A Functional Role for Intrinsic Disorder in the Tau-Tubulin Complex

A functional role for intrinsic disorder in the tau-tubulin complex

Ana M. Meloa,b, Juliana Coraorc,1,2, Garrett Alpha-Cobba,1, Shana Elbaum-Garfinklea,1,3, Abhinav Nathd, and Elizabeth Rhoadesb,4

aDepartment of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT 06520; bDepartment of Chemistry, University of Pennsylvania, Philadelphia, PA 19104; cDepartment of Physics, Yale University, New Haven, CT 06520; and dDepartment of Medicinal Chemistry, University of Washington, Seattle, WA 98195-7610

Edited by Gregory A. Petsko, Weill Cornell Medical College, New York, NY, and approved November 3, 2016 (received for review June 23, 2016) Tau is an intrinsically disordered protein with an important role microtubules (23, 24), suggesting possible mechanisms for the in maintaining the dynamic instability of neuronal microtubules. loss of microtubule stability. Despite intensive study, a detailed understanding of the func- Tau consists of a microtubule binding region (MTBR) composed tional mechanism of tau is lacking. Here, we address this of imperfect repeats R1–R4 (25), a flanking basic proline-rich re- deficiency by using intramolecular single-molecule Förster Reso- gion (P1 and P2) that enhances microtubule binding and assembly nance Energy Transfer (smFRET) to characterize the conforma- (9), and an N-terminal projection domain with putative roles in tional ensemble of tau bound to soluble tubulin heterodimers. microtubule spacing (26), membrane anchoring (27), and tau di- Tau adopts an open conformation on binding tubulin, in which merization (28) (Fig. 1). Alternative splicing results in the expres- the long-range contacts between both termini and the microtu- sion of six isoforms of tau in the adult human brain, with up to two bule binding region that characterize its compact solution struc- N-terminal inserts and three or four repeats in the MTBR. Each ture are diminished. Moreover, the individual repeats within the repeat consists of an interrepeat or linker region and the conserved microtubule binding region that directly interface with tubulin sequence. Contrasting models of tau in complex with microtubules expand to accommodate tubulin binding, despite a lack of exten- suggest binding on the outer surface either along or across proto- sion in the overall dimensions of this region. These results sug- filaments (29, 30) or inside the microtubule lumen (31). Both the gest that the disordered nature of tau provides the significant negatively charged C-terminal tails of tubulin (32) and the interface flexibility required to allow for local changes in conformation α β while preserving global features. The tubulin-associated confor- between / -subunits (33) have been identified as important for tau mational ensemble is distinct from its aggregation-prone one, high- binding. Although significant work has focused on this interaction, lighting differences between functional and dysfunctional states relatively little is known about the structural features of tau when of tau. Using constraints derived from our measurements, we con- associated with tubulin or microtubules. This lack of detail is, in struct a model of tubulin-bound tau, which draws attention to the part, because tau is intrinsically disordered in solution (34, 35) and importance of the role of tau’s conformational plasticity in function. appears to remain largely disordered even on binding, making its structural characterization challenging. single-molecule FRET | intrinsically disordered proteins | microtubule- Single-molecule Förster resonance energy transfer (smFRET) is associated protein | tauopathies | Alzheimer’s disease a powerful approach for probing conformations of large, dynamic proteins, including tau (22, 36). In addition to providing insight into icrotubules play an important role in a host of critical cellular structural features of such proteins, the picomolar concentrations of Mprocesses, including positioning of organelles, maintaining cell shape, axonal transport, and cell division (1–3). Their constit- Significance uent subunits, α/β-tubulin heterodimers, polymerize head to tail to form linear protofilaments that subsequently associate laterally into Tau is a neuronal microtubule-associated protein linked to nu- microtubules. Microtubules are dynamic structures with biological merous neurodegenerative disorders, including Alzheimer’sdis- functions that are dependent on their ability to grow and shrink in a ease. Several lines of evidence support tau aggregation as well temporally and spatially controlled manner in response to cellular as loss of its native interactions with microtubules as contrib- signals. Critical to microtubule function is that they undergo non- uting to pathology. Here, we explored the largely overlooked equilibrium phases of elongation and depolymerization, termed first step of microtubule assembly, namely the interaction of dynamic instability (4). Given the importance of dynamic instability to tau with soluble tubulin heterodimers. Using single-molecule microtubule function, it is highly regulated by a family of nucleotide- Förster Resonance Energy Transfer (smFRET), we determine the independent microtubule-associated proteins (5). topological features of tau in complex with tubulin. Our results Tau is a microtubule-associated protein first characterized as contrast differences in tau isoforms and underscore the im- a promoter of microtubule assembly both in vitro (6) and in vivo portance of conformational flexibility in tau function. (7). Numerous studies have helped provide details on its role in – Author contributions: A.M.M., J.C., G.A.-C., S.E.-G., and E.R. designed research; A.M.M., modulating microtubule growth dynamics (8 11). Tau is found J.C., G.A.-C., and S.E.-G. performed research; A.M.M., J.C., G.A.-C., S.E.-G., A.N., and E.R. primarily in the axons of neurons (12, 13). It is thought that tau analyzed data; A.N. carried out simulations; and A.M.M. and E.R. wrote the paper. may have an important role in regulating axonal transport (14, 15), The authors declare no conflict of interest. and it has been reported to inhibit the activity of several micro- – This article is a PNAS Direct Submission. tubule motor proteins (16 18). Aggregation of tau is linked to a 1J.C., G.A.-C., and S.E.-G. contributed equally to this work. number of neurodegenerative disorders termed tauopathies, 2Present address: Harvard Medical School, Cambridge, MA 02115. among them Alzheimer’s disease (reviewed in ref. 19). Addi- 3Present address: Department of Chemical and Bioengineering, Princeton University, tionally, the loss of native, functional interactions between tau Princeton, NJ 08544. and microtubules and subsequent destabilization of microtu- 4To whom correspondence should be addressed. Email: [email protected]. bules are thought to contribute to pathology (20, 21). Mutations edu. linked to disease have been observed to enhance binding of This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. tau to soluble tubulin (22) as well as disrupt interactions with 1073/pnas.1610137113/-/DCSupplemental.

