Isoforms of tau protein from mammalian brain and avian erythrocytes: structure, self-assembly, and elasticity

PSF suppresses tau exon 10 inclusion by interacting with a stem-loop structure downstream of exon 10

Microtubule binding protein Tau has been implicated in a wide range of neurodegenerative disorders collectively classified as tauopathies. Exon 10 of the human tau gene, which codes for a microtubule binding repeat region, is alternatively spliced to form Tau protein isoforms containing either four or three microtubule binding repeats, Tau4R and Tau3R, respectively. The levels of different Tau splicing isoforms are fine-tuned by alternative splicing with the ratio of Tau4R/Tau3R maintained approximately at one in adult neurons. Mutations that disrupt tau exon 10 splicing regulation cause an imbalance of different tau splicing isoforms and have been associated with tauopathy. To search for factors interacting with tau pre-messenger RNA (pre-mRNA) and regulating tau exon 10 alternative splicing, we performed a yeast RNA-protein interaction screen and identified polypyrimidine tract binding protein associated splicing factor (PSF) as a candidate tau exon 10 splicing regulator. UV crosslinking experiments show that PSF binds to the stem-loop structure at the 5' splice site downstream of tau exon 10. This PSF-interacting RNA element is distinct from known PSF binding sites previously identified in other genes. Overexpression of PSF promotes tau exon 10 exclusion, whereas down-regulation of the endogenous PSF facilitates exon 10 inclusion. Immunostaining shows that PSF is expressed in the human brain regions affected by tauopathy. Our data reveal a new player in tau exon 10 alternative splicing regulation and uncover a previously unknown mechanism of PSF in regulating tau pre-mRNA splicing.

RNA helicase p68 (DDX5) regulates tau exon 10 splicing by modulating a stem-loop structure at the 5' splice site

Regulation of tau exon 10 splicing plays an important role in tauopathy. One of the cis elements regulating tau alternative splicing is a stem-loop structure at the 5' splice site of tau exon 10. The RNA helicase(s) modulating this stem-loop structure was unknown. We searched for splicing regulators interacting with this stem-loop region using an RNA affinity pulldown-coupled mass spectrometry approach and identified DDX5/RNA helicase p68 as an activator of tau exon 10 splicing. The activity of p68 in stimulating tau exon 10 inclusion is dependent on RBM4, an intronic splicing activator. RNase H cleavage and U1 protection assays suggest that p68 promotes conformational change of the stem-loop structure, thereby increasing the access of U1snRNP to the 5' splice site of tau exon 10. This study reports the first RNA helicase interacting with a stem-loop structure at the splice site and regulating alternative splicing in a helicase-dependent manner. Our work uncovers a previously unknown function of p68 in regulating tau exon 10 splicing. Furthermore, our experiments reveal functional interaction between two splicing activators for tau exon 10, p68 binding at the stem-loop region and RBM4 interacting with the intronic splicing enhancer region.

The solution structure of the C-terminal segment of tau protein

Pathological changes in the microtubule associated protein tau, leading to tau-containing filamentous lesions, are a major hallmark common to many types of human neurodegenerative diseases, including Alzheimer's disease (AD). No structural data are available which could rationalize the extensive conformational changes that occur when tau protein is converted to Alzheimer's paired helical filaments (PHF). The C-terminal portion of tau plays a crucial role in the aggregation of tau into PHF and in the truncation process that generates cytotoxic segments of tau. Therefore, we investigated the solution structure of the hydrophobic C-terminal segment 423-441 of tau protein (PQLATLADEVSASLAKQGL) by 1H 2D NMR spectroscopy. The peptide displays the typical NMR evidence consistent with a alpha-helix geometry with a stabilizing C-capping motif. The reported data represent the first piece of structural information on an important portion of the molecule and can have implications towards the understanding of its pathophysiology.

Structure of tau protein and assembly into paired helical filaments

Over the past few years the systematic investigation of paired helical filament assembly from tau protein in vitro has become feasible. We review our current understanding of the structure and conformations of tau protein and how this affects tau's assembly into the pathological paired helical filaments in Alzheimer's disease.

