Abstract

EXPERIMENTAL PROCEDURES

Tau Expression, Purification, and Fibrillization—The MTBR (amino acids 243–375) of full-length (P10636-8) wild-type tau (a gift from Dr. Virginia Lee) was subcloned into pRK172. HA-tagged MTBR Tau contains the sequence YPYDVPDYA on its C terminus. Recombinant Tau MTBR was prepared as described previously from Rosetta (DE3)pLacI competent cells (Novagen), exploiting the heat stability of Tau protein followed by cation exchange chromatography (19). Single-use aliquots were stored at -80 °C in 10 mm HEPES and 100 mm NaCl (pH 7.4). To induce fibrillization of Tau monomer, the MTBR was incubated at room temperature without agitation in 5 mm dithiothreitol, 10 mm HEPES (pH 7.4), 100 mm NaCl, and 150 μm arachidonic acid (Sigma). Incubation times from 3 to 24 h produced similar results.

Atomic Force Microscopy and Antibody Decoration—Tau fibrillization reactions were incubated on freshly cleaved mica (Ted Pella, Inc.) for 2 min. The sample was then rinsed twice with 200 μl of water. After washing, samples exposed to antibody decoration were incubated with 20 μg/ml Tau5 antibody (BD Biosciences) for 2 min. The sample was rinsed twice with water and left to dry for at least 1 h prior to tapping mode atomic force microscopy (Digital Instruments).

Western Blots—HEK293 or C17.2 cells were seeded 300,000 or 75,000 cells/well, respectively, in a 24-well plate. The following day, cells were transfected with 1.2 μg of full-length wild-type Tau-YFP or MTBR-YFP using Lipofectamine2000 (Invitrogen) according to the manufacturer's recommendations. After 15 h, cells were replated 1:4 in a 24-well plate, allowed to re-adhere for 5 h, and treated with 0.4 μm Tau buffer, monomer, or aggregates. After 15 h, cells were harvested with 0.25% trypsin for 3 min, pelleted, and lysed in 40 μl of 1% Triton in PBS plus a protease inhibitor tablet (Roche Applied Science) (lysis buffer). As an additional control, untreated cells were treated with an equivalent concentration of aggregates after lysis. Soluble and insoluble fractions of cell lysates were obtained by centrifugation of lysates at 14,000 rpm for 10 min at 4 °C. Insoluble pellets were washed twice with lysis buffer and resuspended in 40 μl of lysis buffer. For whole cell lysates, cells were harvested and syringe-lysed in PBS plus a protease inhibitor tablet. Triton-soluble and -insoluble fractions were resolved by SDS-PAGE on a 7.5% gel, whereas whole cell lysates were resolved by SDS-PAGE on a 4–15% gradient gel (Bio-Rad). Following transfer to a nylon membrane (Millipore), blots were probed with either the Tau5 antibody (1:5,000; BD Biosciences), which recognizes an epitope in the polyproline region of Tau protein (intracellular transfected Tau) that is not present in the MTBR (extracellular Tau), or the GFP sc-8334 antibody (1:2,000; Santa Cruz Biotechnology). Blots were stripped and reprobed with the I-19-R anti-actin antibody (1:2,000; Santa Cruz Biotechnology). HA Y.11 (1:2,000, Covance) was used to probe for MTBR-HA Tau.

