Monoclonal Antibodies to A lzheimer Neurofibrillary Tangles

Monoclonal Antibodies

1. Ident iJcation

of

to A lzheimer Neurofibrillary Tangles Polypeptides

S.-H. YEN, PhD, A. CROWE, and D. W. DICKSON, MD

From the Department of Pathology (Neuropathology), Albert Einstein College of Medicine, Bronx, New York

Ten monoclonal antibodies to Alzheimer neurofibrillary tangles (ANTs) were produced by immunizing mice with a brain homogenate from senile dementia of the Alzheimer type (SDAT). In methanol-fixed isolated neuronal perikarya, six of these antibodies reacted with nearly every ANT, three recognized 70-88% of ANTs, and one bound to less than 30% of ANT. In paraffin sections, three of the antibodies did not bind to tangles that had been fixed in formalin, three stained weakly, and four reacted with tangles in tissues that had been in formalin for more than a decade. Immunoblotting of brain homogenates showed that all but one antibody reacted with proteins from SDAT samples insoluble in SDS and too large to

enter even the 3% polyacrylamide stacking gel. Polypeptides extractable by Tris buffer of molecular weight 58, 66, and 70 kd were detected in both normal and SDAT brains by two antibodies and only in SDAT brain by two other antibodies. One antibody did not show any reaction on the immunoblot. The results demonstrate that the epitopes recognized by these antibodies are not identical and that ANTs contain unique antigenic determinants as well as determinants in common with normal brain. Whether the unique determinants are acquired during tangle development or are essential in tangle formation remains to be investigated. (Am J Pathol 1985, 120: 282-291)

THE TWO MAJOR histopathologic lesions found in Alzheimer's disease and senile dementia of the Alzheimer type (SDAT) are neurofibrillary tangles (ANTs) and neuritic plaques. 1-4 The tangle and some neurites in the plaque contain bundles of paired 10-nm filaments crossing each other at about every 80 nm (PHFs). The PHF appears to be a very stable structure. It is insensitive to non-ionic and ionic detergents, not affected by reagents which break the disulfide or hydrogen bond, and resistant to proteolytic degradation.5'6 A major problem in studying ANTs is the lack of biochemical methods of purifying them to homogeneity, although many contaminants can be removed by SDS extraction. The relationship between ANTs and normal cytoskeletal components has been studied by immunocytochemistry. Monoclonal and polyclonal antibodies raised against neurofilament (NF), vimentin, or microtubuleassociated protein have been shown to bind tangles in tissue sections and isolated preparation.'-" However, a large number of antibodies to these proteins show no cross-reactivity with tangles. Preextraction of tangles with SDS abolishes the antigenic site for one but not other anti-NF antibodies. 1213 Whether NF antigens are an integral part of ANTs remains a subject of further investigation.

In one study, rabbit antiserum to partially purified PHF recognizes several proteins of molecular weight 45,000-70,000 daltons on immunoblots of PHF-enriched samples14; in another study, antiserum to PHFs reacts only with proteins excluded at the top of the polyacrylamide stacking gel. The application of antisera to ANT or partially purified PHF fractions, as in all studies with polyclonal sera, is limited because of the uncertainty of whether they might contain other antibodies unrelated to ANTs. This problem can be overcome with the use of monoclonal antibodies. In this report, we describe the production of such monoclonal antibodies to ANT.

Materials and Methods The immunogens were prepared from autopsy tissue of human brain from a subject with SDAT. Two grams Supported by NIH Grants AG01136 and AG-04145 (S.-H. Y.) and NIH Training Grant NS-07098-05 (D. W. D.). Accepted for publication April 3, 1985. Address reprint requests to Dr. S.-H. Yen, Department of Pathology (F-144), Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461.