14336–14341 | PNAS | December 13, 2016 | vol. 113 | no. 50 www.pnas.org/cgi/doi/10.1073/pnas.1610137113 Downloaded by guest on October 2, 2021 (tau2N4R17/244)(Fig.1C) and the C terminus and MTBR (tau2N4R354/433)(Fig.1D). Importantly, these results show that the surprisingly close contact between the N and C termini of tau that is observed in solution (36, 37) is lost on binding to tubulin (Fig. 1A). Similarly, both termini, which also exhibit a relatively close interaction with the MTBR in solution (36), extend away from this central domain in the tubulin-bound state (Fig. 1 C and D), re- vealing that tau adopts an overall open structure under these conditions. Subsequently, the loss of these long-range interactions exposes the MTBR and its flanking proline-rich region, promoting binding to tubulin.

Local Rearrangements Throughout the MTBR. Because the MTBR directly interfaces with tubulin, constructs spanning this region and overlapping all four repeats within this region of tau2N4R were measured (Fig. 2 A–F). The construct tau2N4R244/354 spans nearly the entire MTBR; it shows only a small change in ETeff on binding to tubulin (Fig. 2F). The lack of change is striking for two reasons. The first is that our previous smFRET measurements of the K16 fragment of tau (consisting of the proline-rich region and MTBR) labeled at the same positions found an expansion of this region on Fig. 1. Schematic of tau2N4R (upper schematic) and tau2N3R (lower sche- ΔET ∼ matic). The functional domains of tau indicated are the projection domain, binding to tubulin (for tauK16, eff 0.07 with the addition of the proline-rich region, and the MTBR. Alternative splicing results into six tubulin) (22). However, in solution, this region of tau is more different tau isoforms (spliced regions marked with dashed lines). Tau2N3R compact in the protein fragment than in the full-length protein is lacking R2. The residues labeled for smFRET measurements are indicated [for tauK16244/354, ETeff = 0.40 (22); for tau2N4R244/354, ETeff = below the schematic, with all numbering is based on tau2N4R. (A–D) smFRET 0.32 ± 0.006]. The tubulin-bound conformations of both K16 and histograms are in the absence (gray) and presence of 10 μM tubulin (blue tau2N4R are comparable (ETeff ∼ 0.33), suggesting that the pre- and red for tau2N4R and tau2N3R, respectively). (A–C) For constructs that viously observed expansion for tauK16244/354 may reflect the display a mean ETeff close to zero in the presence of tubulin, the histograms were obtained by alternating laser excitation to exclude the contribution of compaction of this region in the protein fragment in the absence donor only-labeled protein (S < 0.8) (SI Materials and Methods). The car- of the flanking N and C termini. Another reason is that the con- toons illustrate the labeling positions of tau. structs probing interrepeat distances within MTBR all show shifts to lower mean ETeff values on tubulin binding, indicating local expansion of these regions (Fig. 2 A–E), despite the lack of change protein required for measurements inhibit both tau self-association in the MTBR overall (Fig. 2F). and tau-mediated polymerization of tubulin (22, 36). Here, we use The subdomain constructs spanning two, three, or all four of the intramolecular smFRET to determine topological features of tau repeats were designed to directly probe the relative relationships bound to soluble tubulin heterodimers. By independently probing between the four repeat regions, enabling us to ascertain local multiple segments within tau, we investigate domain-specific con- conformational features. Although each of the subdomain conformational changes within the context of the full-length protein that structs adopts a slightly more extended conformation on binding provide insight into tau function. We find that binding to tubulin tubulin, the magnitudes of the shifts are not the same for all (Table requires an expansion of the relatively compact ensemble observed 1). To compare the extent of these changes, the mean ETeff values for tau in solution, because long-range interactions between the were first converted to root-mean-square (rms) end-to-end dis- termini and the MTBR are eliminated.Furthermore,bindingisac- tances using a Gaussian chain polymer model (38). We note that companied by local expansions of individual repeats within the the conversion of ETeff values to meaningful distances for a sys- MTBR without changing the overall dimensions, suggesting signifi- tem as complex as tau-tubulin is challenging. Indeed, for several of cant flexibility within this domain in its bound state. These results our constructs, shoulders are apparent in the ETeff histograms on were used to create a model of tubulin-bound tau, which illustrates binding to tubulin (Fig. 2), whereas our analysis is derived from how tau’s conformational plasticity enables accommodation of mul- fitting the main peak of the histogram. These asymmetries likely tiple tubulin dimers. reflect the heterogeneous nature of the tau-tubulin complex, which could reflect tau binding to different numbers of tubulin dimers or Results different topologies of tau in its bound state. Moreover, prior work Loss of Interactions Between the N and C Termini. The longest iso- from our laboratory (39) indicates that tubulin-bound tau likely form of tau, 2N4R (Fig. 1), was labeled at 11 sets of residues with consists of interspersed structured and disordered domains, further donor (Alexa 488) and acceptor (Alexa 594) fluorophores. complicating this calculation. On the basis of comparisons of ide- smFRET measurements were made in the absence and presence of alized α-helical and Gaussian coil end-to-end distances (Fig. S3), tubulin heterodimers (SI Materials and Methods). The smFRET we estimate that our use of an analytical polymer model is unlikely histograms from measurements of tau constructs that probe the to introduce significant errors in distance calculations: the end-to- ∼ α

