In Vivo Cleavage of α2,6-Sialyltransferase by Alzheimer β-Secretase*

β-Site amyloid precursor protein-cleaving enzyme 1 (BACE1) is a membrane-bound aspartic protease that cleaves amyloid precursor protein to produce a neurotoxic peptide, Aβ, and is implicated in triggering the pathogenesis of Alzheimer disease. We previously reported that BACE1 cleaved rat β-galactoside α2,6-sialyltransferase (ST6Gal I) that was overexpressed in COS cells and that the NH2 terminus of ST6Gal I secreted from the cells (E41 form) was Glu41. Here we report that BACE1 gene knock-out mice have one third as much plasma ST6Gal I as control mice, indicating that BACE1 is a major protease which is responsible for cleaving ST6Gal I in vivo. We also found that BACE1-transgenic mice have increased level of ST6Gal I in plasma. Secretion of ST6Gal I from the liver into the plasma is known to be up-regulated during the acute-phase response. To investigate the role of BACE1 in ST6Gal I secretion in vivo, we analyzed the levels of BACE1 mRNA in the liver, as well as the plasma levels of ST6Gal I, in a hepatopathological model, i.e. Long-Evans Cinnamon (LEC) rats. This rat is a mutant that spontaneously accumulates copper in the liver and incurs hepatic damage. LEC rats exhibited simultaneous increases in BACE1 mRNA in the liver and in the E41 form of the ST6Gal I protein, the BACE1 product, in plasma as early as 6 weeks of age, again suggesting that BACE1 cleaves ST6Gal I in vivo and controls the secretion of the E41 form.

␤-Site amyloid precursor protein-cleaving enzyme 1 (BACE1) is a membrane-bound aspartic protease that cleaves amyloid precursor protein to produce a neurotoxic peptide, A␤, and is implicated in triggering the pathogenesis of Alzheimer disease. We previously reported that BACE1 cleaved rat ␤-galactoside ␣2,6-sialyltransferase (ST6Gal I) that was overexpressed in COS cells and that the NH 2 terminus of ST6Gal I secreted from the cells (E41 form) was Glu 41 . Here we report that BACE1 gene knock-out mice have one third as much plasma ST6Gal I as control mice, indicating that BACE1 is a major protease which is responsible for cleaving ST6Gal I in vivo. We also found that BACE1-transgenic mice have increased level of ST6Gal I in plasma. Secretion of ST6Gal I from the liver into the plasma is known to be up-regulated during the acute-phase response. To investigate the role of BACE1 in ST6Gal I secretion in vivo, we analyzed the levels of BACE1 mRNA in the liver, as well as the plasma levels of ST6Gal I, in a hepatopathological model, i.e. Long-Evans Cinnamon (LEC) rats. This rat is a mutant that spontaneously accumulates copper in the liver and incurs hepatic damage. LEC rats exhibited simultaneous increases in BACE1 mRNA in the liver and in the E41 form of the ST6Gal I protein, the BACE1 product, in plasma as early as 6 weeks of age, again suggesting that BACE1 cleaves ST6Gal I in vivo and controls the secretion of the E41 form.
A characteristic feature of Alzheimer disease (AD) 1 is deposition of amyloid ␤-peptide (A␤) in the brain, which is impli-cated in the pathogenesis of AD (1). A␤, a 39 -43-residue peptide, is generated from the amyloid precursor protein (APP) by the action of ␤-and ␥-secretases. APP is a type I transmembrane glycoprotein which is present in the trans-Golgi network and endosomes. Cleavage of APP by ␤-secretase (BACE1 (␤-site APP-cleaving enzyme 1)) initially produces a soluble NH 2 -terminal fragment (APPs␤) and a 12-kDa COOH-terminal fragment (C99) that remains membrane-bound. Subsequently, C99 is cleaved by ␥-secretase in its transmembrane region, resulting in production of the pathogenic A␤ peptide (2,3). Alternative cleavage of APP by ␣-secretase within the A␤ sequence produces a soluble NH 2 -terminal fragment (APPs␣) and a 10-kDa membrane-bound COOH-terminal fragment (C83) (4,5). C83 is also cleaved by ␥-secretase to produce the nonpathogenic p3 peptide. BACE1, a pepsin-like membrane-bound aspartic protease, was recently identified as ␤-secretase (6 -10). A close homologue of BACE1, designated BACE2, was found to share 60% similarity in amino acid sequence with BACE1 (10 -13). BACE1 knock-out mice completely lack A␤ production in the brain (14,15), indicating that BACE1 carries majority of ␤-secretase activity in the brain. BACE2 may be important in Down syndrome pathology, because the enzyme is encoded by chromosome 21 (12) and its expression is elevated in trisomic brains (16).
