or BLP2 mRNA by in situ hybridization histochemistry. Darker areas correspond to areas of higher expression of mRNA. The images are not normalized relative to each other. Y2H response with A␤ (⌬86-⌬tm2). From the carboxyl end, one-half of the tm1 domain could be deleted with no effect on Y2H interaction (⌬60-1 ⁄2tm). However, complete removal of the first tm domain (⌬60-⌬tm1) resulted in loss of activity. These results indicate that crucial A␤ binding determinants within BBP are contained within the segment delineated by ⌬60 and 1 ⁄2tm. The elimination of the A␤ association upon further deletion from either terminus could result from structural disruption of a single binding site by a distant deletion, or more likely, suggests that two important contacts with A␤ exist (between ⌬60 and ⌬86, and ⌬tm1 and 1 ⁄2tm), each necessary but not sufficient for binding alone.
Binding of BBP to A␤ in Vitro-The amino-terminal A␤binding segment of BBP (between the clone 14 and 1 ⁄2tm sites; see Fig. 4A) was expressed in bacteria with amino-terminal His 6 and c-Myc tag sequences. This protein was purified and utilized in in vitro binding assessments with synthetic A␤ peptide. In pull-down experiments, BBP was mixed with disaggregated A␤, and proteins were immunoprecipitated with the anti-A␤ antibody 6E10. Denatured protein complexes were subsequently evaluated using a Ni 2ϩ -peroxidase conjugate to detect the His 6 tag of the BBP protein. As controls, A␤ was omitted or 50-fold excess reverse A␤ 40 -1 was added as a potential competitor of binding. Associated BBP protein could be readily observed in the A␤ immunocomplexes, and the signal was not affected by the addition of reverse peptide (Fig. 4B). These results demonstrate that BBP binds non-fusion A␤ peptide, complementing the Y2H results. Nonspecific binding was determined by measuring BBP bound to wells lacking A␤, and those values were subtracted from total binding to obtain specific binding. D, comparison of BBP binding to aggregated or disaggregated A␤ in vitro. A␤ peptide was disaggregated by suspension in hexafluoro-isopropyl alcohol. A portion of disaggregated peptide was subsequently aggregated as described. These peptide samples were immobilized and incubated with vehicle (PBS ϩ 5% BSA) or vehicle plus purified BBP. Values represent the mean absorbance of triplicate wells, with standard deviation. E, demonstration of specificity of binding. Human A␤ 1-40 , reverse A␤ 40 -1 , or rodent A␤ were added to Myc-tagged BBP (150 nM) over a range of competitor concentrations. Nonspecific binding was subtracted. The IC 50 value for free A␤ 1-40 was ϳ300 nM.
A solid phase assay was developed to initiate pharmacological analyses of BBP/A␤ binding. A␤ was immobilized in 96well plates, and bound BBP was measured with an anti-Myc antibody. Specific binding of BBP to A␤ was determined by subtracting nonspecific binding in wells lacking A␤ peptide. The binding of BBP protein to A␤ exhibited saturable, high affinity characteristics (K d ϭ 150 nM, Fig. 4C). For the binding experiments, A␤ peptide was disaggregated to provide a homogeneous preparation. However, disaggregated A␤ has much reduced neurotoxicity in cell culture. Only A␤ aggregates possess substantial toxic properties (33). BBP binding to wells coated with disaggregated or aggregated A␤ was compared to determine whether BBP could bind neurotoxic-aggregated A␤. BBP bound to A␤ in vitro regardless of the peptide state prior to application (Fig. 4D). BBP binding to A␤ in solution (Fig. 4C) was also independent of peptide aggregation state (data not shown). Binding could be competed by free A␤ with an IC 50 value of ϳ300 nM (Fig. 4E). BBP binding to A␤ was not affected by the addition of peptide of reverse sequence. Rodent A␤ peptide, differing from the human sequence by three amino acid substitutions (positions 5, 10, and 13), also failed to substantially inhibit BBP binding to human A␤ (Fig. 4E). Interestingly, the rodent peptide demonstrates reduced neurotoxicity and an absence of binding to human brain homogenates (33).
