An Abnormahty of Plasma Amyloid Protein Precursor in Alzheimer's Disease Ashley I. Bush, FRANZCP,"? Scott Whyte, FRACP," Linda D. Thomas, BSc(Hons)," Timothy G. Williamson, BSc(Hons)," Cees J. Van Tiggelen, MD,X Jon Currie, FRACP," David H . Small, PhD," Robert D. Moir, BSc(Hons),* Qiao-Xin Li, PhD," Baden Rumble, PhD," Ursula Monning, PhD,S Konrad Beyreuther, PhDJ and Colin L. Masters, MD"

PA4 amyloid deposition in the brain, which is characteristic of Alzheimer's disease (AD), may result from either overexpression of the amyloid protein precursor (APP) or failure of APP to be correctly processed. A blood marker reflecting this abnormal metabolism would be of diagnostic value and would provide a means of monitoring the efficacy of therapeutic interventions. We analyzed immunoblots of plasma APP enriched by heparin-Sepharosechromatography from patients with moderate to severe AD dementia (n = 34) and control subjects (n = 77) and found an approximately 50% increase in the proportion of 130-kd APP species in patients with AD (p < 0.001), no difference in the 110-kd form, a 15 to 30% decrease in the 65-kd form (p < 0.001), and a 20 to 35% decrease in the proportion of 42-kd APP (p < 0.001). These species of APP were soluble, lacked the carboxyl terminus, and the 110- and 42-kd species were shown to be consistent with degradation products derived from the 130-kd species. A comparison of levels of 130-kd plasma APP from moderately to severely demented patients with AD and control subjects distinguished the two groups with a specificity of 87.0% and a sensitivity of 79.4%. Bush AI, Whyte S, Thomas LD, Williamson TG, Van Tiggelen CJ, Currie J, Small DH, Moir RD, Li Q-X, Rumble B, Monning U, Beyreuther K, Masters CL. An abnormality of plasma amyloid protein precursor in Alzheimer's disease. Ann Neurol 1992;32:57-65

Alzheimer's disease (AD) is a dementia characterized by the deposition of PA4, a 4-kd protein that accumulates by polymerization into amyloid fibrils in the brain 11, 2). PA4 is released by the abnormal processing of amyloid precursor protein (APP). The prototype of APP is the form with 695 amino acids (APP&, which has the structural domains of an integral transmembrane protein 131. Alternatively, spliced amyloidogenic forms include APP751,with a Kunitz protease inhibitory (KPI) domain 14, 51; APP,14, with an OX-2 domain insert 161; and APP,,,, with both the KPI domain and an OX-2 domain 171. At least three mechanisms could lead to the generation of PA4. One possibility is that aberrant cleavage of APP may be caused by a defect in the constitutive proteolysis known to occur within the PA4 domain 18, 91. Although candidate enzymes for APP processing have been described 110, 111, an association between an APP-cleaving protease activity and clinical AD has yet to be shown. Conversely, an increase in substrate

may overwhelm the constitutive pathway and provide APP for an amyloidogenic pathway. Such a mechanism is likely in Down syndrome (DS), in which an extra copy of chromosome 21 is associated with increased expression of APP and premature deposition of PA4 112). A third mechanism could involve a modification to the APP molecule itself, which then results in an alteration of the constitutive proteolysis. Such a mechanism is possible in the closely related disorder of Dutch congophilic angiopathy, which is associated with a mutation within the PA4 domain of APP 113, 141, and in forms of famhal AD in which there are mutations close to the carboxyl terminus of the PA4 peptide 115- 191. Circulating forms of APP are concentrated in platelets 120-221, whereas lower amounts occur in plasma 112, 20, 23) and resting T lymphocytes 1241. We report evidence illustrating that the levels and processing of plasma APP may be altered in AD. Furthermore, we show that the profile of plasma APP in AD can be

From the *Department of Pathology, the University of Melbourne, the Neuropathology Laboratory, Mental Health Research Institute Of victoria, and the tDepamnent Of Hospital, Parkvdle, Victoria, Australia; and the $Center for Molecu. lar Biology, University of Heidelberg, Heidelberg, Germany.

Received Sep 10, 1991, and in revised form Jan 2, 1992. Accepted for publication Jan 2, 1992. Address correspondence to Dr Masters, Depmment of pathology, Universiryof Melbourne, Parhde, 3052, Australia,

Copyright 0 1992 by the American Neurological Association 57

changed to approach that of control profiles by the action of a serine protease activity that copurifies with APP during heparin-Sepharose chromatography.

tion. The protein A-Sepharose was pelleted by centrifugation (10,000 g; 3 min), and antiserum was then added to the separated supernatant.