long-range interactions between the N and C termini and between end distance of an 15-residue -helical segment is comparable BIOPHYSICS AND each of the termini and the MTBR are shown in Fig. 1 A, C,andD. with that of a Gaussian random coil of similar length (39). Fur- COMPUTATIONAL BIOLOGY In the presence of 10 μM tubulin, where tau is essentially com- thermore, to account for asymmetries or shoulders in the smFRET pletely bound (Figs. S1 and S2), large shifts toward lower mean histograms, we also compared the (nonparametric) arithmetic ET energy transfer efficiencies ( eff) were observed for each of these means of the ETeff values in a peak with the peak positions esti- constructs. For tau2N4R17/433, spanning nearly the entire length of mated from Gaussian fits (SI Materials and Methods). The directly ET ± the protein, the mean eff shifts from 0.24 0.006 in the absence calculated mean ETeff values and thus, the calculated distances, of tubulin to <0.1 in the presence of tubulin (Fig. 1A). This overall differ only slightly from the fit-derived peak positions (Table S1), expansion is also reflected in shifts in ETeff values for constructs suggesting that Gaussian fitting does not introduce large errors in probing the relationship between the N terminus and MTBR the distance calculations, even for asymmetric distributions. As

Melo et al. PNAS | December 13, 2016 | vol. 113 | no. 50 | 14337 Downloaded by guest on October 2, 2021 Fig. 2. smFRET histograms of constructs probing the MTBR of (A–F) tau2N4R and (G–I) tau2N3R in the absence (gray) and presence of 10 μM tubulin (blue and red for tau2N4R and tau2N3R, respectively). The bar plot shows quantification of the change in conformation of the MTBR on binding to tubulin (%Δrms/ residue). Constructs spanning two repeats show greater change per residue than those probing three or four repeats for both tau2N4R (blue) and tau2N3R (red). The two repeat-spanning constructs with the largest per-residue change all contain R3, a region that we have suggested is particularly important for tau binding to soluble tubulin. The brackets identify the number of repeats probed.

such, all distance analysis discussed here was based on the Gaussian These data suggest that, although the individual repeat polymer model. sequences undergo an expansion to accommodate binding to The percentage changes in rms in the presence of tubulin tubulin, there is significant flexibility between the bound regions, relative to the absence of tubulin (%Δrms) were calculated and which allows for accommodation of these local changes without normalized by the number of residues between the probes in causing a significant change in the global features of the MTBR. each construct (SI Materials and Methods).AsshowninFig.2, Moreover, the greater expansions in repeats R2–R3 and R3–R4 the %Δrms/residue values are largest for two repeat-spanning are compatible with our previous results, which supported R3 in constructs, with the greatest changes in tau2N4R291/322 and combination with R2 or R4 as the primary regions responsible for tau2N4R322/354 probing R2–R3 and R3–R4, respectively. The binding to soluble tubulin (22). remaining adjacent repeat probe, tau2N4R244/291 (R1–R2), shows The proline-rich region of tau N-terminally flanks the MTBR significantly smaller change than these other two constructs, more (Fig. 1) and significantly increases both microtubule binding comparable with the change seen in the three repeat constructs affinity and assembly activity of tau (9). A construct that spans the tau2N4R244/322 and tau2N4R291/354 (R1–R3 and R2–R4, respectively). entire proline-rich region (P1 and P2) (Fig. 1), tau2N4R149/244,