BBP Confers Cellular Sensitivity to A␤-Potential involvement of BBP in A␤ toxicity was investigated in cultured human SH-SY5Y cells transfected with vector or pBBP. The efficiency of transfections was 40 -50% (determined by independent transfection of pEGFP). Samples were treated with aggregated A␤ peptide for 48 h and evaluated for viability. Under these experimental conditions, A␤ treatment had no significant toxic effect in control samples (Fig. 5A). However, transfection with  's t test). B, A␤-induced apoptosis in cells transfected with pBBP is transduced through G proteins. SH-SY5Y cells were transfected with pEGFP plus pBBP or pBBP-R3 E expression plasmids. Samples were treated with 10 M A␤, and nuclear morphologies were evaluated in transfected (EGFP ϩ ) cells as described in the text. One pBBP sample was simultaneously treated with pertussis toxin (PTX) at 100 ng/ml to obtain the value labeled pBBPϩPTX. Values are the means of duplicate samples of Ͼ100 EGFP ϩ cells, with S.D. The asterisk indicates significant (p Ͻ 0.01; Yates G-test) effect of pBBP versus vector. C, expression of BBP and BBP-R3 E protein. The image shows an anti-BBP immunoblot of membrane fractions isolated from cells transiently transfected with vector, pBBP or pBBP-R3 E. D, the BBP-mediated response to A␤ is caspase-dependent. Nt2 stem cells were transfected with pEGFP plus vector or pBBP and treated with 10 M A␤. Duplicate pBBP samples were also treated with 25 M BOC-Asp(Ome)-fluoromethylketone (BAF), a nonspecific caspase inhibitor. Samples were scored for apoptotic nuclei and significance determined as described in the legend to B. E, BBP-specific apoptotic response to A␤ is selective for aged (i.e. aggregated) human peptide. Nt2 stem cells were transfected with pEGFP plus vector or pBBP. Samples were treated for 48 h with the indicated peptide at 10 M and examined for nuclear morphology as pBBP resulted in a significant increase in sensitivity to A␤, with an average loss of 22% of total cells (Fig. 5A), indicating that expression of BBP stimulated sensitivity to A␤. Neurons exposed to toxic aggregated A␤ exhibit characteristics of apoptosis before dying (11-13, 15-17, 34). To determine whether BBP-specific A␤ toxicity includes apoptotic events, nuclear morphology assays were conducted. SH-SY5Y cells were doubly transfected with pEGFP plus test plasmids, treated with toxic A␤, and nuclear morphologies of transfected cells were evaluated by fluorescence microscopy following staining with a Hoechst chromatin dye. Included in these experiments was a BBP expression plasmid mutated to substitute glutamate for the arginine in the DRF motif. The corresponding R3 E substitution has been shown to eliminate the activity of 7-tm domain GPCRs (35,36). Transfection with pBBP resulted in a substantial and significant increase in pyknotic nuclei, and this response was prevented by the R3 E substitution (Fig. 5B). An anti-BBP immunoblot of cell lysates is shown to demonstrate that the R3 E substitution does not alter protein expression (Fig. 5C). The absence of a response in the pBBP-R3 E sample suggested that BBP modulates A␤ toxicity by coupling to heterotrimeric G proteins. To further investigate this possibility, samples were treated with the G␣ i/o inhibitor pertussis toxin. This treatment eliminated cellular sensitivity to A␤ via BBP (Fig. 5B). The same results were observed in transfected Nt2 stem cells. Furthermore, Nt2 stem cells transfected with pBBP were treated with the non-selective caspase inhibitor BOC-Asp(Ome)-fluoromethylketone (BAF) to evaluate the involvement of caspases. Treatment with BAF abrogated the induction of nuclear condensation mediated by A␤ in BBP-transfected cells (Fig. 5D). These data were replicated in SH-SY5Y cells. These findings demonstrate that BBP mediates A␤-induced apoptosis by a G protein-regulated caspase-dependent signaling pathway in neurotypic cells.
It is only aged (i.e. aggregated) preparations of human A␤ that elicit substantial toxicity on primary neurons; disaggregated human peptide or aggregated rodent peptide confer greatly reduced toxicity (33,37,38). Cells transfected with pBBP exhibited the same selectivity for A␤ preparations, failing to show effects with disaggregated A␤, aged reverse peptide, or aged A␤ composed of the rodent sequence (Fig. 5E). The absence of a response to A␤ composed of the rodent sequence correlates with the inability of human BBP to interact with this peptide in binding assays (Fig. 4E). These data demonstrate that selectivity for peptide state and type leading to BBP/A␤ toxicity in cell culture matches that required for A␤ toxicity in neurons. Of further note, A␤ toxicity is specific for only the BBP subtype, as no change in apoptotic response to A␤ was observed in cells transfected with BLP1 or BLP2 expression plasmids (data not shown).
Evaluation of Endogenous BBP Activity-The BBP-R3 E variant is unable to mediate an apoptotic response to A␤. Transient transfection assays were utilized to determine whether BBP-R3 E could act as a dominant negative protein, which, if so, would then allow for the possibility of assessing endogenous BBP activities in human neurons. Nt2 stem cells were transfected with pEGFP plus equal quantities of mixed DNAs consisting of either vector, vector plus pBBP, vector plus pBBP-R3 E, or both pBBP plus pBBP-R3 E. These samples were challenged with A␤, and transfectants were scored for nuclear morphology. As shown previously, BBP stimulated A␤-mediated apoptosis, and protein containing the R3 E substitution was inactive. Cells transfected with pBBP plus pBBP-R3 E exhibited the negative phenotype (Fig. 5F), demonstrating that the BBP-R3 E inactive variant is phenotypically dominant over wild-type protein.
