Vicky
Ury-Thiery
,
Yann
Fichou
,
Isabel
Alves
,
Michael
Molinari
,
Sophie
Lecomte
and
Cécile
Feuillie
*
Univ. Bordeaux, CNRS, Bordeaux INP, CBMN, UMR 5248, F-33600 Pessac, France. E-mail: cecile.feuillie@u-bordeaux.fr
First published on 22nd August 2024
The Tau protein is implicated in various diseases collectively known as tauopathies, including Alzheimer's disease and frontotemporal dementia. The precise mechanism underlying Tau pathogenicity remains elusive. Recently, the role of lipids has garnered interest due to their implications in Tau aggregation, secretion, uptake, and pathogenic dysregulation. Previous investigations have highlighted critical aspects: (i) Tau's tendency to aggregate into fibers when interacting with negatively charged lipids, (ii) its ability to form structured species upon contact with anionic membranes, and (iii) the potential disruption of the membrane upon Tau binding. In this study, we examine the disease-associated P301L mutation of the 2N4R isoform of Tau and its effects on membranes composed on phosphatidylserine (PS) lipids. Aggregation studies and liposome leakage assays demonstrate Tau's ability to bind to anionic lipid vesicles, leading to membrane disruption. Attenuated total reflection Fourier-transform infrared spectroscopy (ATR-FTIR) reveals the accumulation of Tau on the membrane surface without protein insertion, structuration, or lipid removal. Plasmon waveguide resonance (PWR) demonstrates a strong binding of Tau on PS bilayers with an apparent Kd in the micromolar range, indicating the deposition of a thick protein layer. Atomic force microscopy (AFM) real-time imaging allows the observation of partial lipid solubilization and the deposition of polymorphic aggregates in the form of thick patches and fibrillary structures resembling amyloid fibers, which could grow from a combination of extracted anionic phospholipids from the membrane and Tau protein. This study deepens our understanding of full-length Tau's multifaceted interactions with lipids, shedding light on potential mechanisms leading to the formation of pathogenic Tau assemblies.
Tau plays a pivotal role in a class of diseases called tauopathies, which includes Alzheimer's disease and frontotemporal dementia.2 In pathological conditions, Tau can misfold, aggregate and accumulate in the form of amyloid aggregates forming neurofibrillary tangles.3 A hallmark of these aggregates is the characteristic cross-β structure found in amyloids, where β-sheets are ordered in an amyloid core. In the case of Tau, the amyloid core encompasses the third and fourth repeat regions as well as part of the C terminus, although the exact location of the amyloid core is disease specific.4 The remainder of Tau forms a fuzzy coat structure, in majority unfolded, with a few short and transient elements of secondary structure.5 Various single mutations present in the microtubule binding domain have been shown to promote Tau pathology in frontotemporal dementia linked with chromosome 17.6 Among these mutations, P301L (proline replaced by leucine) is one of the most aggressive and widely studied. At a molecular level, this substitution was shown to promote aggregation by modulating the conformation of the protein and exposing hydrophobic regions.7,8
In vitro, polyanionic co-factors such as heparin are widely used to induce the aggregation of Tau.9 If heparin is the most used co-factor in vitro, others, like RNA10 or negatively charged lipids,3,11–13 also have the propensity to induce the aggregation of Tau. The lipid membrane and its interactions with Tau are increasingly studied, as Tau-lipid interactions could be involved not only in Tau aggregation but also in Tau-induced cell toxicity and Tau spreading.14 Tau has been shown to interact with the plasma membrane of neurons.15,16 Indeed, intracellular Tau can be translocated directly through the membrane in a free form and taken up by nearby cells via different pathways.12,15–17 Tau can also directly interact with lipids such as Phosphatidylinositol 4,5-bisphosphate (PIP2),16 cholesterol, and sphingomyelin15 in cells. Interestingly, lipids were found to be enriched in purified brain homogenates containing paired helical filaments (PHF) from Alzheimer's disease patients.18 With Tau neurofibrillary tangles being intracellular, the effect of anionic lipids, predominant on the inner leaflet of neuronal membranes – especially phosphatidylserine (POPS) – is of particularly interest. Studies have demonstrated that patients with Alzheimer's disease display modifications in both the proportion and content of cellular lipids.19 This includes an increase in POPS content, particularly in the hippocampus and temporal lobe,20 areas correlated with the progression of Alzheimer's disease,21 making it a lipid target of choice.
