19 January 2001. ADF/Cofilin Inclusions in Alzheimer's Disease
James Bamburg presented a poster (Abstract 2821) on the formation of cellular inclusions composed of actin-depolymerizing factor (ADF) and the cofilin family of proteins (collectively known as AC) and of their putative involvement in neurodegeneration. AC proteins regulate actin filament assembly and dynamics. Actin polymerization into filaments is a highly regulated process, in which ATP-bound actin monomers are added to the so-called plus end (or barbed end) of filaments, and disassembly of ADP-actin occurring at the minus end (or pointed end), in a process known as treadmilling. This process is essential for numerous cellular processes, including cell motility and growth cone advance. AC proteins are actin-binding proteins that increase the rate of actin filament treadmilling by enhancing depolymerization of actin from the minus end and also by binding actin filaments and severing them. Previous studies had shown that when eukaryotic cells are stressed (by heat-shock or osmotic stress), AC proteins form nuclear inclusions. Depletion of ATP levels in fibroblasts also promotes the formation of AC inclusions in the cytoplasm. Bamburg surmised that in Alzheimer's disease (AD) AC might form inclusions similar to those formed by tau and other proteins. His group tested this theory by using immunohistochemistry of human brains to detect AC proteins. They found that the frontal cortex and hippocampus of brains from individual afflicted with AD contained numerous AC inclusions (in all 10 cases examined) and that these inclusions were absent in brains from individuals not afflicted with AD. In AD brains, AC were localized to Hirano bodies and to neuritic processes associated with amyloid plaques.
The structure of these AC inclusions in brain was not determined, nor is it known whether full-length or proteolytic fragments of the protein accumulate in the lesions. It is also not known if AC inclusions are a feature of other neurodegenerative disorders. Bamburg's group next investigated the cellular conditions that favor formation of AC inclusions. They found that rod-like inclusions of AC form within 10 minutes when cortical and hippocampal neuronal cultures are treated with agents that lower ATP levels. These inclusions disappeared rapidly (within five to 10 min) upon restoration of cellular ATP levels. They also found that treatment of cell cultures with agents that induce oxidative stress, such as H2O2, nitric oxide, and excess glutamate, caused formation of AC inclusions.
They next studied what effect addition of Aβ peptides had on AC rod formation. Both Aβ1-40 and Aβ1-42 peptides that had been preincubated for three days to form fibrilogenic Aβ, caused AC rod formation after 24 hours when added to E18 hippocampal rat cultures. However, soluble Aβ peptides did not induce such inclusions within the same time frame. Interestingly, a shorter Aβ1-28 peptide, which was non-neurotoxic in their cultures, was still able to induce AC rods, suggesting that rod formation is independent of neurotoxicity. The exact mechanism by which AC proteins form rods is not fully understood, although decreased phosphorylation of the proteins appears to facilitate assembly. Bamburg also indicated that the persistence of rods in neurites appeared to occlude microtubules, which probably affects axonal transport and could explain the loss of growth cones in these cultures. He speculated that AC inclusions in brain might similarly cause synaptic loss, ultimately leading to neurodegeneration. Ref: Nature Cell Biol;2:628-636.
Reagent tip: An antibody to cofilin is available from Cytoskeleton Inc. It is a polyclonal rabbit antibody raised against a C-terminal peptide of human cofilin. For almost everyone looking at human brain samples, the cofilin antibody from Cytoskeleton should be adequate. The Bamburg lab also has raised a rabbit antibody against chick ADF that recognizes both human ADF and human cofilin, both of which are in the inclusions. In nonhuman species (e.g., rat), ADF predominates in the rods and the cofilin staining is much weaker. For further information, contact: email@example.com
Competition of tau isoforms for microtubule binding
tau is a microtubule (MT)-binding protein in which mutations have been shown to cause frontal temporal dementia with parkinsonism (FTDP). (See Tau Mutation directory.) Tau is expressed as six alternatively spliced isoforms: three of the tau isoforms contain three MT-binding repeats (3R) and the other three contain four MT-binding repeats (4R), all of which are located in the COOH-terminus of the protein. The 3R isoforms of tau are generally expressed earlier in development than the 4R isoforms. FTDF-associated mutations in tau cluster in the MT-binding repeats, or in introns, and favor splicing of the 4R isoform over the 3R isoform. However, the mechanism by which these mutations cause disease still remains unresolved. Kenneth Kosik's lab (Abstract 1884) used the green fluorescent protein (GFP) and two color variants -Cyan (CFP) and yellow (YFP)- to tag tau proteins at their NH2-termini to analyze their MT-binding properties. Overexpression of these tagged tau isoforms in NIH3T3 cells induced MT bundling while lower levels of expression (inferred from lower GFP fluorescence) caused more diffuse tau localization. To measure MT-binding, pairs of CFP-tagged and YFP-tagged tau isoforms were cotransfected into cells and their MT binding efficiency in vivo was determined using a fluorescent index (FI), which measures the amount of CFP-fluorescence associated with a section of MT divided by the YFP-fluorescence associated with that section of MT. In the studies presented by Kosik, CFP was fused to 3R and 4R tau, and YFP was fused to 4R tau containing either the FDTP-associated mutations G272V, P301L, V337M, or R406W. Kosik's group found that wild-type 4R tau can displace either wild-type 3R tau or 4R tau carrying the FDTP-associated mutations from MT. It remains to be determined whether the fluorescent tags have altered, in any way, the binding properties of the tau proteins to which they were fused.
