Using the yeast two hybrid system, a mouse embryo cDNA library was screened for proteins that interact with the C-terminus of the human [beta]-amyloid precursor protein ([beta]PP). A fusion protein was identified that interacts specifically with the cytoplasmic domain of [beta]PP and does not interact with the [beta]-amyloid region. The protein encoded by this partial mouse cDNA is identical to the C-terminus of the rat Fe65 protein. This mouse protein also interacts with the homologous C-terminal domains of the mouse amyloid precursor-like proteins, APLP1 and APLP2. These conserved cytoplasmic regions contain a common amino acid motif, Asn-Pro-Thr-Tyr, which has previously been shown to influence both the secretion and internalization of [beta]PP. Fe65 has been implicated in regulatory and cell signaling mechanisms because it contains two different motifs involved in protein binding, a WW domain (a variant of Src homology 3 domains) and a phosphotyrosine interaction domain (PID). Interestingly, the PID domain binds to the same motif present in the conserved cytoplasmic domains of the [beta]PP and [beta]PP-like proteins. RNA analyses reveal that Fe65 is predominantly expressed in brain and in the regions most affected by Alzheimer's disease (AD)-associated neuropathology. The human Fe65 mRNA was cloned from a fetal brain cDNA library. The message encodes a protein of 735 amino acids that is 95% identical to the rat Fe65 protein. The human Fe65 gene was mapped on human metaphase chromosomes to band 11p15 using fluorescence in situ hybridization.
Mutations in the amyloid precursor proteingene (APP) coding for the [beta]-amyloid precursor protein ([beta]PP) have been identified in autosomal dominant forms of familial Alzheimer's disease (AD) (1 -3 ). These mutations result in altered processing of [beta]PP and overproduction of the [beta]-amyloid peptide (A[beta]) which accumulates in senile plaques in the brains of AD patients (4 -6 ). Current views of the pathogenesis of AD emphasize an important role for the amyloidogenicity of the 39-43 amino acid A[beta] fragments of [beta]PP (7 ,8 ). Gene action modulating the levels of expression, proteolytic processing and intracellular trafficking of wild type [beta]PP may influence the rates of A[beta] deposition and AD progression. Two additional AD genes (AD3/S182/PS-1 and AD4/STM2/PS-2) (9 -11 ) have been implicated in [beta]PP processing. For example, the AD3 mutation causes both an increase in the secretion of amyloidogenic forms of A[beta] (12 ) and a decrease in the age at which dementia occurs (13 ).
[beta]PP was originally described as a cell surface receptor with a single transmembrane domain (14 -16 ). It is post-translationally modified and a fraction of full-length forms is then transported to the plasma membrane where it undergoes a complex series of events that involve secretion or internalization, degradation, and recycling of various proteolytic fragments (4 ,17 ). It must therefore interact with a variety of different proteins that determine its fate. These proteins could be involved in mechanisms such as membrane trafficking, signal transduction, or proteolytic processing. In order to identify a subset of such proteins (those that specifically interact with the amyloidogenic C-terminal domain of human [beta]PP) (18 ) we screened a mouse embryo cDNA library using the yeast two hybrid system for proteins that interact specifically with the C-terminus of human [beta]PP.
A particularly strong protein-protein interaction led to the identification of a mouse cDNA whose predicted protein sequence is identical with the last one-third of the rat Fe65 protein. This protein was originally described as a neural-specific transcriptional activator with homology to the DNA binding domain of retroviral integrases (19 ). Two database searches for conserved protein-binding motifs (20 ,21 ) have identified two different protein binding domains, WW and PID types, within the Fe65 protein. These particular domains (see Discussion) are common among proteins known to function in receptor-mediated, membrane-associated signal transduction mechanisms.