The mechanism of interactions between Tau protein and negative lipids is far from being fully understood. Existing literature on Tau's interaction with negatively charged lipids has allowed to bring partial understanding of the processes at play. Lipid-induced protein aggregation has been reported in in vitro studies of Tau interactions with negatively charged lipid vesicles,22 with formation of protein/lipid complexes.12,22,23 Full-length Tau 2N4R was shown to form pore-like structures when in interaction with phosphatidylethanolamine (PE)/phosphatidylglycerol (PG) planar bilayers.24 Insertion into the membranes is however not always observed, as in the case of K19 interacting with PS-containing membranes.23 Structural reorganization was observed upon membrane binding: K19, a truncated version of Tau composed of the R1, R3 and R4 domains, for instance has been shown to adopt β-structures upon binding to membrane vesicles composed of dimyristoylphosphatidylcholine (DMPC)/dimyristoylphosphoserine (DMPS) at a molar ratio of 4:
1,25 but has been shown to fold into α-helices upon binding to palmitoyl-oleoyl phosphatidic acid acid (POPA)/palmitoyl-oleoyl phosphatidylcholine (POPC) membranes26 or POPS-containing SUVs.27 Compaction of the full-length Tau protein has also been reported upon selective insertion into DMPG monolayers.28 Tau-induced membrane disruption has been observed in some cases, for instance for K18, a fragment of Tau composed of the four repeat regions, on fluid supported membranes,29 or for full-length Tau on DMPG monolayers.28 Some contradiction remains regarding the disruptive capacities of Tau's interaction with membranes, which seem to depend on lipid composition.25,29 Membrane curvature could also be involved, as K18 was reported to not disrupt the bilayer integrity of PS/PC vesicles,22 but to partly solubilize PS-containing supported membranes locally.29 Overall, many studies have focused on the interactions between membranes and short fragments of Tau, like the microtubule binding domain K18,22,29 the K19 construct,25–27 the microtubule binding domain with the proline-rich P2 region P2R,23 or individual repeat domains.30 The N-terminal region of the protein could nonetheless play a crucial role in Tau – membrane interactions, as indicated by cellular studies,31 and must be considered in Tau assembly and membrane interaction studies.32 As the overall charges of K18 or K19 (respectively +10 and +7 at physiological pH) are higher compared to full-length Tau 2N4R (+2), one can also raise the question of the driving force of the interaction between full-length Tau and membranes compared to shorter fragments.
Investigating the interaction of full-length Tau with POPS, predominant anionic lipid of the cytosolic inner leaflet of the plasma membrane, offers a promising lead to better understand the interaction of Tau with negatively charged lipids from the cell interior, while deepening the understanding of the mechanisms of aggregation and toxicity of Tau on neuronal membranes. In this study, the interactions between the Tau protein (2N4R) harboring the mutation P301L (Tau-P301L) were investigated on pure POPS or DOPS bilayers. Various biophysical techniques were used to probe the interaction between PS bilayers and Tau-P301L protein at the micro and nanoscale to better decipher the mechanisms by which Tau might induce damages on lipid membranes and aggregate in close contact of the membrane. Thioflavin T fluorescence assays and transmission electron microscopy confirmed the formation of amyloid fibers in presence of POPS liposomes. Calcein leakage assays showed a disruptive effect of Tau-P301L on POPS liposomes. Polarized attenuated total reflection Fourier-transform infrared spectroscopy (ATR-FTIR) spectroscopy and Plasmon waveguide resonance (PWR) revealed the accumulation of Tau-P301L on supported POPS membranes. Finally, Atomic Force Microscopy (AFM) revealed different morphologies of aggregated Tau P301L in contact with supported DOPS membranes.