Materials and Methods

Western Blotting

Patient Selection Patients with AD met National Institute of Neurological Disease and Stroke/Altheimer’s Disease and Related Disorders Association (NINDUADRDA) clinical criteria [25], had Mini-mental state examination {26} scores of less than 17, and were excluded if they had clinical findings consistent with peripheral vascular disease or mixed vascular dementia and AD. Patients with familial AD were excluded from this study, as were patients with AD in whom clinical onset was before the age of 55 years. Age-matched control subjects each underwent a Mini-mental state examination and were excluded if they scored less than 29. The different neurological diagnoses used to determine control subjects with nondemented neurological disease (n = 6 ) were epilepsy, multiple sclerosis, hydrocephalus, and three with cerebrovascular disease; this group was included as a small initial survey for the specificity of the changes seen in the A D group and as a preliminary to a more exhaustive survey of acute and chronic neurological and non-neurological disease that is beyond the scope of this initial report. Ail volunteers were ambulatory, in stable health, and not suffering any acute illness at the time of the study. Subjects were excluded from the control group if they had a family history of any heritable dementing disorder.

Western blotting procedures were as described by Bush and associates [20), modified by transferring for 5 hours at 0.6 A and blocking with 3% (wh) bovine serum albumin (Fraction V; Sigma, St Louis, MO) in Tris-buffered saline (2 hours, room temperature (RT]). The efficiency of protein transfer was estimated by blotting onto two stacked nitrocellulose filters (0.2 pm pore diameter) and staining the gels for untransferred protein with Coomassie blue following the blotting step. This process indicated that no immunoreactive APP passed through the first filter under these conditions, but that the efficiency of protein transfer was approximately 90% below 46 kd and gradually decreased to approximately 50% with increasing molecular weight between 46 and 130 kd. Blots were probed with mouse monoclonal antibody (mAb) 2 2 C l l (Boehringer-Mannheim, Munich, Germany), which recognizes an epitope on the amino terminus of APP 1281; it was diluted to 170 ng/mL in blocking buffer. Plasma samples in this series of experiments were separated on 8.5% polyacrylamide gels. Using these conditions, the 110-kd APP immunoreactive band previously reported {20] resolved into a doublet. For the purpose of our analysis, however, we regarded the sum of the signals generated by the doublet as belonging to the one 110-kd region.

Partial Purification of A P P from Plasma

Refectance Analysis of Blots

Blood (20-40 mL) was drawn from fasting individuals with a 2 1-gauge needle into heparinized collection tubes and centrifuged at 1,500 g for 15 minutes. The plasma fraction was separated from the blood cell pellet and centrifuged at 19,000 g for 25 minutes at 4°C to remove any debris. Processing and assay of all plasma specimens were performed with the operator unaware of the diagnostic category of the sample. To detect APP by Western blotting, APP was partially purified from plasma by heparin-Sepharose chromatography. Plasma (2.5 mL) was loaded onto a 0.25-mL. bed volume heparin-Sepharose (Pharmacia, Uppsala, Sweden) column (8 x 5 mm) pre-equilibrated with buffer 1 (175 mmollL NaCl, 50 mmoVL Tris-HC1 [pH, 7 . 4 3 at 4°C. The column was then washed with 3.25 mL buffer 1, and the APP was eluted with 750 pL elution buffer (550 mmol/L NaC1, 50 mmoVL TrisHCl [pH, 7.41). Protein concentration was determined with a bicinchoninic acid protein assay using bovine serum albumin standards (271.

Reflectance analysis was used to quantitate amounts of APP on Western blots. Reflectance was assayed by video-capture with a Videk Megaplus camera (Kodak, Canandaigua, NY) operated by PixelTools vl. 1 (Perceptics, Knoxville, TN). Quantitation was then performed with Image v1.29 software (W. Rasband, National Institutes of Health Research Services Branch, National Institute of Mental Health, Bethesda, MD), which facilitated precise alignments of the individual blot lanes with the reflectance profile and the setting of exclusion limits of individual peaks in the four regions of interest at 130, 110, 65, and 42 kd. The integrated reflectance (area under the curve) was thus computed for each of the four peaks in every sample. The values obtained were linear with concentration for each region over a range of heparin-sepharose eluate loads (20-90 pg protein). To compare the relative amounts of the four APP derivatives, the relative proportion of band immunoreactivity to total lane immunoreactivity was determined in each plasma sample and then averaged to give the values presented in the Table and in Figure 1. Independent samples Student’s t-tests between pooled control and A D groups were performed for the four immunoreactive bands at a significance level of p = 0.0125 (0.05 divided by the number of comparisons 2291). Where these comparisons were significantly different, the test of simple effects [29] followed by Scheffk [30] post hoc comparisons between all pairs of diagnostic groups (i.e., patients with AD, other neurological disease control subjects, normal younger adult control subjects, and nondemented, age-matched control subjects) were then performed.

