Kinetics of amyloid beta-protein degradation determined by novel fluorescence- and fluorescence polarization-based assays.

Proteases that degrade the amyloid (cid:1) -protein (A (cid:1) ) are important regulators of brain A (cid:1) levels in health and in Alzheimer’s disease, yet few practical methods exist to study their detailed kinetics. Here, we describe robust and quantitative A (cid:1) degradation assays based on the novel substrate, fluorescein-A (cid:1) -(1–40)-Lys-biotin (FA (cid:1) B). Liquid chromatography/mass spectrometric analysis shows that FA (cid:1) B is hydrolyzed at closely similar sites as wild-type A (cid:1) by neprilysin and insulin-degrading enzyme, the two most widely studied A (cid:1) degrading proteases. The derivatized peptide is an avid substrate and is suitable for use with biological samples and in high throughput compound screening. The assays we have developed are easily implemented and are particularly useful for the generation of quantitative kinetic data, as we demonstrate by determining the kinetic parameters of FA (cid:1) B degradation by several A (cid:1) degrading proteases, including plasmin, which has not previously been characterized. The use of these assays should yield additional new insights into the biology of A (cid:1) -degrading proteases Bis-Tris- propane, KH 2 PO no salt or 100 m M NaCl and pre- pared at the pH optimum in the BRENDA enzyme data base (pH (cid:1) 7.4 in all cases). All proteases performed optimally in 50 m M HEPES, 100 m M NaCl, 0.05% bovine serum albumin, pH 7.4 (Buffer A); thus, these conditions were used for all subsequent degradation assays. Liquid Chromatography / Mass Spectrometry— Reverse phase high performance liquid chromatography was done on a Vydac diphenyl column (5 (cid:3) m; 1 (cid:2) 150 mm; Hesperia, CA), using a linear gradient (20–80%) of acetonitrile in 0.1% trifluoroacetic acid over 60 min. All components were monitored by UV and visible light absorbance using a Surveyor PDA (three channel outputs at 214, 259, and 495 nm) with a scan rate of 1 Hz and scan step of 1 nm. A nanospray ion source with a 100- (cid:3) m inner diameter-fused silica capillary was used to introduce the column effluent into an LCQDeca Mass Spectrometer, analyzed in a positive ion mode with a scan range of 200–2500 atomic mass units, using the following source conditions: capillary temperature, 250 °C; spray voltage, 4.56 capillary voltage, 46.11 sheath gas, 86.96 (arbitrary units); auxiliary gas, 2.68 (arbitrary units).

allelic association between late onset AD and polymorphisms near or within the IDE gene on chromosome 10q (6 -8).
Progress in elucidating the mechanisms underlying the production of A␤ from its protein precursor, APP, by the ␤and ␥-secretases has depended critically on the development of sensitive, reliable, and accessible assays for quantifying A␤ levels in biological samples. More recently, assays for directly measuring the activity of the secretases have been described (9 -11), enabling significant progress in the biochemical characterization of these proteases and the identification and characterization of small-molecule inhibitors. In contrast, relatively few techniques and assays for studying A␤ degradation have been reported, and the most commonly used general methods (e.g. measurement of radiolabeled peptides by trichloroacetic acid precipitation or HPLC) are cumbersome and ill suited for accurate quantification of kinetic constants or for high throughput assays. To overcome these difficulties, we have developed novel degradation assays based on a derivatized A␤ peptide, fluorescein-A␤-(1-40)-Lys-biotin (FA␤B). This substrate is efficiently degraded by several known A␤-degrading proteases and shows a similar inhibition profile as wild-type A␤ in biological samples. The assays we describe are sensitive, quantitative, and easily implemented, with one version requiring no specialized equipment other than a fluorometer. In addition, we describe a versatile fluorescence polarization (FP)-based assay that is both highly quantitative and suitable for high throughput compound screening. We use these assays to quantify the kinetics of A␤ degradation by IDE, NEP, and plasmin.