282

Vol. 120 * No. 2

of gray matter were minced and immersed in 10 ml of 0.05 M Tris, pH 7.5. The tissue was disrupted by repeated pipetting, and the suspension was centrifuged at 10,000g for 10 minutes. The pellet contained numerous tangles that could be identified by fluorescent microscopy of a thioflavine-S-stained aliquot adfixed to a glass slide.7 One-half of the pellet was fixed with buffered 2% paraformaldehyde for 15 minutes, then washed with phosphate-buffered saline (PBS). The fixed and unfixed tangle pellets were combined and resuspended in Freund's complete adjuvant. Three female BALB/c mice were immunized intraperitoneally with 100 ;A of the tangle preparation. The animals were boosted 4 weeks later. Subsequent boosting was carried out over a period of 3 months. Each animal received five injections. The immunized mice were bled from the tail vein, and the sera were tested for reactivity against isolated perikarya from SDAT. The titer of one mouse reached 1:1000, and it took 5 months for the titer to drop to 1:200. This mouse was boosted once daily for 3 days before the fusion. The mouse was sacrificed, and the spleen cells were mixed with NSO mouse myeloma cells at a 4:1 ratio. Fusion was promoted by incubation of cell suspensions with polyethylene glycol.'5 The resulting hybridomas were distributed into 96-well tissue culture plates at a density of 2.5 x 105 cells/ml and grown in HAT medium.'5 After substantial cell proliferation, the supernatant was collected from the wells and screened for anti-tangle antibody. The positive hybrids were cloned twice and tested again for anti-tangle activity.

Screening of Anti-ANT Antibody The screening method was the same as that described previously.16 Neuronal perikarya were isolated from SDAT brain and smeared on microscopic slides. After drying at room temperature, the slides were fixed with absolute methanol for 5 minutes, washed with PBS, and incubated with hybridoma supernatant. Normal mouse serum served as control. The bound immunoglobulin was detected by rhodamine-conjugated rabbit antimouse immunoglobulin (1:300). The slides were then double-stained with 0.001% thioflavine S in 10% formalin. This reagent, by a yet unidentified mechanism, binds to tangles and emits green fluorescence under ultraviolet illumination.717

Characterization of Anti-ANT Antibody The antibodies that were positive in the screening step (above) were then tested on 1) frozen sections from SDAT, normal human, and rat brains, 2) formalin-fixed paraffin sections from SDAT and normal human brains

ALZHEIMER NEUROFIBRILLARY TANGLE

283

and tissues from human kidney, lymph node, liver,

spleen, heart, intestine, peripheral nerve, thyroid, pancreas, adrenal and muscle, 3) cultured rat cerebellar cells, and 4) HeLa cells. The cerebellar and HeLa cells were cultured on cover glasses by the method described previously."6 The frozen tissue sections or cultured cells were fixed with methanol at - 20 C for 5 minutes before immunofluorescent staining with undiluted monoclonal hybridoma supernatant and rhodamineconjugated rabbit anti-mouse immunoglobulin. The paraffin sections were stained with the peroxidase-antiperoxidase (PAP) method (see following paper). Some sections were treated briefly with trypsin.23

Immunoblot

Gray matter was dissected from normal and SDAT brains, and each sample was minced into small pieces. Some of them were homogenized in 5 volumes of 0.05 M Tris, pH 7.6 (sample A), and others were homogenized in 2% sodium dodecylsulfate (SDS) containing 0.05 M Tris, pH 7.6 (sample B). The homogenate was centrifuged at 10,000g for 10 minutes. The supernatant was assayed for protein content according to the method of Lowry et al."9 The pellet from sample A was dissolved in 2% SDS in 0.05 M Tris, pH 7.6, and then centrifuged. The resulting pellet and the pellet obtained from sample B were washed three times with buffered 2% SDS for removal of the trapped supernatant and then resuspended in 500,1A of 2%7o SDS. Bovine and rat brain microtubule preparations were prepared by the method of Murphy et al,20 and the heat-stable microtubule associated proteins were prepared by the method reported by Herzog and Weber.2' Rat and human brain filament fractions were prepared from spinal cords by the method of Chiu and Norton.22 Each preparation was analyzed by gel electrophoresis on 5-1507o or 10% SDS polyacrylamide gels using the system described by Laemmli.23 The amount of sample loaded on each gel lane was 100 ,g for brain supernatant, 20 WI for brain pellet, 80 ,g for microtubule preparation, and 50 ,g for brain filament fraction. Some gels were stained with Coomassie blue dye; others were transferred to nitrocellulose paper with the use of the Transblot apparatus of Bio-Rad. The transfer was carried out in 2.5 mM Tris, 19.2 mM glycine buffer, pH 8.3, and 20% methanol at room temperature, overnight, with a constant current of 125 mA. The nitrocellulose paper, containing the transferred proteins, was incubated with 5% bovine serum albumin in PBS for 90 minutes and then with undiluted monoclonal antibody for 2 hours at room temperature. After washing with PBS, the paper was incubated with biotinylated horse anti-mouse immunoglobulin, washed with PBS, and

284

AJP * August 1985

YEN ET AL

I

II

i A

I

ii i I

i i

.. ..