Immunoprecipitation Heparin-Sepharose eiuates (650 pg protein) were diluted with 50 mmol/L Tris-HC1 (pH 7.4) to reduce the concentration of NaCl to 175 mmol/L and were immunoprecipitated with rabbit polyclonal antisera (5 pL) according to the method described by Bush and associates {20}, modified by preincubating the samples with 10 mg protein A-Sepharose (Pharmacia) for 1 hour at 4°C to reduce nonspecific absorp-

58 Annals of Neurology

Vol 32 No 1 July 1992

Ratios of Plasma APP F o m Analyzed by Image Capture in A D and Control Subjects" Percentage of Total APP Mean Age +- SD AD (n = 34) Pooled control subjects (n = 77) Neurological control subjects (n = 6 ) Age-matched control subjects (n = 46) Younger adult control subjects (n = 25)

67.9 61.5

* 8.9 * 16.6

64.0

2

6.8

* 10.2 41.5 * 10.9 70.6

130 kd

110 kd

36.5 5 10.6 23.9 +- 6.9b

24.7 23.6

25.0

&

8.3'

24.0 +- 7.2' 23.4

* 6.3'

S

SD

65 kd

* 6.6 * 5.3 (NS) 22.7 * 5.8 25.5 * 5.8 20.2 * 3.6

42 kd

21.5 26.8

5

7.1

17.4 2 6.0 25.7 5 7.2b

31.0

5

3.2'

21.3

25.0

* 5.8 (NS)

30.0

L

* 6.2b

6.6'

* 10.6 (NS 25.4 * 6.4' 27.4

-C

7.5'

"Heparin-Sepharose eluates of AD and control plasma samples were analyzed by Western blotting with mAb 22C11, and the intensities of th bands at 130, 110, 65, and 42 kd were measured by computer-assisted image capture analysis. The relative amounts of the four APP band: as percentages of total lane signal, were determined in each plasma sample and averaged to give the values presented. Independent sample t-tests between pooled control and AD groups were performed in the four regions at a significance level of p = 0.0125. Comparisons wer significant for three regions: 130,65, and 42 kd (two-taiied, p < 0.001). Comparisons between all pairs of diagnostic groups (patients with All other neurological disease control subjects, normal younger adult control subjects, and non-demented, age-matched control subjects) were the performed for the values obtained for the 130-, 65-, and 42-kd bands using a Scheffk test at a significance level of p < 0.05. bt-test, p < 0.001. 'Significant by Scheffk test, p < 0.05. APP

=

amyloid protein precursor; AD = Alzheimer's disease; NS = not significant; mAb

=

monoclonal antibody.

A

B

Fig 1. Analysis of amyloid protein precursor (APP) in control and Alzheimer's disease (AD) plasma by Western blotting. Heparin-Sepharose eluates from plasma (65 pg protein) were analyzed by 8.5% sodium dodecyl sulfate polyacrylamih gel electrophoresis and Western blotting with monoclonal antibody 22CI 1. The relative molecular mass of standard protein markers (Rainbow Stanhrds, Amersham, UK) are shown on the left.

APP immunoreactive bands of 130, 110, 65, and 42 kd are indicated by arrows on the right. Only the relative abundances of the 130- and 42-kd APP forms, as in the samples illustrated, could be used t o visibly discriminate between patients with A D compared with both (A) nondemented el&& control subjects and (B) n o m l younger control populations.

Bush et al: Abnormal Processing of APP in Alzheimer's Disease 59

h s a y of APP-degrading Protease Aliquots (500 pL) of eluates from each heparin-Sepharose column were desalted using a 1.7-mL Sephadex G25 (Pharmacia) column (8 x 34 mm) equilibrated with 175 mmoVL NaCI, 50 mmoVL Tris-HC1, 1 mmol/L CaCl,, and 1 mmoV L MgCl, (pH, 7.4), in the presence or absence of 20 pmoV L ZnC1, at 4"C, and the protein concentration was adjusted to 0.80 mg/mL with the same buffer. The samples were then incubated at 37°C for 2 hours; an aliquot (80 G)was then removed, and the protein in each aliquot was precipitated with chloroformlmethanol(1:4 vlv), boiled in sodium dodecyl sulfate (SDS) sample buffer, and analyzed by Western blotting using mAb 22C11. The effects of inhibitors of various classes of proteases were assayed by adding them to these incubation mixtures and observing their influence on the degradation of 130-kd APP. A sample (500 pL) of a heparin-Sepharose eluate from the plasma of a normal younger adult control subject was desalted into protease assay buffer (175 mmoYL NaCI, 50 mmoYL Tris-HC1, 1 mmolfl. CaCI,, 1 mmoYL MgCI,, 20 pmoVL ZnC1, [pH, 7.4)). Aliquots (containing 65 pg protein) were diluted to 0.80 mdmL with the same buffercontaining protease inhibitor, then incubated for 2 hours at 37°C. Protein was precipitated in each sample by the addition of chlorofordmethanol and analyzed by Western blotting. The final concentrations of inhibitors in the incubation mixtures were ethylene diamine tetraacetic acid (EDTA) (1 mmoYL), diisopropyl fluorophosphate (DFP) (1 mmoYL), aprotinin (10 pg/mL), N-ethylmaleimide (NEM) (1 mmoYL), pepstatin A (10 pg/mL), a1-antichymotrypsin (0.4 mghL), and soybean trypsin inhibitor (SBTI) (1 mg/mL). The effects of AI3Cl(2Opmol/L) and heparin (20 UImL)on APP proteolysis were also analyzed.