RNA stimulates aggregation of microtubule-associated protein tau into Alzheimer-like paired helical filaments

The microtubule-associated protein tau is the main component of the paired helical filaments (PHFs) of Alzheimer's disease, the most common senile dementia. To understand the origin of tau's abnormal assembly we have studied the influence of other cytosolic components. Here we report that PHF assembly is strongly enhanced by RNA. The RNA-induced assembly of PHFs is dependent on the formation of intermolecular disulfide bridges involving Cys322 in the third repeat of tau, and it includes the dimerization of tau as an early intermediate. Three-repeat constructs polymerize most efficiently, two repeat constructs are the minimum number required for assembly, and even all six full-length isoforms of tau can be induced to form PHFs by RNA.

Oxidation of cysteine-322 in the repeat domain of microtubule-associated protein tau controls the in vitro assembly of paired helical filaments

One of the hallmarks of Alzheimer disease is the pathological aggregation of tau protein into paired helical filaments (PHFs) and neurofibrillary tangles. Here we describe the in vitro assembly of recombinant tau protein and constructs derived from it into PHFs. Though whole tau assembled poorly, constructs containing three internal repeats (corresponding to the fetal tau isoform) formed PHFs reproducibly. This ability depended on intermolecular disulfide bridges formed by the single Cys-322. Blocking the SH group, mutating Cys for Ala, or keeping tau in a reducing environment all inhibited assembly. With constructs derived from four-repeat tau (having the additional repeat no. 2 and a second Cys-291), PHF assembly was blocked because Cys-291 and Cys-322 interact within the molecule. PHF assembly was enabled again by mutating Cys-291 for Ala. The synthetic PHFs bound the dye thioflavin S used in Alzheimer disease diagnostics. The data imply that the redox potential in the neuron is crucial for PHF assembly, independently or in addition to pathological phosphorylation reactions.

Assembly and bundling of marginal band microtubule protein: role of tau

Microtubule protein extracted from dogfish erythrocyte cytoskeletons by disassembly of marginal bands at low temperature formed linear microtubule (MT) bundles upon reassembly at 22 degrees C. The bundles, which were readily visible by video-enhanced phase contrast or DIC microscopy, increased in length and thickness with time. At steady state after 1 hour, most bundles were 6-11 microns in length and 2-5 MTs in thickness. No inter-MT cross-bridges were visible by negative staining. The bundles exhibited mechanical stability in flow as well as flexibility, in this respect resembling native marginal bands. As analyzed by SDS-PAGE and immunoblotting, our standard extraction conditions yielded MT protein preparations and bundles containing tau protein but not high molecular weight MAPs such as MAP-2 or syncolin. In addition, late fractions of MT protein obtained by gel filtration were devoid of high molecular weight proteins but still produced MT bundles. The marginal band tau was salt-extractable and heat-stable, bound antibodies to mammalian brain tau, and formed aggregates upon desalting. Antibodies to tau blocked MT assembly, but both assembly and bundling occurred in the presence of antibodies to actin or syncolin. The MTs were "unbundled" by subtilisin or by high salt (0.5-1 M KCl or NaCl), consistent with tau involvement in bundling. High salt extracts retained bundling activity, and salt-induced unbundling was reversible with desalting. However, reversibility was observed only after salt-induced MT disassembly had occurred. Reconstitution experiments showed that addition of marginal band tau to preassembled MTs did not produce bundles, whereas tau presence during MT reassembly did yield bundles. Thus, in this system, tau appears to play a role in both MT assembly and bundling, serving in the latter function as a coassembly factor.

Localization of tau and other proteins of isolated marginal bands

To determine which proteins were associated with and intrinsic to the marginal band (MB) of microtubules (MTs), we studied protein components of MBs isolated from nucleated erythrocytes by differential detergent solubilization of the membrane skeleton (MS). MBs isolated from dogfish erythrocytes contained major proteins in the tubulin M(r) range. A high molecular weight protein of approximately 290 kD that bound antibody to syncolin and to heat-stable brain MAPs was present in the whole cytoskeleton. However, most of it was solubilized by the MB isolation medium, together with the MS. Dogfish erythrocyte cytoskeletons and isolated MBs were examined with polyclonal and monoclonal antibodies against mammalian brain tau and chicken erythrocyte tau. As shown by immunofluorescence and immunoblotting, these antibodies bound to proteins in the 50 to 67 kD range, located along the length of isolated MBs. Two-dimensional SDS-PAGE revealed isolated MB proteins of pI approximately 6.8 in the same molecular weight range, as well as alpha- and beta-tubulin with pI approximately 5.4. Subtilisin or high-salt treatment of isolated MBs resulted in unbundling of MTs, indicating involvement of MAPs. MBs isolated from chicken erythrocyte cytoskeletons also contained tau as shown by anti-mammalian brain tau immunofluorescence. Both chicken and dogfish isolated MBs also bound phalloidin, but the binding was usually discontinuous and, for any given MB, matched the pattern of anti-syncolin binding. Both syncolin and F-actin were part of the MS remnant remaining after MT disassembly, supporting their assignment to a specialized MS region at the MB/MS interface. In contrast, tau protein appears to be intrinsic to the MB, where it may have an MT stabilizing and bundling function.

