Molecular Studies Define the Primary Structure of α1-Antichymotrypsin (ACT) Protease Inhibitor in Alzheimer’s Disease Brains

An α1-antichymotrypsin-like serpin has been implicated in Alzheimer’s disease (AD) based on immunochemical detection of α1-antichymotrypsin (ACT) in amyloid plaques from the hippocampus of AD brains. The presence of neuroendocrine isoforms of ACTs and reported variations in human liver ACT cDNA sequences raise the question of the molecular identity of ACT in brain. In this study, direct reverse transcription-polymerase chain reaction and cDNA sequencing indicate that the hippocampus ACT possesses the reactive site loop that is characteristic of serpins, with Leu as the predicted P1 residue interacting with putative chymotrypsin-like target proteases. The deduced primary sequence of the human hippocampus ACT possesses more than 90% homology with reported primary sequences for the human liver ACT. Moreover, identical ACT primary sequences deduced from the cDNAs were demonstrated in the hippocampus of control and AD brains. Northern blots showed that ACT mRNA expression in hippocampus was 900 times lower than that in liver. Also, hippocampus and liver ACT proteins demonstrated differential sensitivities to deglycosylation. Overall, reverse transcription-polymerase chain reaction combined with cDNA and primary sequence analyses have defined the molecular identity of human hippocampus ACT in control and AD brains. The determined reactive site loop domain of hippocampus ACT will allow prediction of potential target proteases inhibited by ACT in AD.

A role for the protease inhibitor ␣ 1 -antichymotrypsin in Alzheimer's disease (AD) 1 has been suggested based on immunochemical detection of ACT in amyloid plaques in brains of AD patients (1,2).
ACT is a member of the serine protease inhibitor family, known as serpins, which typically possesses a reactive site loop (RSL) domain that interacts with target proteases (9,10). Recently, molecular cloning has identified isoforms of ACT in bovine neuroendocrine tissues of adrenal medulla and pituitary that differ in their RSL domains (3,4). Differences in RSL predict that these ACT isoforms inhibit different target proteases; indeed, expression of these isoforms have demonstrated the protease-specific nature of these ACT isoforms. 2 Furthermore, variations in the deduced primary sequences of human liver ACT cDNAs have been reported (5)(6)(7)(8). Because these ACT isoforms are all recognized by anti-ACT sera, these observations raise the question of the molecular identity of ACT-like immunoreactivity in Alzheimer's disease brains. However, the primary structure of ACT encoded by human brain ACT cDNA has not been elucidated.
Therefore, the goal of this study was to determine the molecular identity of the ACT cDNA expressed in AD and normal hippocampus, a brain region abundant in amyloid plaques in AD, as well as to characterize the brain ACT. Direct reverse transcriptase polymerase chain reaction (RT-PCR) and DNA sequence analyses have defined the primary sequence of ACT in Alzheimer's and normal brains. Moreover, the defined primary sequence of the human hippocampus ACT cDNA sequence resolves its identity compared with reported variations in human liver ACT cDNA sequences (5)(6)(7)(8). Primary sequence comparisons indicate that the human hippocampus and liver ACTs resemble one another with greater than 90% homology. Further analyses of hippocampus and liver ACTs were performed with respect to transcription initiation sites and expression of the ACT gene as well as the glycoprotein nature of ACT. This study has, thus, defined the primary sequence and characteristics of ACT expressed in control and Alzheimer's disease brains.

RT-PCR and DNA Sequencing of ACT cDNAs from Hippocampus of Alzheimer's Disease and Normal
Brains-To obtain the segment of the hippocampus ACT cDNA corresponding to the predicted open reading frame (ORF) encoding the primary sequence of the hippocampus ACT, RT-PCR and DNA sequence analysis of overlapping 790-and 502-bp 5Ј and 3Ј cDNA fragments, respectively, of human hippocampus ACT cDNA from AD and normal brains was performed (see Fig. 1). In addition, RT-PCR also generated a 296-bp DNA fragment that represents the 3Ј-untranslated region (UTR) that overlaps with the 502-bp cDNA fragment. PCR primers were designed based on reported human liver ACT cDNA sequences (5)(6)(7)(8). RT-PCR of poly(A) ϩ RNA from normal and Alzheimer's hippocampus was conducted three times, each time with RNA isolated from a separate sample of tissue; two to four subclones from each PCR reaction were analyzed by DNA sequencing.