The mouse Fe65 fusion protein was also shown to bind to the homologous C-terminal regions of the evolutionarily conserved [beta]PP-like proteins, mouse APLP1 and APLP2. These observations confirm recent findings from a reciprocal two-hybrid study in which the C-termini of [beta]PP and APLP1 were identified in a screen for proteins that interact with the rat Fe65 protein (22 ). Alignment of the mouse APLPs with the human [beta]PP C-terminal sequences involved in Fe65 binding reveals a conserved protein binding motif that has been implicated in signal transduction (20 ,23 ,24 ) and reinternalization of [beta]PP via clathrin-mediated endocytosis (4 ,25 -27 ). These findings suggest that Fe65 may influence [beta]PP trafficking and that [beta]PP may be involved in Fe65-mediated signal transduction pathways.
The mouse APLPs, like [beta]PP, are transcriptionally active in brain (28 ,29 ). We find that Fe65 has a similar spatial pattern of expression in mouse brain, and is abundantly expressed in human brain. We have characterized a novel human Fe65 cDNA which is highly homologous to the rat sequence; however, the human sequence predicts a unique 78 amino acid N-terminal extension not present in the rat protein. The human Fe65 gene maps to human chromosome 11p15.
The specificity of the protein-protein interaction between the mouse Fe65 fusion protein and the [beta]C region of [beta]PP was examined by subdividing the [beta]C region of the original Lex-[beta]C fusion protein into A[beta] and C-terminal domains (Fig. 1 A). The haploid L40 yeast strain containing the pVP16-cDNA clone was mated with haploid JC1 strains containing Lex-[beta]C, Lex-A[beta], Lex-C48, and Lex-lamin fusion protein expression vectors. Diploid yeast were selected and the [beta]-galactosidase filter assay was performed in order to assess transcriptional activation of the Lex A-driven [beta]-galactosidase reporter gene due to interactions between LexA-[beta]PP fusion proteins and the VP16-mouse Fe65 fusion protein, clone 230 (Fig. 1 B). The Lex-C48 clone began to show [beta]-galactosidase activity within 10 min of the addition of substrate; Lex-[beta]C became visible within 20-30 min, as seen in previous experiments. There was no detectable activity after 2 days, however, with the Lex-A[beta] and Lex-lamin clones. These results clearly demonstrate that Fe65's interaction with [beta]PP occurs within the cytoplasmic C-terminal 48 amino acids. There is no apparent specificity for binding within the region containing the [beta]-amyloid peptide.
The mouse APLP1 and APLP2 proteins share a large amount of sequence similarity with [beta]PP in their N-terminal and C-terminal domains (28 ,29 ). The region of these proteins corresponding to the A[beta] domain of [beta]PP, however, is less well conserved (Fig. 2 A). The C-terminal domains of the [beta]PP-like proteins mouse APLP1 and APLP2 directly aligning with the [beta]C region of [beta]PP were cloned, in frame, with the LexA protein in order to create Lex-APLP1 and Lex-APLP2 fusion protein expression constructs. Haploid L40 yeast strains containing the Fe65 pVP16-cDNA clone or pVP16-perlecan were mated with haploid JC1 strains containing Lex-[beta]C, Lex-APLP1 and Lex-APLP2 fusion protein expression vectors. Diploid yeast were selected and the [beta]-galactosidase filter assay was performed in order to assess transcriptional activation of the Lex A-driven [beta]-galactosidase reporter gene due to interactions between Lex-APP gene family C-terminal fusion proteins and the VP16-mouse Fe65 fusion protein, clone 230 (Fig. 2 B). During this filter assay, [beta]-galactosidase activity was observed within 10-15 min with the Lex-APLP2 fusion protein. Lex-[beta]C became visible within 20-30 min, and the Lex-APLP1 fusion protein construct developed more slowly, becoming appreciable 1 h after the addition of the [beta]-galactosidase substrate. There was no activity observed for 2 days with any of the Lex A-fusion proteins when combined with the pVP16-perlecan construct whereas each of the Lex-APP gene family fusion proteins became saturated with visible [beta]-galactosidase activity within 2-3 h when combined with the VP16-Fe65 fusion protein.
