Abbreviations used : AD, Alzheimer's disease ; AF150(S), 1-methylpiperidine-4-spiro-(2'-methylthiazoline) ; apoE, apolipoprotein E ; mAb, monoclonal antibody.
Address correspondence and reprint requests to Dr. D. M. Michaelson at Department of Neurobiochemistry, George S. Wise Faculty of Life Sciences, Tel Aviv University, Ramat Aviv 69978, Israel.
Abstract : Apolipoprotein E (apoE)-deficient mice have memory deficits that are associated with synaptic loss of basal forebrain cholinergic projections and with hyperphosphorylation of distinct epitopes of the microtubuleassociated protein tau. Furthermore, treatment of apoEdeficient mice with the M1 selective agonist 1-methylpiperidine-4-spiro-(2'-methylthiazoline) [AF 150(S)] abolishes their memory deficits and results in recovery of their brain cholinergic markers. In the present study, we used a panel of anti-tau monoclonal antibodies to further map the tau epitopes that are hyperphosphorylated in apoE-deficient mice and examined the effects of prolonged treatment with AF 150(S). This revealed that tau of apoE-deficient mice contains a distinct, hyperphosphorylated “hot spot” domain which is localized N-terminally to the microtubule binding domain of tau, and that AF150(S) has an epitope-specific tau dephosphorylating effect whose magnitude is affected by apoE deficiency. Accordingly, epitopes which reside in the hyperphosphorylated “hot spot” are dephosphorylated by AF 150(S) in apoE-deficient mice but are almost unaffected in the controls, whereas epitopes which flank this tau domain are dephosphorylated by AF150(S) in both mice groups. In contrast, epitopes located at the N and C terminals of tau are unaffected by AF150(S) in both groups of mice. These findings suggest that apoE deficiency results in hyperphosphorylation of a distinct tau domain whose excess phosphorylation can be reduced by muscarinic treatment.
Mass spectroscopy analysis and the development of specific anti-tau monoclonal antibodies (mAbs) led to the identification and mapping of ~20 distinct Ser and Thr residues that are hyperphosphorylated in AD tau and many of which reside near Pro residues. In vitro studies revealed that tau can be phosphorylated by proline-directed kinases and by second messenger-dependent kinases and that it can be dephosphorylated by several protein phosphatases (for review, see Billingsley and Kincaid, 1997). Complementary in vivo studies revealed that inhibition of distinct protein kinases results in tau dephosphorylation (Munoz-Montano et al., 1997) and that inhibition of protein phosphatase activities results in tau hyperphosphorylation (Arendt et al., 1995). The in vivo mechanisms that regulate the activities of distinct brain protein kinases and phosphatases and thus determine the pattern and extent of tau phosphorylation in vivo are not, however, fully understood.
Recent tissue culture experiments suggest that the level of tau phosphorylation can also be controlled by cell surface neurotransmitter receptors. For example, it has been shown that muscarinic stimulation of PCl2 cells transfected with the gene for the M1 muscarinic receptor results in time- and dose-dependent tau dephosphorylation that can be blocked by the muscarinic antagonist atropine (Sadot et al., 1996). This inverse relationship between muscarinic activation and tau phosphorylation suggests a link between the cholinergic signal transduction system and the neuronal cytoskeleton (Sadot et al., 1996). This provides a mechanism by which cholinergic hypofunction, such as that observed in AD (Francis et al., 1994), can affect the homeostasis of the cytoskeleton and thereby enhance neurodegeneration.
We have previously shown that apolipoprotein E (apoE)-deficient mice, whose apoE gene has been knocked out, have memory deficits that are associated with specific impairments in forebrain cholinergic projections (Gordon et al., 1995 ; Chapman and Michaelson, 1998) and that these impairments can be reversed by chronic treatment with the M1 muscarinic agonist 1-methylpiperidine-4-spiro-(2'-methylthiazoline) [AF150(S)] (Fisher et al., 1998). Furthermore, it was shown that tau of apoE-deficient mice contains hyperphosphorylated epitopes, e.g., the epitope recognized by mAb AT8 (Genis et al., 1995). These findings provide a unique system for studying the relationship, in vivo, between cholinergic hypofunction and tau phosphorylation and the extent to which it can be modulated pharmacologically.
In the present study we investigated the effects of chronic treatment with the muscarinic agonist AF150(S) on the levels of tau phosphorylation of apoE-deficient mice and the extent to which they correlate with the previously published, cognitive therapeutic effects of AF150(S). This was performed by immunoblot assays using a panel of mAbs that are directed against distinct phosphorylated and nonphosphorylated tau epitopes and by comparing their immunoreactivities toward tau of apoE-deficient and control mice before and following treatment with AF150(S).