Cells were initially inoculated into 1 L of LB medium with kanamycin supplementation at 50 μg mL−1. Cultures were maintained at 37 °C with shaking at 200 rpm until they reached an OD600 of 0.6–0.8. Protein expression was induced by adding 1 mM isopropylß-D-488 thiogalactoside (IPTG, Sigma Aldrich), followed by further incubation at 37 °C and 200 rpm for 3 hours. Cells were harvested by centrifugation at 5000g for 20 minutes at 4 °C. The resulting cell pellets were resuspended in a lysis buffer (50 mM Tris-HCl pH 7.4, 100 mM NaCl, 10 mM Imidazole, 0.5 mM DTT, 0.1 mM EDTA), supplemented with 1 mM PMSF, 20 μg ml−1 DNase, 10 mM MgCl2, and 1 Pierce protease inhibitor tablet (Thermo Fisher EDTA-Free, Sigma 88266), and subsequently stored at −20 °C for later use. Cell lysis was initiated with the addition of lysozyme (2 mg ml−1), and the cell suspension was agitated on an orbital shaker at room temperature for 30 minutes. To disrupt the cells, a freeze–thaw cycle was repeated three times using liquid nitrogen. After cell debris removal by centrifugation at 10000 rpm for 10 minutes at 4 °C, the supernatant was heated at 90 °C for 12 minutes, followed by cooling on ice for 20 minutes before another centrifugation at 10
000 rpm for 10 minutes at 4 °C to eliminate precipitated proteins. The supernatant, containing Tau protein, was loaded onto a 5 ml Ni-NTA agarose column pre-equilibrated with buffer A (20 mM sodium phosphate pH 7.0, 500 mM NaCl, 10 mM imidazole, 100 μM EDTA). The column was washed successively with 10 column volumes (CV) of buffer A, 8 CV of buffer B (20 mM sodium phosphate pH 7.0, 1 M NaCl, 20 mM imidazole, 100 μM EDTA), and 2 CV of buffer A before elution with 8 CV of buffer C (20 mM sodium phosphate pH 7.0, 500 mM imidazole, 0.5 mM DTT, 100 mM NaCl). Elution progress was monitored using absorbance at 280 nm and SDS-PAGE. Peak fractions were combined and concentrated through centrifugal filters (MWCO 10 kDa, Sigma) for subsequent gel filtration chromatography on a Superdex 200 pg HiLoad 16–600 column equilibrated with a final buffer (20 mM HEPES pH 7.0, 100 mM NaCl). Fractions containing purified protein were further concentrated, and the final protein concentration was determined via UV-vis absorption at 274 nm, employing an extinction coefficient of 7.5 cm−1 mM−1. The proteins were aliquoted into 100 μL tubes and stored at −20 °C for future use.
Small unilamellar vesicles (SUVs), used for aggregation and supported lipid bilayers (SLB) formation, were obtained by sonication of the lipid suspensions until clarity was achieved.
For calcein leakage experiments, large unilamellar vesicles (LUVs) containing calcein were obtained using lipid films of pure POPS, re-suspended in 20 mM HEPES pH7, with 70 mM calcein, and subjected to five freeze/thaw cycles. Subsequently, lipid suspensions were extruded 21 times using an Avanti Polar Lipids extruder set, through a membrane with a pore size of 200 nm. To separate LUVs containing calcein from the free dye, samples were passed through a Sephadex G50 size exclusion column that had been pre-equilibrated with 20 mM HEPES pH7 150 mM NaCl. The concentration of phospholipids was then determined using phosphorus quantification according to a previous protocol.34 The LUVs containing calcein were then stored at room temperature and protected from light for a maximum of 2 weeks.