Although we did not perform quantitative liquid [beta]-galactosidase assays on extracts of yeast containing these two-hybrid interactions, the time-dependent kinetics of initial color development during the [beta]-galactosidase filter assay may somewhat approximate the relative strength of binding. These assays were repeated more than six times and the relative timing of these events was consistently proportional. These results demonstrate that the ability of the Fe65 protein to bind to the C-terminal region of [beta]PP is conserved within the gene family members APLP1 and APLP2. These results may also suggest that a hierarchy of Fe65 binding exists with APLP2 having the strongest, [beta]PP being slightly weaker, and APLP1 having the weakest interaction.
In order to support further the in vivo yeast data for the existence of an interaction between the C-terminus of [beta]PP and the VP16-cDNA mouse clone 230 fusion protein, glutathione-agarose affinity chromatography was used to determine whether these proteins can be co-precipitated in vitro. GST and GST-C48 fusion proteins were expressed in E. coli and purified using glutathione-agarose affinity chromatography. Equal amounts of in vitro translated (35S-labeled) VP16 or VP16-Fe65 protein were incubated with GST and GST-C48 bound to glutathione-agarose beads or incubated with anti-VP16 antibodies bound to protein A-agarose beads. After 2 h of incubation at 4oC, the protein-protein complexes were washed, and the retained 35S-labeled proteins were electrophoretically separated on a 14% Tricine-SDS-polyacrylamide gel and processed for autoradiography (Fig. 3 , right panel). GST protein and anti-VP16 antibodies were used as negative and positive control reagents for precipitating 35S-labeled VP16 proteins. GST was not able to significantly precipitate VP16 or VP16-Fe65 fusion proteins whereas the anti-VP16 antibodies were able to immunoprecipitate both forms of 35S-labeled VP16 protein. GST-C48 had no apparent affinity for the VP16 protein alone, but showed substantial affinity for the VP16-Fe65 fusion protein. Therefore, the interaction (originally identified using the yeast two-hybrid transcriptional reporter system) between the C-terminus of human [beta]PP and the mouse Fe65 protein can be demonstrated to occur at the biochemical level under in vitro conditions.
Total RNA was isolated from various human and mouse tissue samples (18 ) in order to estimate relative levels of mouse Fe65 mRNA expression in the adult soma. Ten [mu]g of total RNA from each tissue was separated on a formaldehyde-agarose gel, blotted, probed with 32P-labeled mouse clone 230 cDNA, and processed for autoradiography (Fig. 4 ). The expression of clone 230 appears to be highly enriched in brain in both human and mouse RNAs. This finding is in agreement with the original description of the rat Fe65 mRNA being expressed primarily in brain (19 ). The mouse probe hybridizes with two major transcripts of approximately 2.3 and 2.8 kb present in mouse brain. Although they appear less intense, two very similarly sized transcripts can also be detected in human brain and, even more weakly, in mouse testis. The longer transcript appears to be more abundantly expressed in brain tissue than the shorter one. In mouse testis the two transcripts appear to be of equal intensity. Other tissues examined exhibit very low levels of expression; thus, only the larger-sized transcript is perceptible. The cross-hybridization of the mouse probe with human RNA under high-stringency conditions suggests conservation exists between the human and rodent sequences.
We have carried out in situ hybridization studies to determine the pattern of Fe65 mRNA expression in the adult murine brain. Oligonucleotide probes and riboprobes were hybridized to sagittal and coronal sections through the adult murine brain. The signals from these probes indicate widespread expression of Fe65 in the brain. The strongest signal was present in the regions with the highest density of neurons including the cerebellum, cortex, hippocampus, medial habenular nucleus and the olfactory bulb.