Total RNA was isolated from frozen hippocampus tissue from AD and normal brains with the TRIZOL TM reagent (Life Technologies, Inc.) (tissues were from the Harvard Brain Tissue Resource Center at McLean Hospital, Belmont, MA and the First Department of Anatomy, Semmelweis University, Budapest, Hungary). Frozen tissue (100 mg aliquots) was pulverized in liquid N 2 , solubilized in 1.0 ml of TRIZOL reagent, extracted with 0.2 ml chloroform, isoamylalcohol (49/1, v/v), and incubated at room temperature for 5 min. The sample was then centrifuged at 12,000 ϫ g at 4°C for 15 min, and the resultant RNA in the aqueous phase was precipitated by isopropanol and resuspended in 50 l of TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA).
Isolation of poly(A) ϩ RNA from total RNA utilized the Poly(A)Tract mRNA Isolation System IV (according to the manufacturer's protocol; Promega). Poly(A) ϩ in the RNA sample (100 g) was annealed to the biotinylated oligo(dT) probe at room temperature. The oligo(dT)poly(A) ϩ hybrid was bound to streptavidin paramagnetic particles and washed with 0.1ϫ SSC (1ϫ SSC, 0.15 M NaCl and 0.015 M sodium citrate), and bound poly(A) ϩ RNA was eluted with diethyl pyrocarbonate-treated water and concentrated by ethanol precipitation. In addition, poly(A) ϩ RNA from normal human hippocampus and liver were purchased (CLONTECH) for RT-PCR.
RT-PCR of overlapping 790-and 502-bp cDNA fragments that span the ORF of the ACT utilized SuperScript II reverse transcriptase (Life Technologies, Inc.) and PCR reagents from Perkin-Elmer (according to the manufacturer's protocols). RT-PCR of a 790-bp 5Ј-fragment of ACT utilized the primers 5Ј-GGTTCTGCCCTGCTGTCCTCTGC-3Ј (sense, primer 1) and 5Ј-GGGAGGATGAAGAGTGCGCTGGC-3Ј (antisense, primer 2). RT-PCR of a 502-bp 3Ј-fragment of ACT utilized the primers 5Ј-CGGGACGAGGAGCTGTCCTGCA-3Ј (sense, primer 3) and 5Ј-TGCTGGGATTGGTGACTTTGCTCAT-3Ј (antisense, primer 4) (primers were from Life Technologies, Inc.). First strand cDNA synthesis utilized 0.5 g of poly(A) ϩ RNA and antisense primer (0.15 M) with SuperScript II reverse transcriptase (200 units) at 50°C for 50 min. After the addition of MgCl 2 to 1.5 mM and sense primer (0.15 M), PCR with Taq polymerase (5 units) was conducted with 40 cycles of 1 min at 94°C, 1 min at 60°C, and 1 min at 70°C, with a final step at 70°C for 7 min. PCR products were analyzed by DNA agarose gel electrophoresis. Attempts to obtain the full-length coding region of the ACT cDNA by one RT-PCR reaction (primers 1 and 4) were not successful, most likely because of the lower efficiency of PCR in amplifying larger DNA fragments. Therefore, the segment of the hippocampus ACT cDNA corresponding to the predicted ORF was assessed by DNA sequence analyses of the overlapping 790-and 502-bp PCR fragments.
To obtain a cDNA fragment representing the 3Ј-UTR of human hippocampus ACT, RT-PCR was utilized to generate a 296-bp fragment that overlaps with the 502-bp PCR-generated fragment. The first strand DNA synthesis utilized 2 g of human hippocampus poly(Aϩ) RNA (CLONTECH) primed with 0.5 g of oligo(dT) 20 and incubated with SuperScript II reverse transcriptase (200 units, Life Technologies, Inc.) at 42°C for 1 h. PCR was then conducted with sense and antisense primers at 0.4 M (primers 5 and 6, respectively) consisting of 5Ј-CAGACACCCAGAACATCTTCTT-3Ј and 5Ј-GGCCAACGAAATTATT-TATTGCTG-3Ј, respectively, Taq polymerase (0.5 unit, Qiagen), and thermocycling consisting of 35 cycles of 50 s at 94°C, 1 min at 42°C, and 1 min at 72°C, with a final incubation at 72°C for 10 min.