Plasma Zinc Assay Zn2+ assays were performed by atomic absorption spectrophotometry according to the method of Davies and colleagues E317.

Results Multiple Molecular Weight F o m s of APP in Plasma Are Recognized by mAb 22C11 on Western Blots To examine forms of APP in plasma, blood was collected from younger adult control subjects and plasma was prepared. The APP in plasma was then partially purified by heparin-Sepharose chromatography and the bound protein analyzed by Western blotting. MAb 22C 11 identified four major immunoreactive bands of APP (130, 110, 65, and 42 kd) in Western blots of human plasma 120). This antibody has been previously used in the identification of APP from platelets 120). The staining of bands by mAb 22C11 on Western blots of heparin-Sepharose eluates could be abolished by preincubating 3 mL. of diluted antibody with 6 pg purified human brain APP (full-length, possessing the intact carboxyl terminus) for 2 hours at RT. The identity of the four major bands labeled by mAb 2 2 C l l as different molecular weight forms of

60 Annals of Neurology Vol 32 No 1 July 1992

Fig 2. Analysis of immunoreactive amyloid protein precursor (APP) in plasma by Western blot analysis. APP immunoreactiue proteins in heparin-sepbarose eluates fmm plasma were analyzed by 8.5% sodium dodeql sugate polyacrylamide gel electrophoresis and Western blotting with monoclonal antibody 22C1I. Lane I: heparin-sepharose eluate of plasma (65 pg protein); lane 2: heparinSepharose eluate immunoprecipitated by antiserum 9013 (raised against full-length human brain APP); lane 3: heparin-Sepharose eluate immunoprecigitated by the prebleed to antiserum 9013; lane 4: heparinSephurose eluate immunoprecipitated by anti-Fd-APP (raised against APP fusion protein). The relative molecular masses of standard protein markers (Rainbmu Standards, Amersham, UK) are shmun on the left. APP immunoreactive bands previously reported (20) of 130, 110 (a doublet), 65, and 42 Rd, are indicated by arrows on the right.

APP was also confirmed by cross-reactivity with antibodies raised against native human brain full-length APP and against synthetic peptides representing APP domains. The immunoprecipitated protein was analyzed by Western blotting with mAb 2 2 C l l . A polyclonal antiserum (90/3)raised against full-length APP purified from human brain, and another polyclonal (anti-Fd-APP) 128) raised against an APPGg5fusion protein, both immunoprecipitated 130-, 110-, and 42-kd bands detected by mAb 2 2 C l l on Western blots. The prebleed from the 90/3 rabbit antiserum did not immunoprecipitate these proteins (Fig 2). An approximately 50-kd broad band detected in Western blots of these immunoprecipitates (Fig 2, lanes 2-4) was the immunoglobulin G ( 1 6 ) heavy chain from the rabbit sera used for immunoprecipitation being weakly recognized by the anti-IgG secondary antibody. Antibody 9013 and its prebleed were also used to directly probe Western blots of heparin-Sepharose eluate (65 pg protein). Unlike its prebleed, 9013 ( I : 10,000 in blocking buffer,

4"C, overnight) detected the 65-kd band recognized by mAb 22C 11 on Western blots of heparin-Sepharose eluate (data not shown). The forms of APP observed in plasma are unlikely to be full-length APP, which is known to be released into the blood by vesiculation of platelet-plasma membranes [20, 211, because the APP signal detected by Western blotting was unaffected by ultracentrifugation of plasma (100,000 g x 1 hr). Two rabbit polyclonal antisera, one raised against a synthetic peptide consisting of the last 43 residues of APP6,, (anti-CT) 120) and the other raised against residues 667 to 676 of APPos (anti-CTII) 112, 201, did not immunoprecipitate any protein from plasma heparin-Sepharose eluates that could be detected by mAb 22C11 on Western blotting, indicating that the APP species detected by mAb 22Cl l on Western blots are unlikely to possess an intact carboxyl terminus or intact PA4 domain. The concentration of 130- and 110-kd APP species in normal plasma was estimated by comparing reflectance levels of these species in Western blots of heparin-Sepharose eluates with levels from a known amount of the same molecular weight soluble APP species purified from normal human brain. Assuming 100% extraction of these APP forms from plasma during chromatography, we estimate the concentration of these species to be 31 to 65 picomolar (range, n = 4) in whole plasma. Abnormal Projile dPlasma APP in AD We surveyed the relative abundance of APP-immunoreactive bands from A D and control plasma samples. The largest apparent changes were an increase in the 130-kd APP band and a decrease in the 42-kd plasma APP band in AD samples compared with samples from all control groups. The control groups consisted of nondemented, age-matched persons (see Fig 1A); normal younger adults (see Fig 1B); and patients with other neurological disease. There was an apparent decrease in the levels of the 65-kd band in patients with AD compared with normal younger adult control subjects, but not compared with nondemented, agematched individuals. There was no consistent difference in the levels of the 110-kd band between patients with AD and control subjects. The total amount of APP immunoreactivity did not differ between patients with AD and control subjects; the ratio of the total immunoreactivity of patients with AD to control subjects was 1.02 0.22 (mean ? standard deviation). Quantitation of these findings by image capture analysis (see Table) showed that the 130-kd band was significantly (t145.831 = 6.34, p < 0.001) increased, and that the 65- and 42-kd bands were significantly (t{109) = - 3 . 9 7 , ~< 0.001; and tIlO91 = - 5.88,p < 0.001, respectively) decreased in patients with AD compared with pooled control subjects (averaged data from other