Immunofluorescence—All cells were grown on polyornithine-coated glass coverslips for microscopy. For visualization of intracellular Tau-YFP and MTBR-YFP, cells were transfected with Tau-YFP or MTBR-YFP. Cells were fixed with -20 °C methanol for 7 min and stained for α-tubulin (1:500; Sigma) followed by secondary staining with a rhodamine-conjugated anti-mouse antibody A546 (1:500; Molecular Probes). To visualize MTBR-AF488 aggregates inside cells, buffer, MTBR monomer, and MTBR aggregates were labeled with 62.5 ng/μl AF488 (Molecular Probes) for 1 h at room temperature and then overnight at 4 °C. Reactions were quenched with 100 mm glycine and added to cells. After 15 h, cells were treated with either PBS or 0.25% trypsin for 3 min and allowed to recover for 5 h before methanol fixation and staining for α-tubulin. Z-stacks were rendered into a three-dimensional image using the NIS-Elements AR 3.0 software (Nikon) from which an apical to distal slice containing the aggregate was obtained. For co-labeling with dextran, C17.2 cells were treated with MTBR-AF488 and 50 μg/ml rhodamine-dextran (M_r 10,000, pH neutral; Molecular Probes) for 4 h. Cells were trypsinized, fixed with 4% paraformaldehyde, and prepared as described above. For extracellular MTBR-HA and intracellular Tau-YFP co-localization studies, C17.2 cells expressing Tau-YFP were incubated with buffer, MTBR-HA monomer, or MTBR-HA aggregates for 15 h. Cells were trypsinized and allowed to recover as described before fixation with 4% paraformaldehyde. Cells were stained with the HA-Y11 antibody (1:1,000; Santa Cruz Biotechnology), followed by secondary labeling with rhodamine-conjugated anti-rabbit antibody (1:500; Molecular Probes). For co-culture experiments, C17.2 cells were transfected separately with MTBR-YFP and SV40-mCherry (a gift of Dr. Anthony Gerber) or Tau-YFP and mCherry. Cells were treated with Tau buffer, 0.4 μm Tau monomer, or 0.4 μm MTBR Tau aggregates. After the indicated co-culture incubation, cells were fixed for 7 min with 4% paraformaldehyde, stained with 4,6-diamidino-2-phenylindole (Sigma), and mounted on glass slides for imaging. Confocal microscopy was performed on a C1sl confocal microscope (Nikon Instruments Inc.).

Flow Cytometry—AF488 containing buffer or MTBR-AF488 was incubated with cells for the time indicated. Cells were harvested with 0.25% trypsin for 3 min and resuspended in PBS plus 10% fetal bovine serum prior to flow cytometry. Cells were counted in a FACSCalibur (BD Biosciences) flow cytometer. Each experiment was repeated four times, and 10,000 cells were counted in each individual experiment. For co-culture experiments, MTBR-YFP or Tau-YFP expressing cells were co-cultured with SV40-mCherry-expressing cells. Tau-YFP expressing cells were treated with buffer, MTBR monomer, or MTBR aggregates for indicated durations prior to flow cytometry. To collect dual positive cells, cells were sorted in a FACSAria cell sorter (BD Biosciences).

Sarkosyl Extraction—1% Sarkosyl-insoluble material was purified from cell lysates from a confluent 5-cm cell culture dish as described previously (20), with some modifications. Cell lysis buffer consisted of 10 mm Tris-HCl (pH 7.4), 0.8 m NaCl, 1 mm EGTA, 5 mm EDTA, 10% sucrose, and a protease inhibitor tablet (Roche Applied Science). Sarkosyl-insoluble pellets were resuspended in 50 μl of 10 mm HEPES and 100 mm NaCl (pH 7.4). When Sarkosyl-insoluble material from cells was used to seed the Tau monomer, the reactions contained 90 μl of 4 μm MTBR Tau monomer, 5 mm dithiothreitol, 100 mm NaCl, 10 mm HEPES, and 10 μl of syringe-sheared Sarkosyl-insoluble material from a 5-cm dish of buffer or MTBR aggregate-treated, Tau-YFP-expressing cells.

RESULTS

Full-length Tau Does Not Aggregate Spontaneously in Cells; Truncated Tau Forms Inclusions—To characterize the cellular activities of Tau, we used four different constructs (Fig. 1A). For expression in mammalian cells, we fused either full-length Tau (441 amino acids) or the microtubule-binding region (amino acids 243–375) to yellow fluorescent protein (YFP) (Tau-YFP and MTBR-YFP). For in vitro Tau fibrillization reactions, we created MTBR Tau with a hemagglutinin tag (MTBR-HA) and untagged MTBR. We expressed Tau-YFP in C17.2 neuronal precursor cells (21) by transient transfection. Tau-YFP co-localized with microtubules (Fig. 1, B and F), as others have reported (22, 23), and did not readily form inclusions. In contrast, MTBR-YFP did not bind microtubules and readily formed inclusions (Fig. 1, C and E) when expressed at a level comparable with Tau-YFP (Fig. 1D). 83% of cells expressing Tau-YFP exhibit co-localization with tubulin, whereas 3.5% of cells expressing MTBR-YFP exhibit co-localization with tubulin (Fig. 1F).