PCR-generated DNA fragments were subcloned (by ligation with T4 DNA ligase) into the PCR 2.0 plasmid vector (Invitrogen) and amplified in XL-1 Blue supercompetent Escherichia coli cells (Stratagene). DNA inserts of appropriate size (assessed by digestion of the plasmid with EcoRI) were subjected to automated DNA sequencing with fluorescentlabeled dideoxynucleotides and the Applied Biosystems 373A automated DNA sequencer, as well as manual DNA sequencing (with the U. S. Biochemical Corp. Sequenase Version 2.0 sequencing kit, Amersham Pharmacia Biotech), as described previously (3,4). Primers for DNA sequencing utilized reverse and forward primers corresponding to vector M13 sequences that flank the DNA insert. DNA sequencing of the 5Ј fragment also required primers corresponding to the ACT cDNA to walk the sequencing along the length of the cDNA. These sequencing primers were designated ACT N1, ACT N2, and ACT N3, which corresponded to 5Ј-TGTCTCTGGGGGCCCATAAT-3Ј, 5Ј-ACGGAGGATGC-CAAGAGGCT-3Ј, and 5Ј-CTTTGACCCCCAAGATACTC-3Ј, respectively. The MacVector DNA sequencing software was used to align DNA sequences from overlapping sequencing reactions. The determined nucleotide sequence of the human hippocampus ACT cDNA has been submitted to GenBank with accession number AF089747.
Northern Analysis and Slot Blots-For Northern blots, total RNA was isolated from hippocampus of Alzheimer's disease and normal aged (50 -80 years old) brains (from the Harvard Brain Tissue Resource Center at McLean Hospital, Belmont, MA) by cesium trifluoroacetate density ultracentrifugation, as described previously (3). The precipitated RNA was resuspended in TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA), subjected to 1.2% agarose/denaturing formaldehyde gel electrophoresis (10 g RNA/lane), and blotted to GeneScreen PlusR membrane (NEN Life Science Products) for Northern analysis. The DNA probe was a XmnI-BglII 652-bp fragment of the human liver ACT cDNA (6), labeled by nick-translation (kit from Stratagene) with [␣-32 P]dCTP (5,000 Ci/mmol, NEN Life Science Products). Hybridization with 32 P probe was conducted in 10% dextran sulfate, 1% SDS, and 1.0 M NaCl at 60°C overnight. The membrane was then washed 2 times in 2ϫ SSC at room temperature for 5 min, 2 times in 2ϫ SSC, 1% SDS at 60°C for 30 min, and in 0.1ϫ SSC at room temperature for 30 min. Autoradiography with Kodak X-Omat AR-5 film was conducted. RNA slot blots were performed with poly(A) ϩ RNA from human hippocampus and liver (purchased from CLONTECH) diluted to 9, 1, 0.1, and 0.1 ng/l. One hundred l of each dilution was combined with 300 l of 6.15 M formaldehyde in 10ϫ SSC and incubated at 67°C for 15 min. Samples were applied onto nitrocellulose membranes under vacuum, washed with 10ϫ SSC, and cross-linked by exposure to UV irradiation. The DNA probe, EcoRI-SalI human ACT cDNA fragment of 943 bp, was labeled to a specific activity of 1 ϫ 10 8 cpm/g with [␣-32 P]dCTP by Stratagene's random priming kit. The membrane was hybridized with the probe (1 ϫ 10 6 cpm/ml) in 5ϫ SSPE, 5ϫ Denhardt's solution, 0.1% SDS, 100 g/ml denatured salmon sperm DNA at 68°C overnight. The membrane was washed three times in 1ϫ SSPE, 0.1% SDS for 15 min at room temperature, twice in 0.1ϫ SSPE, 0.1% SDS at 42°C for 15 min, in 0.1ϫ SSPE, 0.1% SDS at 60°C for 5 min, and then in 0.1ϫ SSPE at room temperature. Autoradiography was performed using Kodak X-Omat AR-5 film.