Oligonucleotides complementary to sequences near the 5' and 3' ends of the human Fe65 mRNA were end-labeled by terminal deoxynucleotidyl transferase and used for in situ analyses on sagittal sections of mouse brain. Both oligonucleotides exhibit similar patterns of hybridization to mouse hippocampal, cortical and cerebellar regions (Fig. 5 A,B). The similar pattern of hybridization signal from both 5' and 3' probes indicates that these oligonucletotides are specific for Fe65 mRNA. Specificity of oligonucleotide labelling is also supported by the observation that the signal was eliminated by the addition of excess unlabeled oligonucleotides to the hybridization mixture (not shown).
To characterize the human Fe65 protein coding sequence, the 531 bp mouse clone 230 cDNA was used to probe a [lambda]ZAPII phage library of human fetal brain cDNAs (Stratagene, La Jolla). In a single screen of 5 * 105 recombinant plaques, 51 primary clones were selected and 30 clones were enriched to purity. PCR was performed using oligonucleotide primers corresponding to the T3 and T7 promoters at the opposing ends of each cDNA in order to determine individual clone sizes. cDNAs were sequenced at their ends to determine their identity and relative overlap, and the largest clones were sequenced in their entirety. Clones containing the longest 5' ends were identified by PCR and sequenced using T3 or T7 primers in combination with reverse primers designed from internal human Fe65 coding sequence.
The human Fe65 mRNA consists of 2859 nucleotides before polyadenylation occurs (GenBank Accession #L77864). Human Fe65 mRNA is 88% similar to the rat Fe65 mRNA. This level reflects high conservation in the protein coding region. The similarly sized 400 nucleotide 3' untranslated regions are less conserved. The 5' untranslated region of the human message extends approximately 240 bases upstream of the longest open reading frame.
The human Fe65 mRNA encodes a putative protein of 735 amino acids (Fig. 6 ). The predicted protein sequence is 95% similar to the predicted 658 amino acid rat Fe65 protein (19 ,30 ). The human and rat protein sequences are essentially identical within their conserved WW and PID domains. The 177 amino acids encoded by the mouse Fe65 cDNA clone, the [beta]PP-interacting region of Fe65, encompass the second more highly conserved PID domain and are also identical with the rat and human proteins (Fig. 6 ). Despite a unique 78 amino acid N-terminal extension in the human Fe65 protein coding sequence (not reported for the rat protein) these proteins appear to be highly conserved between species.
The Fe65 gene was mapped on human chromosomes by FISH using a 2.5 kb human Fe65 cDNA probe. Of 72 cells examined, 24 (35%) had signals on both chromatids of the distal end of chromosome 11 at band 11p15 (Fig. 7 ). There was no significant hybridization to other chromosomes.
We have identified a fragment of a mouse protein (Fe65) that interacts strongly with the cytoplasmic domain of the human [beta]-amyloid precursor protein using the yeast two hybrid system. The sequence encoded by this mouse cDNA clone is identical to a region of the rat Fe65 protein. This mouse protein was also shown to interact with conserved cytoplasmic domains of the [beta]PP-like proteins, APLP1 and APLP2. Northern analyses of RNA derived from human and mouse tissues reveals the expression of two transcripts of 2.8 and 2.3 kb in brain. The longer transcript is also observed in other tissues examined but at a lower level of expression. The size of the longer transcript agrees with the 2859 bp sequence we determined for the human message. This message contains an open reading frame of 2205 bp which encodes a 735 amino acid protein. The protein sequence of human Fe65 is 95% identical to the rat Fe65 protein. In addition, the human protein which is encoded by a gene on chromosome 11 band p15 contains an additional 78 amino acids at its amino terminus. In situ analyses reveal that Fe65 is expressed in areas of the brain with the highest neural density, including the hippocampus and the cortex, two important regions of the brain involved in AD-associated neuropathology.