The sensor consisted of a prism coated with a 50 nm-thick layer of silver and overcoated with a 460 nm-thick silica layer. The sensor, in direct contact with the sample cell, was positioned on a rotating table supported by a corresponding motion controller (Newport, Motion controller XPS; ≤1 mdeg resolution).
To form a supported lipid bilayer (SLB) on the sensor's surface, we introduced 250 μl of a POPS SUV suspension at a concentration of 2 mg ml−1, supplemented with 1 mM CaCl2 (freshly added), into a Teflon compartment in contact with the silica-coated prism (cell sample compartment). The spontaneous formation of a lipid bilayer occurred after 30-minutes liposome incubation period, after which the cell was rinsed to remove excess of lipids. The absence of angle shift in the TIR angle after rinsing when comparing with the angle observed with buffer, confirms that all excess liposomes has been removed from the bulk.
Subsequently, Tau-P301L was introduced into the cell incrementally from 0.1 to 1 μM. As the sensor is mounted parallel to the cell sample, there is no problem regarding potential protein settling on the sensor surface. The modifications in the plasmon resonance characteristics reflecting protein/lipid interactions were monitored over time. Here we mainly focus on the shift in the resonance angle position that is correlated with mass deposition, reflecting the quantity of bound protein.
Prior to sample measurements, background spectra for both s and p polarizations were recorded. Subsequently, 20 μl of buffer (20 mM Hepes at pH 7 with 100 mM NaCl) was added to the liquid cell created using an o-ring, and the buffer spectra were recorded for both s and p polarizations for later subtraction from the sample spectra.
20 μl of a POPS SUV suspension at a concentration of 1 mg ml−1, in buffer supplemented with 1 mM CaCl2, were deposited and incubated for 5 minutes to facilitate SLB formation. The SLB was then rinsed three times by the buffer to remove any remaining SUV. Polarized ATR-FTIR spectra of the SLB were recorded before Tau-P301L addition, as well as after a 3 h incubation, either alone for control conditions, or with 1 μM of Tau-P301L directly added inside the liquid cell. Final s and p polarized ATR-FTIR spectra after incubation were collected following three rinses with buffer. Each experiment was conducted three times.
Polarized ATR-FTIR spectra of SLB alone and Tau protein interacting with the SLB were processed using Omnic Software (Thermo Fisher Scientific), by subtracting the buffer spectrum at each polarization. Additionally, the baseline was corrected.
AFM imaging was performed using the Dimension FastScan setup (Bruker) operating in PeakForce Quantitative Nano-Mechanics (PF-QNM) mode in liquid environment at room temperature (∼20 °C). SNL-C probes (Bruker, Silicon tips on silicon nitride cantilevers) were individually calibrated for each experiment, with a nominal spring constant of 0.24 N m−1, a resonance frequency of 56 kHz, and a tip radius ranging from 2 to 12 nm. SLB were imaged with a scan rate of 0.8–1 Hz, a resolution of 256 × 256 pixels, and a constant setpoint force of approximately 0.5 nN. To ensure the stability of the SLB, an area was imaged for 1 to 3 hours before introducing the protein at the desired concentration directly inside the liquid droplet. Images were recorded immediately after the injection of Tau every 3–5 minutes until the system stabilized, typically within 1–2 hours. Different concentrations were tested to maintain consistency with other experiments in terms of protein/lipid ratio, and 400 or 500 nM were chosen to enable the observation of defects without introducing a mechanism that is too rapid. Image analysis was carried out using Gwyddion, and the sizes were determined using the cross-section tool. Each experiment was repeated 3 times.