A series of extensive studies showed that ␥-secretase activity is performed by a protein complex that includes presenilins (PS1 or PS2) (17). ␥-Secretase is essential, not only for A␤ production, but also for neuronal development, due to its cleavage of a Notch receptor (3,18). Nevertheless, development of inhibitors to ␥-secretase is a promising approach for treating Alzheimer disease, as is also true for ␤-secretase inhibitors (19).
ST6Gal I (␤-galactoside ␣2,6-sialyltransferase) is a type II membrane protein that is localized in the trans-Golgi network. It catalyzes ␣2,6-sialylation of Gal␤1,4-GlcNAc structures on N-glycans. ST6Gal I is highly expressed in the liver and is expressed in most other tissues to some extent (20). The majority of serum ST6Gal I is secreted from the liver (21,22), and secretion is enhanced during acute-phase hepatic reactions (23,24). We previously found that BACE1 is involved in the cleavage and secretion of ST6Gal I, at least in cultured cells (25,26), leading to an assumption that BACE1 cleaves ST6Gal I in the liver and triggers its secretion into plasma.
To confirm in vivo cleavage of ST6Gal I by BACE1, in the present study we analyzed plasma ST6Gal I levels in BACE1deficient and BACE1-transgenic mice. We also used a mutant animal strain, the Long-Evans Cinnamon (LEC) rat, to analyze the mechanisms of ST6Gal I secretion in hepatopathological conditions. The LEC rat, a model of Wilson disease, has a deletion in the gene for the copper-transporting ATPase gene (ATP7B) (27)(28)(29). Golgi-localized ATP7B is involved in copper secretion into the plasma, which is coupled with ceruloplasmin synthesis and biliary copper excretion (30). Like patients with Wilson disease, LEC rats suffer from toxic accumulation of copper in the liver and eventually develop hepatitis and then hepatocellullar carcinoma (31). This rat strain is often utilized for studying the pathogenesis of hepatitis and hepatoma. On the assumption that ST6Gal I secretion is stimulated in the hepatopathological model of LEC rats and that such stimulation is related to the level of BACE1 activity, we analyzed the expression profiles of mRNAs of ST6Gal I and BACE1 in the liver of LEC rats and also the secretion of ST6Gal I into plasma.

EXPERIMENTAL PROCEDURES
Materials-Male LEC and Wistar rats, maintained in specific-pathogen-free conditions, were purchased from Charles River Japan Inc. (Yokohama, Japan) and the Shizuoka Agricultural Cooperative Association for Laboratory Animals (Shizuoka, Japan), respectively. Samples of LEA (Long-Evans Agouti) rats were kindly provided by Dr. Noriyuki Kasai, Tohoku University. BACE1 knock-out mice and their control mice from Amgen were maintained in the Charles River Laboratories, Inc. (Wilmington, MA). BACE1-transgenic mice were prepared by using an overexpression cassette, in which the chicken ␤-actin promoter drived the expression of human BACE1 (32). The sources of materials used in this work were as follows: tissue culture media and reagents, including Dulbecco's modified Eagle's medium and William's E medium, were from Invitrogen; Expre 35 S 35 S protein labeling mix was from PerkinElmer Life Sciences; recombinant N-glycosidase F from Roche Diagnostics, CMP-[ 14 C]NeuAc and protein A-Sepharose Fast Flow were from Amersham Biosciences; affinity Gel-CDP from Merck/EMD Bioscience (San Diego, CA); protein molecular weight standards were from Bio-Rad; and all other chemicals from Sigma or Wako Chemicals (Osaka, Japan). Protein concentration was determined with Pierce BCA protein assay reagents. Anti-BACE1 antibody was purchased from Mo-BiTec Co. (Göttingen, Germany). We used anti-ST6Gal I polyclonal antibodies to detect ST6Gal I from plasma and to immunoprecipitate ST6Gal I proteins from cell lysates and media of hepatocytes (33). E41 antibody that specifically recognizes ST6Gal I starting at Glu 41 was obtained from IBL Co. (Fujioka, Japan).