MATERIALS AND METHODS
Animal and tissue preparation
Fourteen-week-old male apoE-deficient (knockout) mice (weighing 20-30 g) and age-matched male control offspring derived from the same parent line (C57BL/6J) were kindly provided by Dr. J. L. Breslow (Plump et al., 1992). Two sets of mice were used. The first set contained six control and six apoE-deficient mice and was used only for comparison of the levels of tau phosphorylation of the two mouse strains. The second set of animals contained eight mice of each strain and was used for examination of the effects thereon of AF150(S). Half of the mice of each strain (n = 4) were treated orally with AF150(S) dissolved in 50 mM phosphate-buffered saline (Brandeis et al., 1995), whereas the other half was sham-treated with phosphate-buffered saline. The drug (0.5 mg/kg) was administered orally every day for 3 weeks, after which the mice were killed, and their brains were rapidly frozen in a mixture of hexane and dry ice until used. The brains were thawed on ice and homogenized (1 : 1 wt/vol) at 4°C in 100 mM 3-(N-morpholino)ethanesulfate (MES) buffer (pH 6.4), which contained 0.5 mM MgCl2, 0.1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 5 μ/ml leupeptin, 5 μg/ml pepstatin, 200 μM phenylmethylsulfonyl fluoride, and 1 μM okadaic acid. The homogenates were then centrifuged (10 min X 1,000 g at 4°C), and the supernatants thus obtained were ultracentrifuged (30 min × 100,000 g at 4°C). The resulting supernatants were collected, boiled for 10 min in 30% (vol/vol) sodium dodecyl sulfate sample buffer (100 mM Tris buffer containing 5% sodium dodecyl sulfate, 10 mM dithiothreitol, and 15% glycerol), and stored in aliquots at -70°C.
Table 1. Anti-tau antibodiesThe locations of epitopes on the long human tau isoform (441 amino acids) and on the long rat tau isoform (432 amino acids) that are recognized by the anti-tau antibodies used in this study are given. The antibodies are identified by their name followed by an indication of whether or not they bind to phosphorylation-independent (Pi) or -dependent epitopes. P+ corresponds to antibodies that bind to phosphorylated epitopes, whereas P- corresponds to antibodies that bind to epitopes in their dephosphorylated state. The positions of the tau epitopes of the mAbs were obtained from the indicated references. Epitopes that have been fully mapped are denoted by S or T for Ser and Thr, followed by their position along tau, whereas for the other epitopes the tau domain in which they reside is indicated. ND, not determined.
Antisera 133 and 134 were kindly provided by Dr. M. Goedert. MAb T46 was from Zymed Laboratories, mAb P262-7 was kindly provided by Dr. F. P. Zemlan, mAb RT97 was from Boehringer, and the SMI mAbs were from Sternberger Monoclonals. All of these reagents were used at a dilution of 1 : 1,000. The mAbs AT8 and AT270 were from Innogenetics and were used at dilutions of, respectively, 1 : 40 and 1 : 200, whereas mAb PHF1, which was kindly provided by Dr. P. Davies, was used at a 1 : 2 dilution.
The amount of protein loaded onto the gels was 40 μg per lane for mAbs AT8 and AT270 and 20 μg per lane for all the other mAbs. Twelve samples were routinely electrophoresed per gel and were reacted, following blotting, with the same mAb. In the case of the first set of mice, this corresponded to six apoE-deficient and six control mice. In the second set of mice, which was used for examination of the effects of AF150(S), this corresponded to six apoE-deficient and six control mice, half of which were treated with AF150(S). This design enabled a quantitative comparison, for each of the mAbs, of the tau immunoreactivities of all the different mouse groups that were tested in each experiment. As the number of mice of each of the groups in the AF150(S) experiment was four, qualitative analysis of the mice was first performed using two gels. Three representative mice from each group were then used for quantitative analysis as outlined above.
In all experiments the blotted samples were first reacted with the anti-nonphosphorylated tau mAb T46, after which the intensities of the resulting tau immunoblot bands were determined using the BIS-202 Bio-Imaging system and the PC gel reader program TINA. In accordance with our previous findings (Genis et al., 1995), this revealed that on average the two mouse groups contained similar levels of T46 immunoreactivity. Furthermore, treatment with AF150(S) also had no effect on the average T46 immunoreactivity of the two mouse groups.