The mouse Fe65 fusion protein was identified in a yeast two hybrid screen by virtue of its interaction with the C-terminal 106 amino acids of human [beta]PP. Not only was this the strongest interacting protein identified in our screen, but upon further examination, it was the only clone that was able to bind specifically to the C-terminal 48 amino acids of human [beta]PP and not to the A[beta] region. These in vivo results in yeast were confirmed in vitro in E.coli extracts using GST affinity chromatography and co-precipitation.
Human [beta]PP and the related mouse APLPs are a family of proteins sharing both sequence and structural similarity. The Fe65 fusion protein interacts specifically with the C-terminus of each of these proteins. The C-termini of the APLP1 and APLP2 proteins are 48% and 70% identical with the C-terminus of human [beta]PP, respectively. The fact that the interaction occurs within a conserved region implies that Fe65 binds to a common protein motif and that Fe65 may play a common role in the biology of these related, membrane-spanning proteins. Our observations suggest that Fe65 has higher affinities for APLP2 and [beta]PP than for APLP1. This may be explained by the greater C-terminal sequence divergence of APLP1 (Fig. 2 A).
We have determined the sequence of the human Fe65 mRNA by sequencing fetal brain cDNAs. The predicted size of the human Fe65 message agrees well with the 2.8 kb transcript observed in northern analyses of RNA derived from adult human and mouse brains. Our northern analyses also detect a novel 2.3 kb Fe65 transcript. The structure of this minor transcript is currently unknown. Two transcriptional start sites have previously been identified within the rat Fe65 gene (30 ). Ribonuclease protection experiments revealed the existence of two messages differing by the addition of a single 24 nucleotide exon to the 5' end of the Fe65 mRNA sequence. This gene also contains a six nucleotide miniexon which is absent from Fe65 mRNA in non-neural tissues (19 ). These findings would predict that the various species of Fe65 mRNAs, although having the potential for being differentially expressed or regulated, would not vary in size by more than approximately 30 nucleotides. These earlier descriptions of rat Fe65 mRNA do not account for the difference in size that we observed (approximately 500 bases) between the two transcripts in mouse and human brain.
Both northern blot analyses and in situ hybridization studies suggest that Fe65 is expressed throughout the murine brain. The pattern of Fe65 expression corresponds to brain regions with the highest density of neurons. This pattern indicates that Fe65 is highly expressed in the regions particularly affected by AD-associated neuropathology, the hippocampus and cortex (32 ,33 ).
The high level of conservation between the mouse, rat, and human Fe65 protein sequences and the conserved interaction of Fe65 with the C-termini of the APP family members both suggest an evolutionarily conserved function for Fe65. In addition, a similar yeast two-hybrid study has identified an association between the C-terminus of [beta]PP and a human Fe65-like protein exhibiting a lesser degree of sequence similarity to rat Fe65 protein (34 ). Therefore, Fe65 appears to be a member of a gene family in which binding to the C-termini of the APP family members is conserved.
The Fe65 protein contains two different consensus motifs for protein binding (20 ,21 ). These domains are present in families of proteins that function in plasma membrane-associated signal transduction mechanisms. The WW domain, a variant of the Src homology 3 domain, is involved in protein-protein interactions with proline-rich ligands (21 ,30 ). The PID, a SH2-like domain, is specific for binding to peptide sequences containing phosphotyrosine residues (23 ). This motif, present twice in Fe65, was used in a reciprocal two-hybrid study (22 ) to screen a human brain cDNA library for proteins that interact with this particular domain of the rat Fe65 protein. Interestingly, this group obtained three interacting clones, two of which were identified as C-terminal fragments of the [beta]PP protein; the third clone was a C-terminal fragment of the human APLP1 protein. These results are in complete agreement with our findings of the conserved binding between the mouse Fe65 protein and the [beta]PP protein family members. Furthermore, an examination of the overlap between the Fe65 domains used in these studies reveals that the second PID domain in Fe65 is sufficient for [beta]PP binding (Fig. 8 ).