A GFP Reporter to Measure Proteasome Function,
Bence et al. (from Ron Kopito's lab; Abstract 615) constructed a modified GFP reporter for monitoring proteasome function. For this application, they fused a 25 amino acid sequence (whose sequence was not disclosed) to the C-terminus of GFP, which decreased the half-life of GFP from ~20 hours to ~25 minutes in transfected HEK293 cells. Presumably the fused sequence contains a ubiquitination target sequence, since treatment of the cells with β-lactone, an inhibitor of the proteasome, stabilized the GFP reporter (as monitored by an increase in fluorescence). Interestingly, in order for GFP to be stabilized, it was necessary to inhibit proteasome activity by at least 75%. Upon proteasome inhibition, the GFP reporter accumulated at the microtubule organizing center into a structure called the aggresome. Kopito's group has generated HEK293 cell lines that stably express this proteasome-sensitive GFP reporter. These cell lines were used to determine what effect overexpression of proteins encoded by exon 1 of the Huntingtin gene (in which polyglutamine (Q) expansion causes Huntington's disease [HD]), containing either 25, 72 or 103 polyQ repeats, or the cystic fibrosis transmembrane receptor (CFTR), a known aggresome inducer, have on proteasome function. They found that huntingtin protein containing 25 polyglutamine repeats (within the normal range of Q repeats; i.e., not associated with causing HD) did not noticeably alter the rate of degradation of the GFP reporter, nor did it alter the reporter's diffuse intracellular localization. In contrast, huntingtin with 103 Q repeats (which is associated with HD) stabilized the GFP reporter, and caused the GFP reporter to accumulate into bright fluorescent inclusions in 20% of the transfected cells. This latter phenotype was also seen upon co-expression of the GFP reporter with the CFTR protein. These results suggest that overexpression of huntingtin with expanded polyQ repeats, or CFTR, compromises proteasome function and that this may be relevant to the etiology by which these mutant proteins cause disease.
Role of Kinesin in Axonal Transport of Amyloid-β Precursor Protein (APP)
Lawrence Goldstein's lab presented two papers on the relationship between amyloid-beta precursor protein (APP) and kinesin. Earlier studies by Edward Koo and Sangram Sisodia had established that APP moves from the cell body to the synapse by fast anterograde transport. Kamal et al. from Goldstein's group investigated the involvement of kinesin in APP transport because the kinesin family of proteins are the main mechanochemical motors that move molecules towards the plus end (anterograde or orthograde direction) of microtubules (MT). Kinesin is structurally similar to the actin based motor, myosin II; it is composed of two heavy chains that contain an N-terminal MT-binding domain and a motor domain, and two light chains that bind the C-terminus of the heavy chains. The kinesin light chains are thought to confer specificity by binding different cargoes. Using antibodies specific for either kinesin heavy chain or to two different kinesin light chains (KLC1 and 2), Kamal showed that APP coimmunoprecipitates with kinesin proteins from mouse brain and sciatic nerve lysates. Moreover, APP cosedimented with KLC1 and KLC2 on sucrose gradients, suggesting that the three proteins are part of a complex. Using in vitro MT-binding assays, Goldstein's group showed that APP cosedimented with MT in mouse brain lysates. In addition using an in vitro assay they found that APP associated with MT upon addition of both hydrolysable and non-hydrolysable analogues of ATP. This behavior is in contrast to that of kinesin, which typically gets trapped on MT by non-hydrolysable ATP. The authors demonstrated that the C-terminal domain of APP was responsible for binding KLC1 and KLC2 fusion proteins using GST-pull down assays. Binding was remarkable strong, with a Kd of ~20 nM. Kamal et al. also showed that axonal transport of APP in sciatic nerves was substantially reduced in KLC1 knockout mice. This may explain, in part, why APP (-/-) and KLC1 (-/-) knockout mice have similar phenotypes. Finally, to test whether APP is a receptor, or merely a cargo of KLC1, the authors studied axonal transport of kinesin chains in APP (-/-) mice. They found that transport of both KLC1 and KLC2 were substantially reduced in the APP knockout mice suggesting that APP may be a receptor for kinesin light chains. Curiously however, KLC1 (-/-) and APP (-/-) mice also had altered transport of GAP43, and why this occurred is not understood (I am not aware of any direct connection between GAP43 and APP). Whether these findings have any relationship to AD etiology is not know. Ref: Neuron;28:449-459.
A second poster from Goldstein's laboratory (Abstract 2859) described the consequences of altering APP expression in Drosophila. Drosophila contains an APP-like (APPL) gene that is most homologous to the human APP695 isoform, except that it lacks the Aβ region of the human protein. Goldstein's group found that both deletion and overexpression of the Drosophila APPL gene resulted in axonal blockage in mutant larvae. Axonal blockage was detected using a GFP reporter targeted to neurons, which accumulated along the length of axonal bundles in structures the authors referred to as "clogs." The authors speculated that these clogs may block axonal transport, eventually leading to cell death. This axonal blockage was similar to that caused by mutations in kinesin and dynein (the anterograde and retrograde motors, respectively), suggesting that APP may indeed be involved in axonal transport. Interestingly, overexpression of APP deleted of its intracellular C-terminal region did not induce clog formation. This may be because this region to APP is important for binding kinesin, as outlined above. Finally, Goldstein's measured cell death in the Drosophila larvae overexpressing APP. They found that overexpression of wild type human APP695 caused apoptosis (detected by Tunnel staining), and that overexpression of the familial AD Swedish APP695 mutation induced even higher levels of apoptosis. They narrowed the region responsible for apoptosis to the C-terminal region of APP. This is consistent with the studies of Yamatsuji et al (Science;272:1349-52) who showed the C-terminal region of APP is involved in apoptosis.-Mervyn Monteiro.