A␤ Peptides-Fluorescein-A␤-(1-40)-Lys-biotin and fluorescein-A␤-
(1-28)-biotin were synthesized by New England Peptide, Inc. (Fitchburg, MA). Biotin was attached to the carboxyl-terminal lysine side chain via an aminocaproic linker, and 5(6)-carboxyfluorescein (Sigma catalog no. C7153) was attached to the amino terminus via a peptide bond. Nonderivatized (wild-type) A␤ peptides were synthesized on an automated peptide synthesizer (Applied Biosystems model 433A) by 9-fluorenylmethoxycarbonyl-based methods. In preparation for quantitative kinetic determinations, aggregated species were removed by centrifuging freshly dissolved peptide (ϳ20 to ϳ150 M in 50 mM Tris-HCl, pH 7.4) at 100,000 ϫ g for 3 h and carefully removing the top two-thirds of the resulting supernatant, which was immediately aliquoted and stored at Ϫ80°C until further use. Peptide concentrations were determined by amino acid analysis.
Amino Acid Analysis-Samples in 50-mm glass tubes were dried in vacuo and then transferred to a hydrolysis vessel (Millipore Corp., Marloborough, MA; part no. 007603). Approximately 300 l of 6 N HCl was added to the vessel, which was then alternatively purged with nitrogen and evacuated three times before being sealed under vacuum. Vapor phase hydrolysis was performed by heating at 110°C for 22 h. Separation and quantitation of amino acids was carried out on a Beckman model 6300 amino acid analyzer. Each sample was analyzed in triplicate. Absolute peptide concentrations typically ranged from one-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
half to one-third of the concentration calculated from lyophilized peptide weight.
Proteases-The cDNA for human IDE beginning with the translation start site at Met 42 was subcloned into the pFastBacHT vector (Invitrogen), which codes for an amino-terminal tag containing a His 6 affinity tag followed by a tobacco etch virus protease cleavage site. Baculovirus production and protein expression in Sf9 moth cells were performed at the Wistar Institute (University of Pennsylvania). Recombinant IDE protein was then purified from snap-frozen Sf9 cell pellets by cobalt metal affinity chromatography according to the manufacturer's recommendations (Clontech), and the affinity tag was removed by treatment with tobacco etch virus protease (1000 units/mg of recombinant protein) for 2 h at 22°C (Invitrogen). The His 6 -containing tag and any residual tagged protein were separated from the cleaved IDE by cobalt affinity chromatography, followed by overnight dialysis against two changes of 20 mM KH 2 PO 4 , pH 7.3, using a dialysis container with a 35-kDa molecular mass cut-off (Millipore). Finally, to separate full-length ϳ113-kDa recombinant IDE from a prominent ϳ70-kDa truncated species, the sample was washed and concentrated using a 100-kDa molecular mass cut-off centrifugal filter column (Amicon). Recombinant IDE was stored in aliquots in 100 mM KH 2 PO 4 , pH 7.3, plus 20% glycerol at Ϫ80°C until further use. Recombinant rabbit secreted NEP (sNEP) produced in P. Pastoris (12) was a gift from Drs. Guy Boileau and Philippe Crine (University of Montreal). Human plasmin was purchased from Calbiochem/Novabiochem. Prior to initiation of quantitative studies, optimal conditions for each protease were determined by testing its activity in various buffers (e.g. HEPES, Tris-HCl, Bis-Trispropane, KH 2 PO 4 ) containing either no salt or 100 mM NaCl and prepared at the pH optimum listed in the BRENDA enzyme data base (pH ϳ7.4 in all cases). All proteases performed optimally in 50 mM HEPES, 100 mM NaCl, 0.05% bovine serum albumin, pH 7.4 (Buffer A); thus, these conditions were used for all subsequent degradation assays.