The PID domain has specificity for the peptide sequence Asn-Pro-X-Tyr(P), a binding motif believed to be dependent upon phosphorylation of the tyrosine residue, and is found in the intracellular domain of some growth factor receptors (24 ). This four amino acid sequence has also been described as a consensus for clathrin-mediated endocytosis (25 ). Interestingly, this sequence is present at the C-terminus of each of the APP family members. Previous studies on [beta]PP processing (4 ,26 ,27 ,36 ,37 ) have implicated this C-terminal domain as being necessary for reinternalization and targeting of [beta]PP through the alternative endosomal-lysosomal pathway that results in the generation of A[beta]. At present, there is no evidence for phosphorylation of the tyrosine residue within the Asn-Pro-Thr-Tyr motif in the [beta]PP protein (Tyr686of [beta]PP695). In addition, it remains to be seen if the Fe65 interaction with [beta]PP is strictly dependent upon these particular amino acid residues. Nevertheless, this interaction implicates Fe65 in the regulation of [beta]PP membrane trafficking events. Site-directed mutagenesis, peptide-inhibition and [beta]PP internalization studies will be required to resolve these issues.
The discovery of interactions between the Fe65 protein and the [beta]PP family of proteins by three independent groups (22 ,34 , and this paper) together with the likelihood that these interactions are involved in [beta]PP processing, trafficking and signal transduction, should stimulate new and exciting research on the pathogenesis of AD. Futhermore the presence of both SH2- and SH3-like domains as well as putative DNA binding- and transcriptional activation-domains within the Fe65 protein suggests that Fe65 may be a new member of the growing family of STAT proteins, whose members function as signal transducers and activators of transcription (38 ,39 ). This would provide a novel role for [beta]PP in neuronal-specific signal transduction and gene regulatory mechanisms. Of particular interest is the possibility that polymorphisms or mutations at the human Fe65 locus may influence individual susceptibility to AD via alterations in such regulation. We are unaware of any data suggesting a link between AD and chromosome 11. However, we are currently investigating this possibility as well as assessing whether the interaction between [beta]PP and Fe65 occurs at physiological concentrations of these proteins inside neuronal cells. We are also investigating whether specific inhibition of this protein interaction affects the secretion and production of the various forms of the [beta]-amyloid peptide in vivo.
The yeast two-hybrid transcriptional reporter selection system as originally described (40 ) and modified (41 ) was used to screen a size-selected mouse embryo (9.5-10.5 d.p.c.) cDNA-fusion library cloned into the pVP16 plasmid (kindly provided by Dr Stanley Hollenberg). VP16-cDNA fusion proteins were selected by virtue of their ability to interact with the 106 C-terminal amino acid residues of the human [beta]PP protein ([beta]C), amino acids 590-695, of [beta]PP695, (14 ). [beta]PP cDNA corresponding to [beta]C was fused, in frame, to the cDNA sequence encoding the first 211 amino acids of the Lex A protein in the expression vector pBTM116. [beta]C-interacting clones were identified by the activation of transcriptional reporters for both histidine prototrophy and [beta]-galactosidase activity in the S.cerevisiae strain L40 (MATa). Yeast strain JC1 (MAT[alpha]) (41 ) was used in mating assays, and the [beta]-galactosidase-filter assay was performed essentially as described (42 ).
LexA-fusion protein expression constructs were also created with human [beta]PP domains A[beta] (56 amino acids, 590 to 645 of [beta]PP695) and C48 (48 amino acids, 648 to 695 of [beta]PP695), and the C-terminal domains of mouse APLP1 and APLP2 (C-terminal 80 amino acids of APLP1, and C-terminal 106 amino acids of APLP2) in the pBTM116 vector. VP16-perlecan (domain V, gift of J. Hassell) fusion protein and Lex-lamin C fusion protein, plasmid pLam5 (40 ), were used as negative controls in mating and reconstitution studies with Lex-[beta]C and VP16-230 plasmids, respectively.