*.:

Figure 1 -Double labeling of isolated perikarya by anti-ANT antibodies (a, Ab 705; c, Ab 215) and thioflavine (b and d). (x 250) Binding of thioflavine is detected by green fluorescence, and bound immunoglobulin is detected with rhodamine-conjugated rabbit anti-mouse immunoglobulin (red) (1:300). Figure 2-Immunoperoxidase labeling of SDAT brain sections with anti-ANT monoclonal antibodies. Numerous tangles (a) and neurites in the neuritic plaque (b) are recognized by the antibodies. (a, x255; b, x320) Fine processes in the area rich in tangles are also stained. Preincubation of sections briefly with trypsin enhances the staining of processes (c). (x 255)

Vol. 120 * No. 2

then incubated with avidin-biotin-peroxidase complex (Vector). Three milligrams of diamino benzidine and 2 pl 30% H202 in 10 ml 0.1 M Tris buffer, pH 7.6, was used as the substrate for peroxidase. Several nitrocellulose strips were treated with alkaline phosphatase by the method described by Sternberger et al24 prior to immunostaining (as above). This treatment has been shown to hydrolyze phosphate residues of the transferred proteins, which allows one to assess whether the antigenic determinant contains phosphate residues. Loss of staining implies that a phosphate group was a component of the epitope, whereas staining of proteins not stained before the treatment implies that a phosphate group may have been "blocking" the antigenic site.

Results Ten monoclonal antibodies (Ab) to neurofibrillary tangles were produced. They are all of the immunoglobulin G class. Immunofluorescent and thioflavine fluorescent double-labeling studies of isolated neuronal perikarya from SDAT showed that six of the antibodies recognize almost all of the tangles, but others bind to only a fraction of tangles (Figure 1). The results are summarized in Table 1. Eight of the antibodies recognized at least 8007o of the tangles. One antibody, Ab 215, recognizes no more than 30/o of the tangles. The difference is probably not due to a difference in antibody titer, since dilution of antibodies (1:10-1:100) which recognize nearly 100% of tangles does not change the fraction of tangles stained; only the fluorescent intensity is diminished. The intensity of immunofluorescence of tangle staining by different antibodies is otherwise similar, on the basis of photographs processed

identically.

Immunofluorescence and Immunoperoxidase Staining of ANTs The antigenic determinants of different anti-ANT antibodies respond differently to fixation and tissue processing. While all antibodies stain ANTs in frozen sections of SDAT brain fixed briefly with methanol, some of them do not recognize tangles in paraffin-embedded tissue that had been in formalin for no longer than 24 hours. This group of antibodies included Ab 117, Ab 215, and Ab-636. Ab 175, Ab 322, and Ab 635 react weakly with tangles in paraffin sections. The antigenic determinants of Ab 705 and Ab 39, 64, and 69, on the contrary, are very stable (Figure 2a). They are detected in tissues that have been in formalin for as long as 17 years. The peroxidase reaction products are detected on


285

Table 1-Some Alzheimer Neurofibrillary Tangles in Isolated Neuronal Perikarya Are Not Recognized by Anti-ANT Antibodies