Liquid Chromatography/Mass Spectrometry-Reverse phase high performance liquid chromatography was done on a Vydac diphenyl column (5 m; 1 ϫ 150 mm; Hesperia, CA), using a linear gradient (20 -80%) of acetonitrile in 0.1% trifluoroacetic acid over 60 min. All components were monitored by UV and visible light absorbance using a Surveyor PDA (three channel outputs at 214, 259, and 495 nm) with a scan rate of 1 Hz and scan step of 1 nm. A nanospray ion source with a 100-m inner diameter-fused silica capillary was used to introduce the column effluent into an LCQDeca Mass Spectrometer, analyzed in a positive ion mode with a scan range of 200 -2500 atomic mass units, using the following source conditions: capillary temperature, 250°C; spray voltage, 4.56 kV; capillary voltage, 46.11 V; sheath gas, 86.96 (arbitrary units); auxiliary gas, 2.68 (arbitrary units).
FA␤B Degradation Assay Using Avidin-Agarose Precipitation (AP)-For progress curves and dose-response experiments, each protease (1-400 nM final concentration) was dissolved in 0.5 ml of Buffer A in 0.65-ml siliconized microcentrifuge tubes (Costar), and the reaction was initiated by adding 100 l of 6 M FA␤B. For kinetic determinations, different concentrations of FA␤B (0.25-150 M final concentration) were dissolved in 0.5 ml of Buffer A, and the reaction was initiated by adding 100 l of diluted protease. At various time points (0, 5, and 10 min), 80 l of the reaction mixture was removed and quenched in 0.72 ml of Buffer A supplemented with the appropriate protease inhibitor (2 mM 1,10-phenanthroline for IDE and NEP or 1 mM phenylmethylsulfonyl fluoride for plasmin). Uncleaved FA␤B substrate was precipitated from each quenched reaction by the addition of suitable quantities of Neutravidin TM -coated agarose beads (Pierce) followed by gentle rocking for 30 min and centrifugation at 14,000 ϫ g for 10 min. The supernatant (containing cleaved fluorescein-A␤ fragments) was carefully transferred in three 200-l aliquots to black 96-well plates (Nunc), and fluorescence intensity (488 excitation, 515 emission) was measured using a Victor2 multilabel plate reader (PerkinElmer Life Sciences). All AP assays were performed at 37°C. Activity was normalized to control reactions containing either no protease or excess protease. Quantitative kinetic data were derived by hyperbolic regression analysis using the computer program HYPER.EXE created by John S. Easterby (University of Liverpool).
FA␤B Degradation Assays Using FP-FP degradation assays carried out at 37°C were performed as above, except that 30-l aliquots were quenched in 170 l of Buffer A supplemented with 1 M egg white avidin (Molecular Probes, Inc., Eugene, OR) plus appropriate protease inhibitor(s). After incubating at room temperature for 15 min, quenched reactions were loaded in quadruplicate onto "nonbinding surface" 384well plates (Corning), and fluorescence polarization (excitation, 488; emission, 515) was determined using a Victor2 multilabel plate reader (PerkinElmer Life Sciences).

Robot-assisted, High Throughput FP FA␤B Degradation Assays-
High throughput experiments were conducted on a customized apparatus (SAGIAN; Beckman) containing a 3-meter rail ORCA robot that integrates a Biomek FX liquid handling station, a SAGIAN core system, and SAGIAN six-plate shaker. Recombinant IDE (0.3-3 nM final concentration) dissolved in Buffer A (20 l/well) was loaded onto "nonbinding surface" 384-well plates (Corning). Following a 20-min incubation period at room temperature, FA␤B (1 M) in Buffer A (20 l/well) was added to initiate the reactions. Reactions were terminated with 10 l/well Buffer A supplemented with 10 mM 1,10-phenanthroline and 5 M avidin. FP values were determined on an LJL Analyst HT (488-nm excitation, 515-nm emission).
Performance of the assay in the high throughput format was evaluated by computing the Z-factor values (13) according to the following formula, where HI and LO designate the S.D. values and X HI and X LO represent the means of data from internal controls representing 100 and 0% activity, respectively. Using this formula, an ideal assay would have a Z-factor value approaching 1.0, and assays with values below ϳ0.5 would generally be considered unreliable.