Pulse-Chase Analysis and Immunoprecipitation of ST6Gal I Using Rat Hepatocytes-Hepatocytes (ϳ5 ϫ 10 6 cells), isolated from 10-weekold male Wistar rats by two-step collagenase perfusion (34), were seeded on collagen type I-coated plates (100-mm, Iwaki, Funabashi, Japan) and were grown in William's E medium containing 5% fetal bovine serum. Metabolic labeling of cells and immunoprecipitation of expressed proteins were performed essentially as described previously, using Expre 35 S 35 S protein label (100 Ci/ml) (25,35). Cells were labeled with 18 h, or labeled with 2 h and chased for 6 or 18 h, in 4 ml of William's E medium containing 5% fetal bovine serum. The collected media were diluted with 4 ml of Pierce binding buffer, and the cells were lysed in immunoprecipitation buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 0.1% SDS). From the cell lysates and media, ST6Gal I proteins were immunoprecipitated with anti-ST6Gal I or E41 polyclonal antibodies. Immunoprecipitated proteins were denatured in Laemili sample buffer containing 5% ␤-mercaptoethanol by boiling for 5 min. Immunoprecipitated proteins were analyzed on SDS-polyacrylamide gels (5-20% gradient of polyacrylamide), and radiolabeled proteins were visualized with a BAS 2000 radioimage analyzer (Fuji Film, Tokyo).
Real-time Quantitative PCR-Total RNA was isolated from rat livers using Trizol reagent (Invitrogen), and 5-10 g of RNA was reversetranscribed with random hexamers by using a Super Script II RT kit (Invitrogen) according to the manufacturer's protocol. The cDNA was then amplified with 900 nM forward primer, 900 nM reverse primer, 250 nM fluorogenic probe, and 25 l of Universal PCR Master Mix (Applied Biosystems) in a total volume of 20 l, using an ABI PRISM 7900HT sequence detection system (Applied Biosystems). The PCR conditions were 1 cycle at 50°C for 2 min, 1 cycle at 95°C for 10 min, and 40 cycles at 95°C for 15 s and 50°C for 1 min. All primers and probes were purchased from Applied Biosystems. The sequences of primers and probes were as follows: ST6Gal I, 5Ј-CAGCAAGCAAGACCCTA-AGGA-3Ј (forward primer), and 5Ј-CTGGAAGGAAGGCTGTGGTTT-3Ј (reverse primer), 5Ј-CCAATCCTCAGTTACCACAGGGTCACAGC-3Ј (probe); BACE2, 5Ј-AGAACGCCAGTCGCTCCTT-3Ј (forward primer), 5Ј-ATTGAAACCAGCTCCCATCATG-3Ј (reverse primer), 5Ј-CACCAT-TCTGCCACAGCTCTACATTCAGC-3Ј (probe). For BACE1 and GAPDH primers and probes, we used Assays-on-Demand Gene Expression Products, and cDNAs were added to the TaqMan Universal PCR Master Mix (Applied Biosystems), which contained all reagents for PCR. The probes for ST6Gal I, BACE1, and BACE2 were labeled with the reporter fluorescent dye FAM. The probes for GAPDH were labeled with VIC at the 5Ј ends and at the 3Ј ends were labeled with the quencher dye TAMRA. The expression levels of target genes were measured in duplicate and were normalized to the GAPDH expression levels.
Characterization of BACE1-deficient and BACE1-transgenic Mice-Genotypes of BACE1-deficient and control mice were confirmed by PCR analysis using genomic DNA prepared from livers and primers (primer 1, 5Ј-TGG ATG CCC ATT CTT TCT GAC GA-3Ј and primer 2, 5Ј-AGG CAC TAA CCA CTA GCC TGT TCA-3Ј). A 770-bp fragment was detected in wild-type mice, and a 3.5-kb fragment was detected in the BACE1-deficient mice. We also examined the level of BACE1 protein in the livers of BACE1-deficient and BACE1-transgenic mice by Western blot analysis. Mouse livers were homogenized with buffer H (50 mM MES, pH 6.0, 1 mM EDTA, 0.15 M NaCl). Microsome fraction (50 or 100 g) that was solubilized with buffer S (50 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 1% Triton X-100) was treated with Laemmli sample buffer (36), subjected to SDS-polyacrylamide gel electrophoresis (5-20% gradient), and then transferred to a nitrocellulose membrane. The membrane was incubated with anti-BACE1 (1:1,000). horseradish peroxidase-goat anti-rabbit IgG (Cappel, 1:1,000) was used as the secondary antibody, and chemiluminescent substrate (Pierce) was used for detection (25).
