Elevation of microtubule-associated protein tau in the cerebrospinal fluid of patients with Alzheimer's disease

The diagnosis of Alzheimer's disease (AD) typically involves lengthy neurologic and psychological testing leading to a diagnosis by exclusion. AD is frequently confused with other distinct neurologic conditions that also result in dementia. An inclusionary biochemical marker for the diagnosis of AD would thus be extremely useful in the clinical evaluation of demented patients. The presence of pathologic lesions diagnostic for the disease begs the question of whether the components of these lesions can form the basis for an antemortem test for AD.

AD is a neurodegenerative disease characterized by senile plaques and neurofibrillary tangles (NFTs) in the brain. NFTs occur intraneuronally and are composed of paired helical filaments, a feature also shared with neuropil threads and dystrophic neurites. Immunocytochemical and biochemical studies ^[1-6] show that the major component of paired helical filaments is the microtubule-associated protein tau in a highly phosphorylated state. These findings in part explain the observation that the antibody Alz-50 reacts specifically with tangles and the protein A68 on immunoblots of AD brain tissue ^[7]. This protein is clearly elevated in AD brain tissue, ^[8] but its presence in CSF is difficult to establish. We now know that A68 is indistinguishable from highly phosphorylated forms of tau ^[9]. In addition, tau levels are elevated in AD brain relative to control brain samples ^[10]. Using an extremely sensitive ELISA, low levels of tau immunoreactivity in CSF are detected in a number of patients with neurologic diseases, including a majority of AD patients ^[11]. In the present study, we use tau-specific monoclonal antibodies to confirm the presence of tau in CSF and to define its utility in the diagnosis of AD.

Methods. Tau reference standards. Human brain and recombinant sources of purified tau were used for characterization of the assay and antibodies. Human tau, purified from AD brain tissue, was a generous gift of Dr. Andre Van de Voorde (Innogenetics, Ghent, Belgium). Recombinant human tau was produced using the previously described baculovirus vector containing the pVL941-tau-4-repeat isoform ^[12]. High levels of tau were expressed and purified from both Sf9 and High Five insect cells (Invitrogen). Maximally expressing cell cultures were harvested, washed once in phosphate-buffered saline (PBS), and chilled on ice. The cells were then sonically disrupted in 0.1 M 2-(n-morpholino)ethane sulfonic acid (MES) pH 6.5, 1 mM ethylene glycol tetra-acetic acid (EGTA), 18 micromolars EDTA, 0.5 mM MgCl₂, 5 micrograms/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). Cell debris was removed by low-speed centrifugation and the supernatant adjusted to 0.75 M NaCl, 2% beta-mercaptoethanol. The samples were boiled 10 minutes in capped tubes, cooled in ice, and clarified by centrifugation at 100,000 x g for 30 minutes. The supernatants were than adjusted to 2.5% perchloric acid and spun for 15 minutes at 13,000 x g. The pellets were subjected to a second cycle of boiling/acid precipitation, and the pooled supernatants were dialyzed against 100 mM KH₂ PO₄ pH 6.9, 2 mM EDTA, 2 mM EGTA, 2 mM beta-mercaptoethanol, and 0.3 mM PMSF.

The recombinant tau was judged to be at least 85% pure by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) stained with Coomassie Blue and was used without further purification. The concentrations of all tau standards were estimated by amino acid analysis. To dephosphorylate tau, an aliquot was dialyzed into 20 mM TRIS-HCl pH 8.6, 2 mM MgCl₂, 1 mM dithiothreitol (DTT), and 10 micromolars ZnCl₂ buffer. To one-half of the sample, alkaline phosphatase (Boehringer Mannheim), 0.1 U/micrograms tau, was added; the other one-half was similarly diluted with buffer alone and the two samples were incubated for 5 hours at 37 degrees C.

Preparation of monoclonal antibodies against tau. Monoclonal antibodies were prepared according to a modification of the method of Kohler and Milstein ^[13]. Tau used in all injections and screening assays was purified from Sf9 cells infected with the tau-containing baculovirus construct. Six-week-old A/J mice were injected with 100 micrograms of purified tau at 2-week intervals. Tau was emulsified in complete Freund's adjuvant for the first immunization and in incomplete Freund's adjuvant for all subsequent immunizations. Serum samples were taken 3 days after the third injection to assess the titer of these animals. The mouse with the highest titer was injected IV with 100 micrograms of tau in 500 microliters of PBS 2 weeks after receiving its third injection. The myeloma fusion occurred 3 days later, using SP2/0 as the fusion partner. Antibodies 16G7 and 8C11 were obtained from this fusion while antibodies 16B5 and 16C5 were isolated from a subsequent fusion.