However, to minimize experimental scatter when the tau immunoreactivities of the different mouse groups toward other anti-tau mAbs were compared, the immunoblots were further calibrated. Accordingly, the amount of material from each individual mouse that was loaded onto the gel was slightly adjusted so that the tau mAb T46 immunoreactivities of the individual mice were all the same. This correction, which was in the range of up to 15% of the amount of tau of the individual mice loaded onto the gel, was then used for all the immunoblots that were subsequently reacted with the other mAbs. Because the T46 immunoblots were routinely assayed as parallel controls, this enabled a quantitative analysis under accurately calibrated conditions of the anti-tau immunoreactivities of each of the antibodies to tau of the apoE-deficient and control mice.
Immunoblots of tau of apoE-deficient and control mice using the phosphorylation-independent antisera 133 and 134 and mAb T46, whose binding to tau is phosphorylation-independent (Kosik et al., 1988 ; Goedert et al., 1989), and of mAb SMI33, which binds to a dephosphorylated tau epitope (Lichtenberg-Kraag et al., 1992), are depicted in Fig. 1. As can be seen, these antibodies reacted similarly with tau of the two mouse strains. Antisera 133 and 134 are directed against, respectively, the N terminus and C terminus of tau (see Table 1). Thus, the results obtained with these antisera as well as with mAb T46 suggest that the total tau levels of apoE-deficient and control mice are similar and that tau is not truncated preferentially in either of them. Furthermore, the finding that tau of apoE-deficient mice and that of control mice react equally with mAb SMI33 implies that the levels of phosphorylation of the epitope recognized by this mAb are the same in the two groups of mice (see Table 1). The apparent molecular weights of tau of the two groups of mice were similar for each of the above antibodies (Fig. 1), suggesting that the total overall levels of tau phosphorylation in the two mouse groups do not differ ostensibly.
The extents to which the levels of phosphorylation of distinct tau epitopes of apoE-deficient and control mice differ and the effects thereon of prolonged treatment with the muscarinic agonist AF150(S) were examined. This was performed by immunoblot assays using nine mAbs that bind to defined and previously mapped phosphorylated tau epitopes (see Table 1). The tau immunoblots thus obtained, which are depicted in Figs. 2 and 3, were first analyzed by comparison of the levels of tau phosphorylation in nontreated apoE-deficient and control mice. This was followed by assessment of the effects thereon of prolonged treatment with AF150(S). Quantification, by computerized densitometry, of the levels of binding of the anti-phosphorylated tau mAbs to tau of apoE-deficient and control mice is presented in Fig. 4. As can be seen, the immunoreactivities of mAbs AT270, AT8, SMI32, and SMI37 toward tau of the apoE-deficient mice were markedly and significantly higher than the corresponding control values (p < 0.01). The anti-tau immunoreactivities of mAbs P262-7 and SMI34 were also higher in the apoE-deficient mice than in the controls. However, the magnitude of this effect [121 ± 4% of control for mAb P262-7 (p = 0.07) and 114 ± 5% of control for mAb SMI34 (p < 0.03)] was smaller than that observed with the mAbs described above. In contrast, mAbs PHF1, SMI31, and RT97 reacted similarly with tau of the two groups of mice. These results and the known location along the tau backbone of the epitopes that are recognized by these mAbs (Table 1) suggest that tau of apoE-deficient mice contains a hyperphosphorylated “hot spot,” e.g., the epitopes of mAbs AT270, AT8, SMI32, and SMI37, which is located in the vicinity of Thr172 and Ser193, i.e., the epitopes of mAbs AT270 and AT8 (Kosik et al., 1988 ; Goedert et al., 1994). Tau epitopes that are located C-terminally to this “hot spot” are also hyperphosphorylated in the apoE-deficient mice, e.g., mAbs P262-7 and SMI34, but to a lesser extent, whereas epitopes situated in either the C-terminal, e.g., mAbs PHF1 and SMI31, or N-terminal, e.g., RT97, domains of tau are unaffected by apoE deficiency and are similarly phosphorylated in the two mouse strains.