The presence of fibers was confirmed by transmission electron microscopy (TEM, Fig. 1C and D). Characteristic amyloid fibers were observed in the presence of both heparin and POPS, displaying long filaments, straight or twisted. No discernible differences in size or morphology were noted based on the cofactors. AFM imaging was used in buffer to characterize the morphologies of POPS-induced Tau-P301L filaments in more details (Fig. 1E). Fibers present lengths of several micrometers, with straight and twisted fibers (Fig. 1E for cross-sections of low and high points of a fiber), as observed for heparin-induced Tau 2N4R fibers.38
As a reference, the aggregation behavior of Tau-WT was also studied (Fig. S1†). In accordance with the results obtained for Tau-P301L, the same increase in fluorescence without lag-time was observed for Tau-WT incubated with heparin, with a half-time of 3 hours but an intensity 3 times lower compared to Tau-P301L. A very low fluorescence intensity was observed with POPS with a lag time of 7 hours, and the maximum fluorescence was threefold lower than that of Tau-P301L, and 4.5 times lower compared to the heparin condition. Furthermore, a greater abundance of fibers was observed in both conditions for Tau-P301L compared to Tau-WT, consistent with the mutant's heightened propensity for aggregation.39 These experiments confirm that POPS is capable of inducing the aggregation of Tau-P301L protein, leading to the formation of fibers. To delve deeper into POPS-induced Tau aggregation and POPS-Tau interactions, we explored the effect of Tau-P301L monomer on POPS lipid membranes.
![]() | ||
Fig. 2 Calcein leakage of POPS liposomes (10 μM) after 3 h of incubation with increasing concentrations of Tau-P301L. Error bars represent standard deviation over 2 repeats. |
A POPS SLB alone shows the antisymmetric and symmetric stretching modes of CH2 groups νas(CH2) and νs(CH2) at 2923 cm−1 and 2850 cm−1 respectively, and the νas(CH3) and νs(CH3) modes less intense appear as shoulders at 2950 cm−1 and 2870 cm−1, respectively. The position of the νas(CH2) at 2923 cm−1 indicates disordered chains as expected for fluid membranes, and a value of absorbance at 1.6 × 10−3 (for p-polarization) is consistent with the formation of a single bilayer on the surface of the ATR crystal.40 The band at 1740 cm−1 assigned to the ν(CO) ester group of the phospholipid (Fig. 3, grey spectra) also confirms the formation of a lipid bilayer. As a control, the stability of the supported POPS bilayers was evaluated by following the absorbance of the band at 2923 cm−1 over time (Fig. 3 inset, grey column). After 3 hours, only a small decrease can be observed, revealing bilayer stability over this timescale.
After 3 hours of incubation with Tau-P301L (Fig. 3), both the absorbance of νas(CH2) and ν(CO) ester bands decreased but in the same way as without protein (Fig. 3), with 80 ± 8% of lipids remaining after Tau incubation, compared to 87 ± 16% in the control condition. This suggest that the integrity of the membrane is not significantly affected by its interaction with the Tau protein. The accumulation of Tau-P301L on the POPS membrane is revealed by the observation of the Amide I and Amide II bands centered at 1640 and 1550 cm−1 respectively (Fig. 3). The broad band centered at 1640 cm−1 indicates the presence of various elements of secondary structure, with a high percentage of random coil, as expected for the intrinsically disordered protein (IDP) Tau protein in buffer.41 Upon increase of the protein concentration to 5 μM (see Fig. S2†), similar effect was observed with no diminution of lipid content, suggesting the absence of membrane damage. An increase in the amount of adsorbed protein was however revealed by the increase in intensity of the amide II band (×4). Moreover, the dichroic ratio (Ab νsCH2 p-pol/Ab νsCH2 s-pol) did not vary with or without the addition of the protein, indicating that the protein did not deeply insert into the lipid membrane.
Overall, ATR-FTIR results suggest that Tau-P301L protein accumulates on the POPS without strong damage on the membrane and without strong modifications of the Tau-P301L structure in a 3 hours timescale.
Taken together, ATR-FTIR and PWR data suggest that Tau-P301L accumulates on the POPS surface with a strong binding to the phospholipid surface, without inserting or damaging the lipids present in the supported bilayer.