Table 1. ETeff and distance calculations for tau in the absence and presence of tubulin +Tubulin No. of † Construct residues ETeff rmsexp (Å) rmsRC*(Å) ETeff rmsexp (Å) %Δrms/residue

2N4R 17/433 417 0.24 ± 0.006 93 181 <0.1 NA NA 17/244 228 0.65 ± 0.003 51 126 <0.1 NA NA 354/433 80 0.56 ± 0.010 57 67 0.47 ± 0.018 65 NC 149/244 96 0.39 ± 0.004 72 75 0.31 ± 0.004 82 NC 184/291 108 0.32 ± 0.010 80 80 0.31 ± 0.011 82 NC 244/354 111 0.32 ± 0.006 80 82 0.33 ± 0.002 79 −0.01 244/322 79 0.41 ± 0.002 70 67 0.37 ± 0.016 74 0.07 291/354 64 0.55 ± 0.004 58 59 0.51 ± 0.003 61 0.08 244/291 48 0.58 ± 0.003 56 49 0.54 ± 0.010 59 0.11 291/322 32 0.78 ± 0.001 42 39 0.72 ± 0.040 46 0.30 322/354 33 0.80 ± 0.001 41 39 0.74 ± 0.025 45 0.30 2N3R 17/433 386 0.23 ± 0.005 95 172 <0.1 NA NA 244/354 80 0.42 ± 0.006 69 67 0.42 ± 0.011 69 0.00 244/322 48 0.62 ± 0.005 53 49 0.56 ± 0.017 57 0.16 322/354 33 0.79 ± 0.003 41 39 0.73 ± 0.020 45 0.30

The mean peak ETeff values in the absence and presence of tubulin converted to rms distances by a polymer model (as described in SI Materials and Methods) are listed. Construct denotes the positions of the fluorophores. Errors denote the SD from the mean of at least three independent measurements. NA, not applicable; NC, not calculated.

*rmsRC is the expected rms for the measurements in the absence of tubulin if tau had the properties of an ideal random coil in a good solvent calculated as described in SI Materials and Methods. †%Δrms/residue is the relative change in the dimensions of tau on binding to tubulin normalized to the number of residues between the fluorophores.

14338 | www.pnas.org/cgi/doi/10.1073/pnas.1610137113 Melo et al. Downloaded by guest on October 2, 2021 Binding to tubulin opens up the compact “S” (36) or “paper- clip” (37) ensemble observed in solution, exposing binding sites within the MTBR. Specifically, the long-range interactions between each of the termini and the MTBR are lost. This expansion is compatible with structural studies, which found that both termini of tau extend away from the microtubule surface when bound (46, 47). Interestingly, the same type of expansion was Fig. 3. Histograms from smFRET measurements of constructs probing the observed previously for tau bound to the molecular aggregation entirety of the proline-rich region (A) or only P2 (B) of tau2N4R in the ab- inducer, heparin (36), suggesting that the open conformation of μ sence (gray) and presence of 10 M tubulin (blue). tau is important both for function as well as in dysfunction. It is also of note that the N terminus of tau, which has the strongest propensity toward intrinsic disorder of all tau domains (36), shows significant shift in mean ET values on binding tubulin eff maintains its extended, disordered character when bound to tu- (Fig. 3A)(ΔET = 0.08). This result shows that, even in the ab- eff bulin (Fig. S4). Our observation supports NMR studies, which sence of a direct interaction between the proline-rich region and suggest that the N terminus retains high flexibility in the tau- tubulin, there is a conformational change in this region relevant microtubule complex (33, 35). We find key differences between to binding and function. In contrast, a construct encompassing the heparin- and tubulin-bound tau conformations as well, P2, the portion of the proline-rich region adjacent to the MTBR showing that, although the binding sites for heparin and tubulin tau2N4R , undergoes only a slight change in conformation on 184/291 or microtubules may be the same (40), the conformational binding tubulin (Fig. 3B)(ΔET = 0.01). P2 has been proposed to eff changes on binding are not simply the result of tau binding to play a key role in binding to microtubules (40), and our data suggest anionic molecules or surfaces but are distinct conformational that binding to tubulin does not require