Tau Aggregates Enter C17.2 Cells—It has been shown previously that aggregated amyloid proteins amyloid β, prion protein, and expanded polyglutamine can gain entry into cells (24–27). Thus, we tested for uptake of recombinant Tau aggregates into C17.2 cells. First, we prepared recombinant MTBR-HA and used arachidonic acid to stimulate its fibrillization, which we detected using AFM (Fig. 2A). We found that these aggregates were completely digested when treated with 0.125% trypsin for 1 min (Fig. 2B). This is 50% of the level of trypsin used to re-plate cells in tissue culture (which are treated for 3 min), indicating relative sensitivity of the extracellular aggregates to proteolysis.

To determine whether Tau aggregates were taken into cells, we labeled MTBR Tau monomer and aggregates with AF488. We exposed C17.2 cells to AF488-containing buffer, AF488 conjugated to MTBR monomer, or MTBR aggregates. After 3 and 9 h, we harvested the cells from the plates by treating cells with 0.25% trypsin for 3 min. We quantified AF488 fluorescence using flow cytometry (Fig. 2C). Fluorescence signal due to background uptake of unconjugated AF488 in the buffer was subtracted from all samples. We observed significantly more uptake of aggregates compared with monomer at both time points (Fig. 2C). Direct visualization of these cells before and after trypsinization confirmed the conclusions derived from the flow cytometry experiment, indicating internalization of the aggregated species by the cells (Fig. 2, D and E). Aggregates of extracellularly derived Tau of various sizes were observed in cultured cells.

To verify that the Tau present after trypsinization was indeed intracellular, we stained MTBR-AF488 aggregate-treated cells for tubulin and visualized the cells using confocal microscopy. In cells that were treated with aggregates and imaged before trypsinization, both intracellular and extracellular membrane-associated aggregates were observed (data not shown). In trypsin-treated cells, however, the internalized aggregates displaced tubulin (Fig. 2E) and were present in the same plane as tubulin (Fig. 2E, bar), confirming their intracellular locale.

Tau Aggregates Enter C17.2 Cells and Co-localize with Dextran—It has been shown previously that exogenously derived amyloid β and prion protein aggregates co-localize with dextran, a marker of fluid-phase endocytosis. Such aggregates did not co-localize with cholera toxin subunit B, a marker of lipid rafts (26). We exposed C17.2 cells to MTBR-AF488 Tau aggregates and dextran. 24% of Tau aggregates co-localized with dextran (n = 3, 200 aggregates counted per experiment) (Fig. 3) but did not co-localize with cholera toxin B (data not shown). This suggests that Tau aggregates enter cells via an endocytic pathway involving engulfment by the cell membrane, rather that the simple penetration of the membrane, as has been proposed previously for other amyloid-forming proteins (28).