Inhibitor Profiles Determined in Rat Brain Membrane Fractions Using FA␤B FP Versus 125 I-A␤-(1-40)-Trichloroacetic Acid Precipitation
Assays-Fresh-frozen rat cerebral hemispheres were homogenized in 8 volumes (w/v) of 0.25 M sucrose in 50 mM Tris-HCl (pH 7.4) in a Potter-Elvehjem homogenizer. After pelleting nuclei and unbroken cells, the supernatant was spun again at 100,000 ϫ g for 1 h. This resulting supernatant was saved as the soluble fraction, and the membrane pellet was washed in 100 mM Na 2 CO 3 (pH 11.3) to linearize microsomes and strip adventitiously associated proteins (14). The membranes were pelleted by another 1-h spin at 100,000 ϫ g, resuspended in 50 mM Tris-HCl (pH 7.4), and sonicated by a 10-s pulse with a model 300 sonic dismembrator (Fisher) set at 60% maximum power. Protein concentrations were determined using a bicinchoninic acid-based protein assay (Pierce). For trichloroacetic acid precipitation-based A␤ degradation assays, 100 pM synthetic human 125 I-A␤-(1-40) (Amersham Biosciences) was incubated at 37°C with 100 g/ml membrane fractions in 50 mM Tris-HCl (pH 7.4) with or without various inhibitors. At various time points (e.g. 3 and 6 h), an aliquot of the sample was added to an equal volume of 15% trichloroacetic acid to precipitate uncleaved peptides. Following centrifugation, radioactivity in the trichloroacetic acid-insoluble pellet (undegraded peptide) and trichloroacetic acid supernatant (degraded peptide fragments) were determined, and the percentage of radiolabeled substrate degraded was calculated. The FP assay was performed as described above using identical quantities of the same brain membranes. Recombinant insulin, purified glucagon, thiorphan, and 1,10-phenanthroline were from Sigma.

Measurement of FA␤B Proteolysis by Fluorescence Polariza-
tion-FP is a sensitive method for measuring the relative mass of fluorescent molecules in solution (19). Stationary fluorescent molecules, when excited with plane-polarized light, emit photons polarized in the same plane as (or at a fixed angle to) the light used for excitation. Fluorescent molecules in solution, by contrast, rotate (or tumble) at a rate that is inversely proportional to their mass. Because there is an appreciable time delay between absorption and emission of photons by the fluorescent molecule, tumbling causes emitted light to be depolarized relative to the plane-polarized light used for excitation. The degree of depolarization (reflecting the average mass of all fluoresceinated species) can be quantified accurately using appropriately equipped fluorometers.
To effectively measure proteolysis using FP, there must be a substantial mass difference between the intact, fluorescently tagged substrate and the cleaved, fluorescently tagged proteolytic fragments. The addition of a biotin moiety at the opposite end of the molecule to the fluorescent tag allows the molecular weight of the intact substrate to be increased (to ϳ70 kDa) by adding excess avidin at the end of the reaction. The molecular weights of the cleaved, fluoresceinated NH 2 -terminal fragments (Ͻ4 kDa), by contrast, remain unaffected by the presence of avidin and tumble rapidly. The relative amounts of cleaved and uncleaved substrate can then be accurately measured using FP. The robust and reproducible nature of our FP-based FA␤B degradation assay is illustrated in Fig. 2A, which shows progress curves with very small S.D. values for the degradation of FA␤B by IDE or by NEP.
Because of concern about the tendency of A␤ peptides to aggregate and/or adsorb nonspecifically to the surfaces of reaction vessels, we also tested a version of the derivatized substrate lacking the hydrophobic carboxyl-terminal 12 amino acids (i.e. fluorescein-A␤-(1-28)-biotin). However, this species proved to be a relatively poor substrate for IDE and other A␤-degrading proteases (data not shown) and was not pursued further. We were able to overcome adsorption problems associated with full-length FA␤B substrate by adding 0.05% bovine serum albumin to the assay buffer and by keeping the substrate concentration above 20 nM (see below).