Detection of ST6Gal I Protein-Plasma from mice (50 l) were diluted with 10 volumes of buffer A (20 mM Tris-HCl, pH 8.0) and then loaded onto a HiTrap Q-Sepharose column (1 ml of gel volume, Amersham Biosciences) using an Akta Prime system. The column was washed with 5 ml of buffer A, and protein was then eluted with a gradient of 0 -0.5 M NaCl for 30 min at a flow rate of 0.5 ml/min. Fractions were collected every minute, and the ST6Gal I elution pattern was monitored by immunostainingwith anti-ST6Gal I antibody. ST6Gal I-containing fractions were pooled and precipitated with ice-cold acetone. Plasma from rats (600 g of protein) were treated by the Montage Albumin Depletion kit from Millipore (Bedford, MA) and then precipitated with ice-cold acetone. To see the level of soluble ST6Gal I, protein samples were suspended in PBST buffer (20 mM sodium phosphate, pH 7.2, 0.15 M NaCl, 0.1% Tween 20), subjected to SDS-polyacrylamide gel electrophoresis (5-20% gradient), and then transferred to a nitrocellulose membrane. The membrane was incubated with anti-ST6Gal I (1:100). horseradish peroxidase-goat anti-rabbit IgG (Cappel, 1:10,000) was used as the secondary antibody, and chemiluminescent substrate (Pierce) was used for detection (25). ST6Gal I detected by anti-ST6Gal I antibody was quantified with a Luminoimage Analyzer LAS-1000 PLUS (Fuji).
For N-glycosidase F treatment, the sample precipitated with acetone was resuspended in 60 l of N-glycosidase F buffer (50 mM sodium phosphate, pH 7.4, 50 mM EDTA, 0.2% SDS, 1% Nonidet P-40, and 1% ␤-mercaptoethanol) and then divided into two. The subsamples were incubated at 37°C for 16 h in the presence or absence of 1 unit of N-glycosidase F. The samples were analyzed by immunoblotting with anti-ST6Gal I antibody.
Partial Purification of ST6Gal I Proteins from Rat Plasma-Plasma (10 ml) from Wistar rats was diluted with 10 volumes of buffer A (50 mM Tris-HCl, pH 7.5) and dialyzed against buffer A. The sample was then loaded onto a Q-Sepharose column (26 mm inner diameter ϫ 100 mm, Amersham Biosciences) using an FPLC system. The column was washed with 250 ml of buffer A, and protein was then eluted with a gradient of 0 -0.6 M NaCl for 120 min at a flow rate of 5 ml/min. Fractions were collected every minute, and the ST6Gal I elution pattern was monitored by immunostaining samples of the eluant with anti-ST6Gal I antibody. The fraction rich in ST6Gal I protein was dialyzed against buffer C (10 mM MES-NaOH, pH 6.0, 0.1 M NaCl) and then loaded onto a 2-ml CDP-gel column. Unbound material was washed out with buffer C, and bound material was eluted with a stepwise gradient of buffer D (10 mM MES-NaOH, pH 6.0, 0.075 N NaCl, 10 mM CDP).