Supernatants from wells containing hybridoma cells were screened for their ability to precipitate Iodine-125-labeled tau. Tau was radio-iodinated using immobilized glucose oxidase and lactoperoxidase according to the manufacturer's instructions (Bio Rad). Briefly, 10 micrograms of purified recombinant tau was radiolabeled with 1 mCi of Na Iodine-125 to a specific activity of 20 microcuries/micrograms protein. 16G7, 8C11, 16B5, and 16C5 were identified as the four highest-affinity monoclonal antibodies specific to tau and were cloned by limiting dilution. The isotypes on all four monoclonal antibodies specific to tau were determined to be gamma 1 kappa.

Immunoblot characterization of tau antibodies. Approximately one-half gram of superior frontal gyrus of both an Alzheimer-diseased brain (from an 81-year-old woman, 3-hour postmortem interval) and of a control brain (from a 73-year-old man, 2-hour postmortem interval) were separately homogenized in 10 volumes of SDS-PAGE sample buffer containing reducing agent, using a Teflon/glass tissue homogenizer. Brain samples were a generous gift from Dr. Joseph Rogers, Sun Health Research Institute, Sun City, AZ. After boiling and centrifugation, the samples were diluted 1:4 with SDS-PAGE sample buffer, and 200 microliters was loaded into 6-cm-wide wells of 10% SDS-polyacrylamide gels (Novex). After electrophoresis, the proteins were transferred to Pro-Blot membranes at 40 V overnight. The membranes were then cut lengthwise into strips and blocked with 3% bovine serum albumin (BSA) in TRIS-buffered saline (50 mM TRIS HCl, pH 8.0, 150 mM NaCl, 0.05% Tween-20; TTBS) at 37 degrees C for 3 hours. One-half of the strips were incubated overnight at 37 degrees C in 100 mM TRIS pH 8.6, 1 mM MgCl₂, 50 micrograms/ml leupeptin, 0.1 mM PMSF, and 50 U/ml of alkaline phosphatase (Boehringer Mannheim, molecular biology grade). The remaining strips were incubated under the same conditions minus the alkaline phosphatase.

Strips were then incubated in 5% nonfat dry milk/TTBS for 1 hour at 37 degrees C and subsequently probed with the various monoclonal antibodies for 1 hour at a concentration of 0.5 micrograms/ml (except for tau-1, which was used at 1 micrograms/ml). Immunoreactive proteins were visualized using the ECL system (Amersham).

Characterization of tau immunoreactivity in CSF. Antibody 16B5 was coupled at a concentration of 5 mg/ml to Actigel resin (Sterogene) and then washed extensively. A 6-ml pool, derived from three CSF samples shown by ELISA to have elevated tau levels, was incubated with 30 microliters of the 16B5 antibody resin and gently mixed overnight at 4 degrees C. The resin was then washed twice (20 mM TRIS-HCl pH 8.0, 0.5 M NaCl, 0.1% NP-40, 5 mM EDTA). To the resin, 30 microliters of 2 x SDS-PAGE sample buffer was added, the sample was boiled and briefly centrifuged, and the supernatant processed with electrophoresis in an 8% SDS polyacrylamide gel. After transfer and blocking with 5% nonfat dry milk, the blot was probed with antibody 16C5 at a concentration of 0.25 micrograms/ml. Standards of recombinant tau (1 to 30 ng) were analyzed on the same gel.

ELISA of tau. The anti-tau monoclonal antibody 16G7 was suspended at 5 micrograms/ml in 100 mM phosphate buffer, pH 8.5, and 100 microliters/well was coated into microtiter plates (Dynatec Microlite 2). The coating was carried out overnight at room temperature. The solution was then aspirated and the plates blocked with 0.25% casein (w/v) in PBS. The anti-tau antibody 16B5 was biotinylated with the N-hydroxysuccinimide ester of biotin following the manufacturer's instructions (Pierce). Samples of either 50 microliters CSF or calibrators (50 microliters of 3 to 1,000 pg/ml human tau) were combined with 50 microliters of the biotinylated anti-tau antibody (0.75 micrograms/ml in PBS-casein, 0.05% Tween 20) into the 16G7-coated wells and incubated overnight at room temperature with constant shaking. The solution was then aspirated and plates washed three times in TTBS. Streptavidin alkaline phosphatase (Boehringer Mannheim) was diluted 1:1,000 in PBS-casein and 0.05% Tween 20, and 100 microliters was added to each well. After incubation for 1 hour at room temperature, the fluid was aspirated and wells washed three times. The chemiluminescent reagent disodium 3-(4-methoxyspiro (1, 2-dioxetane-3,2 sup '-tricyclo (3.3.1.1 ^[3,7]) decan)-4-yl) phenyl phosphate (AMPPD, Tropix) and an enhancer, Emerald green (Tropix), were diluted 1:1,000 and 1:100 respectively in 1 M diethanolamine buffer containing 1 mM MgCl₂, 0.02% NaN₃ pH 10. Then 100 microliters was added per well and the plates were read after 30 minutes in a Dynatech ML 1000 chemiluminometer. The data reported here used human tau isolated from brain as the calibrator.