Quantitative analysis of the effects of prolonged treatment of the two mouse strains with the M1 muscarinic agonist AF150(S) on tau immunoreactivity is depicted in Fig. 5. As can be seen in Fig. 5A and B, the tau immunoreactivities of mAbs RT97, PHF1, and SMI31, which reside, respectively, in the N- and C-terminal tau domains, were unaffected by AF150(S) in either apoE-deficient or control mice. In contrast, AF150(S) reduced the tau immunoreactivities of the “hot spot” mAbs AT8, SMI32, and SMI37 of apoE-deficient mice by ~30-50% but had an insignificant effect on the corresponding control tau epitopes (Fig. 5A and B). Tau epitopes flanking the above were dephosphorylated significantly by AF150(S) in both mouse strains. Thus, the mAb AT270 epitope, which resides at the N-terminal edge of the “hot spot,” was dephosphorylated by 25 ± 7% in the apoE-deficient mice and by 37 ± 7% in the controls, whereas the epitopes of mAbs P262-7 and SMI34, which reside at the C-terminal tail of the “hot spot,” were thus dephosphorylated in the apoE-deficient mice by, respectively, 69 ± 7 and 57 ± 7% and in the control mice by, respectively, 36 ± 17 and 54 ± 4%.
The extent to which the AF150(S) treatment abolishes the hyperphosphorylation of apoE-deficient mice was examined by comparing the levels of their tau phosphorylation following treatment with those of the nontreated controls. As shown in Fig. 5C, the extent of dephosphorylation of the “hot spot” epitopes for AT8 and SMI37 of the apoE-deficient mice following treatment with AF150(S) was such that their resulting immunoreactivities were similar to those of the untreated control mice, i.e., 111 ± 7 and 81 ± 12%. The same trend was observed with the “hot spot” epitopes AT270 and SMI32, except that in this case the AF150(S)-induced dephosphorylation reduced but did not abolish the excess tau phosphorylation of the apoE-deficient mice. Accordingly, the tau immunoreactivities of the epitopes for mAbs AT270 and SMI32 of the apoE-deficient mice decreased following treatment with AF150(S) from, respectively, 210 ± 4 and 284 ± 3% to 157 ± 15 and 210 ± 4% of the nontreated controls (compare Figs. 4 and 5C). It is interesting that the P262-7 and SMI34 tau epitopes, which were only slightly hyperphosphorylated in the untreated apoE-deficient mice (Fig. 4), were markedly dephosphorylated following treatment with AF150(S) to, respectively, 42 ± 9 and 50 ± 5% of the nontreated control (Fig. 5C).
The present study revealed that tau of apoE-deficient mice contains a hyperphosphorylated “hot spot” domain that is localized at approximately amino acids 170-200 (Table 1 and Fig. 6) and is composed of the epitopes of mAbs AT270, AT8, SMI32, and SMI37 (Fig. 4). This “hot spot” is flanked on its C-terminal side by epitopes, i.e., mAbs P262 and SMI34, that are also hyperphosphorylated in apoE-deficient mice, but to a lesser extent. In contrast, the levels of phosphorylation of epitopes residing near either the N terminus, i.e., RT97, or the C terminus, i.e., PHF1 and SMI31, of tau were unaffected by apoE deficiency. This study also revealed that prolonged treatment of apoE-deficient and control mice with the M1 muscarinic agonist AF150(S) results in selective dephosphorylation of distinct tau epitopes. Accordingly, AF150(S) selectively dephosphorylates tau “hot spot” epitopes of apoE-deficient but not of control mice, i.e., mAbs AT8, SMI32, and SMI37 ; it dephosphorylates in both mouse strains epitopes that flank the tau “hot spot” at its N-terminal, e.g., mAb AT270, and C-terminal, e.g., mAbs P262-7 and SMI34, sides, and it has no effect, in either of the groups of mice, on the extents of phosphorylation of the N- and C-terminal tau domains. It should be noted that the phosphorylated tau epitope of mAb SMI37 has not yet been mapped. However, as the phosphorylated epitopes affected by both apoE deficiency and AF150(S) seem to cluster (Figs. 4 and 6), we have tentatively mapped the SMI37 epitope near those of mAbs AT8 and SMI32 (Fig. 6).