*

neurological disease control subjects, normal younger adult control subjects, and nondemented, age-matched control subjects). Because readings of the 130-kd APP band gave heterogeneous variances according to an F test, a separate variance estimate was undertaken, which adjusted the degrees of freedom (df) to yield fractional df for analysis. These findings were also supported by the results of a Mann-Whitney U test (a nonparametric analogue of the t-test) {32), which showed the differences in the 130- and 42-kd bands in patients with AD compared with pooled control groups to be significant at the p < 0.0001 level and the difference in the 65-kd band to be significsant at the p < 0.001 level. In patients with AD, there was a 53% increase in the proportion of the 130-kd form, a 20% decrease in the 65-kd form, and a concomitant 32% decrease in the 42-kd form. The immunoreactivity pattern was more evenly distributed between the APP species in the pooled control group. This trend was maintained throughout the comparisons made of the patients with AD with the three control subgroups, where analysis of variance on simple effects indicated significant differences between patients with AD and the three control groups in the 130-kd (F13, 107) = 18.17, p < O.OOl), 65-kd (FE3, 107) = 8.70, p < O.OOl), and 42-kd (Fr3, 107) = 13.09,p < 0.001) bands. Further post hoc analysis confirmed the significance of the difference between the patients with AD and each control group in the 130-kd region (Fig 3A) and between the patients with AD and the younger adult and elderly control groups in the 42-kd region (Fig 3B) according to a Scheffk procedure conducted at the p < 0.05 level of significance. The levels of the 65-kd band were found to be significantly lowered in patients with AD compared with younger adult and neurological disease control groups, but not compared with age-matched elderly control subjects, therefore limiting the clinical usefulness of 65 kd plasma APP levels in discriminating patients with AD. There were no significant differences between the mean reflectance proportions of the three control groups in either the 130- or the 42-kd regions. The proportion of 110-kd APP was notably constant between all diagnostic groups. Further statistical analysis is being reserved for studies of the clinical predictive value of these observations. h e r Molecakzr Weight Plasma APP Species Can Be Generatedfrom the 230-kd Species b~u Serine Protease The existence of a different profile of lower molecular weight forms of APP in AD plasma may be consistent with a defect in the proteolysis of APP. To examine whether such a defect in proteolytic activity is reflected in plasma, we studied the ability of proteases copurifying with APP from heparin-Sepharose to cleave APP.

Bush et ak Abnormal Processing of APP in Alzheimer's Disease 61

t

.

i

.

i

i

!

I

I

0 Y-a ddts

&xmtch?d contrds

n-25

n-46

Neurdogical contrds n - 6

A0

n-34

A Fag 3. Rejectance analysis of immunoblots comparing Alzheimer's disease (AD) and control plasma amyloid protein precursor (APP). The distribution of plasma APP immunoreactivity was analyzed & rejlectance as detailed in the Table. A significant diffence between the A D and the normal age-matched control groups was observed only in the Levels of 130- and 42-kd species of APP. The test of significance was the Scheffe'procedure conducted at the p < 0.05 level. Solid lines indicate the means for each group. Asterisks indicate a significant dgference between a control group and the A D group means. (A)Proportions of 130-kd APP in the A D group and the individual control groups. The mean proportion of 1.30-kdAPP species was significantly (approximately 50%) greater in the A D group compared with each control group. The dotted line indicates a suggested threshold to distinguish the A D group, with 87.0% specifcity and 79.4% sensitivity within these sample groups. Diagnostic pwer was 84.7%. (B) Proportions of 42-kd APP concentrations in the A D group compared with the individual control groups. The mean proportion of the 42-kd A P P species was significantly h e r (range, approximately 20 to 35%) in the A D group compared with the younger adult and the agematched control groups. The dotted line indicates a threshold that distinguishes the A D group, with 80.5% specificity and 73.5% sensitivity within these sample groups. Diagnostic power was 78.4%.