Exogenous Tau Aggregates Induce Aggregation of Intracellular Full-length Tau-YFP—Tau fibrils propagate in vitro via seeded polymerization (29). We thus tested whether extracellular Tau aggregates that enter cells would trigger the misfolding of intracellular Tau-YFP. We transiently transfected HEK293 cells with Tau-YFP. We treated these cells with buffer, MTBR monomer, or MTBR aggregates. After 15 h, cells were harvested and lysed in 1% Triton, and extracts were fractionated by centrifugation at 15,000 × g. After SDS-PAGE, Western blot was performed using the Tau5 antibody, which recognizes a motif in the polyproline region of full-length Tau that is not present in the MTBR. Cells treated with aggregates, but not buffer or monomer, contained detergent-insoluble Tau-YFP (Fig. 4A). We repeated this experiment in C17.2 neural cells, including a negative control in which we mixed MTBR aggregates with cell lysate containing Tau-YFP prior to SDS-PAGE to rule out induction of aggregation following lysis (Fig. 4B). To confirm Tau-YFP aggregation in C17.2 cells, detergent-free whole cell lysates were created and resolved by Western blot with the Tau5 antibody. We observed Tau-YFP aggregates at the top of the well when cells were treated with MTBR aggregates but not after buffer or monomer exposure (Fig. 4C). To determine the specificity of aggregate induction, we treated Tau-YFP-expressing cells with aggregates of Huntingtin exon 1 containing 53 glutamines. Exogenous aggregated Huntingtin had no effect on intracellular Tau-YFP aggregation based on Western blot (data not shown).

Taken together, these experiments indicated that extracellular Tau aggregates enter cells and induce misfolding of intracellular Tau, but left uncertain whether this was because of a direct association of exogenous and endogenous protein. Thus, we used double label fluorescence confocal microscopy to test for co-localization of exogenous MTBR-HA and endogenous Tau-YFP. After treatment with MTBR-HA aggregates for 15 h and visualization via confocal microscopy, we observed intracellular inclusions consisting of both exogenously and endogenously derived protein, most of which co-localized (Fig. 4D). 74% of aggregates observed were composed of both MTBR-HA and Tau-YFP; 23% of aggregates were composed only of Tau-YFP; and 3% of aggregates were composed only of MTBR-HA (Fig. 4E). Tau-YFP aggregates observed in the absence of MTBR-HA co-localization may arise from self-seeding or may contain MTBR-HA seeds that are too small to detect via immunofluorescence.

Extracellular Tau Aggregates Induce Fibrillization of Intracellular Tau-YFP—To rule out the possibility that exogenously derived aggregated Tau caused disordered aggregation of intracellular Tau-YFP, rather than fibrillization, we purified induced Tau-YFP aggregates from cells. We treated C17.2 cells expressing Tau-YFP with buffer or aggregates, and we examined the Sarkosyl-insoluble fraction (20) from cell lysates using AFM. No fibrillar material was observed except in cells exposed to exogenous MTBR Tau aggregates (Fig. 5A). To confirm that these fibrils were derived from intracellular full-length Tau-YFP, and not the exogenous MTBR Tau aggregates or other proteins, we used antibody decoration (30) with the Tau5 antibody. We observed an increase in fibril diameter along the length of the fibrils in Tau5-treated extracts, indicating the purified fibrils were composed of Tau-YFP (Fig. 5B). We next used Sarkosyl-insoluble Tau-YFP purified from C17.2 cells to seed the fibrillization of recombinant MTBR in vitro, confirming the conversion of Tau-YFP to a fibrillar form that is competent to seed aggregation of Tau monomer (Fig. 5C). Antibody decoration of the seeded fibrils using the Tau5 antibody produced isolated densities on the fibrils, confirming the source of the seed as Tau-YFP (Fig. 5D). Thus, exposure to exogenous Tau aggregates induces intracellular Tau-YFP to form fibrils that are competent to seed further aggregation.