Genomic Blot-Genomic DNA from the midbrain of normal adults (from the Harvard Brain Tissue Resource Center at McLean Hospital, Belmont, MA) was isolated by cesium chloride equilibrium density gradient centrifugation, as described previously (3). An aliquot of genomic DNA (10 g) was digested with 40 units of EcoRI, BamHI, KpnI, or HindIII and subjected to 0.8% DNA-agarose gel electrophoresis. The gel was dried at 60°C for 1-2 h and subjected to in situ hybridization (3) with human liver ACT cDNA of 1.5 kilobase (EcoRI digests) (ACT cDNA from Dr. Harvey Rubin, University of Pennsylvania) (6). The probe was labeled by nick translation (kit from Stratagene) with [␣-32 P]dCTP (5,000 Ci/mmol, NEN Life Science Products) to a specific activity of 1 ϫ 10 8 cpm/g of DNA. Unincorporated [␣-32 P]dCTP was removed by G-50 gel filtration. Hybridization was conducted by incubating the genomic DNA gel with ACT cDNA probe (at 5 ϫ 10 6 cpm/ml) in 5ϫ SSPE, 5ϫ Denhardt's solution, 0.1% SDS, 100 g of salmon sperm DNA at 60°C overnight, followed by washing in 1ϫ SPPE, 0.1% SDS at 60°C for 2 h, and a final wash in 0.1ϫ SSPE, 0.1% SDS at 60°C for 1 h. The genomic blot was then subjected to autoradiography with Kodak X-Omat AR-5 film.
Primer Extension-Primer extension utilized poly(A) ϩ RNA from human hippocampus and liver (purchased from CLONTECH) with a primer corresponding to 5Ј-CTGCCTCAGGGAGCTGGA-3Ј (primer A, see Fig. 1). The primer was labeled by T4 polynucleotide kinase with 5 pmol of [␥-32 P]ATP (5,000 Ci/mmol, NEN Life Science Products) to a specific activity of 1 ϫ 10 9 cpm/g. Unincorporated [␥-32 P]ATP was removed by G-25 gel filtration. The 32 P-labeled primer (1 ϫ 10 7 cpm) was annealed to 5.0 g of poly(A) ϩ RNA by heating at 65°C for 1.5 h in 1ϫ SuperScript II buffer (from Life Technologies, Inc.) and then cooled to room temperature. Primers were extended with SuperScript II reverse transcriptase (200 units, Life Technologies, Inc.) at 48°C for 90 min in 0.5 mM dNTP, 50 mM Tris-HCl, pH 8.3, 2 mM MgCl 2 , 27 mM KCl, 10 mM dithiothreitol, and 120 M actinomycin D (Boehringer Mannheim). RNase A (10 g from Boehringer Mannheim) was added, and incubation continued at 37°C for 15 min. Sodium acetate was added to a final concentration of 0.27 M, and the sample was subjected to phenol/ chloroform extraction, followed by chloroform extraction. The aqueous phase was precipitated with ethanol, and the resultant sedimented cDNA was resuspended in 10 l of 40 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , 50 mM NaCl, 33% formamide, 0.05% bromphenol blue, and 0.05% xylene cyanol FF. This sample was electrophoresed on an 8.0% acrylamide-bis-acrylamide 19:1 (w/w) 7 M urea sequencing gel in 1ϫ TBE buffer (0.1 M Tris-HCl, 83 mM boric acid, 1 mM EDTA). After electrophoresis, the gel was soaked for 15 min in 5% acetic acid, 15% methanol to remove urea and was then dried for 1.5 h at 80°C under vacuum. Autoradiography was performed with Kodak X-Omat AR-5 film.
Deglycosylation of Hippocampus ACT-Proteins were extracted from frozen hippocampus (approximately 4 g of tissue) from Alzheimer's disease or normal brains (tissues were from the Harvard Brain Tissue Resource Center at McLean Hospital, Belmont, MA) by the TRIZOL TM reagent (Life Technologies, Inc.). For deglycosylation of protein extracts from hippocampus and of human liver ACT (from Athens Research and Technology Biochemicals), ACT samples were incubated with N-glycosidase F (0.2 units, Boehringer Mannheim) at 37°C for 18 h in buffer (25 l total volume) consisting of 20 mM sodium phosphate, pH 7.2, 10 mM sodium azide, 50 mM EDTA, and 0.5% (w/v) octyloglucoside. Samples were then subjected to Western blots, as described previously (11,12), with affinity-purified anti-ACT IgGs (1:100 final dilution), detected with anti-rabbit goat IgGs conjugated to alkaline phosphatase (from Promega, performed according to the manufacturer's procedure).