Plasma ST6Gal I Levels Were Decreased in BACE1-deficient
Mice-ST6Gal I is highly expressed in the liver (20), and a membrane-bound form of ST6Gal I is proteolytically cleaved in the trans-Golgi network and then secreted into the plasma (22,33). We previously reported that secretion of ST6Gal I was markedly increased by overexpression of BACE1 in COS cells (25), indicating that BACE1 cleaves ST6Gal I at least in the cultured cells. To demonstrate cleavage of ST6Gal I by BACE1 also in vivo, we utilized BACE1-deficient mice for analyzing plasma ST6Gal I levels, expecting ST6Gal I cleavage and secretion to be lacking or diminished. At first we examined genotypes of BACE1-deficient and control mice by PCR (Fig. 1A), and then confirmed the absence or presence of BACE1 protein in the microsome fraction prepared from the liver (Fig. 1B). From mouse plasma, ST6Gal I was partially purified or enriched by HiTrap Q anionic exchange chromatography. Western blot analysis of ST6Gal I-enriched fraction revealed that ST6Gal I levels in BACE1-deficient mice were decreased to ϳ30% that of control mice (Fig. 1C), suggesting that BACE1 is a major protease that cleaves ST6Gal I to induce its secretion in vivo and that the remaining plasma ST6Gal I in BACE1deficient mice may be cleaved by some other protease(s). We next studied whether the elevated level of BACE1 could cause the elevation of the plasma ST6Gal I in vivo. To do so, we have generated the two lines of BACE1-transgenic mice that express human BACE1 under the control of actin promoter for ubiquitous expression. We detected high BACE1 expression in the liver of both of two lines (Fig. 2, upper panel). Both of BACE1transgenic mice had elevated levels of plasma ST6Gal I as compared with the control mice (Fig. 2, lower panel).
ST6Gal I Secretion in Rat Plasma-We previously reported that rat ST6Gal I secreted from COS cells has a sequence of Glu 41 -Phe 42 -Gln 43 at its amino terminus (E41 form) (25,35). Overexpression of BACE1 in COS cells markedly increased the secretion of the E41 form (25). We found that BACE1 first cleaved rat ST6Gal I between Leu 37 and Gln 38 to generate the sequence Gln 38 -Ala 39 -Lys 40 -Glu 41 -Phe 42 -Gln 43 at the amino terminus, and then the three terminal amino acids, Gln-Ala-Lys, were trimmed by a luminal aminopeptidase(s) to produce the E41 form that is secreted from the cells (26).
For further clarification of the in vivo cleavage and secretion of ST6Gal I, we looked for the E41 form in rat plasma. Western blot analysis using anti-ST6Gal I antibody revealed two isoforms of soluble ST6Gal I in the plasma of Wistar rats (Fig. 3). After N-glycosidase F treatment to remove N-glycans, the two isoforms were still detected as a pair of bands on an immunoblot, but with slightly increased mobility, suggesting that the difference between the isoforms is not due to the presence of N-glycan structures but rather to the position of the cleavage sites. The two isoforms were separated by anion-exchange FPLC chromatography (Fig. 4A). These isoforms were subjected to immunostaining in which we used an E41-antibody that specifically recognizes the E41 form (25,26). E41-antibody reacted only with the higher molecular weight isoform, suggesting that it is the E41 form, most likely produced by BACE1 (Fig. 4B). The lower molecular weight isoform (LMW form) has a distinct cleavage site, possibly produced by some other protease(s).
To characterize these isoforms further, we used CDP-hex-anolamine-agarose column chromatography, which is often used for purifying sialyltransferases (33,37). The E41 form bound to the affinity column and was eluted with CDP, whereas the LMW form did not bind to the column (Fig. 5A). The result suggests that the LMW form lacks affinity for the donor substrate, CMP-sialic acid, and does not have catalytic activity. The E41 form and the LMW form, which were separated each other by the anion-exchange column chromatogra-FIG. 1. Decrease of plasma ST6Gal I in BACE1-deficient mice. A, genotypes of BACE1-deficient and wild-type mice are shown. The exon 2 that contains an amino-terminal active site motif of BACE1 was replaced with a HIS-NEO selectable marker in BACE1-deficient mouse. Arrows represents primers for PCR analysis. B, BACE1-deficient (Ϫ/Ϫ) and wild-type (ϩ/ϩ) mice were identified by PCR analysis (upper panel). BACE1 protein in microsome fraction (100 g of protein) prepared from liver homogenates was analyzed by immunostaining with anti-BACE1 antibody (lower panel). C, plasma ST6Gal I in BACE1-deficient and wild-type mice. ST6Gal I in mouse plasma (50 l) was partially purified by HiTrap Q anion-exhchange chromatography using an Akta prime system (upper panel). ST6Gal I-containing fraction was monitored by immunostaining with anti-ST6Gal I antibody. Soluble ST6Gal I eluted at ϳ0.3 M NaCl was pooled as indicated with the bar. An aliquot of the pooled fraction (0.6% each) was subjected to immunostaining with anti-ST6Gal I antibody (middle panel) and ST6Gal I signals were quantified (lower panel). The percentage of plasma ST6Gal I is the average (ϮS.E.) of 10 mice of each genotype, with the mean for wild-type mice being set to 100 (*, p Ͻ 0.005).
phy, were subjected to sialyltransferase assay. Indeed, the LMW fraction contained less than one-tenth the activity that the E41 fraction had (Fig. 5B). The activity in the former fraction is possibly due to the trace contamination of the E41 form. Alternatively, the LMW form may carry a little activity.