a significant conformational differences between the aggregation-prone and functional forms change in this region. of tau. Although heparin binding is accompanied by a pro- nounced compaction of the MTBR (36), tubulin binding results Comparison of 2N3R and 2N4R Isoforms. Tau isoforms containing F I three repeats in the MTBR (lacking R2) (Fig. 1) have lower in no significant change (Fig. 2 and for tau2N4R and microtubule binding affinities (9, 11) and assembly activity (41) and tau2N3R, respectively). The appearance of different conformational features in the presence of different binding partners is stabilize microtubules more weakly than tau2N4R (42), suggesting consistent with the concept that tubulin may effectively serve as functional and potentially structural consequences of R2. Addi- a chaperone by binding to excess free tau and consequently, tionally, several mutations associated with tauopathies disrupt the inhibiting tau aggregation. ratio of 3R and 4R isoforms, suggesting that dysregulation of the Our focus on constructs that probe the relationship between expression of tau isoform levels is relevant to pathology (43, 44). specific repeat regions within the MTBR allows us to develop a In solution and similar to tau2N4R (36), tau2N3R displays a rel- more detailed picture of this functionally important region. In a atively compact structure, with a significantly higher mean ET eff previous study, we hypothesized that the primary binding site within than expected (Table 1). The construct tau2N3R has a mean 17/433 tau for tubulin includes R3 (22). This hypothesis is supported by our ET = 0.23 ± 0.005 (Fig. 1B); detectable energy transfer for these eff results here, where greater extensions are observed for constructs probes separated by 386 residues suggests relatively close contact including R3 than for those lacking it (Fig. 2). We expect tau to be between the protein termini. Additionally, as observed for associated with multiple tubulin dimers under the conditions of our tau2N4R, the MTBR is relatively extended in solution, with di- measurements (39). For tau2N4R, R2 or R4 (or potentially both) mensions comparable to a random coil in a good solvent (36) SI Materials and Methods provides additional binding sites; for tau2N3R, which lacks R2, the (Table 1 and ). additional binding site is R4. It is worth noting that a tubulin On binding to tubulin, tau2N3R becomes significantly more ex- B ET < binding site centered on R3 is in contrast to what has been pro- panded (Fig. 1 )( eff 0.1). The changes within the MTBR posed for tau binding to microtubules, where the region spanning parallel those observed in tau2N4R; namely, although there are shifts R1 and R2 is thought to be critical (11). One possible outcome of in the mean ETeff values for the subdomain constructs to smaller G H separate binding sites for tubulin and microtubules is that it allows values (i.e., longer distances) on binding to tubulin (Fig. 2 and ), tau to associate with both simultaneously, which may have impli- overall, the dimensions of the MTBR do not undergo a significant cations in tau function. Alternatively, the tau-tubulin complex may change (Fig. 2I). Moreover, the greatest magnitude change was in a – represent an earlier intermediate in the microtubule polymerization two-repeat construct probing R3 R4 (tau2N3R322/354)(Fig.2). pathway. Through association with multiple tubulin dimers, tau may Interestingly, the other two-repeat construct-probed tau2N3R244/322 (R1–R3) shows a smaller expansion than R3–R4 but greater than the comparable tau2N4R construct (R1–R2) (Fig. 2). These data suggest that in tau2N3R, where R2 is lacking, the N-terminal region of the MTBR must undergo a more significant expansion to accommodate binding to tubulin. Discussion Although tau’s role as a regulator of microtubule dynamic in-