pVP16 cDNA library clones were amplified using the polymerase chain reaction (PCR) (43 ). The 5' oligonucleotide primer was derived from the DNA sequence directly upstream of the NotI cloning site within the VP16 cDNA sequence (5'-GAG TTT GAG CAG ATG TTT A-3'). The 3' oligonucleotide primer used was the forward M13 Universal primer (5'-GTT GTA AAA CGA CGG CCA GT-3'), which is cloned in the reverse orientation at the 3' end of the VP16 cDNA sequence in the pVP16 plasmid. Dideoxy-DNA sequencing (44 ) was performed using these PCR primers by facilities at the Department of Genetics at University of Pennsylvania and the Division of Biological Sciences at the University of Montana. Database searches were performed using the BCM Search Launcher for Nucleic Acid Sequence Searches from the Human Genome Center, Baylor College of Medicine, Houston TX (www site http://dot.imgen.bcm.tmc.edu:9331/seq-search/nucleic_acid-search.html). Sequence alignments and statistical analyses were performed using the WISCONSIN PACKAGE of GCG, Version 8.1-UNIX (Genetics Computer Group, Inc., Madison, Wisconsin)
Glutathione S-transferase (GST) expression vector pGEX-4T-2 (Pharmacia Biotech) was used to create a fusion protein between GST and the 48 C-terminal amino acids of the human [beta]PP protein (GST-C48, amino acids 590-695, of [beta]PP695). GST and GST-C48 were expressed in E.coli and purified using glutathione-agarose affinity chromatography as previously described (45 ). VP16 protein and VP16-clone 230 fusion protein were labeled with 35S-methionine during in vitro translation using the TNT T7-coupled transcription-translation reticulocyte lysate system (Promega). Equal amounts of 35S-labeled VP16 or VP16-230 reticulocyte lysate were diluted into 200 [mu]l of binding buffer (50 mM Tris, pH 7.5, 150 mM NaCl) and incubated for 2 h at 4oC with affinity purified GST or GST-C48 fusion protein bound to glutathione-agarose beads, or anti-VP16 rabbit polyclonal antibodies (Upstate Biotechnology Inc., Lake Placid, NY) bound to protein A-agarose beads. After the incubation period, protein-protein complexes were washed three times in binding buffer and the retained 35S labeled proteins were then boiled in SDS sample buffer and resolved on a 14% Tricine-SDS- polyacrylamide gel (46 ). The gel was stained with coomassie blue to assess the relative levels of GST and GST-C48 used in the precipitation step and then soaked in EN3HANCEtm (DuPont) for 30 min, dried, and subjected to X-ray autoradiography.
Northern blot analyses were performed essentially as previously described (18 ). For northern blots, 10 [mu]g of total RNA was purified from various human and mouse tissues, separated on a 1.2% agarose gel, and blotted on to nylon Hybond-C membrane (Amersham). Northern blots were probed with randomly primed 32P-labeled mouse clone 230 cDNA. All hybridizations were performed under high-stringency conditions (47 ).
In situ hyridization assays using radiolabeled probes were carried out according to methods previously described for cRNA probes (48 ). Twenty [mu]m sections were cut from quick frozen brains extracted from (C57BL/6 * DBA/2) F1 hybrid mice and thaw-mounted on to RNAse-free silanized slides. Oligonucleotides complementary to the 5' and 3' ends of the human Fe65 coding sequence were synthesized (5' probe, 5'-GGC CTC CTC CGC CAA GGT CAA GGT CAC ATT GCG ATT CTG GTC-3'; 3' probe, 5'-CAA GTT TAG AGT GGT CCA GGG AGA GTC CAT TTA CCA AGC AGC G-3'; Integrated DNA Technologies, IA) and labeled with 35S-ATP using terminal deoxynucleotidyl transferase (Promega, WI). Approximately 1.8 * 106 d.p.m. of 5' oligonucleotide or 1.3 * 106 d.p.m. of 3' oligonucleotide (2 pmol/ml) was applied to individual slides in 45 [mu]l of hybridization buffer. Slides were incubated in moist chambers overnight at 30oC, and high stringency post-hybridization washes were carried out at 58oC.