For affinity purification of anti-ACT for Western blots, IgGs (immunoglobulins) were purified from anti-ACT serum (ART Biochemicals) by protein A-Sepharose chromatography (according to the manufacturer's protocol, Amersham Pharmacia Biotech). Anti-ACT IgGs were then affinity-purified on an ACT-Sepharose affinity column. The ACT affinity column was obtained by covalently linking ACT (from human liver, ART Biochemicals) to CNBr-activated Sepharose (Amersham Pharmacia Biotech, according to the manufacturer's instructions). The anti-ACT IgGs were bound to the ACT affinity column in equilibration buffer consisting of 60 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20. After washing with equilibration buffer, anti-ACT IgGs were eluted with elution buffer 1 (50 mM Tris-HCl, pH 7.5, 750 mM NaCl, 0.05% Tween 20), followed by elution buffer 2 (0.1 M citric acid-NaOH, pH 4.0, 750 mM NaCl, 0.05% Tween 20). Fractions containing anti-ACT IgGs were detected by binding to ACT in enzyme-linked immunosorbent assay, performed as described previously (13). Affinity-purified anti-ACT IgGs were concentrated by a Centricon-30 apparatus (Pierce).

RT-PCR of Hippocampus ACT cDNA Defines the ORF En-
coding the ACT Primary Sequence-RT-PCR was used to determine the primary sequence of ACT expressed in hippocampus from AD and normal brains. Primers for RT-PCR were designed based on reported homologous sequences of neuroendocrine and liver ACTs (3)(4)(5)(6)(7)(8) for amplification of the mature coding region of ACT (without the NH 2 -terminal signal sequence) (Fig. 1). RT-PCR and DNA sequence analyses for each overlapping DNA segment were performed from three separate tissue samples of control hippocampus and from three separate samples of AD hippocampus. Poly(A) ϩ RNA from each tissue sample was subjected to RT-PCR to amplify 3Ј and 5Ј regions of the ACT cDNA. From each RT-PCR reaction, DNA inserts from two to four subclones were subjected to DNA sequence analyses.
The predicted 790-bp and 502-bp DNA fragments represent-ing 5Ј and 3Ј domains of the ORF of the ACT cDNA were generated from AD and normal hippocampus as well as from liver poly(A) ϩ RNA (Fig. 2). Southern blots confirmed that these 790-and 502-bp PCR-amplified DNA fragments hybridized with the human liver ACT cDNA as probe (data not shown). DNA sequence analyses of these RT-PCR-generated DNA fragments, with alignment of overlapping 5Ј and 3Ј DNA fragments, provided the nucleotide and deduced the primary sequence of the hippocampus ACT cDNA from AD and normal brains (Fig. 3). Importantly, the ACT cDNAs isolated from AD and normal hippocampus were identical in nucleotide and deduced primary sequences for the mature ACT protein translation product. It is predicted that the mature ACT in hippocampus may begin with His or Asn at its NH 2 terminus, consistent with removal of the NH 2 -terminal signal peptide sequence (14,15) to form functional ACT. Similarly, mature ACT from liver and plasma begins with His (residue ϩ1) or Asn (ϩ3) (6,16). The human hippocampus ACT possesses the RSL domain within the COOH-terminal region of the inhibitor (Fig. 3) that is characteristic of serpins. The hippocampus ACT possesses Leu-Ser as the predicted P 1 -P 1 Ј residues that are presumably recognized and cleaved by putative chymotrypsin-like target brain proteases. In addition, the glycoprotein nature of hippocampus ACT is indicated by consensus glycosylation sites of Asn-Xaa-Ser/Thr at residues 10, 70, 83, 104, 163, and 248 (Fig. 3).