When plasma APP from heparin-Sepharose eluates was incubated at 37°C for 18 hours, the APP was degraded slowly, with loss of the 130-kd band and accentuation of lower bands (110 and 42 kd). This finding indicated that the lower molecular weight forms of APP in plasma could be degradation products of the 130-kd form, and that proteolysis of the 130-kd form in an AD preparation changes the plasma APP profile toward that of control subjects. Proteolysis was accelerated in the presence of Zn*+;the reaction yielded the same products within 2 hours (Fig 4). The Zn2+ concentration used to stimulate proteolysis was 20 pmoY L, within the range of the normal human plasma con62 Annals of Neurology Vol 32 No 1 July 1992

Younger adults

Agctnatckd controis

Neurological controls

AD

n-25

n-46

n-6

n-34

B

centration. There was no clear difference between patients with AD and control subjects in the ability of Zn2+ to stimulate APP breakdown. Identical degradation profiles were found in both groups. Zn2+enhanced APP proteolysis of a younger adult control preparation over 2 hours was completely inhibited by EDTA, heparin, and the serine protease inhibitors aprotinin, DFP, SBTI, and incompletely inhibited by a,-antichymotrypsin. AI,Cl, the cysteine-protease inhibitor N-ethylmaleimide, and the acid-protease inhibitor pepstatin A did not influence the reaction (data not shown). To test the possibility that a zinc deficiency might contribute to decreased APP proteolysis in AD, we assayed total 2nz+ levels in plasma from fasting patients with AD (14.8 ? 2.8 pmoVL; n = 17) and found no significant differences from levels in fasting healthy elderly control subjects (14.8 & 2.8 pmoVL; n = 40). To determine whether APP might be processed in whole plasma, we incubated fresh plasma from a young adult control subject at 37°C over a period of seven days and assayed the plasma APP by heparin-Sepharose chromatography. No significant degradation of the 130-kd APP form was observed over this period, indicating that constitutive processing of APP does not occur in plasma. Discussion This study shows that there is an increase in the 130-kd form and a decrease in the 65- and 42-kd forms of APP in the plasma of moderately to severely demented patients with AD. These observations may form the basis for a peripheral biochemical marker for AD. A comparison of levels of 130-kd plasma APP between these patients with AD and control subjects showed that a threshold of 30% total APP immunoreactiviry

Fig 4. Idntification of an amyloid protein precursor (APP)degrading protease in plasma. APP purified by heparin-sepharose chromatography of plasma fmm patients with Alzheimer's disease (AD) and control subjects was incubated at 3 7°C in saline buffer in the presence or absence of Zd'. Samples (6s pg protein)from each incubation were analyzed by electrophoresis on 8.5 % polyacrylamide gels and Western blotting with monoclonal antibody 22C11. The relative molecular masses of stanakrd protein markers (Rainbow Standzrds, Amersham, UK)are shown on the kfi. APP immunoreactive bands of 130, 110, 65, and 42 Rd are indicated by arrows on the right. Illustrated are samples representative of 6 patients with A D and G normal young adult control subjects.

could distinguish the AD group with a sensitivity of 79.4% and a specificity of 87.0%. A plasma APP profile of less than 20% total APP for the 42-kd species could distinguish the AD group with a sensitivity of 73.5% and a specificity of 80.5%. The overlapping values in the AD and control groups could be explained by incorrect clinical diagnosis of AD, cases of subclinical AD being detected in control subjects, and the possibility that subgroups of AD (e.g., early onset AD) are not represented by the same biochemical lesion detected in this study. The diagnostic power of this test within these sample groups at the thresholds recommended is 84.7% for 130-kd plasma APP and 78.4% for 42-kd APP. Prospective studies are now in progress to determine the predictive value of the plasma APP profile for clinical or pathological outcome.

Our data are at variance with those of Podlisny and colleagues [23}, who reported no qualitative or quantitative differences in a soluble APP band of similar molecular weight (125 kd) in AD plasma. Several reports have now documented changes in cerebrospinal fluid levels of APP in patients with AD, but a consensus has yet to emerge on the direction of these changes 133-351. Palmert and associates 1333 describe alterations in the relative amounts of APP bands of 125, 105, and 25 kd, which could also be attributed to altered processing of APP. The possibility also remains that the altered plasma APP profile seen in patients with AD could reflect a change in APP heparin-binding affinity. Further studies are currently being undertaken to determine the validity of this hypothesis. Our estimates of the concentration of 130- and 110kd plasma APP (approximately 50 picomolar) agree with previous estimates of plasma KPI-containing APP of similar molecular weight [36), suggesting, like others 1231, that KPI-containing APP may be the predominant form in plasma. Therefore, our observations are consistent with an increased production of fully processed 130-kd KPI-containing APP in AD. Several lines of evidence support the possibility that an overproduction of KPI-containing APP or an increase in the ratio of KPI-containing APP to APP,, is associated with PA4 amyloidogenesis. In DS, both APP messenger RNA (mRNA) and protein levels are increased and are associated with the invariable premature onset of AD [5, 12). In sporadic AD, the proportion of 01-containing APP mRNA is increased 15, 37, 381. Increased mRNA and expression of KPI-containing APP released by lymphoblastoid cells in patients with familial AD have also been shown to be accompanied by aberrant intra-PA4 proteolysis [39]. Finally, it has been shown that PA4 immunoreactive deposits develop in transgenic mice overexpressing APP,,I in the brain 1401. The major APP fragments in plasma could be derived from the 130-kd form of APP. The decrease in the 42-kd APP species in A D plasma suggests that inhibition of constitutive proteolysis of 130-kd APP may occur in the disease condition. Abnormal APP processing could result either from an alteration in APP itself or from an alteration in the activity of a protease that constitutively hydrolyzes APP. It is also possible that an increase in KPI-containing APP inhibits its own constitutive catabolism, resulting in a reduction in the amount of the 42-kd APP fragment appearing in AD plasma. The precise mechanism and location of this altered processing, as well as the site of production of plasma APP, remain to be elucidated. The physiological relevance of the proteolytic mechanism copurifying with APP on heparin-Sepharose chromatography is s t d to be determined. Its identity may yield clues to the role of APP in blood and the