stability is well-established, the molecular mechanism by which it BIOPHYSICS AND promotes the assembly of tubulin into protofilaments and mi- COMPUTATIONAL BIOLOGY crotubules is still poorly understood. To obtain a complete de- scription of this polymerization process, it is essential to describe the first step of the largely overlooked interaction of tau with soluble tubulin. Additionally, as much as 60% of tubulin exists in Fig. 4. Model of the MTBR of tau2N4R bound to tubulin under nonassembly conditions. A single conformation was selected for illustrative soluble form in the cytoplasm (45), suggesting that it is a readily purposes, with two views shown. The model shows how tau’s flexibility al- available binding partner, although little is known about the in- lows it to accommodate binding to two or more tubulin dimers (shown as teraction of its soluble forms with tau or even other microtubule- the surface representation of Protein Data Bank ID code 1TUB). The repeats associated proteins. are colored as follows: R1, blue; R2, red; R3, green; and R4, orange.

Melo et al. PNAS | December 13, 2016 | vol. 113 | no. 50 | 14339 Downloaded by guest on October 2, 2021 function as a scaffold that decreases the critical concentration for alignment of the resulting structures to calculate an “average” microtubule assembly. (Fig. S5). For Fig. 4, we selected a single tau structure, which was We find that binding to tubulin imposes local expansions of consistent with our FRET constraints, and placed it in proximity individual repeats within tau’s MTBR without a significant to tubulin dimers, which allows us to visualize how binding to change in the overall dimensions (Fig. 2). On the surface, this multiple dimers might be accommodated. Inherent to our model observation seems contradictory. However, it has been proposed is that the MTBR does not fold in a single stable structure but that tau’s interaction with microtubules is mediated by an array of instead, retains its intrinsically disordered features. Importantly, short motifs within the MTBR, with flanking residues remaining the formation of such a “fuzzy complex” (50) allows binding to flexible (33). If a similar model holds for tau bound to multiple several tubulin dimers and emphasizes the importance of con- tubulin dimers, then we can rationalize that binding to tubulin may formational flexibility in tau function. cause local expansion in regions involved directly in binding, Binding of tau to soluble tubulin is relevant and may also be whereas the flexibility of the flanking regions allows the overall important to the pathological role of tau in disease (22). Here, we dimensions of the MTBR to be maintained. Indeed, because the probe the conformational features of tau bound to tubulin. We MTBR is already fairly expanded in solution, with dimensions find a loss of long-range interactions that expose the MTBR; the close to those of a theoretical random coil, there may be no strong MTBR itself undergoes local conformational changes to accom- driving force for it to further expand on binding tubulin. modate its interaction with tubulin. Our data support a model Additionally, our results exclude two extreme conformations. whereby tau retains much of its intrinsic disorder, providing a (i) The distances calculated from smFRET measurements are flexible scaffold with multiple tubulin or microtubule binding sites. incompatible with a linear structure, because the average length This feature is noteworthy, because like many intrinsically disor- – ∼ of the construct spanning R1 R4 ( 79 Å) is considerably smaller dered proteins, tau’s interactions with its binding partners are – – ∼ than the sum of the distances of the R1 R2 and R2 R4 ( 120 Å) heavily regulated by posttranslational modifications (51). Estab- – – ∼ ii or the R1 R3 and R3 R4 ( 120 Å) constructs. ( ) We do not lishing that tau retains much of its disordered character when find evidence of the U-turn conformation within the MTBR re- binding tubulin supports that tau may be susceptible to post- cently proposed for tau bound to a single-tubulin dimer (48). translational modification even in its bound state, which may have Such a conformation would be expected to result in very high implications for understanding tau’s roles in function and disease. FRET efficiencies in both our tau2N4R184/291 and tau2N4R244/291 constructs, which we do not observe (Figs. 2 and 3). This dif- Materials and Methods ference may not be terribly surprising, because our conditions of Protein Expression and Labeling. Tau2N4R and tau2N3R were expressed and high-tubulin (10 μM) and low-tau (∼30 pM) concentrations purified as described previously (22, 36). For smFRET constructs, site-specific would favor occupying all possible binding sites on tau and differ labeling was performed by introducing cysteines at desired locations for significantly from the protein concentrations required for the reaction with maleimide fluorophores Alexa Fluor 488 and Alexa Fluor 594 NMR measurements in that study. However, we note that ti- (Life Technologies). Tubulin was purified as described previously. Details are tration over a range of tubulin concentrations results in apparent in SI Materials and Methods. two-state conformational transition of tau (Fig. S2), without evidence of a highly bent intermediate. smFRET Measurements. smFRET experiments were carried out on a laboratory- “ ” built instrument based on an inverted Olympus IX-71 Microscope (Olympus) as One must tread carefully in proposing structural models for ∼ dynamic, flexible, and heterogeneous protein assemblies, such as described previously (22, 36). Each smFRET histogram was acquired using 50 pM tau in the absence or presence of 10 μM tubulin. All measurements were taken the one observed for tau and soluble tubulin. Although we do not at 20 °C and under nonassembly conditions (in 20 mM Tris buffer, 50 mM want to promote the idea that a single, well-defined structure NaCl, pH 7.4). For constructs where the data peak could not be distinguished exists for tau when in a tau-tubulin complex, models are useful in from the zero peak, measurements were performed using alternating laser

visualizing what such an assembly may look like. Here, we cre- excitation (Fig. S6). A Gaussian coil polymer model was used to convert ETeff ated a topological model of tau bound to soluble tubulin (Fig. 4). values into rms distances (38). Details are in SI Materials and Methods. An ensemble of possible tau conformations was generated using PyRosetta with experimentally constrained Monte Carlo simu- ACKNOWLEDGMENTS. We thank Melissa Birol for helpful discussion about lations as described previously (49). The simulations were con- protein modeling, Lester Binder for the tau plasmid, and Lynne Regan for the pET- HT vector. This work was supported by funds from the Beverly and Raymond strained by the pairwise distance distributions from our MTBR Sackler Institute (J.C. and E.R.), NIH Institutional Training Grants T32GM067543 (to constructs (SI Materials and Methods) and also, considering G.A.-C.) and 5T32GM007223 (to S.E.-G.), National Science Foundation Grant MCB partial helical structure in R3 (39). This helical region allows for 0919853 (to E.R.), and NIH Grants R01NS079955 (to E.R.) and RF1AG053951 (to E.R.).