Nt2 stem cells can be differentiated into cells possessing the morphological, genetic, and physiological properties of neurons by treatment with retinoic acid (26). BBP mRNA levels were evaluated in Nt2 stem cells and neurons, and a Ͼ20-fold increase in BBP gene expression was observed in the differentiated cells (Table I). Stem cells and neurons were transfected with pEGFP plus vector, pBBP, or pBBP-R3 E, and examined for A␤-induced apoptosis. Results are shown in Table I. Nt2 stem cells became sensitive to A␤ either by differentiation into neurons or by transfection with pBBP. Transfection of neurons with pBBP did not have an additive effect. Transfection of neurons with the pBBP-R3 E dominant negative variant substantially reduced the induction of apoptosis by A␤ exposure, presumably by inhibiting the activity of the endogenous BBP protein. These data indicate that the BBP protein plays a central role in A␤-induced apoptosis in human neurons. DISCUSSION BBP, BLP1, and BLP2 constitute a novel family of proteins containing a module related to the G protein-coupled receptor superfamily. The normal physiological activities of these proteins remain to be elucidated, but the demonstration that the BBP subtype can modulate the apoptotic response to A␤ through a G protein and caspase-dependent mechanism suggests important regulatory roles for BBP/BLPs in cell death or proliferation signal cascades. The involvement of heterotrimeric G proteins in the regulation of apoptotic pathways has been reported (39 -42) but remains only partially characterized. It is possible that BBP/BLPs act as key components of integral membrane protein complexes, providing the molecular trigger for G protein activation in the context of other proteins serving structural roles. We have preliminary evidence that BBP can associate with the amyloid precursor protein APP, which has been shown to modulate apoptosis through binding to heterotrimeric Go protein (41)(42)(43). It has not yet been determined whether the A␤ and APP association domains overlap within BBP. A simple model would place a BBP/APP heteroduplex as the source of a G protein regulated signal transduction cascade modulating apoptosis with the two proteins each providing functional components more commonly contained within a single 7-tm domain protein. The biophysical mechanism by which the binding of aggregated A␤ to such a BBP/APP complex might induce apoptosis remains to be determined, but the numerous correlations between this description of BBP as a regulator of pertussis-sensitive G protein-dependent apoptosis and the findings of Nishimoto and co-workers (41-42) suggest a common pathway. Recently, it was shown that a cell surface protein complex containing APP binds fibrillar A␤ and may contribute to its toxicity (44). It will be important to determine whether BBP is also a component of this complex. The search for molecular mechanisms underlying Alzheimer's disease became more sharply focused with the discovery that the biochemical basis of inherited, aggressive forms of the disease was a phenotypic change in A␤ peptide, manifested as increased generation of total peptide or elevated production of the more amyloidogenic 42-residue form relative to shorter species (1). However, the subsequent mechanisms of neurotoxicity mediated by A␤ are reported to include many different biochemical perturbations, making the elucidation of the critical initiating events challenging. The cumulative data pertaining to BBP are consistent with a possible activity for this protein in these early events. The gene is prominently expressed in neurons in a regional pattern consistent with the known pathophysiology of Alzheimer's disease. BBP binds A␤ peptide with high affinity and selectivity in vitro. In cell culture, the BBP protein selectively responds to aggregated human A␤, failing to respond to disaggregated or rodent peptides. These findings correlate with descriptions of sharply reduced A␤-mediated toxicity on primary neuronal preparations if the peptide is not multimeric or is composed of the rodent amino acid sequence (33,37,38). It is important to note that both aggregated and disaggregated A␤ bound BBP in vitro, yet only A␤ aggregates were able to induce BBP-specific cell toxicity. Demonstration of a physical association between BBP and A␤ in vivo is lacking in this report, leaving open the possibility that BBP expression potentiates A␤ toxicity by a mechanism that does not include direct binding between the molecules. However, the apparent discrepancy between the in vitro binding of disaggregated A␤, and the absence of toxicity in cell culture invokes a potential molecular model for the requisite aggregation of A␤ to achieve toxicity. Ligand-mediated receptor clustering is a common mechanism of signal activation. As one example, parallel dimers of the nerve growth factor protein activate TrkA by bridging the extracellular domains of two receptor proteins, thereby stimulating intrinsic tyrosine kinase activity (45). Similarly, we speculate that A␤ aggregates may promote oligomerization, and consequent activation, of BBP by a bridging mechanism unachievable by A␤ monomers. Central to implicating BBP as a molecular target of A␤ was the finding that a signaling-deficient variant of BBP could block the activity of native BBP in human Nt2 neurons, inhibiting the induction of apoptosis by A␤. These data strongly suggest that the BBP protein regulates neuronal apoptosis initiated by A␤. The discovery of BBP introduces an important new molecule to be considered in the complex pathophysiology of Alzheimer's disease and presents a promising new target in the intensive search for novel therapeutic approaches.