Hyperphosphorylation of tau in apoE-deficient mice may be due to distinct enzymatic changes that shift the balance between protein kinase and phosphatase activities. Alternatively, it may be due to changes in the accessibility of distinct tau epitopes to neuronal kinase and phosphatase activities. In vitro studies revealed that hyperphosphorylated tau epitopes recognized by “hot spot” mAbs, e.g., AT8 and AT270, can be phosphorylated and dephosphorylated by several proline-directed kinases and protein phosphatases (Ishiguro et al., 1991 ; Drewes et al., 1992 ; Gustke et al., 1992 ; Vulliet et al., 1992 ; Biernat et al., 1993 ; Billingsley and Kincaid, 1997). However, each of these enzymatic activities also affects tau epitopes whose extents of phosphorylation are minimally affected in the apoE-deficient mice, e.g., the epitopes of mAbs P262-7, SMI34, PHF1, SMI31, and PHF1 (Morishima-Kawashima et al., 1995 ; Billingsley and Kincaid, 1997). Thus, inferring from these in vitro findings to the situation in vivo, we may conclude that tau hyperphosphorylation in apoE-deficient mice is probably not due to distinct changes in tau kinases and phosphatase activities but rather to alterations in the accessibility of tau to these enzymes. This may result either from direct tau conformational changes or from alterations in the interactions of tau with other cellular constituents. One such protein may be apoE itself, which binds to tau in vitro (Strittmatter et al., 1993). The epitopes close to the tau-apoE binding domain, e.g., mAbs P262-7 and SMI34, are less hyperphosphorylated than those of the “hot spot” and reside C-terminally to it, e.g., mAbs AT270, AT8, and SMI32 (see Figs. 4 and 6). Thus, the effects of apoE deficiency on the accessibility of tau epitopes to phosphorylating enzymes may be mediated by a mixed mechanism that includes conformational changes of tau as well as alterations in its interactions with other cellular constituents such as apoE. The effects of apoE deficiency on tau phosphorylation develop postnatally (data not shown), which suggests that they may be related to other age-dependent neuronal changes such as synaptic loss (Masliah et al., 1995) which are observed in these mice.
The present in vivo effects of AF150(S) on the levels of tau phosphorylation are in accordance with recent in vitro studies revealing that M1 muscarinic activation of PC12 M1 cells in culture causes tau dephosphorylation (Sadot et al., 1996). These muscarinic effects may be mediated by changes either in protein phosphatase/kinase activities or in the accessibility of tau to these enzymatic activities. As discussed above, the “hot spot” tau epitopes, e.g., mAbs AT8, SMI32, and SMI37, of the nontreated apoE-deficient mice seem to be more accessible to phosphorylating reactions than the corresponding control tau epitopes (Figs. 4 and 6). This increased accessibility could explain why these epitopes are affected by treatment with AF150(S) much more markedly in apoE-deficient than in control mice. The finding that other phosphorylated epitopes, except those that reside near the N and C terminals of tau, are dephosphorylated by AF150(S) in both groups of mice is consistent with the interpretation that muscarinic activation alters the balance between kinase/phosphatase activities. Taken together, the presently observed selective effects of muscarinic activation on the levels of tau phosphorylation may thus be due to two interacting effects, namely, apoE deficiency-mediated increased accessibility of a distinct tau domain and muscarinic-driven dephosphorylating shift in the kinase/phosphatase balance. It is not yet known whether these muscarinic effects occur primarily in neurons that contain muscarinic receptors or whether they are mediated by secreted and transmitted signals.
We have recently shown that AF150(S) reverses the memory and cholinergic deficits of apoE-deficient mice (Fisher et al., 1998). Examination of the extents of correlation between the levels of phosphorylation of the “hot spot” mAbs of AF150(S)-treated and nontreated apoE-deficient and control mice (Figs. 4 and 5) and their individual working memory performances (Fisher et al., 1998) revealed that they are inversely related (R = 0.64 for mAb AT270, p < 0.05 ; R = 0.62 for mAbs SMI370 and SMI32, p < 0.05). This suggests that tau hyperphosphorylation is a suitable and useful biochemical parameter for assessing the cognitive performance of apoE-deficient mice and presumably of other models of neuronal dysfunction.
It is of interest to note that although there is agreement in the literature regarding the existence of cognitive deficits in apoE-deficient mice (Gordon et al., 1995 ; Krugers et al., 1997 ; Krzywkowski et al., 1997 ; Oitzl et al., 1997), the specificity of their neurochemical deficits varies (Gordon et al., 1995 ; Krzywkowski et al., 1997), and in one such line of apoE-deficient mice, mAbs AT8 and AT270 revealed no tau hyperphosphorylation (Mercken and Brion, 1995). A possible interpretation of these varying observations is that the neuronal effects of apoE deficiency depend on genetic background. Such an interpretation is in accordance with AD data revealing that although apoE4 is a risk factor for the disease, not all subjects who carry this allele develop the disease (Roses, 1994, 1997).
In conclusion, the present study shows that tau of apoE-deficient mice contains a cluster of hyperphosphorylated epitopes and that treatment of these mice with the M1 muscarinic agonist AF150(S) triggers the dephosphorylation of distinct tau epitopes and thereby partially abolishes the tau hyperphosphorylation of apoE-deficient mice. These findings provide a new model system for studying the mechanisms underlying tau phosphorylation in vivo and for evaluating the functional and neuropathological ramifications of tau hyperphosphorylation.