1. de Forges H, Bouissou A, Perez F (2012) Interplay between microtubule dynamics and 11. Goode BL, Chau M, Denis PE, Feinstein SC (2000) Structural and functional differences intracellular organization. Int J Biochem Cell Biol 44(2):266–274. between 3-repeat and 4-repeat tau isoforms. Implications for normal tau function 2. Ferreira JG, Pereira AL, Maiato H (2014) Microtubule plus-end tracking proteins and and the onset of neurodegenetative disease. J Biol Chem 275(49):38182–38189. their roles in cell division. Int Rev Cell Mol Biol 309:59–140. 12. Litman P, Barg J, Rindzoonski L, Ginzburg I (1993) Subcellular localization of tau 3. Kapitein LC, Hoogenraad CC (2015) Building the neuronal microtubule cytoskeleton. mRNA in differentiating neuronal cell culture: Implications for neuronal polarity. Neuron 87(3):492–506. Neuron 10(4):627–638. 4. Mitchison T, Kirschner M (1984) Dynamic instability of microtubule growth. Nature 13. Hirokawa N, Funakoshi T, Sato-Harada R, Kanai Y (1996) Selective stabilization of tau in 312(5991):237–242. axons and microtubule-associated protein 2C in cell bodies and dendrites contributes to 5. Akhmanova A, Steinmetz MO (2008) Tracking the ends: A dynamic protein network polarized localization of cytoskeletal proteins in mature neurons. JCellBiol132(4):667–679. controls the fate of microtubule tips. Nat Rev Mol Cell Biol 9(4):309–322. 14. Ebneth A, et al. (1998) Overexpression of tau protein inhibits kinesin-dependent 6. Weingarten MD, Lockwood AH, Hwo SY, Kirschner MW (1975) A protein factor es- trafficking of vesicles, mitochondria, and endoplasmic reticulum: Implications for sential for microtubule assembly. Proc Natl Acad Sci USA 72(5):1858–1862. Alzheimer’s disease. J Cell Biol 143(3):777–794. 7. Drubin DG, Kirschner MW (1986) Tau protein function in living cells. J Cell Biol 103(6 15. Terwel D, Dewachter I, Van Leuven F (2002) Axonal transport, tau protein, and Pt 2):2739–2746. neurodegeneration in Alzheimer’s disease. Neuromolecular Med 2(2):151–165. 8. Drechsel DN, Hyman AA, Cobb MH, Kirschner MW (1992) Modulation of the dynamic 16. Vershinin M, Carter BC, Razafsky DS, King SJ, Gross SP (2007) Multiple-motor based instability of tubulin assembly by the microtubule-associated protein tau. Mol Biol transport and its regulation by Tau. Proc Natl Acad Sci USA 104(1):87–92. Cell 3(10):1141–1154. 17. Dixit R, Ross JL, Goldman YE, Holzbaur ELF (2008) Differential regulation of dynein 9. Gustke N, Trinczek B, Biernat J, Mandelkow EM, Mandelkow E (1994) Domains of tau and kinesin motor proteins by tau. Science 319(5866):1086–1089. protein and interactions with microtubules. Biochemistry 33(32):9511–9522. 18. Vershinin M, Xu J, Razafsky DS, King SJ, Gross SP (2008) Tuning microtubule-based 10. Trinczek B, Biernat J, Baumann K, Mandelkow EM, Mandelkow E (1995) Domains of transport through filamentous MAPs: The problem of dynein. Traffic 9(6):882–892. tau protein, differential phosphorylation, and dynamic instability of microtubules. 19. Brunden KR, Trojanowski JQ, Lee VM (2009) Advances in tau-focused drug discovery Mol Biol Cell 6(12):1887–1902. for Alzheimer’s disease and related tauopathies. Nat Rev Drug Discov 8(10):783–793.