Bush et al: Abnormal Processing of APP in Alzheimer's Disease 63

nature of APP processing in general. To the best of our knowledge, no mammalian Zn2+-stimulated serine protease has been previously described; however, the possibility that this activity may represent more than one enzyme or a Zn2+-stimulated cofactor must also be considered. The difference in plasma APP levels in AD could not be attributed to a change in total plasma zinc concentration, because total Zn2+levels in plasma from fasting patients with AD were not significantly different from levels in fasting elderly control subjects. Although plasma APP was seen to be degraded by a protease when heparin-Sepharose eluates were incubated at 37”C, our data indicate that this mechanism is not active in whole plasma. This study was supported by funds from the National Health and Medical Research Council, the Victorian Health Promotion Foundation, and the Aluminium Development Council. Prof Beyreuther is supported by the Deutsche Forschungsgemeinschaft and the Bundesministerium fiir Forschung und Technologie. We thank Dr Michael Berndt, Department of Medicine, Westmead Hospital, Westmead, New South Wales, for his helpful discussions. We also thank Drs A. Wooton, D. Deam, and S. Ratnaike, Department of Clinical Biochemistry, Royal Melbourne Hospital for plasma Zn2+analysis; Dr Kurt Naujoks, Boehringer-Mannheim, for providing the mAb 22C11; Mr Dean McKenzie, Department of Psychological Medicine, Monash Universiry, for statistical advice; Mr Allyn Radford for assistance with image capture analysis and photography; Ms Valcy Malone for coordination of volunteers; and Mr Stuart Portbury for technical assistance.

References 1. Masters CL, Simms G, Weinman NA, et al. Amyloid plaque core protein in Alzheimer disease and Down syndrome. Proc Natl Acad Sci USA 1985;82:4245-4249 2. Glenner GG, Wong CW. Alzheimer’s disease: initial report of the purihcation and characterization of a novel cerebrovascular amyloid protein. Biochem Biophys Res Commun 1984;120: 885-890 3. Kang J, Lemaire H-G, Unterbeck A, et al. The precursor of Alzheimer’s disease amyloid A4 protein resembles a cell-surface receptor. Nature 1987;325:733-736 4. Ponte P, Gonzalez-DeWhitt P, Schilling J, et al. A new A4 amyloid mRNA contains a domain homologous to serine proteinase inhibitors. Nature 1988;331:525-527 5. Tanzi RE, McClatchey AI, Larnperti ED, et al. Protease inhibitor domain encoded by an amyloid protein precursor mRNA associated with Alzheimer‘s disease. Nature 1988;3 11:528-530 6. Golde TE, Esms S, Usiak M, et al. Expression of p amyloid protein precursor mRNAs: recognition of a novel alternatively spliced form and quantitation in Alzheimer’sdisease using PCR. Neuron 1989;4:253-267 7. Kitaguchi N,Takahashi Y,Tokushima Y,et al. Novel precursor of Alzheimer‘s disease amyloid protein shows protease inhibitory activity. Nature 1988;311:530-532 8. Sisodia SS, Koo EH, Beyreuther K, et al. Evidence that p-amyloid protein in Alzheimer’s disease is not derived by normal processing. Science 1990;248:492-495 9. Esch FS, Keim PS, Beattie EC, et al. Cleavage of amyloid P peptide during constitutive processing of its precursor. Science 1990;248:1122-1124 10. Ishiura S, Tsukahara T, Tabira T , et al. Identification of a puta-