The deduced primary sequence of human hippocampus ACT cDNA resembles previously reported variant human liver ACT

FIG. 1. Strategy for RT-PCR of ACT cDNA from hippocampus. Primers 1 and 2 and primers 3 and 4 were designed
to amplify 5Ј and 3Јdomains of the hippocampus ACT cDNA, respectively. The overlapping 5Ј and 3Ј domains were predicted to include the RSL and the NH 2 terminus of mature, processed ACT, which begins at the COOH terminus of the signal peptide sequence. Primers 5 and 6 allowed amplification of the 3Ј-UTR of the cDNA. Primer A was used in primer extension analyses of ACT gene transcripts.
sequences (5)(6)(7)(8). The hippocampus ACT differs by 37 residues from the human liver cDNA sequence reported by Chandra et al. (5) at residues 79 -93, 100 -105, and 398 -400 as well as by 10 residues at the COOH terminus. In addition, the hippocampus ACT, compared with the liver ACT cDNA reported by Chandra et al. (5), possesses different amino acids for residues 69, 199, and 338. However, the human hippocampus ACT cDNA sequence is identical to the human liver ACT cDNA reported by Rubin et al. (6) and others (7,8). These results have, therefore, defined the primary sequence of human hippocampus ACT and have resolved its similarity to human liver ACT.
Northern Analysis-Northern blots were performed to com-pare ACT mRNAs in human hippocampus and liver (Fig. 4). The hippocampus and liver ACT mRNAs were similar in electrophoretic mobility corresponding to 1.5 and 1.6 kilobases, respectively; these results also show slight differences in apparent size of the ACT mRNAs from these two tissues. It is noted that no significant differences in ACT mRNA levels were detected in hippocampus from AD (n ϭ 5) and normal (n ϭ 8) brains.
Further studies compared the 3Ј-UTR region of the hippocampus ACT cDNA with that of the liver ACT cDNA. RT-PCR of hippocampus poly(A ϩ ) RNA was performed (primers 5 and 6, Fig. 1) to generate a 296-bp DNA fragment encoding the 3Ј-UTR region (data not shown). DNA sequence analysis indi- The open reading frame domains of ACT cDNAs from normal and Alzheimer's disease brains were identical. The DNA sequence of the 3Ј-UTR domain of the hippocampus ACT cDNA (from normal brain) was also determined. Alignment of the DNA sequences determined for overlapping 5Ј and 3Ј PCR fragments indicates the human hippocampus ACT cDNA (shown in bold for nucleotides 111 to 1576), whose deduced primary sequence (shown in bold for residues Ϫ7 to 398) corresponds to the mature ACT protein. The positions of primers 1-6 used in RT-PCRs are shown by dotted lines with arrows. Arrows above the His (ϩ1) and Asn (ϩ3) residues indicate the predicted NH 2 terminus of the mature ACT, which lacks the signal sequence. The RSL domain is boxed, with the predicted P 1 residue as Leu underlined. Consensus glycosylation sites are indicated by asterisks under the Asn residues as possible sites of glycosylation. The predicted 5Ј-region analyzed by primer extension is shown (not bold) for nucleotides 1-110 (6, 17). cated that the 3Ј-UTR of the hippocampus ACT cDNA (shown in Fig. 3) is nearly identical to the reported human liver ACT cDNAs (5,6), with greater than 98% homology in nucleotide sequence within the 296-bp 3Ј-UTR region. It is possible that slight differences in apparent electrophoretic mobility of hippocampus and liver mRNAs may be explained by possible differences in polyadenylation or features that influence electrophoretic mobility such as variations between Northern blots.
Northern blots indicated large differences in ACT mRNA levels in liver and hippocampus. Semiquantitative slot blots (Northern blots) with specified amounts of poly(A) ϩ RNA showed that liver contains approximately 900-fold higher levels of ACT mRNA compared with that in hippocampus (Fig. 5). These differences in ACT gene expression suggest specialized functions of ACT in brain compared with liver.
Genomic Blot and Primer Extension-The hippocampus ACT cDNA identified by RT-PCR indicates the existence of the human ACT gene. Further assessment of the ACT gene(s) was obtained by genomic blots. Human genomic DNA was digested with restriction enzymes and probed with the human liver ACT cDNA (6). The genomic blot (Fig. 6) demonstrated the presence of the ACT gene(s) as several DNA bands after digestion of genomic DNA, indicating the presence of at least one or possi-bly several ACT genes.
To compare transcriptional initiation sites of ACT gene expression in human hippocampus and liver, primer extension analyses were performed. Primer extension of poly(A) ϩ RNA utilized a primer corresponding to a 5Ј region of the ACT cDNA (Fig. 1), and the extended cDNA was sized on a DNA sequencing gel. For both hippocampus and liver, primer extension resulted in a 58-bp cDNA (Fig. 7), indicating identical transcription initiation sites for ACT gene expression in hippocampus and liver. Based on these primer extension results and the ACT gene sequence (18), the corresponding the 5Ј nucleotide sequence of the hippocampus ACT cDNA is illustrated in Fig. 3.