Immunoglobulin

% of tangle stained

39 64 69

Gl Gl G2B

117 175 215 322 635 636 705

G3 Gl G3 G3 G3 G2A

95 94 100 97 85 25 93 80 75 97

Antibody

class

filamentous structures. In addition to the tangles in perikarya, the reaction products are located on neurites in the neuritic plaque (Figure 2b) and many fine processes in the neuropil of areas where tangles are abundant (Figure 2c). The staining of these processes was less prominent without pretreatment of tissue sections with trypsin (Figure 2a). The white matter, with axons rich in neurofilaments, and regions of SDAT brain devoid of tangles are not stained by the antibodies (Figure 3). Although it is difficult to quantitate the number of tangles in any tissue section stained by a given antibody, comparison of adjacent SDAT brain sections stained with different anti-ANT antibodies showed that Ab 39, 69, and 705 stained more tangles than Ab 175 and Ab 635. Absence of ANT Antigen in Normal Brains, Nonneuronal Tissues and Cells To determine whether the ANT antigens are present only in SDAT brain, we used paraffin sections of brain from an aborted fetus, a normal newborn, young subjects aged 27 and 33, and elderly subjects aged 77, 92, and 103 which were free of neurologic disease. The hippocampal sections from elderly subjects had been previously shown by thioflavine staining to contain a few tangles. In brain sections from normal young subjects no peroxidase reaction product is detected (Figure 4). In normal elderly subjects, antibody binding is present only in a few neurons of hippocampus and entorhinal cortex (Figure 5). All nonneuronal human tissues from a 70-year-old neurologically normal woman who died of a myocardial infarction are negative for immuno-

peroxidase staining (Figure 6).

Brain sections obtained from young adult rats do not react with the antibodies. None of our anti-ANT anti-

bodies, except Ab 322, reacted with HeLa cells (Figure 7) or cultured rat cerebellar cells, which contain astrocytes and fibroblasts. The antigens detected by Ab 322 are in clusters located densely around nuclei and thinly

286

YEN ET AL

AJP * August 1985

I Figure 3-White matter fiber tracts are negative with anti-ANT antibodies. (x 96) Figure 4-Brain sections from normal young subject are not stained by the monoclonal anti-ANT antibodies. (x 85) Figure 5-A few hippocampal neurons in normal 103-year-old subject bind to anti-ANT antibodies. (x240) Figure 6-No antibody binding is detected in kidney sections from a normal 70-year-old subject. (x96) Figure 7-Most anti-ANT antibodies do not react with HeLa cells (a). One does (b). (x 700)

at the periphery of the cell. The pattern is different from that of intermediate filaments, microtubules, or microfilaments.

Immunoabsorption The specificity of anti-ANT antibody was examined by absorption studies with the use of various preparations including SDAT brain homogenate, bovine and

rat microtubule preparations, and rat and human spinal cord filament fraction. The microtubule preparation, as reported previously, contains mostly tubulin. The spinal cord filament fractions consists mainly of neurofilament triplet proteins and glial filament proteins. The SDAT brain homogenate contains many more different polypeptides than either the spinal cord filament or microtubule preparation (Figure 8). Incubation of monoclonal supernatant for 1 hour at room

Vol. 120 * No. 2

Figure 8-Gel electrophoretic profile of molecular weight standards (a), 10,000g supernatant of Tris buffer homogenate from SDAT brain (b), SDSsoluble 10,000g supernatant from SDAT brain (being extracted with Tris buffer) (c), SDS-insoluble SDAT brain homogenate (d), human spinal cord filament preparation (e), and human brain microtubule fraction (f). a-d were run on 10% gel and e and f were run on 5-15% gradient gel. Figure 9-Absorption of anti-ANT antibodies with ANT preparation removes the antibodies that recognize tangles. (x 240)

temperature with

1 mg of microtubule preparation or brain filament proteins does not remove the anti-tangle activity, whereas incubation with 10 jig of SDAT brain homogenate with the antibody removes partially the anti-tangle activity. Complete absorption could be achieved with 100 jig of SDAT brain homogenate (Figure 9).