Intracellular Tau-YFP Aggregates Transfer between Co-cultured Cells—We next tested whether intracellular Tau aggregates transfer between co-cultured cells. We separately transfected cells with either MTBR-YFP or mCherry (a red fluorescent protein variant) and then co-cultured them for 24 h. As a control, we cultured MTBR-YFP and mCherry cells separately and mixed them immediately prior to flow cytometry. Co-cultured cells had significantly more YFP/mCherry dual positive cells (Fig. 6A). Based on four independent experiments, 0.48% of mixed cells (i.e. background) scored as dual-fluorescent versus 1.6% of co-cultured cells (Fig. 6B). Dual positive cells were collected and viewed via fluorescence microscopy. This confirmed the presence of MTBR-YFP inclusions inside cells expressing mCherry (Fig. 6C), indicating that transfer of MTBR-YFP aggregates had occurred. To determine whether the transfer of Tau aggregates between co-cultured cells is inducible, we co-cultured cells expressing full-length Tau-YFP and mCherry and initiated misfolding of Tau-YFP by exposing cells to MTBR Tau aggregates for 48 h (Fig. 7A). Treatment with aggregated species caused Tau-YFP aggregation and transfer of these species to 1% of the mCherry cells (Fig. 7B). We sorted dual positive cells and imaged them with fluorescence microscopy. This confirmed the presence of Tau-YFP inclusions in mCherry-expressing cells (Fig. 7C). Simple cell fusion was unlikely to have occurred at a significant rate, because buffer and monomer treated cells did not score positive. Thus, spontaneously formed MTBR-YFP Tau intracellular inclusions or induced full-length Tau-YFP inclusions spontaneously transfer between co-cultured cells.

DISCUSSION

We have tested the hypothesis that aggregated extracellular Tau enters cells and transmits a misfolded state specifically to intracellular Tau. We used recombinant MTBR Tau prepared in vitro to form aggregates, which were readily taken up by cultured cells. Full-length Tau-YFP protein does not readily aggregate within the cell, but internalized aggregates induce its fibrillization. When purified from cells, Tau-YFP is competent to seed the fibrillization of Tau monomer in vitro. Finally, we observed that aggregated intracellular Tau transfers between co-cultured cells. We are uncertain how an aggregate taken into an endocytic vesicle might ultimately gain access to protein located in the cytoplasm, and whether aggregates that transfer from one cell to another are capable of inducing further intracellular aggregation. Aggregate uptake and induced aggregation in neighboring cells was recently suggested by co-culture of polyglutamine-expressing cells, although this study failed to document direct transfer of aggregated protein (27). Relatively low efficiencies of cell-to-cell aggregate transfer and extracellular-to-intracellular seeding may make it difficult to document these events in vitro.

It is unknown why misfolded amyloid protein accumulates progressively throughout the brain in tauopathy and other neurodegenerative diseases. These data provide a plausible cellular mechanism and are consistent with the hypothesis that Tau aggregates propagate protein misfolding within the brain. Such aggregates could be released from dead or dying cells. Alternatively, they might move between cells via an exocytic process, as has been suggested for prion protein (31). Following uptake by healthy neurons, the aggregates might stimulate further misfolding of an otherwise stable intracellular protein, in the manner of prions. The exocytosis model is more consistent with the observation that spread of Tau pathology traditionally affects networks of neurons that are synaptically connected (8), although both mechanisms could act in concert to transfer Tau aggregates between cells. These ideas await testing in vivo.We predict that introduction of exogenous fibrillar protein into the brain of a susceptible animal, like prions, will induce further aggregation of endogenous protein and progressive pathology.

We have recently shown that wild-type Tau is capable of conformational diversity that depends on templated conformation change (32). These data, along with our present study, suggest a mechanism to explain disease-specific cellular vulnerabilities to misfolded protein based on fibrillization rate and propensity for propagation of a given misfolded species. Indeed, splicing variants or post-translational modifications of Tau (e.g. phosphorylation) could render a subgroup of neurons particularly vulnerable to spontaneous or seeded protein misfolding. Thus the propagation of misfolded species through the brain would reflect a combination of neuronal proximity, connectivity, and the particular Tau moieties expressed within the involved regions. Finally, propagation via progressive, templated misfolding suggests a general pathogenic mechanism for other neurodegenerative diseases linked to amyloid protein aggregation.

Acknowledgments

Notes

^*This work was supported, in whole or in part, by National Institutes of Health Grant R01 NS50284-03 from NINDS (to M. I. D.) and a National Institutes of Health grant from NINDS (training grant to B. F.). This work was also supported by grants from the Sandler Family Supporting Foundation, the Taube Family Foundation Program in Huntington Disease Research, and the Muscular Dystrophy Association.

Footnotes

References