ST6Gal I Secretion in Rat Hepatoytes-Cao et al. (38) showed that ST6Gal I is expressed in hepatocytes but not in nonparenchymal cells such as biliary epithelial cells, Kupfer cells, and
Ito cells in the liver, suggesting that the plasma isoforms originate from hepatocytes. Thus, we prepared primary cultured hepatocytes from the livers of Wistar rats and metabolically labeled them with [ 35 S]methionine. ST6Gal I proteins were immunoprecipitated from the cell lysates or media with anti-ST6Gal I antibody. We detected two bands corresponding to soluble ST6Gal I in the culture media (Fig. 6A). The molecular weights of the two ST6Gal I isoforms in the media were similar to those of soluble ST6Gal I in plasma (Figs. 3 and 6A). The two isoforms secreted form the cells had higher molecular weight than the cellular form. This may be due to glycosylation difference as proposed for N-acetylglucosaminyltransferase I (39). The higher molecular weight isoform was immunoprecipitated with anti-E41 antibody (data not shown), suggesting that the higher molecular weight form corresponds to the E41 form and the lower molecular weight form has a different cleavage site. Next, we performed a pulse-chase experiment to see difference in the secretion of these isoforms. When hepatocytes were pulse labeled and chased for 6 h, the majority of the soluble ST6Gal I isoform was the lower molecular weight form (ϳ91%) (Fig.  6A). After 18-h chase, the lower molecular weight form was a minor component (ϳ8%) (Fig. 6B). Thus the lower molecular weight form appeared first, degraded rapidly, and then later higher molecular weight form was generated, suggesting that the lower molecular weight form is not a secondary product derived from the higher molecular weight form (E41 form).
ST6Gal I Cleavage and Secretion in LEC Rats-ST6Gal I activity in the plasma increases in various pathological conditions such as hepatitis (23) and cancer (40 -42). To analyze ST6Gal I isoforms in hepatopathological conditions, we used LEC rats, which have a genetic mutation in the copper-transporting ATPase (ATP7B), resulting in toxic accumulation of copper in the liver. At first we measured ␣2,6-sialyltransferase activity of plasma of LEC and control Wistar rats. As shown in  (#1 and #6). BACE1 proteins in the microsome fraction (50 g of protein), prepared from liver homogenates of BACE1-transgeninc and control mice, were analyzed by immunostaining with anti-BACE1 antibody (upper panel). ST6Gal I in plasma was partially purified and its level was quantified by immunostaining with anti-ST6Gal I antibody.
FIG. 3. Isoforms of soluble ST6Gal I from rat plasma. An albumin-depleted plasma sample (100 g of protein before albumin depletion) from Wister rats was subjected to N-glycosidase F treatment. The reaction products were subjected to Western blot analysis with anti-ST6Gal I antibody.  Fig. 3A was subjected to CDP-gel column chromatography. The applied sample and the eluant were analyzed by immunostaining with either anti-ST6Gal I antibody (left) or anti-E41 antibody (right). B, the E41 and the LMW fractions were assayed for sialyltransferase activity, at equal ST6Gal I concentrations. Sialyltransferase activities are relative to that of the E41 fraction. Fig. 7A, LEC rats showed higher sialyltransferase activity in plasma than Wistar rats. The activity was increased at 6 and 8 weeks of age, while that of Wistar rat was stable. The result suggests a high level of the E41 form in LEC and its increase during the hepatopathological conditions. We then analyzed plasma ST6Gal I isoforms of LEC rats by Western blot analysis using anti-ST6Gal I antibody. LEA rats were used as the wildtype controls. The ratio of the E41 form to total ST6Gal I (E41 ϩ LMW form) in LEC rats was higher than that in Wistar rats at all ages analyzed. The E41 ratio in LEC rats increased at ages of 6 and 8 weeks, whereas that in LEA rats exhibited little change along with age (Fig. 7B). Wistar rats, another control strain, also exhibited very little change in the isoform pattern (data not shown).