The deposition of a DOPS SLB resulted in a homogeneous bilayer with an approximate thickness of 3.0 ± 1.1 nm, as illustrated by the cross-section of a defect in Fig. 5A and D. To ensure the stability of the SLB, areas of the DOPS bilayer were imaged over time to monitor any time-dependent disruption and ensure that no AFM tip-induced disruption occurred (Fig. S6†). DOPS bilayers remained stable over the experimental timescale of 1 to 3 hours, showing no additional holes or significant modifications.
Fig. 5 presents a sequence of images obtained for the interaction of a DOPS bilayer with 400 nM of Tau-P301L as a function of time. Following the addition of Tau-P301L in the experimental volume, significant changes were immediately observed. After 6 min (Fig. 5B and E), pre-existing defects in the bilayer have widened, suggesting partial solubilization of the membrane at the edges of the initial defects. Interestingly, after incubation with Tau-P301L, the height difference between the mica and the remaining lipid layer is only about 1.5 nm, rather than the expected 3–3.5 nm for the SLB thickness, as shown in Fig. 5B and E at the location of the green arrowhead. This height would be consistent with a monolayer of lipids, hinting at membrane thinning and lipid removal. In addition to an apparent membrane disruption and loss of lipids, the rapid apparition of both intermediate height and high large flat patches of approximately 7.8 ± 0.5 nm and 16.5 ± 0.8 nm thickness was observed, depositing onto the remaining lipid layer, and reminding of multiple lipid bilayers.46 This suggests a possible re-deposition of solubilized phospholipids, associated or not with the protein on the mica and remaining lipid layer. Observed heights for the flat patches would be consistent with the deposition of 2 to 5 bilayers. Over time, the deposited patches appear to break and reorganize, as smaller patches were visualized after 2 hours (Fig. 5C). The major effect of Tau-P301L on the membrane occurred within the first 30 minutes of interaction and subsequently stabilized rapidly without further changes (Fig. 5C).
The observed effect on the DOPS surface appears to be correlated with the concentration of Tau-P301L added to the bilayer surface. Higher concentrations of Tau-P301L led to comparable effects regarding membrane disruption as well as deposition of flat thick patches, but different morphologies in the deposited patches were observed. Fig. 6 presents a sequence of images happening before and after addition of 500 nM of Tau-P301L to a DOPS bilayer. Initially, the SLB appeared homogeneous (Fig. 6A). After 15 min of incubation with Tau-P301L at 500 nM, thick flat patches of 15–20 nm were observed, but also elongated structures that seem to extend from the patches (see green arrowheads on Fig. 6B and C). Moreover, these filamentous structures exhibited growth over short timescales, with significant growth in 8 minutes evidenced by arrowheads on Fig. 6C. This suggests that these filaments could be Tau aggregates in the process of fiber formation.
This concentration-dependent effect was also noted with the incremental addition of proteins, with defects widening at 300 nM with holes consistent with the removal of only a monolayer of lipids (Fig. S7†), patch redeposition at 400 nM (Fig. 5 and Fig. S8B†), and the growth of fibrillar structures at 500 nM (Fig. 6 and Fig. S8C†). Two types of filaments were observed, presenting thicknesses of ∼5 nm (Fig. 6B, C, E and F) or ∼13 nm (Fig. S8D and E†). In comparison to POPS-induced Tau filaments obtained in solution, that showed a thickness of ∼17 nm (Fig. 1E), the filaments observed in interaction with DOPS supported bilayers appear to be much thinner. However, these lower thicknesses could be consistent with protofibrils, that could assemble into thicker filaments later in the fibrillization process, as was observed previously for heparin-induced filaments.38
Overall, AFM data suggest that the incubation of Tau on the anionic SLB leads to partial solubilization of the membrane, with patchy aggregates deposition on the remaining bilayer, and eventual growth of fibrillary structures extruding from the patches.
Our study aims at providing a comprehensive answer to all these questions for the interaction of P301L-Tau, a more aggressive form of full-length Tau, with negatively charged lipids. As the inner neuronal membrane contains a high content of POPS of about 10–20% of its total lipid content,20,49,50 we chose to focus on this anionic lipid. An array of biophysical methods was used to study the interaction at the global and local scale, to investigate whether Tau-P301L is capable of interacting with the POPS with focus on the strength of the interaction and the mechanisms involved, and to assess if this interaction can impact the lipid organization of the membrane.
