Evidence That Tumor Necrosis Factor α Converting Enzyme Is Involved in Regulated α-Secretase Cleavage of the Alzheimer Amyloid Protein Precursor*

The amyloid protein, Aβ, which accumulates in the brains of Alzheimer patients, is derived by proteolysis of the amyloid protein precursor (APP). APP can undergo endoproteolytic processing at three sites, one at the amino terminus of the Aβ domain (β-cleavage), one within the Aβ domain (α-cleavage), and one at the carboxyl terminus of the Aβ domain (γ-cleavage). The enzymes responsible for these activities have not been unambiguously identified. By the use of gene disruption (knockout), we now demonstrate that TACE (tumor necrosis factor α converting enzyme), a member of the ADAM family (a disintegrinand metalloprotease-family) of proteases, plays a central role in regulated α-cleavage of APP. Our data suggest that TACE may be the α-secretase responsible for the majority of regulated α-cleavage in cultured cells. Furthermore, we show that inhibiting this enzyme affects both APP secretion and Aβ formation in cultured cells.

cleavage). The enzymes responsible for these activities have not been unambiguously identified.
In most cells in culture, a fraction (10 -30%) of all APP undergoes ␣-cleavage (4 -7). This results in the secretion of the large extracellular domain of APP into the medium. This secreted APP (APP s ) is a major APP-related species found in cerebrospinal fluid and brain homogenates (8,9) and is thought to interact with components of the extracellular matrix and with receptors on cells. In cultured cells it has been shown that the fraction of APP that is converted to APP s can be increased by activating second messenger cascades including those involving protein kinase C, protein kinase A, mitogen-activated protein kinase, protein phosphatase 1, protein phosphatase 2B (calcineurin), and calcium (4 -6, 10 -14). In most cells in culture, activating protein kinase C causes the majority (80 -95%) of the APP to undergo ␣-cleavage ("regulated" ␣-cleavage). Stimulation of APP s formation and secretion by activating second messenger cascades is not due to the phosphorylation of APP (15,16) but may be due to protein phosphorylation leading to alterations in the trafficking of APP (17) or in the activity of an ␣-secretase. Importantly, stimulating ␣-cleavage of APP leads to a significant decrease in A␤ formation (18 -20).
The potential importance of regulated cleavage of APP is indicated by the ability of acetylcholine, a critical neurotransmitter altered in Alzheimer's disease, working through muscarinic receptors, to stimulate regulated cleavage (4,10,21,22). Activation of other metabotropic receptors also leads to activation of regulated secretion of APP (4,21,23,24). Regulated cleavage of APP appears to occur in vivo under conditions in which protein kinase C is activated. These conditions include depolarization-induced release of acetylcholine in brain slices (25,26), as well as pharmacological, chronic activation of protein kinase C in rodent brain (27). The potential importance of the regulated cleavage pathway of APP makes it worthwhile to identify the enzyme(s) involved in this pathway.
Membrane-bound tumor necrosis factor ␣ (TNF-␣), like APP, is a transmembrane protein that can undergo proteolysis to release the extracellular domain as soluble TNF-␣. The generation of soluble TNF-␣ from membrane-bound TNF-␣ involves an enzyme activity denoted TNF-␣ converting enzyme (TACE) (28,29), which is a member of the ADAM family (a disintegrin and metalloprotease family) of proteases (30,31). Members of this family of proteases contain an autoinhibitory domain that must be removed for activity, a proteolytic domain, a disintegrin domain, a cysteine-rich domain, and most importantly for APP, a transmembrane domain.
Immunex compound 3 (IC3) (28), an effective inhibitor of ADAM family proteases as well as matrixin family proteases, had previously been shown to inhibit regulated secretion of APP (32,33). In this study, we determined whether TACE might play a role in the regulated ␣-cleavage of APP.

EXPERIMENTAL PROCEDURES
Materials-Cells stably expressing APP were the gift of Dr. E. H. Koo. Antibodies 22C11 and 6E10 were purchased from commercial sources.
Methods-Immunoblotting and metabolic labeling followed by immunoprecipitation were carried out as described previously (18,34).
The exact composition of the digests was determined by liquid chromatography/mass spectrometry (LC/MS), using a 1-mm inside diameter Vydac C18 column with a flow rate of 50 l/min and a linear gradient of 2% acetonitrile/min in 0.1% trifluoroacetic acid. MS was carried out by directing 10% of the HPLC effluent to the electrospray source of a Finnigan TSQ700 triple quadrupole mass spectrometer (San Jose, CA).