Actin depolymerisation induces process formation on MAP2-transfected non-neuronal cells

We have previously shown that microtubules in nonneuronal cells form long, stable bundles after transfection with the embryonic neuronal microtubule-associated protein MAP2c. In this study, we found that treating MAP2c-transfected cells with the actin depolymerising drug cytochalasin B led to the outgrowth of microtubule-containing processes from the cell surface. This effect was specific to MAP2c and did not occur in untransfected cells whose microtubules had been stabilised by treatment with taxol. The outgrowth and retraction of these processes during repeated cycles of cytochalasin addition and removal was followed by video time-lapse microscopy and was suggestive of a physical interaction between compressive forces exerted by the MAP2c-stabilised microtubule bundles and tensile forces originating in the cortical actin network. We suggest that MAP2c confers three properties on cellular microtubules that are essential for process outgrowth: stability, bundling and stiffness. The latter probably arises from the linking together of neighbouring tubulin subunits by three closely spaced tubulin-binding motifs in the MAP2 molecule that limits their motion relative to one another and thus reduces the flexibility of the polymer. Similar multimeric tubulin-binding domains in other proteins of the MAP2 class, including tau in axons and MAP4 in glial cells, may play the same role in the development and support of asymmetric cell morphology. Axial bundles of microtubules are found in growing neurites but not in growth cones, suggesting that the regulated expression of these MAP-induced properties makes an important contribution to the establishment of a stable process behind the advancing growth cone.

Interaction of brain mitochondria with microtubules reconstituted from brain tubulin and MAP2 or TAU

To explore the behaviour of microtubule-associated proteins, MAP2 and TAU in the interactions of mitochondria with microtubules, an homologous acellular system has been reconstituted with organelles isolated from rat brain. We have established a quantitative in vitro binding assay based on the cosedimentation of 125I-labeled microtubules with mitochondria. We found that binding of microtubules to mitochondria was concentration dependent and saturable. Binding was insensitive to ATP. A comparison of taxol-stabilized microtubules prepared from MAP-free tubulin or tubulin coated with TAU or MAP2 showed that the microtubule-associated proteins diminished, or reduced to background levels, the formation of complexes with mitochondria. In contrast, the amount of MAP-free taxol microtubules that cosedimented with mitochondria increased two- and six-fold when mitochondria were coated with MAP2 or TAU. These studies suggest that the two major brain MAPs could have a crosslinking or a spacing role, depending on their organelle localization.

Solubilization and fractionation of paired helical filaments

Paired helical filaments isolated from brains of two different patients with Alzheimer's disease were extensively treated with the ionic detergent, sodium dodecyl sulphate. Filaments were solubilized at different extents, depending on the brain examined, thus suggesting the existence of two types of paired helical filaments: sodium dodecyl sulphate-soluble and insoluble filaments. In the first case, the number of structures resembling paired helical filaments greatly decreased after the detergent treatment, as observed by electron microscopy. Simultaneously, a decrease in the amount of sedimentable protein was also observed upon centrifugation of the sodium dodecyl sulfate-treated paired helical filaments. A sodium dodecyl sulphate-soluble fraction was isolated as a supernatant after low-speed centrifugation of the sodium dodecyl sulphate-treated paired helical filaments. The addition of the non-ionic detergent Nonidet-P40 to this fraction resulted in the formation of paired helical filament-like structures. When the sodium dodecyl sulphate-soluble fraction was further fractionated by high-speed centrifugation, three subfractions were observed: a supernatant, a pellet and a thin layer between these two subfractions. No paired helical filaments were observed in any of these subfractions, even after addition of Nonidet P-40. However, when they were mixed back together, the treatment with Nonidet P-40 resulted in the visualization of paired helical filament-like structures. These results suggest that at least two different components are needed for the reconstitution of paired helical filaments as determined by electron microscopy. The method described here may allow the study of the components involved in the formation of paired helical filaments and the identification of possible factors capable of blocking this process.