Finally, in previous studies, structuration was observed only for shorter constructs of Tau (K18, K19, P2R…).12,23,25–27 However, the full-length Tau protein, particularly 2N4R, possesses a significant portion of disordered regions.2,51 Our results show thatTau-P301L in interaction with POPS is mainly in random coil (IDP) and accumulates on the surface of the POPS membrane. Adsorption could take place via 3 small α helices, as shown for K19 interacting with PS-containing membranes, but it is impossible to detect this in the midst of a large majority of unstructured amide groups.
Indeed, our ATR-FTIR and PWR results support the idea that P301L induced perturbation rather than destruction of the POPS membrane. ATR and PWR data point also to a strong binding and accumulation of the protein on the bilayer surface, but without significant membrane disruption or lipid loss. How can we explain the discrepancy between what happens on a flat lipid membrane vs. what happens with lipid vesicles? One could invoke membrane curvature. For α-synuclein, the effect of membrane curvature was studied with varied lipid vesicle compositions,52 revealing a much stronger affinity of the protein for small vesicles of POPS:
POPC (1
:
1) compared to large ones. If SUV present more packing defects that LUVs,53 we could attribute this difference in interaction to the accessibility of the hydrophobic tails of the lipids, made more easily available in a vesicular context. In the case of positively charged P301L-Tau, it would be consistent with a shift of the interaction mechanism from an electrostatic to a hydrophobic interaction. However, some local membrane disruption was also observed by AFM imaging. The fact that we do not observe pore-like structure tends to suggest that the membrane leakage observed in liposomes is linked to accumulation of peptides on their surface.
AFM allowed us to go further in the understanding of the processes at play, by observing the local effects of Tau-P301L interaction with PS membranes at the nanoscale, using DOPS bilayers. Though differing in one of their fatty acid chains, with 1 additional unsaturation and 2 additional carbons for DOPS compared to POPS, both lipids share the same anionic headgroup and are fluid at room temperature, with melting temperatures of −11 and 14 °C for DOPS and POPS, respectively.43 Polarized ATR-FTIR data (Fig. S3†) confirms that both DOPS and POPS bilayers are fluid in our experimental conditions. We also showed (Fig. S4†) that both bilayers behave similarly in interaction with Tau-P301L, thus validating the use of DOPS for AFM experiments. Our AFM results suggest that Tau-P301L has two effects on DOPS membranes, the disruption of the supported membrane, and a simultaneous self-assembly on the DOPS surface.
The bilayer solubilization effect observed in our experiments is consistent with previous reports for the interaction of shorter Tau fragment K18 with mixed POPC:POPS or POPC:PIP2 membranes.29 In the case of membranes containing PIP2 especially, large portions of the membrane showed similar thinning as in Tau-P301L/DOPS interaction after incubation with the protein.29 The lower thickness observed locally after incubation with the protein could be consistent with a remaining monolayer of lipids in both cases. When defects are present in the bilayer, solubilization seems to be starting at the defects, supporting a hydrophobic interaction with the hydrophobic tails of the lipids. However, we note that in some cases, holes are forming upon protein addition in areas of the bilayer that did not contain defects (Fig. S7†).