Measurement of FA␤B Proteolysis by Avidin-Agarose Precipitation-
The FA␤B substrate can be incorporated into another fluorescence-based degradation assay we developed that does not require the specialized equipment used to measure FP. In this assay, avidin-conjugated agarose beads are used to centrifugally separate uncleaved FA␤B from the fluoresceinated (nonbiotinylated) amino-terminal proteolytic fragments; the amount of fluorescence remaining in the supernatant corresponds directly to the amount of proteolysis. As shown in Fig.  2B, this AP method yielded virtually identical results to the FP-based method in parallel experiments (cf. Fig. 2, A and B; see also Fig. 4, A and B). Whereas the AP assay has the advantage of requiring no specialized equipment, the requirement for a centrifugation step precludes its use in high throughput compound screening; consequently, our further work focused on the validation and miniaturization of the FP assay.
Specificity of the FA␤B Substrate in Biological Samples-To test the FP FA␤B degradation assay with biological material,

FIG. 1. Fluorescein-A␤-(1-40)-Lysbiotin (FA␤B) is cleaved at similar sites as wild-type A␤-(1-40). Amino acid sequence of A␤-(1-40) depicting scissile bonds of FA␤B (upper arrowheads)
and wild-type A␤-(1-40) (lower arrowheads) cleaved by IDE (A) or NEP (B), as determined by liquid chromatography/ mass spectroscopy, is shown. The proteolytic fragments identified by liquid chromatography/mass spectrometry are listed in Tables I and II. we determined the inhibition profile of FA␤B using bicarbonate-washed rat brain membrane fractions, which contain a wide array of proteases, and compared the result with that determined using a standard 125 I-A␤-(1-40)/trichloroacetic acid precipitation assay (4,20). As we have recently reported elsewhere (4), the majority of 125 I-A␤-(1-40) degradation in membrane fractions is competitively inhibited by insulin or glucagon, which are avid IDE substrates, whereas a smaller fraction is inhibited by thiorphan, which is a potent inhibitor of NEP (Fig. 2C). 1,10-Phenanthroline, a broad spectrum inhibitor of zinc-metalloproteases, also strongly inhibited 125 I-A␤-(1-40) degradation, as expected. Hydrolysis of FA␤B, determined in the same biological samples using the FP assay, exhibited a highly similar inhibition profile (Fig. 2D). Thus, FA␤B is an avid substrate for multiple known A␤-degrading proteases and is suitable for in vitro degradation assays on biological samples. Validation of the FP Assay-To directly test the linearity of the FP assay, we determined the polarization values produced by aliquots of FA␤B (1 M) containing predetermined percentages of uncleaved substrate and substrate previously digested with IDE or NEP. As shown in Fig. 3A, the change in polarization values, normalized to the maximum obtained with each protease, varied in a nearly linear fashion with percentage of hydrolyzed substrate. Because quantitative assays depend on accurate determinations of the initial velocities of substrate hydrolysis, which are typically determined from the first 10 -20% of a progress curve, we quantified the deviation from linearity observed in this range. The FP assay was found to overestimate the initial velocity by ϳ10% for hydrolysis by either protease (Fig. 3A). Hence, this figure was used as a correction factor in subsequent quantitative determinations (Fig. 4A).
During the development of the FP assay, we initially sought to maintain a low (5-20 nM) substrate concentration before reading the polarization values, since such levels are typically used for FP applications. This constraint precluded attempts to miniaturize the assay for high throughput screening, because proteolysis was inefficient and highly variable at such low substrate concentrations and because protocols involving further dilution of the substrate were undesirable. However, additional testing revealed that the FP assay was actually most linear, in terms of absolute polarization values, in the range of substrate concentrations between 100 and 1000 nM (Fig. 3B). The working range could be extended further, provided the resultant data were normalized to empirically determined maximal and minimal polarization values at each substrate concentration (see below), and in practice, such normalization was routinely used. This range of suitable substrate concentrations in the FP assay had several practical benefits, particularly for kinetic analyses requiring variable concentrations of substrate.