For riboprobe analyses, mouse Fe65 cDNA clone 230 was subcloned between the Sp6 and T7 RNA polymerase promoters in the pGEM-7Zf(+) vector (Promega, WI). Sp6-sense and T7-antisense riboprobes were transcribed from linearized vectors using 35S-UTP and the appropriate RNA polymerase enzyme (Promega, WI). Approximately 1.2 * 107 d.p.m. of sense probe or 7.7 * 106 d.p.m. of antisense probe was applied to each slide at a concentration of 2 pmol/ml in hybridization buffer. Slides were incubated in moist chambers overnight at 55oC, and high stringency post-hybridization washes were carried out at 62oC. After stringency washes, sections were dehydrated and apposed to film (Hyperfilm [beta]Max; Amersham, IL). Autoradiographs were digitized using the MCID computerized image-analysis system (Imaging Research Inc., Ontario).
An oligo(dT) and randomly primed human fetal brain cDNA library constructed in bacteriophage [lambda]ZAPII (Stratagene, La Jolla) was screened in E.coli strain XL1-blue (49 ) with the 531 bases of the mouse cDNA fragment originally isolated using the two hybrid system. Sequence determinations and database searches were performed as described above and in the text.
A 2.5 kb human fetal brain Fe65 cDNA clone, clone HB6, was labeled with biotin-11-dATP and hybridized to metaphase chromosomes derived from a normal male individual using the method of Edelhoff et al. (50 ). Hybridized chromosome spreads were processed for immunofluorescence, visualized and photographed using fluorescence photomicroscopy.
We would to thank Dr Stanley Hollenberg for his guidance and for generously providing the mouse embryo cDNA library in the plasmid vector pVP16 and the L40 S.cerevisiae reporter strain; Drs Stanley Fields and Paul Bartel for the pBTM116 vector; Dr John Hassell for providing the perlecan cDNA; Dr Enrique Villacres for the human fetal brain cDNA [lambda]ZAPII library; Dr Junko Oshima for her assistance in final sequence determinations; and David Adler for his advice and expertise in bioinformatics. This work was supported in part by NIH Grants AG10917 (G.M.M. and M.A.M.), AG00057 (G.M.M., P.I.; Q.H., M.G.H. and B.L.S.), the University of Washington's Nathan Shock Center for Excellence in the Basic Biology of Aging-Yeast Genetics Core AG13280 (Peter S. Rabinovitch, P.I.; S.L.B. and M.D.G.), GM07454 (George Stamatoyannopoulos, P.I.; M.B.D.), AG12850 (K.F.), NS33606 (M.A.M.), GM46883 (C.M.D.) and the March of Dimes Birth Defects Foundation I-0409 (C.M.D.).
Human Molecular Genetics
Pages
Introduction
Results
The C-terminal domain of [beta]PP interacts with the mouse Fe65 protein
The C-terminal, cytoplasmic domain of human [beta]PP interacts specifically with mouse Fe65
Conservation of binding among the C-termini of the APP-like proteins and Fe65
In vitro interaction of the C-terminus of human [beta]PP with mouse Fe65 protein
Distribution of Fe65 mRNA in human and mouse tissues
Expression pattern of Fe65 mRNA in mouse brain
Identification of the human Fe65 messenger RNA
Chromosome localization of the human Fe65 gene
Discussion
Materials And Methods
Yeast two-hybrid system
DNA sequencing and database searching
GST- and amyloid-affinity chromatography
Northern analysis
In situ hybridization assays
Isolation and sequencing of human Fe65 cDNA clones
Fluorescence in situ hybridization
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Thu Oct 31 15:27:49 GMT 1996. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996