To understand the molecular mechanism of the elevation of the plasma E41 form in LEC rats, we analyzed the levels of the mRNAs for BACE1 and ST6Gal I in the liver. The level of mRNA for BACE2, a homologue of BACE1, was also analyzed. Real-time PCR analysis revealed that LEC rats exhibited significant elevation of BACE1 mRNA at the ages of 6 and 8 weeks, whereas control Wistar rats did not show such significant elevation (Fig. 8). The levels of BACE2 and ST6Gal I mRNAs were very stable in both rat strains. These results suggest that the high levels of E41 form in LEC plasma at the ages of 6 and 8 weeks can be attributed mainly to enhanced transcription of the BACE1 gene in the liver. DISCUSSION BACE1 plays a critical role in the generation of amyloid ␤-peptide, the deposition of which is an initial pathological change occurring in Alzheimer disease. Inhibiting BACE1 activity may therefore be a promising way to treat the disease. This therapeutic strategy is supported by the observation that BACE1-deficient mice are viable and fertile (14,15). However, Harrison et al. (43) have recently reported that the mice exhibit some neurological abnormalities and behavioral changes. The exact molecular mechanisms for developing these phenotypes are not known, but the phenotypes may arise from impaired processing of APP or other BACE1 substrates. So far, ST6Gal I (25,26) and P-selectin glycoprotein ligand-1 (PSGL-1) (44) have been reported as BACE1 substrates other than APP. In vivo cleavage of these substrates and their biological relevance need to be carefully examined in relation to the phenotypes of BACE1-deficient mice. Searching for additional physiological substrates of BACE1 may also be required for understanding the molecular mechanisms behind the development of these phenotypes.
In the present study, we demonstrated that BACE1 is responsible for the secretion of ST6Gal I from the liver into the plasma. This finding indicates that BACE1 can function not only in the brain but also in the liver, although the level of BACE1 mRNA in liver is much lower than that in brain (6). It is of interest that BACE1, membrane-bound protease could act on both type I (APP) and type II (ST6Gal I) membrane proteins. The stem region of ST6Gal I may be flexible enough to access  (n ϭ 3). B, proteins in albumin-depleted plasma samples from LEC and LEA rats were precipitated with acetone and suspended in PBST buffer. The sample (50 g of protein before albumin depletion) from each animal was then analyzed by immunostaining with anti-ST6Gal I antibody. Typical immunoblots from LEC and LEA rats (4-, 6-, and 8-week-old males) are shown in the upper panel. The amounts of the E41 form and LMW form of ST6Gal I in plasma samples from LEC (n ϭ 8) and LEA (n ϭ 3) rats were estimated with a Luminoimage analyzer LAS-1000 PLUS (Fuji), and the ratio of the E41 form to total ST6Gal I (E41 form ϩ LMW form) is shown in the lower panel as mean Ϯ S.E. into a catalytic pocket of BACE1 in an appropriate orientation. There is another example showing that a membrane-tethered protease can process both types of substrate proteins, i.e. TACE (tumor necresis factor ␣-converting enzyme) can cleave EGF (epidermal growth factor) receptor (type I protein) and pro-TNF␣ (type II protein). As mentioned above, Lichtenthaler et al. (44) reported that BACE1 cleaves PSGL-1, which bears sialyl-LewisX glycans (Sia␣2,3Gal␤1,4(Fuc␣1,3)GlcNAc-) and mediates leukocyte trafficking, suggesting a possible function of BACE1 in the immune system. Cleavage of PSGL-1 by BACE1 may be an important down-regulation mechanism for controlling leukocyte migration during acute inflammation. In vivo cleavage of physiological substrates by BACE1 in nonneuronal tissues are important issues that need to be addressed.