High Throughput FP Degradation Assay-To test the performance of the FP assay in a high throughput, robot-assisted format, various concentrations of recombinant IDE (0.3-3 nM final concentration) were transferred robotically to 384-well plates containing 0.2 l of Me 2 SO (0.5% final concentration). The reactions were initiated by adding FA␤B (0.5 M final concentration) and terminated at various times by the addition of assay buffer containing 1,10-phenanthroline and avidin. The assay performed superiorly in this format (Fig. 3C), routinely yielding Z-factor values greater than 0.8 (see "Materials and Methods").
Kinetic Analysis of FA␤B Degradation by IDE, NEP, and Plasmin-The AP and FP assays were used to obtain quantitative kinetic data for three different A␤-degrading proteases: IDE, NEP, and plasmin. Data for endothelin-converting enzyme-1 and -2 were not determined due to lack of availability of the purified enzyme. Because even freshly dissolved A␤ peptides consist of unknown fractions of oligomerized species, we were careful to first remove aggregated FA␤B peptides by high speed centrifugation and then quantify the absolute concentration of peptide by amino acid analysis (see "Materials and Methods"). Lineweaver-Burk plots of these kinetic determinations are illustrated in Fig. 4, A-C, whereas quantitative data derived by hyperbolic regression analysis of the raw data are provided in Table III protease, very similar quantitative kinetic data were obtained with the AP and the FP assays, and the data were in even closer agreement after correcting for the slight overes-timation in initial velocity that occurs with the FP assay (see above and Figs. 3A and 4A).
Applications of the FA␤B Degradation Assays-To highlight the utility of the FP assay in generating highly reliable quantitative data, we performed several demonstration experiments using medium throughput protocols optimized for 96-well plates. Fig. 5A shows raw progress curves from a typical ex- , and plasmin (C). Data were determined by FP (q) and AP (OE) degradation assays. Open circles in A represent FP data before correction for minor nonlinearity in FP progress curves (see Fig. 3A). Kinetic parameters, determined by hyperbolic regression analysis, are given in Table III. periment that determined the IC 50 for inhibition of IDE-mediated FA␤B degradation by 1,10-phenanthroline. The dose-response relationship, determined from the average of three such experiments, is illustrated in Fig. 5B. Fig. 5C shows Lineweaver-Burk plots for inhibition of FA␤B degradation by insulin, an avid substrate of IDE. Note that these results are characteristic of competitive inhibition, as expected. For comparison, Fig. 5D shows comparable data for inhibition by Nethylmaleimide, an irreversible thiol-alkylating agent, and this experiment yielded the expected profile for a noncompetitive inhibitor. Collectively, these experiments show that the FP FA␤B degradation assay is highly quantitative, reproducible, and capable of revealing insights into the detailed kinetics of A␤-degrading proteases.

DISCUSSION
Evidence emerging from numerous laboratories has highlighted the importance of A␤-degrading proteases in regulating steady-state cerebral A␤ levels in vivo (2)(3)(4)(5). Inherited defects in one or more of these proteases could explain A␤ accumulation in some cases of AD, but regardless of whether such genetic connections are found, these enzymes provide potentially attractive drug targets for treating all cases of the disease. However, traditional methods for studying A␤ degradation (e.g. trichloroacetic acid precipitation assays with iodinated A␤, HPLC analysis of synthetic radiolabeled or unlabeled A␤, and autoradiography of metabolically labeled A␤) are ill suited for high throughput compound screening or routine quantitative assays. To meet this important need, we have developed and characterized versatile and quantitative assays that utilize the derivatized A␤ peptide, FA␤B.