14340 | www.pnas.org/cgi/doi/10.1073/pnas.1610137113 Melo et al. Downloaded by guest on October 2, 2021 20. Ballatore C, Lee VM, Trojanowski JQ (2007) Tau-mediated neurodegeneration in 39. Li XH, Culver JA, Rhoades E (2015) Tau binds to multiple tubulin dimers with helical Alzheimer’s disease and related disorders. Nat Rev Neurosci 8(9):663–672. structure. J Am Chem Soc 137(29):9218–9221. 21. Winklhofer KF, Tatzelt J, Haass C (2008) The two faces of protein misfolding: Gain- 40. Mukrasch MD, et al. (2007) The “jaws” of the tau-microtubule interaction. J Biol and loss-of-function in neurodegenerative diseases. EMBO J 27(2):336–349. Chem 282(16):12230–12239. 22. Elbaum-Garfinkle S, Cobb G, Compton JT, Li XH, Rhoades E (2014) Tau mutants bind 41. Goedert M, Jakes R (1990) Expression of separate isoforms of human tau protein: tubulin heterodimers with enhanced affinity. Proc Natl Acad Sci USA 111(17): Correlation with the tau pattern in brain and effects on tubulin polymerization. 6311–6316. EMBO J 9(13):4225–4230. 23. Hasegawa M, Smith MJ, Goedert M (1998) Tau proteins with FTDP-17 mutations have 42. Panda D, Samuel JC, Massie M, Feinstein SC, Wilson L (2003) Differential regulation of a reduced ability to promote microtubule assembly. FEBS Lett 437(3):207–210. microtubule dynamics by three- and four-repeat tau: Implications for the onset of 24. Bunker JM, Kamath K, Wilson L, Jordan MA, Feinstein SC (2006) FTDP-17 mutations neurodegenerative disease. Proc Natl Acad Sci USA 100(16):9548–9553. compromise the ability of tau to regulate microtubule dynamics in cells. J Biol Chem 43. Hutton M, et al. (1998) Association of missense and 5′-splice-site mutations in tau with 281(17):11856–11863. the inherited dementia FTDP-17. Nature 393(6686):702–705. 25. Butner KA, Kirschner MW (1991) Tau protein binds to microtubules through a flexible 44. Spillantini MG, et al. (1998) Mutation in the tau gene in familial multiple system array of distributed weak sites. J Cell Biol 115(3):717–730. tauopathy with presenile dementia. Proc Natl Acad Sci USA 95(13):7737–7741. 26. Chen J, Kanai Y, Cowan NJ, Hirokawa N (1992) Projection domains of MAP2 and tau 45. Hiller G, Weber K (1978) Radioimmunoassay for tubulin: A quantitative comparison of determine spacings between microtubules in dendrites and axons. Nature 360(6405): the tubulin content of different established tissue culture cells and tissues. Cell 14(4): 674–677. 795–804. 27. Brandt R, Léger J, Lee G (1995) Interaction of tau with the neural plasma membrane 46. Sillen A, et al. (2007) NMR investigation of the interaction between the neuronal mediated by tau’s amino-terminal projection domain. J Cell Biol 131(5):1327–1340. protein tau and the microtubules. Biochemistry 46(11):3055–3064. 28. Feinstein HE, et al. (2016) Oligomerization of the microtubule-associated protein tau 47. Chung PJ, et al. (2015) Direct force measurements reveal that protein Tau confers is mediated by its N-terminal sequences: Implications for normal and pathological tau short-range attractions and isoform-dependent steric stabilization to microtubules. action. J Neurochem 137(6):939–954. Proc Natl Acad Sci USA 112(47):E6416–E6425. 29. Al-Bassam J, Ozer RS, Safer D, Halpain S, Milligan RA (2002) MAP2 and tau bind 48. Gigant B, et al. (2014) Mechanism of Tau-promoted microtubule assembly as probed longitudinally along the outer ridges of microtubule protofilaments. J Cell Biol 157(7): by NMR spectroscopy. J Am Chem Soc 136(36):12615–12623. 1187–1196. 49. Nath A, et al. (2012) The conformational ensembles of α-synuclein and tau: Com- 30. Santarella RA, et al. (2004) Surface-decoration of microtubules by human tau. J Mol bining single-molecule FRET and simulations. Biophys J 103(9):1940–1949. Biol 339(3):539–553. 50. Tompa P, Fuxreiter M (2008) Fuzzy complexes: Polymorphism and structural disorder 31. Kar S, Fan J, Smith MJ, Goedert M, Amos LA (2003) Repeat motifs of tau bind to the in protein-protein interactions. Trends Biochem Sci 33(1):2–8. insides of microtubules in the absence of taxol. EMBO J 22(1):70–77. 51. Wang Y, Mandelkow E (2016) Tau in physiology and pathology. Nat Rev Neurosci 32. Serrano L, Montejo de Garcini E, Hernández MA, Avila J (1985) Localization of the 17(1):5–21. tubulin binding site for tau protein. Eur J Biochem 153(3):595–600. 52. Kapanidis AN, et al. (2004) Fluorescence-aided molecule sorting: Analysis of structure 33. Kadavath H, et al. (2015) Tau stabilizes microtubules by binding at the interface and interactions by alternating-laser excitation of single molecules. Proc Natl Acad Sci between tubulin heterodimers. Proc Natl Acad Sci USA 112(24):7501–7506. USA 101(24):8936–8941. 34. Cleveland DW, Hwo SY, Kirschner MW (1977) Physical and chemical properties of 53. Castoldi M, Popov AV (2003) Purification of brain tubulin through two cycles of po- purified tau factor and the role of tau in microtubule assembly. J Mol Biol 116(2): lymerization-depolymerization in a high-molarity buffer. Protein Expr Purif 32(1): 227–247. 83–88. 35. Mukrasch MD, et al. (2009) Structural polymorphism of 441-residue tau at single 54. Trexler AJ, Rhoades E (2010) Single molecule characterization of α-synuclein in ag- residue resolution. PLoS Biol 7(2):e34. gregation-prone states. Biophys J 99(9):3048–3055. 36. Elbaum-Garfinkle S, Rhoades E (2012) Identification of an aggregation-prone struc- 55. Metskas LA, Rhoades E (2015) Conformation and dynamics of the Troponin I C-ter- ture of tau. J Am Chem Soc 134(40):16607–16613. minal domain: Combining single-molecule and computational approaches for a dis- 37. Jeganathan S, von Bergen M, Brutlach H, Steinhoff HJ, Mandelkow E (2006) Global ordered protein region. J Am Chem Soc 137(37):11962–11969. hairpin folding of tau in solution. Biochemistry 45(7):2283–2293. 56. Kohn JE, et al. (2004) Random-coil behavior and the dimensions of chemically un- 38. O’Brien EP, Morrison G, Brooks BR, Thirumalai D (2009) How accurate are polymer folded proteins. Proc Natl Acad Sci USA 101(34):12491–12496. models in the analysis of Förster resonance energy transfer experiments on proteins? 57. Teraoka I (2002) Polymer Solutions: An Introduction to Physical Properties (Wiley, J Chem Phys 130(12):124903. New York). BIOPHYSICS AND COMPUTATIONAL BIOLOGY

Melo et al. PNAS | December 13, 2016 | vol. 113 | no. 50 | 14341 Downloaded by guest on October 2, 2021