287

Immunoblots The samples used for immunoblots included 1) 10,000g supernatant of Tris buffer homogenates from SDAT brain, 2) 10,000g supernatant of SDS homogenates from normal and SDAT brains, 3) SDS-insoluble material of homogenates from normal and SDAT brains, 4) microtubule fractions prepared from rat and bovine brains, and 5) filament fractions prepared from rat and human spinal cords. The gel electrophoretic pattern of these samples is shown in Figure 8. Most of the proteins found in SDS soluble brain homogenate could be extracted by 0.05 M Tris buffer, pH 7.6. The majority of polypeptides in SDS-insoluble brain homogenates of SDAT or normal brain were excluded at the top of the stacking gel after electrophoresis, whereas several protein bands, staining weakly by Coomassie blue dye, are detected in the separation gel. The results of immunblotting are shown in Figures 10-13. All but one of our anti-ANT antibodies react with SDS-insoluble SDAT brain proteins excluded at the top of the stacking gel and show no reaction with insoluble normal brain proteins. Two antibodies, in addition, recognize three polypeptides of molecular weight 58, 66, and 70 kd in buffer- or SDS-soluble SDAT samples. Two react with soluble protein of similar molecular weights from both normal and SDAT brains. The relative intensity of staining of these bands depend upon which antibody was applied and varies in different SDAT brains. Based on the immunoblots, our anti-ANT antibodies can be classified into five groups. Group 1 recognizes only SDS-insoluble proteins from SDAT excluded at the top of the gel. They are Ab 39, 69, 64, and 635. Group 2 reacts with insoluble proteins from SDAT and sometimes with SDS-soluble proteins from SDAT brains. This group includes Ab 117 and 175. Group 3 stains insoluble proteins from SDAT and SDS- or buffersoluble proteins from normal and SDAT brains. This group of antibodies also reacts weakly with a 70 kd protein in spinal cord filament preparation (Figure 13) and brain microtubule fraction (not shown). This group includes Ab 636 and 322. Ab 705 belongs to Group 4, which recognizes insoluble and soluble proteins from SDAT. The staining pattern, however, is different from that shown by antibodies of group 2. Ab 705 binds to elements distributed diffusely over the nitrocellulose strip giving an appearance of smearing; no bands can be identified clearly. Ab 215 belongs to Group 5, which does not react with either normal or SDAT brain proteins. Incubation of transferred proteins with alkaline phosphatase does not affect the immunostaining of proteins excluded at the top of gel from SDAT nor the soluble

288

YEN ET AL

AJP * August 1985

ba-SDS-insoluble homogenate from SDAT brain. Figure 10-Immunoblots showing the antigenic specificty of monoclonal anti-ANT antibodies. d -1 O,OOOg supernatant from normal brain c-10,000g supernatant from SDAT brain homogenate. SDS-insoluble homogenate from normal brain. homogenate. The antibodies applied to the blots are Ab 39 (a), Ab 69 (b), Ab 64 (c), Ab 635 (d), Ab 117 (e), Ab 175 (f), Ab 636 (g), Ab 322 (h), Ab 705 (i) Ab 215 (j), and molecular weight standard (k) where applicable.

proteins of molecular weight 58 to 68 kd from normal or SDAT brains (Figures 11 and 12). This treatment, however, abolishes or blocks the staining of smears by Ab 705.

Discussion The data presented here show that monoclonal antibodies to neurofibrillary tangles of Alzheimer type could be produced by the immunization of mice with homogenized SDAT brain. Since no detergent or other harsh treatment is involved in preparing the antigen, more ANT antigenic determinants are likely to be preserved for eliciting an antibody response than tangle antigens purified by SDS extraction. The treatment of a portion of the antigen preparation with paraformaldehyde may have contributed to the production of antibodies which recognize antigenic determinants in tissue stored in formalin for over a decade. This type

of antibody will be useful for retrospective studies in which the tissue processing cannot be controlled (see our companion report). The proportion of tangles recognized by our antiANT antibodies is not identical. Whereas most of the anti-ANT antibodies react with more than 80O1o of tangles, Ab 215 binds to less than 3007o of tangles. Our previous studies showed that about 50% of ANTs share an antigenic determinant with vimentin. 6 These results suggest that the compositions of any two ANTs even from the same patient are not necessarily the same. The antibodies that react with all tangles are probably recognizing a stable or mature determinant, whereas Ab 215 may be reacting with a site that is transitional or labile and that may be altered or removed during the process of tangle formation. The presence of some ANT antigenic determinants (or components), however, might be affected by factors such as the local environment, the size of the neuron, or the type of neuron the ANT

Vol. 120

*

No. 2


289

vessels, nor with white matter fiber tracts in either SDAT or normal brains. The staining pattern suggests that the antigenic determinants recognized by our monoclonal antibodies are not likely to be identical to neurofilament proteins. The lack of staining of normal neurons and nonneuronal tissues suggests that the antigenic sites are not likely to be shared by tubulin, a major protein subunit of microtubules. Other supportive evidence is the lack of binding between most of our anti-ANT antibodies and HeLa cells. Although Ab 322 reacts with cultured cells, the staining pattern bears no resemblance to that by anti-tubulin antibody. We do not know which subcellular organelle carries the 322 antigen. Further studies of immunolabeled cells at the electron-microscopic level might provide the answer. An intriguing finding in our immunoperoxidase labeling of tissue sections is the presence of ANT anti-