Several lines of evidence indicate that degradation of FA␤B is similar to that of wild-type A␤. First, the derivatized peptide is avidly degraded by several different proteases that have been shown to degrade A␤: IDE, NEP, and plasmin. Second, liquid chromatography/mass spectrometric analysis shows that wild-type and derivatized A␤ are cleaved at nearly identical sites by IDE and NEP, the two most extensively characterized A␤-degrading proteases. Third, degradation of FA␤B and 125 I-A␤-(1-40) show closely similar inhibition profiles when incubated in biological samples containing an array of competing proteases. These findings suggest that the fluorescein and lysine-biotin moieties do not significantly alter the cleavage specificity of the substrates by the A␤-degrading proteases investigated here.
Additional support for the validity of our FA␤B substrate comes from analysis of quantitative data from kinetic determinations. Independent analyses using the FP-and AP-based assays yielded quantitatively similar estimates of kinetic parameters for hydrolysis, and this pertained for all three A␤degrading proteases. Moreover, kinetic determinations using FA␤B (Table I)   3.1 ϫ 10 8 2.5 ϫ 10 8 6.5 ϫ 10 6 6.2 ϫ 10 6 9.8 ϫ 10 6 8.6 ϫ 10 6 a Data calculated using correction factor derived from Fig. 3A.
FIG. 5. Quantitative applications of the FA␤B degradation assay. A, typical progress curves in an experiment determining the IC 50 of 1,10-phenanthroline on IDE-mediated hydrolysis of FA␤B. B, the average of three experiments such as that shown in A. C, Lineweaver-Burk plots of kinetic data for the hydrolysis of FA␤B by IDE without (f) or with (E) insulin (100 nM), an avid competitive substrate. D, Lineweaver-Burk plots of kinetic data for inhibition by N-ethylmaleimide (30 M), an irreversible, noncompetitive inhibitor. these values are lower than the published values, this discrepancy disappears after it is noted that about one-half to twothirds of the lyophilized peptide was pelleted by high speed centrifugation and thus in an aggregated state that is incapable of degradation by IDE (see Ref. 23). Indeed, when freshly dissolved substrate was used, we obtained kinetic parameters that closely matched the published values for IDE (k m ϭ 2.11 Ϯ 0.20 M). These results suggest that caution should be exercised in interpreting kinetic data for A␤ degradation when the peptide's aggregation state has not been characterized.
Our liquid chromatography/mass spectrometric analysis revealed that NEP-mediated hydrolysis of A␤ (both wild-type and derivatized) occurred at substantially more sites than previously reported by Howell et al. (18), with considerable heterogeneity observed at the carboxyl terminus of the peptide (Fig.  1B). This discrepancy may be explained by the fact that the latter study focused only on the initial peptide fragments detectable upon hydrolysis with NEP. However, it is also possible that our analysis was more sensitive, given the fact that the LC and MS analyses were coupled within the same instrument. Relative to IDE, it appears that NEP shows less cleavage specificity within A␤. In addition, the k m for hydrolysis of FA␤B by NEP (ϳ14 M) is more than 17 times larger than that for IDE (ϳ0.8 M; see Table I). We note that comparison with reported k m values for A␤ hydrolysis by NEP is complicated by the fact that we used a secreted form of rabbit NEP, whereas published values for wild-type A␤ degradation were determined using membrane-anchored NEP (24). However, the secreted NEP used in this study has been directly compared with membrane-anchored NEP, and found to exhibit similar kinetics parameters using the preferred substrate, [D-Ala 2 ,D-Leu 5 ]enkephalin (12). To our knowledge, the k m value for A␤ degradation by plasmin has not been determined previously.
In conclusion, we have developed sensitive and quantitative assays for the study of the proteolytic degradation of A␤. The assays are simple, highly reproducible, inexpensive, and suitable for high throughput screening. Implementation of these assays should yield novel insights into the detailed kinetics of various A␤-degrading proteases and facilitate the identification of novel small molecule pharmacophores.