CSF samples. CSF was taken from 181 patients from either Harbor-UCLA Medical Center (Los Angeles, CA) or Segovia General Hospital, Neurology Department (Segovia, Spain). CSF was obtained by lumbar puncture in accordance with the ethical standards of the responsible committee on human experimentation and the Helsinki Declaration of 1975, as revised in 1983. Seventy-one of these patients were diagnosed as having probable AD following the NINCDS-ADRDA criteria. The remaining 110 individuals were diagnosed as having various conditions and were divided into three categories as follows: (1) controls--this group included individuals with diagnosed migraine headaches (n = 6), vascular headaches (n = 7), depression (n = 2), and acute cranial neuropathy (n = 11); (2) patients with non-AD dementia--multi-infarct dementia (n = 8), multiple system atrophy (n = 4), ischemia (n = 1), frontal lobe degeneration (n = 5), Creutzfeldt-Jakob disease (n = 1), and other mixed dementias (n = 6); (3) neurologic controls without dementia--ALS (n = 20), stroke (n = 16), lacunar infarct (n = 1), multiple sclerosis (n = 14), spinal tumor (n = 1), neurosyphilis (n = 1), pneumonia (n = 1), polyneuropathy (n = 3), leukosis (n = 1), and chronic meningitis (n = 1). The CSF samples were stored at either -40 degrees C or -70 degrees C until assayed. The larger pool of CSF used for some of the depletion studies described in this report was obtained from Universal Reagents. All statistical analyses employed one-way ANOVA and the software program NStat (Graphpad Software Inc., San Diego, CA).

Results. Tau immunoassay and characterization in CSF. Tau purified from brains of patients with AD ^[14] and recombinant tau obtained as described in the Methods section were quantitated by amino acid analysis and compared in this assay. Both human and recombinant tau are recognized with very similar affinities in the low nanomolar range by these two antibodies (data not shown). The detection limit of both forms of tau is approximately 25 pg/ml. To assess whether the phosphorylation state of tau affected the reactivity seen in the ELISA, several control experiments were conducted. The ratio of immunoreactivities of recombinant tau and alkaline phosphatase-treated tau were compared. Tau produced by the baculovirus expression system in Sf9 cells has been shown to be highly phosphorylated ^[15]. In agreement with this, we observed a large increase in gel mobility after alkaline phosphatase treatment (data not shown). Nonetheless, no significant difference was observed in the tau immunoassay after alkaline phosphatase treatment of either tau source (data not shown). These data suggest that phosphorylation state is not a factor in the measurement of tau in the immunoassay.

Further evidence of the specificity and lack of phosphorylation state sensitivity of these tau antibodies is shown in figure 1. As can be seen when comparing antibody reactivity with brain homogenates containing tau, no difference is detected by dephosphorylation of the blot strips, in contrast to the increase in immunoreactivity seen with tau-1 (lane 10 of figure 1), an antibody known to be dependent on a dephosphorylated epitope ^[16]. The lack of reactivity with other microtubule-associated proteins is also evident by the absence of bands in regions in the samples that represent other molecular weights.

The specificity of the tau ELISA is further supported by immunoblot analysis of the material immunoaffinity-captured from CSF. The CSF sample pool used in this experiment contained an estimated 5 ng of tau. The resultant immunoblot of the captured material has an intensity and gel mobility consistent with this ELISA prediction (figure 2).

Analysis of CSF tau levels. CSF obtained by lumbar puncture from 181 patients, 71 diagnosed as having probable AD and the remaining 110 as having various other conditions, was assayed for tau immunoreactivity (figure 3). Individuals were divided into four groups depending upon their diagnosis. Four broad groups were chosen: patients with AD, controls, patients with non-AD dementia, and control patients with other neurologic disorders. Control patients were defined as those individuals who were cognitively normal and without evidence of a metabolic, inflammatory, or ischemic disorder (eg, migraine headache). The non-AD dementia group was composed of demented patients with a diagnosis other than AD. Finally, the neurologic control group was represented by a large number of assorted disorders not defined by the first three groups. The average CSF tau level in the AD group was essentially twice those of both the control and neurologic control groups (figure 3). In fact, figure 3 shows that while a significant number of AD patients had levels of tau exceeding 400 pg/ml (n = 28), no control patients did, and only a single patient with non-AD dementia did. Although the average levels of CSF tau were markedly elevated in AD, a significant fraction of AD patients (n = 25) did have CSF tau levels below 300 pg/ml (figure 3). With regard to the few non-AD patients with high CSF tau levels, two of the patients with high false-positive levels (>400 pg/ml) in the demented group were diagnosed as having frontal lobe dementia. Two other patients in this group with somewhat elevated CSF tau levels were diagnosed as having multi-infarct dementia (380 and 360 pg/ml). In the neurologic control group patients with high false-positive levels were diagnosed as having ALS (780, 350, and 500 pg/ml), stroke (408 and 340 pg/ml), and neuropathy (562 pg/ml). These occurrences were somewhat unusual because most patients with ALS, stroke, or neuropathy tested low for CSF tau (figure 3). Comparing the AD patient group with each of the other groups individually by one-way ANOVA, the elevated tau levels were highly statistically significant (p < 0.001 for all comparisons).