Figure 11-Preincubation of transferred SDS-insoluble SDAT brain homogenate with alkaline phosphatase does not alter the staining pattern of immunoblots. a-Ab 117. b-Ab 64. c-Ab 69. d-Ab 705. Figure 12-Immunoblotting of Tris-buffer-extracted SDAT brain homogenate that was pretreated with alkaline phosphatase and stained with monoclonal anti-ANT antibodies. a-Ab 64. b-Ab 635. c-Ab 117. d-Ab 175. e-Ab 322. f-Ab 636. g-Ab 705. The smear stained by Ab 705 (seen in Figure 1 Oc) is no longer detected. Figure 13-Ab 322 reacts weakly with a 66 kd protein present in human spinal cord filament preparation (f). Other anti-ANT antibodies tested (Ab 69, Ab 64, Ab 117, Ab 175, and Ab 705) show no reaction with brain filament proteins. h-Molecular weight standards.

is in. The reactions of our anti-ANT antibodies to fixative are not identical, which suggests again that these antibodies do not recognize identical determinants. The minimum number of amino acids that determine an epitope recognized by a monoclonal antibody has been estimated to be 7-10 residues. Thus, a complicated macromolecular structure such as the neurofibrillary tangle may conceivably contain many distinct epitopes. A series of monoclonal antibodies such as ours and those reported by Wang et al25 may allow a finer immunochemical "dissection" of the ANT than has been possible with the use of the conventional antisera. In SDAT brain sections, anti-ANT antibodies bind to tangles in neuronal perikarya and to neurites in the neuritic plaques, but not to amyloid in plaques or blood

gen in many fine processes in the area of SDAT brain that contains tangles. Numerous ultrastructural studies of SDAT brain have shown that PHFs are distributed mostly in cell bodies, in neurites of neuritic plaque, and only occasionally in axons and small processes. The number of the small neuronal processes detected by our anti-ANT antibodies is more than that identified in thioflavine stain or ultrastructural investigations. It is tempting to speculate that the ANT antigens present in the fine processes are not in the filamentous (or polymerized) form. Nevertheless, we have not ruled out the possibility that ultrastructural examination may not be as sensitive as the immunoperoxidase labeling in detecting a low concentration of tangles (number of tangles per square area). The specificities of our anti-ANT antibodies are demonstrated clearly by immunoabsorption with different preparations. SDAT brain homogenates containing 0.1 mg protein have a sufficient amount of ANT antigen to remove anti-ANT antibodies from 100-200 ,u of hybridoma supernatant, but ten times as much protein from a microtubule sample or a spinal cord filament preparation does not. While 80% of the proteins in the microtubule sample are tubulin, and at least 30% of the proteins in spinal cord filament preparations are neurofilament proteins, no more than 1% of the brain homogenates are ANTs, according to our morphologic data (unpublished observation). The results of absorption studies, therefore, are consistent with the interpretation that our anti-ANT antibodies do not recognize a determinant shared by NFs or tubulin. Wang et al25 reported recently that monoclonal antibodies raised against PHFs extracted by SDS did not react with NFs or microtubles. They did not characterize the biochemical property of ANTs nor ANT-related antigen(s). Our immunoblotting studies show that of our 10 monoclonal antibodies 7 react with substances that are