Deglycosylation of ACT-Multiple glycosylation sites (Asn-Xaa-Thr/Ser) are indicated by the hippocampus ACT cDNA, suggesting that brain ACT exists as a glycoprotein. Therefore, FIG. 4. Northern blot of ACT mRNA in hippocampus from AD and normal hippocampus as well as liver. Northern blots of total RNA isolated from hippocampus of normal (N) and AD brains (10 g of RNA each, lanes 1 and 2, respectively) and liver poly(A) ϩ RNA (1 g, lane 3) probed with the human ACT cDNA (6), as described under "Experimental Procedures." Autoradiography of Northern blots (15 h exposure to x-ray film) (lanes 1-3) showed ACT mRNA in hippocampus and a high level of ACT mRNA in liver. Intact ribosomal RNAs were detected by ethidium bromide staining of the RNA samples on denaturing formaldehyde gels (data not shown). kb, kilobases.
FIG. 5. Slot blot of ACT mRNA in hippocampus and liver. Slot blots of poly(A) ϩ RNA (with the indicated amounts of RNA) from hippocampus and liver were performed to compare ACT mRNA levels in these two tissues. Hybridization of slot blots with ACT cDNA as probe was performed identically as described for Northern blots of ACT mRNA (Fig. 4). Northern blots were subjected to autoradiography (15 h exposure to x-ray film) for detection of ACT mRNA.  1 and 2, respectively) was conducted with 32 P-labeled primer 5Ј-CTGCCTCAGGGAGCTGGA-3Ј. The 32 P-extended cDNA was analyzed on 8% acrylamide, bis-acrylamide, 7 M urea DNA sequencing gels, with detection of the extended cDNA by autoradiography, as described under "Experimental Procedures." Arrows indicate the radiolabeled cDNAs of 58 bp obtained by primer extension. the extent of glycosylation of ACT in hippocampus was examined by deglycosylation with N-glycosidase F, which cleaves the N-glycan linkage of glycoproteins between Asn residues and the carbohydrate chain (19). Western blots (Fig. 8) showed hippocampus ACT of 60 -65 kDa in normal and AD brains, which was deglycosylated by N-glycosidase F to a band of approximately 46 kDa. The deglycosylated 46-kDa hippocampus ACT is consistent with the theoretical molecular weight of the ACT polypeptide calculated from its primary sequence, deduced from its cDNA. The liver ACT of 66 -75 kDa was slightly larger than the hippocampus ACT. Deglycosylation of liver ACT to a 46-kDa polypeptide indicates that both liver and hippocampus ACT consist of similar molecular weight polypeptide backbones. Differences in apparent molecular weights of hippocampus and liver ACTs and the similar molecular weights of their polypeptide backbones suggest that the two forms of ACT may undergo different types of glycosylation. Overall, these results indicate that the hippocampus and liver ACTs are both glycoproteins. DISCUSSION Immunochemical detection of the protease inhibitor ACT in amyloid plaques in the hippocampus region of AD brains suggests a role for a protease inhibitor in AD (1,2). Recent identification of isoforms of neuroendocrine ACTs in bovine adrenal medulla and pituitary (3,4) and reports of variations in the primary sequences of isolated human liver cDNAs (5)(6)(7)(8) lead to the question of the molecular identity of ACT in AD and normal brains. In this study, direct RT-PCR and DNA sequence analyses of hippocampus ACT cDNAs from AD and normal brains show that the hippocampus ACT possesses the reactive site loop that is characteristic of serpins, with Leu as the predicted P1 residue for inhibition of putative brain chymotrypsin-like proteases. The hippocampus ACT cDNAs from control and AD brains were identical in nucleotide and deduced primary sequences and resemble the human liver ACT with greater than 90% homology. Further comparison of ACT in hippocampus and liver showed that ACT gene expression in these two tissues utilizes identical transcription initiation sites. However, significantly lower levels of ACT mRNA are expressed in hippocampus compared with liver. In addition, the hippocampus ACT protein appears to be differentially glycosylated compared with liver ACT. These studies have defined the primary sequence and molecular characteristics of hippocampus ACT expressed in control and Alzheimer's disease brains.