No significant correlation was observed (r² < 0.2, results not shown) between tau and score on the Mini-Mental State Examination (MMSE) in AD patients. This suggests elevation of tau in CSF in AD may occur early and persist in the course of the disease. Correlation of tau levels with age was also examined for all groups and not found to be significant (r² < 0.06 in all groups, results not shown). During the course of this study, autopsy studies were performed on two of the AD patients, and their brain sections were examined for the presence of plaques and tangles using histochemical techniques. Both brains showed high numbers of NFTs (data not shown), and both antemortem CSF specimens had elevated tau levels (1,190 pg/ml and 500 pg/ml).

Discussion. The present study strongly suggests that CSF tau levels are elevated in patients with probable AD relative to other patient populations. The critical non-AD dementia groups analyzed in this study were age-matched relative to the AD group yet nevertheless demonstrated a clearly lower level of CSF tau. While the two other groups studied in this report were somewhat younger, age appeared to have no effect on CSF tau. Although these two groups are worthy of comparison, in a clinical setting these patients would not be part of the group for which a diagnosis to exclude AD is sought. There were elevated levels of tau in several patients with early stages of AD (MMSE score >= 25), a period where the diagnosis is most difficult. Since high levels of tau in CSF were present in the two patients who had abundant NFTs at autopsy, there might also be a correlation between the levels of tau released in CSF and the extent of NFT pathology. We do not have a clear explanation for the subset of AD patients who had low levels of tau in CSF, but these patients potentially might have low numbers of tangles. Terry et al ^[17] showed that in 60 patients over age 74 with diagnosed AD, 30% did not have abnormally elevated numbers of neocortical tangles nor dystrophic neurites thought to contain tau. This research group ^[18] has recently advanced these findings with the observation that most, but not all, of the patients with low numbers of NFTs have the Lewy body variant type. Follow-up autopsy studies will determine whether the demented patients with low tau CSF values in the present work have pathology that corresponds to the Lewy body variant. The converse issue of the few non-AD demented patients with high CSF tau levels will also require autopsy studies to confirm the absence of AD pathology in those patients.

It is presently unclear whether the elevation of CSF tau is a result of dying neurons, dystrophic neurites, or NFTs. One or more of these components might contribute to elevated tau in CSF. It is unlikely that other sources, such as choroid plexus or peripheral cells, contributed to our measurements since the tau that we detected is the molecular size of the neuronal tau form. Tau in CSF (figure 2) has much less heterogeneity than that detected with the same antibodies against brain homogenates (figure 1). We do not know if this is due to a preferential release of certain transcriptional or post-translational tau forms. The slightly faster migration of CSF tau compared with the recombinant tau standard is consistent with either less extensive phosphorylation or shorter splice variants being predominant in CSF.

Our results are consistent with, but different from, those reported by Vandermeeren et al ^[11] on a smaller group of AD patients. They also observed an increase in tau levels in AD but found an increased number of false-positive values in various other disease conditions. Importantly, the vast majority of their false positives were in younger patients (less than 40 years old) with inflammatory conditions, a patient population easily distinguishable from suspected AD cases. The range of tau in CSF is also significantly higher in our study; this appears to be due to differences in the tau standards used in the two assays (Andre Van de Voorde, PhD, personal communication). Variation would be expected between the CSF tau values in these studies due to differences in the standards used.

The present study is encouraging and suggests that the overall increase of tau in CSF of AD patients might result in a reliable marker to aid in the diagnosis of AD. Further studies will be required to assess fully the clinical utility of CSF tau measurements and what relation, if any, exists between CSF tau levels and the tau-related pathology of the brain.

Acknowledgments

The authors wish to thank Paula Southwick and Robert Wolfert for careful reading of the manuscript. Acknowledgment is also given to Ruth Motter for coordination of the clinical samples, and to Kristen Philipkoski for careful preparation of the manuscript.