290

YEN ET AL

present only in SDAT brains. Our results are comparable to those of studies using polyclonal antibodies raised against ANT enriched fraction prepared by SDS extraction. 12 Group 1 antibodies react only with insoluble SDAT proteins. The antigenic determinants recognized by these antibodies are more stable in response to tissue processing and prolonged fixation in formalin than those of the other antibodies. These determinants are detected in tangles of other neuropathologic conditions (see our companion report). The molecule(s) that carries these determinants, therefore, might be important for making abnormal neuronal filamentous structures. Group 2 antibodies recognize polypeptides of molecular weight 58, 66, and 70 kd, in addition to those proteins excluded by the gel from SDAT, but not normal brains. Although the 66 kd band is detected readily by immunoblots, it is not obvious on gels stained with Coomassie blue, which indicates that this protein represents only a small portion of the total soluble proteins in SDAT brain. The binding of antibodies to these proteins is detected consistently in preparations from one SDAT brain, but not in preparation from another SDAT brain extracted in a similar way. We do not know whether this variation represents a difference in the extent of the disease, treatment, and/or the patient's terminal condition. Group 3 antibodies react with the same three bands as the previous group. Two major differences, however, become apparent. First they are present in both SDAT and normal brain samples. Second, the 66-kd and 70kd proteins are also present in spinal cord filament preparation and in microtubule samples. This raises the possibility that the determinants recognized by Group 3 antibodies may be shared to some extent with NFs, because the molecular weight of one of the NF triplet proteins is about 70 kd and the microtubule samples could have been contaminated with a small amount of neurofilament. If this is so, then the concentration of these determinants is probably very low, because more 70-kd protein is stained with Coomassie blue than by Group 3 antibodies on immunoblots. The possibility that 66- and 70-kd proteins may be microtubuleassociated proteins was explored by immunoblotting of heat-stable microtubule proteins. Although no binding is detected between the heat-stable proteins and antibodies, we could not rule out the possibility that heat treatment (100 C, 4 minutes) might have affected the determinants. The reaction of Ab 705 (Group 4) on immunoblots is similar to that seen with antisera to the PHF preparation. 12"4 Iqbal et al14 suggested recently that the smearing substances are aggregated PHF proteins. Our data demonstrate that the smear indeed is related to

AJP * August 1985

tangles. We report here for the first time that the determinant in the smearing substance is sensitive to alkaline phosphatase, which suggests that a phosphate residue may be a component of the antigenic site. The insoluble material, excluded at the top of stacking gel, is less sensitive to the enzyme, perhaps because the determinant is not as accessible as those in the smear. The lack of binding between Ab 215 (Group 5) and brain proteins may be due to the high sensitivity of this determinant to SDS or other reagents used in the electrophoresis. Alternatively, it could be due to an insufficient amount of determinants present in the homogenate, because only a small fraction of tangles contains this determinant. The basis for neurofibrillary tangle formation in Alzheimer's disease and senile dementia of the Alzheimer type is unkown. The study of Sajdel-Sulkowska et a126 showed that apparently no new proteins are synthesized by messenger RNAs isolated from SDAT brains. It is conceivable that the degradation of one or several normal neuronal proteins might be defective in the SDAT patients, and subsequent events such as phosphorylation, cross-linking, and other mechanisms are involved in transforming the otherwise soluble normal brain constituents into a stable, abnormal filamentous structure. The three proteins mentioned above might be part of the initial building blocks for PHF. Isolation of tangle precursors or associated proteins and most of all the pure tangle fraction would be essential for understanding the nature of the histologic hallmarks of the Alzheimer's disease and SDAT. Monoclonal antibodies might become useful reagents in the purification of soluble and insoluble tangle components and in determining the role of the synthesis of abnormal proteins in tangle formation.

References 1. Wisniewski HM, Terry RD: Neuropathology of the aging brain, Neurobiology of Aging. Edited by RD Terry, S Gershon. New York, Raven Press, 1976, pp 265-280 2. Oyanagi S: Electron-microscopic observations on the

3.

4. 5.

6.

brains of patients with senile dementia: Conversion of neurofilaments to twisted tubules and interrelations between Alzheimer's neurofibrillary tangles and Pick's bodies. Adv Neurol Sci Jpn 1974, 18:77-88 Shibayama H, Kitoch J: Electron-microscopic structure of the Alzheimer's neurofibrillary changes in case atypical senile dementia. Acta Neuropathol (Berl) 1978, 41: 229-234 Yagishita S, Itoh Y, Amano N: Ultrastructure of neurofibrillary tangles in progressive supranuclear palsy. Acta Neuropathol (Berl) 1979, 48:27-30 Selkoe DJ, Ihara Y, Salazar FJ: Alzheimer's disease insolubility or partially purified paired helical filaments in sodium dodecyl sulfate and urea. Science 1982, 215: 1243-1245 Yen SH, Kress Y: The effect of chemical reagents or proteases on the ultrastructure of paired helical filaments,

Vol. 120 * No. 2

7.

8. 9.

10.

11.