Alzheimer-like paired helical filaments and antiparallel dimers formed from microtubule-associated protein tau in vitro

Recent evidence from several laboratories shows that the paired helical filaments of Alzheimer's disease brains consist mainly of the protein tau in an abnormally phosphorylated form, but the mode of assembly is not understood. Here we use EM to study several constructs derived from human brain tau and expressed in Escherichia coli. All constructs or tau isoforms are rodlike molecules with a high tendency to dimerize in an antiparallel fashion, as shown by antibody labeling and chemical crosslinking. The length of the rods is largely determined by the region of internal repeats that is also responsible for microtubule binding. One unit length of the repeat domain (three or four repeats) is around 22-25 nm, comparable to the cross-section of Alzheimer PHF cores. Constructs corresponding roughly to the repeat region of tau can form synthetic paired helical filaments resembling those from Alzheimer brain tissue. A similar self-assembly occurs with the chemically cross-linked dimers. In both cases there is no need for phosphorylation of the protein.

Domain structure and antiparallel dimers of microtubule-associated protein 2 (MAP2)

We have studied the microtubule-associated protein MAP2 from porcine brain and its subfragments by limited proteolysis, antibody labeling, and electron microscopy. Two major chymotryptic fragments start at lys 1528 and arg 1664, generating microtubule-binding fragments of Mr 36 kDa (303 residues, analogous to the "assembly domain" of Vallee, 1980) and 18 kDa (167 residues). These fragments can be labeled with the antibody 2-4 which recognizes the last internal repeat of MAP2 (Dingus et al., 1991). The epitope of another monoclonal antibody, AP18 (Binder et al., 1986), was mapped to the first 151 residues of MAP2. The interaction with AP18 is phosphorylation dependent; dephosphorylated MAP2 is not recognized. Intact MAP2 forms rod-like particles of 97 nm mean length, similar to Gottlieb and Murphy's (1985) observations. Both antibodies bind near an end of the rod, suggesting that the sequence and the structure are approximately colinear. There is a pronounced tendency for MAP2 to form dimers whose components are nearly in register but of opposite polarity. MAP2 can also fold in a hairpin-like fashion, generating 50-nm rods, and it can self-associate into oligomers and fibers. The 36-kDa microtubule-binding fragment also has a rod-like shape; its mean length is 49 nm, half of the intact molecule, even though the fragment contains only one-sixth of the mass. The antibody 2-4 decorates one end of the rod, similar to the intact protein. The fragment also forms antiparallel dimers, but its tendency for higher self-assembly forms is much lower than with intact MAP2.

The effect of estramustine derivatives on microtubule assembly in vitro depends on the charge of the substituent

Estramustine, and derivatives of estramustine with a charged substituent at position 17 on the estrogen moiety, have been investigated for their effects on bovine brain microtubules in vitro. The negatively charged estramustine phosphate has been found previously to be a microtubule-associated protein (MAP)-dependent microtubule inhibitor [Wallin M, Deinum J and Fridén B, FEBS Lett 179: 289-293, 1985]. In the present study the binding of estramustine phosphate to MAP2 and tau was investigated. Both these MAPs were found to have two to three binding sites for estramustine phosphate which is compatible with the reported number of basic amino acid repeats of these MAPs, considered to be the ultimate tubulin binding domains. The Kd for the binding of estramustine phosphate to MAP2 was estimated to be 20 microM at 4 degrees, and for the binding of tau, 200 microM. The rate of dissociation was very low (T1/2 greater than 2 hr), which indicates that the binding of estramustine phosphate may stabilize the protein-drug complex by changing the protein conformation. Two new negatively charged estramustine derivatives, estramustine sulphate and estramustine glucuronide, were found to be similar MAP-dependent microtubule inhibitors. The concentration for 50% inhibition of assembly was 100 microM for the sulphate derivative, the same as found previously for estramustine phosphate, and 250 microM for the more bulky estramustine glucuronide. A positively charged derivative, estramustine sarcosinate, did not inhibit microtubule assembly or alter the composition of the coassembled MAPs. The morphology of the microtubules was, however, affected. The uncharged estramustine bound to both tubulin and MAPs, but no effects were seen on microtubule assembly, the composition of coassembled MAPs or the microtubule morphology. Our results suggest that only negatively charged estramustine derivatives have a MAP-dependent microtubule inhibitory effect. The two new negatively charged derivatives could therefore be valuable tools in the study of tubulin-MAP interactions. The results also confirm that these interactions between tubulin and MAPs are mainly electrostatic.