We found two isoforms of soluble ST6Gal I in rat plasma: one is the E41 form having sialyltransferase activity and the other is the LMW form losing enzyme activity. Our results suggest that the LMW form is missing some important domain for catalytic activity, i.e. part of the catalytic and/or substratebinding domain. We are currently unable to identify a protease(s) that is involved in formation of the LMW form. Purification of the LMW form and subsequent sequence analysis will provide information on the substrate specificity of the protease. Molecular cloning of the protease cDNA will be required for understanding the entire process of in vivo cleavage and secretion of ST6Gal I. Two ST6Gal I isoforms showing different molecular weights were detected in rat plasma, whereas an apparent single band was observed in mouse plasma. The reason why we did not detect two bands in mouse plasma may be due to the amino acid difference at the position 41, i.e. Leu 37 -Gln 38 -Ala 39 -Lys 40 -Glu 41 Phe 42 -Gln 43 in the rat ST6Gal I and Leu 37 -Gln 38 -Ala 39 -Lys 40 -Val 41 -Phe 42 -Gln 43 in the mouse ST6Gal I. Rat ST6Gal I was first cleaved by BACE1 between Leu 37 and Gln 38 to generate the sequence Gln 38 -Ala 39 -Lys 40 -Glu 41 -Phe 42 -Gln 43 at the amino terminus, and then the three terminal amino acids, Gln 38 -Ala 39 -Lys 40 , were trimmed by a luminal aminopeptidase(s) to produce the E41 form (26). In the trimming process, the presence of Glu 41 appeared to be a signal for preventing further trimming. Mouse ST6Gal I, which lacks the Glu residue, would be further trimmed by the aminopeptidase, and the trimmed product may overlap with the other ST6Gal I isoform on a immuoblot (Fig. 1C).
LEC rats were initially found as a natural mutant that spontaneously develops jaundice around the age of 16 weeks, then they turned out to be a good model of Wilson disease. Like patients with Wilson disease, LEC rats have a deletion in the gene for the copper-transporting ATPase (ATP7B), which transports copper to the lumen of the secretory compartment to supply it to various copper-dependent enzymes such as Cu,Znsuperoxide dismutase, ceruloplasmin, and cytochrome c oxidase. High copper and iron levels in the liver and low copper concentrations in the plasma of LEC rats appear to cause several biochemical abnormalities, including decreased levels of Cu, Zn-superoxide dismutase activity (SOD) (45), cytochrome P450 isozymes (46), and S-adenosylmethionine synthase (␥-GTP) (47), as well as induction of N-acetylglucosaminyltransferase III activity (48), even before hepatitis occurs (ϳ16 weeks). We demonstrated that BACE1 mRNA transcription in the liver is significantly elevated in 6-and 8-week-old LEC rats, suggesting that elevation of BACE1 mRNA and plasma ST6Gal I can be used as early markers for hepatic stress prior to onset of hepatitis. Even though the level of BACE1 mRNA was higher at 8 weeks than 6 weeks of age, the ST6Gal I E41 form showed a maximum at 6 weeks. Although the expression level of BACE1 would be a major factor that controls the secretion of E41 form, other regulatory mechanisms may also affect the secretion, e.g. sorting of ST6Gal I to a particular subcellular compartment, in which the substrate ST6Gal I meet BACE1 protease. Such a mechanism other than BACE1 expression may also regulate the production of E41 form. Kaplan et al. (23) reported that serum ST6Gal I levels are enhanced during acute hepatitis, and the enhancement is attributed to elevation of ST6Gal I mRNA in the liver. Dalziel et al. (24) then reported that hepatic acute phase induction of ST6Gal I mRNA is controlled by a liverspecific promoter-regulatory region (P1) of ST6Gal I gene. Therefore, it will be necessary to examine the levels of ST6Gal I mRNA together with those of BACE1 mRNA in the liver during the hepatitis stage (after the age of 20 weeks) in LEC rats.
In conclusion, we showed that BACE1 is responsible for cleaving ST6Gal I in vivo and that increase of ST6Gal I secretion in the early hepatopathological condition of LEC rat is attributed mainly to up-regulation of BACE1 mRNA transcription in the liver. The amounts of mRNAs of BACE1 and GAPDH were analyzed by real-time PCR using the standard curve method according to the instructions of Applied Biosystems. The amounts of mRNAs for ST6Gal I and BACE2 relative to that for GAPDH were determined by the comparative C T method (following the instructions in User Bulletin 2, Applied Biosystems). All values were normalized to the level of GAPDH mRNA. Data are presented as the mean Ϯ S.E. (n ϭ 3) for each strain (LEC and Wistar) (*, p Ͻ 0.01).