Regarding Tau self-assembly on the DOPS surface, two types of aggregates are observed, patches and fibrillary structures originating from the patches. Our findings suggest that the morphology of the deposit is concentration-dependent, and that varying concentrations of Tau can elicit different effects on the membrane surface, with fibers only observed at concentrations of 500 nM and higher. Flat patches of amorphous aggregates of 2N4R Tau have been previously reported on membranes composed of brain total lipid extracts by Mari et al., for concentrations ranging from 100 to 300 nM.55 The surface area of the reported “flat islands” increases over time and is dependent on the concentration of Tau. However, the reported patches were much thinner than the ones we report, with thicknesses of ∼2 nm, and showed the presence of protrusions. This could be partly explained by the lipid composition of the supported membranes considered. In the present study, we use 100% DOPS, whereas Mari et al. used brain total lipid extract, which although overall negatively charged is heterogeneous in composition. Therefore, a charge-mediated interaction of Tau with membranes would be much stronger in fully anionic membranes, and could lead to bigger aggregates. Our observations of fibers on PS membranes is consistent with previous observations of K18 interactions on POPC:
PIP2 (4
:
1) SLB,29 where fibers appeared after 6 hours of incubation and also seemed to emerge from patchy aggregates on the membranes. However, fibers appeared much faster in the case of Tau-P301L interacting with DOPS, i.e. within 15 minutes of incubation, with subsequent observable fiber growth in a 1-hour timescale, although the concentration of protein was 2 times lower (500 nM vs. 1 μM in a previous study29). The faster growth we observed in the case of Tau-P301L could be attributed to the use of pure anionic membranes in our case, which could accelerate the impact of Tau on the membrane, as opposed to mixed SLB with zwitterionic lipids.29
Our observations that fibers assemble on negatively charged lipid surfaces that show membrane disruption and potential lipid loss suggests that lipid removal and fiber growth are two linked processes, which would agree with previous reports indicating recruitment of PIP2 inside a K18 fiber.56
The incubation of Tau-P301L with small liposomes of POPS resulted in the aggregation of Tau, leading to the formation of amyloid fibers. This aggregation was closely associated with liposome leakage, that could be attributed to a different interaction mechanism due to membrane curvature, potentially linked to easier access to hydrophobic tails of lipids in a vesicular context. Through the utilization of ATR-FTIR and PWR techniques, which provide a global scale assessment of the sample, we observed a robust and irreversible binding of Tau to the POPS surface without the emergence of microdomains, insertion in the bilayer, or lipid removal. This binding appears to create a homogeneous and thick layer of protein on the surface, resembling the carpet mechanism (Fig. 7A) previously identified in other amyloids.57,58 Carpeting is consistent with the morphologies observed by AFM imaging on DOPS bilayers, where the formation of large flat patches was observed. Additionally, we found no evidence of a predominant protein structure, indicating the potential coexistence of different species. This potential polymorphism was further confirmed at the nanoscale by AFM, where both thick patches and fibrillary structures were observed. The fibrillary forms resembled amyloid fibers, and appeared to extend and grow from the flat patches at higher concentrations of Tau-P301L (Fig. 7B). Considering the role of the anionic membrane in interacting with Tau, it has the ability to concentrate Tau at its surface, possibly triggering various disruptive mechanisms that result in the extraction of anionic lipids and the accumulation of polymorphic aggregates possibly composed of both proteins and lipids.
We propose that Tau-P301L interacts with PS-containing membranes by first establishing electrostatic interactions, that leads to accumulation of the protein in a thick layer without insertion. Carpeting leads to partial membrane disruption, a process that depends on membrane curvature, possibly accelerated by hydrophobic interactions with the lipid tails. Because ATR-FTIR did not show lipid depletion, we propose that the deposited flat patches could be a mixed composition of solubilized phospholipids and proteins, which could reconcile ATR-FTIR and AFM results, considering the timescales of both experiments. Above a threshold concentration of Tau-P301L, the aggregate structure shifts towards fibrillary structures that would be consistent with amyloid fibers, and would be consistent with the observed amyloid fibers obtained after incubation of Tau-P301L with small liposomes of POPS. Such objects are extremely interesting, both in secondary structure and chemistry, and a nanoscale investigation using AFM coupled to infrared spectroscopy could provide complementary information to confirm these fibers hold a beta-sheet content consistent with amyloids, and assess whether they contain lipids.
M. M., C. F. and S. L. managed the project. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4nr01343c |
This journal is © The Royal Society of Chemistry 2024 |