RESULTS
As a direct and definitive test for an involvement of TACE in the ␣-secretase pathway, we examined basal and regulated APP secretion in primary embryonic fibroblasts derived from control mice or mice in which the TACE gene had been disrupted by homologous recombination (knockouts) (28) (Fig. 1). In cells derived from control mice, regulated secretion was stimulated 300 -500% by activation of protein kinase C by phorbol 12-myristate 13-acetate (PMA). Inhibition of protein phosphatases 1 and 2A by okadaic acid had similar effects (not shown). These effects were blocked by IC3. In contrast, in cells derived from knockout mice, there was no increase in the formation and secretion of APP s caused by the addition of PMA ( Fig. 1) or okadaic acid (not shown). The levels of cell-associated, full-length APP were identical in the two cell populations, indicating that TACE knockout cells are not deficient in expressing APP (not shown). As an additional control, basal and PMA-activated protein kinase C activities were determined and were identical between the cell populations (not shown). The nearly total lack of PMA-and okadaic acid-induced cleavage and secretion of APP indicates that TACE (or a substrate of TACE) is the predominant ␣-secretase for regulated secretion in this system. Note that basal formation and secretion of APP s was unaffected in cells derived from knockout mice supporting the existence of two classes of secretases.
To begin to distinguish whether it is TACE itself that acts on APP or whether TACE acts via an intermediary, we tested whether recombinant TACE catalytic domain (35) could cleave a synthetic peptide, acetyl-VHHQKLVFFA-amide, encompassing the sequence of the ␣-secretase site of APP. TACE was able to cleave the synthetic peptide between Lys and Leu, at the APP ␣-secretase site (37-39) (Fig. 2), indicating that TACE has the potential to appropriately cleave APP. It should be noted that the ␣-secretase activity responsible for basal secretion of APP (which may not be TACE, see Fig. 1) acts at a site largely, but not completely, determined by the distance of the site from the membrane (40, 41). The amino acid requirements for sub-strate cleavage by TACE have not been fully determined although it is clear that TACE functions most effectively when it and its substrate are appropriately oriented in the membrane.
In addition to their effects on the formation and secretion of APP s , second messenger cascades are also able to regulate the formation and secretion of A␤. For example, activation of protein kinase C and/or inhibition of protein phosphatase 1 leads to increased formation and secretion of APP s together with a decrease in the formation and secretion of A␤ (18 -20). It has been hypothesized that the effects of these compounds are due to limiting amounts of substrate (APP) (7, 18), such that by FIG. 2. Cleavage by TACE of a peptide encompassing the ␣-cleavage site of APP. Acetyl-VHHQKLVFFA-amide was incubated with buffer alone (upper panel) or with the catalytic domain of TACE (lower panel), followed by HPLC to separate the reaction products, as described under "Methods." Peak 1 is the intact peptide, peak 2 is LVFFA-amide, and peak 3 is acetyl-VHHQK. The very minor UVabsorbing species generated were not detected by the alternate LC/MS procedure (see "Methods"). Data are representative of three experiments.

FIG. 3. Regulated formation and secretion of A␤ and its modulation by IC3
. CHO cells, stably expressing human APP 751 , were metabolically labeled for 2 h with [ 35 S]methionine followed by a 2-h chase period in which the cells were incubated in the presence of excess unlabeled methionine. During the chase period, cells were incubated in the absence or presence of PDBu and/or IC3. After incubation, secreted APP (APP s ) and A␤ were immunoprecipitated using antibodies 6E10 (which reacts with ␣-cleaved APP s and A␤) and 4G8 (which reacts with A␤). Upper panel, the precipitates were resolved on 10 -20% Tris-Tricine gels, dried, and subjected to quantitative storage phosphorautoradiography. Lower panel, means Ϯ S.E. of three independent experiments performed in duplicate are shown. *, p Ͻ 0.005; **, p Ͻ 0.05.

FIG. 1. Regulated secretion of APP in primary embryonic fibroblasts derived from wild-type, but not TACE-deficient, mice.
Primary embryonic fibroblasts derived from wild-type mice or from mice in which the TACE gene had been disrupted (knockout), were incubated for 2 h in the absence or presence of PMA and/or IC3. The amounts of secreted APP (APP s ) were then determined by immunoblotting 50-l aliquots of the culture medium with antibody 22C11. Data are representative of three experiments performed in duplicate. Note that all samples were run in duplicate on the same gel to facilitate comparisons (duplicate lanes were excised from the final image for simplicity).
stimulating the formation and secretion of APP s there is per force less A␤ formed. To test this hypothesis, cells stably expressing APP 751 were metabolically (pulse-chase) labeled and examined for effects of phorbol 12,13-dibutyrate (PDBu) and/or IC3 on APP s and A␤ formation and secretion, as determined by quantitative immunoprecipitation (Fig. 3). Importantly, the effects of PDBu on A␤ formation were completely blocked by IC3. This paralleled the effects of IC3 on regulated secretion of APP s , consistent with the hypothesis detailed above. Interestingly, not only were the effects of PDBu on A␤ formation blocked by IC3 but there was a significant increase in A␤ formation in the presence of PDBu and IC3. This indicates that there is a heretofore masked pathway in which activation of protein kinase C can lead to increased formation and secretion of A␤. As a final point, the antibodies used for immunoprecipitation of APP s (6E10) recognize the portion of APP between the ␤and ␣-cleavage sites. Thus, it is the ␣-secretase-derived species of APP s that is formed in the presence of PDBu and blocked by IC3. Similar results were observed with cells expressing mutant (NL) APP or when PMA (1 M) was used in place of PDBu (not shown).

DISCUSSION
In this study, we have examined the role of TACE in regulated ␣-secretion. Pharmacological manipulation of TACE leads to altered ␣-cleavage of APP and altered A␤ formation. More importantly, disruption of the TACE gene completely abolishes regulated ␣-cleavage in cultured cells. This result is consistent with TACE being an ␣-secretase, a premise further supported by the ability of TACE to cleave the APP synthetic peptide at the appropriate site. We cannot, however, rule out the possibility that TACE is acting on a separate protein which then affects APP cleavage.
It is important to note that basal formation and secretion of APP s was unaffected in cells derived from knockout mice. This supports the existence of two classes of secretases, one involved in basal secretion and one (perhaps TACE) involved in regulated secretion. Note, however, that in cultured CHO cells some basal secretion seems to be inhibited by IC3. Finally, by the use of IC3, we provide additional evidence that the effects of protein kinase C activation on A␤ formation are secondary to effects on the formation and secretion of APP s and identify a novel effect of protein kinase C on increasing A␤ formation under restricted conditions. Alterations in TACE activity during aging may contribute to amyloid formation. In addition, activating TACE by pharmacological manipulation might prove beneficial in Alzheimer's disease. Furthermore, homologs of TACE might be involved in basal secretion of APP s and in ␤-cleavage of APP.