12.

13. 14.

15.

16.

Biological Aspects of Alzheimer's Disease. Edited by R. Katzman. Cold Spring Harbor, Cold Spring Harbor Laboratory, 1983, pp 155-165 Yen SH, Gaskin F, Terry RD: Immunocytochemical studies of neurofibrillary tangles. Am J Pathol 1981, 104:77-89 Grundke-Iqbal I, Johnson AB, Wisniewski HM, Terry RD, Iqbal K: Evidence that Alzheimer neurofibrillary tangles originate from neurotubules. Lancet 1979, 1:578-580 Gambetti P, Autillo-Gambetti L, Shecket G: Immunological and silver staining characteristics of neurofibrillary tangles of Alzheimer type. Exp Brain Res Suppl 1982, 5:15-19 Bignami A, Selkoe DJ, Dahl D: Amyloid-like (congophilic) neurofibrillary tangles do not react with neurofilament antisera in Alzheimer's cerebral cortex. Acta Neuropathol (Berl) 1984, 64:243-250 Anderton BH, Breinburg D, Downes MJ, Green PJ, Tomlinson BE, Ulrich J, Wood JN, Kahn J: Monoclonal antibodies show that neurofibrillary tangles and neurofilaments share antigenic determinants. Nature 1982, 298: 84-86 Ihara Y, Abraham C, Selkoe, DJ: Antibodies to paired helical filaments in Alzheimer's disease do not recognize normal brain proteins. Nature 1983, 304:7272-7230 Johnson AB: On the relationship of neurofilaments and Alzheimer's neurofibrillary tangle (Abstr). J Neuropathol Exp Neurol 1984, 43:347 Grundke-Iqbal I, -Iqbal K, Tung Y-C and Wisniewski HM: Alzheimer paired helical filaments: Immunochemical identification of polypeptides. Acta Neuropathol (Berl) 1984, 62:259-267 Kwang S-P, Yelton DE, Scharff M: Production of monoclonal antibodies, Genetic Engineering. Edited by JK Setlow and A Hollaender. New York, Plenum Press, 1980, pp 31-46 Yen S-H, Gaskin F, Fu SM: Neurofibrillary tangles in senile dementia of the Alzheimer type share an antigenic


-17. 18. 19.

20. 21.

22.

23. 24.

25.

26.

291

determinant with intermediate filaments of the vimentin class. Am J Pathol 1983, 113:373-381 Schwarz P: Amyloid degeneration and tuberculosis in the aged. Gerontologia 1972, 18:321-362 Sternberger LA: Immunocytochemistry. 2nd edition. New York, John Wiley & Sons, 1979 Lowry OH, Rosebrough NJ, Farr AL, Randall RJ: Protein measurement with the folin phenol reagent. J Biol Chem 1951, 193:265-275 Murphy B: Cytoskeletal proteins, isolation and characterization, Methods of Cell Biology. Vol 24A. Edited by L Wilson. New York, Academic Press, 1982, pp 31-49 Herzog W, Weber K: Fractionation of brain microtubuleassociated protein isolation of two different proteins which stimulates tubulin polymerization in vitro. Eur J Biochem 1978, 92:1-8 Chiu FC, Norton WT, Fields KL: The cytoskeleton of primary astrocytes in culture contains actin, glial fibrillary acidic protein and the fibroblast-type filament protein, vimentin. J Neurochem 1981, 37:147-155 Laemmli UK: Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970, 227:680-685 Sternberger LA, Sternberger NH: Monoclonal antibodies distinguish phosphorylated and non phosphorylated forms of neurofilament in situ. Proc Natl Acad Sci USA 1983, 80:6126-6130 Wang GP, Grundke-Iqbal I, Kascsak RJ, Iqbal K, Wisniewski HM: Alzheimer neurofibrillary tangles: Monoclonal antibodies to inherent antigen(s). Acta Neuropathol (Berl) 1984, 62:268-275 Sajdel-Sulkowska EM, Coughlin JF, Staton DM, Marotta, CA: In vitro protein synthesis by messenger RNA from Alzhemier's disease brain, Banbury Report. Vol 15, Biological Aspects of Alzheimer's Disease. Edited by R Katzman. Cold Spring Harbor, Cold Spring Harbor Laboratory, 1983, pp 193-200