Before this study, ACT in human hippocampus was thought to resemble human liver ACT based on recognition of the brain ACT with antibodies against human liver ACT (1, 2). However, several reports have indicated variations in nucleotide and deduced primary sequences for full-length and partial human liver ACT cDNAs (5)(6)(7)(8). In this study, direct DNA sequence analyses of the hippocampus ACT cDNA indicates that its deduced primary sequence differs by 37 residues compared with the human liver ACT cDNA reported by Chandra et al. (5). However, the hippocampus ACT cDNA is identical to human liver cDNAs characterized by other groups (6 -8). These results indicate that the ACT expressed in human hippocampus and human liver ACT (5)(6)(7)(8) are nearly identical in nucleotide and deduced primary sequences.
The hippocampus ACT possesses the RSL domain (Fig. 3, boxed region) that participates in the specificity of the serpin to regulate target proteases. The predicted Leu-Ser as P 1 -P 1 Ј residues are known to inhibit chymotrypsin, suggesting that ACT may inhibit a brain chymotrypsin-like protease. It is noteworthy that cross-class inhibition of cysteine proteases by serpins occurs. For example, the interleukin 1␤-converting enzyme (20,21) and caspase cysteine proteases (22,23) are inhibited by the CrmA serpin encoded by the cowpox virus (24). In addition, ACT has been demonstrated as a potent inhibitor of a cysteine protease, known as prohormone thiol protease (PTP), that is involved in pro-neuropeptide processing (25). It may, therefore, be logical to predict that ACT in Alzheimer's disease or normal brains may regulate a serine or cysteine protease.
The characterization of a single hippocampus ACT cDNA is consistent with the demonstration of the human ACT gene(s), as demonstrated by genomic blots. Furthermore, primer extension analyses showed that ACT gene expression in hippocampus and liver utilizes the same transcription initiation site. However, the hippocampus possesses significantly lower levels of ACT mRNA than liver, with differences of approximately 900-fold. These findings suggest differential tissue regulation of ACT gene expression in hippocampus compared with liver. In addition, examination of ACT mRNAs by Northern blots showed that hippocampus and liver ACT mRNAs were close in size, 1.5-1.6 kilobases, and possess nearly identical 3Ј-UTRs.

FIG. 8. Deglycosylation of ACT in hippocampus of Alzheimer's disease and normal brains as well as in liver.
Deglycosylation by N-glycosidase F of ACT in tissue extracts from hippocampus of AD and normal brains as well as human liver ACT was assessed by Western blots with anti-ACT serum. Panel A shows ACT in normal (N) and AD hippocampus (lanes 1 and 3, respectively). ACT in these tissues was also incubated with (ϩ) N-glycosidase F (lanes 2 and 4,  respectively). Panel B shows human liver ACT without and with N-glycosidase F treatment (lanes 1 and 2, respectively). Subsequent to translation of the ACT mRNA, ACT undergoes posttranslational modification as a glycoprotein. Differential glycosylation demonstrated by sensitivity to N-glycosidase F suggests that there may be different types of glycosylation for the ACT in hippocampus compared with liver. Hippocampus and liver ACT proteins appear as 60 -65-kDa and 66 -75-kDa proteins, respectively, on SDS-polyacrylamide gel electrophoresis. After deglycosylation by N-glycosidase F, the ACT polypeptide backbone is represented by a 45-46-kDa band in both tissues. The apparent molecular mass of deglycosylated ACT is consistent with the theoretical molecular mass of mature ACT of 46,248 daltons, calculated from the ACT cDNA.
The cellular localization of ACT is an important consideration for future studies of proteases that may be regulated by ACT, because protease inhibitor(s) and target protease(s) must be co-localized to allow their interaction in vivo. The NH 2terminal signal sequence of ACT suggests cellular routing of the ACT to the secretory pathway. The predicted transport of ACT within neurosecretory vesicles from neuronal cell bodies along the axon to nerve terminals is supported by the disappearance of ACT from nerve terminals following axotomy (26). Subsequent to secretion, extracellular ACT accumulates with ␤-peptide in amyloid plaques of AD brains (1,2). Target proteases regulated by ACT in brain may be colocalized with intracellular or extracellular ACT. Knowledge of the molecular identity of ACT obtained in this study will allow future identification of target proteases that may be regulated by ACT in normal and AD brains.