64 Annals of Neurology Vol 32 N o 1 July 1992

tive amyloid A4-generating enzyme as a prolyl endopeptidase. FEBS Lett 1990;260:131-134 11. Small DH, Moir RD, Fuller SJ, et al. A protease activity associated with acetylcholinesterase releases the membrane-bound form of the amyloid protein precursor of Alzheimer’s disease. Biochemistry 1991;30:10795-10799 12. Rumble B, Retallack R, Hilbich C, et al. Amyloid A4 protein and its precursor in Down’s syndrome and Alzheimer’s disease. N Engl J Med 1989;320:1446-1452 13. Van Broeckhoven C, Haan J, Bakker E, et al. Amyloid P protein precursor gene and hereditary cerebral hemorrhage with amyloidosis (Dutch). Science 1990;248:1120-1 122 14. Levy E, Carman MD, Fernandez-Madrid IJ, et al. Mutation of the Alzheimer’s disease amyloid gene in hereditary cerebral hemorrhage, Dutch Type. Science 1990;248:1124-1126 15. Goate R, Chartier-Harlin M-C, Mullan M, et al. Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer’s disease. Nature 1991;349:704-706 16. Chartier-Harlin M-C, Crawford F, Houlden H, et al. Early-onset Alzheimer’s disease caused by mutations at codon 717 of the P-amyloid precursor protein gene. Nature 1991;353:844-846 17. Naruse S, Igarashi S, Aoki K, et al. Mis-sense mutation Val-+Ile in exon 17 of amyloid precursor protein gene in Japanese familial Alzheimer’s disease. Lancet 1991;337:978-979 18 Murrell J, Farlow M, Ghem B, Benson MD. A mutation in the amyloid precursor protein associated with hereditary Alzheimer’s disease. Science 1991;254:97-99 19. Lucotte G, Beriche S, David F. Alzheimer’s mutation. Nature 1991;351:530 20. Bush AI, Martins RN, Rumble B, et al. The amyloid precursor protein of Alzheimer’s disease is released by human platelets. J Biol Chem 1990;265:15977-15983 21. Cole GM, Galasko D, Shapiro IP, Saitoh T. Stimulated platelets release P-protein precursor. Biochem Biophys Res Commun 1990;170:288-295 22. Van Nostrand WE, Schmaier AH, FarrowJS, Cunningham DD. Protease Nexin-I1 (amyloid P-protein precursor): a platelet alpha-granule protein. Science 1990;248:745-748 23. Podlisny MB, Mammen AL, Schlossmacher MG, et al. Detection of soluble forms of the P-amyloid precursor protein in human plasma. Biochem Biophys Res Commun 1990;167: 1094- 1101 24. Monning U,Konig G, Prior R, et al. Synthesis and secretion of Alzheimer amyloid PA4 precursor protein by stimulated human peripheral blood leucocytes. FEBS Leu 1990;277:26 1-266 25. McKhann G, Drachman D, Folstein M, et al. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA work group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s disease. Neurology 1984;34:939-944 26. Folstein MF, Folstein SE, McHugh PR. Mini-mental state: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198 27. Smith PK, Krohn RI, Hermanson GT, et al. Measurement of protein using bicinchoninic acid. Anal Biochem 1985;150: 76-85 28. Weidemann A, Konig G, Bunke D, et al. Identification, biogenesis and localization of precursors of Alzheimer‘s disease A4 amyloid protein. Cell 1989;57:115-126 29. Maxwell SE, Delaney HD. Designing experiments and analysing data. Belmont, CA: Wadsworth, 1990:177-179 30. Scheff6 H. The analysis of variance. New York: Wiley, 1959: 66-72 31. Davies IJT, Musa M, Dormandy TL. Measurements of plasma zinc. J Clin Pathol 1968;21:359-365 32. Siege1 S, Castellan NJ. Non-parametric statistics for the behavioural sciences, ed 2. New York: McGraw-Hill, 1988:128-139 33. Palmert MR, Usiak M, Mayeux R, et al. Soluble derivatives of

the P amyloid protein precursor in cerebrospinal fluid: alterations in normal aging and in Alzheimer's disease. Neurology 1990;40:1028-1034 34. Kitaguchi N, Tokushma Y,Oishi K, et al. Determination of amyloid P protein precursors harboring active form of proteinase inhibitor domains in cerebrospinal fluid of Alzheimer's disease patients by crypsin-antibodysandwich ELISA. Biochem Biophys Res Commun 1990;166:1453-1459 35. Prior R, Monning U, Schreiter-Gasser U, et al. Quantitative changes in the amyloid PA4 precursor protein in Alzheimer cerebrospinal fluid. Neurosci Lett 1991;124:69-73 36. Van Nostrand WE, Schmaier AH, Farrow JS, et al. Protease Nexin-2/amyloid P-protein precursor in blood is a plateletspecific protein. Biochem Biophys Res Commun 1331;175: 15-2 1

37. Johnson SA, Rogers J, Finch CE. APP-695 transcript prevalence is selectively reduced during Alzheimer's disease in cortex and hippocampus but not in cerebellum. Neurobiol Aging 1991; 10755-760 38. Neve RL, Rogers J, Higgins GA. The Alzheimer arnyloid precursor-related transcript lacking the @/A4sequence is specifically increased in Alzheimer's disease brain. Neuron 1990;5: 329-338 39. Matsumoto A, Fujiwara Y.Abnormal and deficient processing of p-amyloid precursor protein in familial Alzheimer's disease lymphoblastoid cells. Biochem Biophys Res Commun 1991; 175:361-365 40. Quon D, Wang Y, Catalan0 R, et al. Formation of P-amyloid protein deposits in brains of transgenic mice. N a m e 1991;352: 239-241

Bush et al: Abnormal Processing of APP in Alzheimer's Disease 65