Quantitative Characterization of Heparin Binding to Tau Protein

Neurofibrillary tangles, principally composed of bundles of filaments formed by the microtubule-associated protein Tau, are a hallmark of a group of neurodegenerative diseases such as Alzheimer disease. Polyanionic cofactors such as heparin can induce Tau filament formation in vitro. Here we quantitatively characterize the interaction between recombinant human Tau fragment Tau244–372 and heparin (average molecular mass = 7 kDa) as well as heparin-induced fibril formation by using static light scattering, isothermal titration calorimetry, turbidity assays, and transmission electron microscopy. Our data clearly show that at physiological pH, heparin 7K, and human Tau244–372 form a tight 1:1 complex with an equilibrium association constant exceeding 106 m−1 under reducing conditions, triggering Tau fibrillization. In the absence of dithiothreitol, heparin shows a moderate binding affinity (105 m−1) to Tau244–372, similarly triggering Tau fibrillization. Further fibrillization kinetics analyses show that the lag time appears to be approximately invariant up to a molar ratio of 2:1 and then increases at larger ratios of heparin/Tau. The maximum slope representing the apparent rate constant for fibril growth increases sharply with substoichiometric ratios of heparin/Tau and then decreases to some extent with ratios of >1:1. The retarding effect of heparin in excess is attributed to the large increase in ionic strength of the medium arising from free heparin. Together, these results suggest that the formation of the 1:1 complex of Tau monomer and heparin plays an important role in the inducer-mediated Tau filament formation, providing clues to understanding the pathogenesis of neurodegenerative diseases.

Microtubule-associated protein Tau is the major protein subunit of neurofibrillary tangles, one of the major neuropathological hallmarks of a group of neurodegenerative diseases such as Alzheimer disease (1,2). Neurofibrillary tangles are mainly composed of bundles of Tau in the form of paired helical filaments, straight filaments, or twisted ribbons (3). Because the accumulation of Tau aggregates correlates well with nerve cell loss and the severity of dementia, formation of a neurofibrillary tangle is considered to be involved in the pathways leading to neuronal death (4 -6), and the characterization of factors involved in Tau fibrillization is of great importance to clarify the etiology of such neurodegenerative diseases.
Tau is among one of the largest proteins without recognizable secondary structure and adopts a natively unfolded structure in solution (7). It binds to microtubules through four repeat domains in its C-terminal part (8). The repeat domain of Tau forms the core of paired helical filaments in Alzheimer disease and also assembles more readily than full-length Tau into bona fide paired helical filaments in vitro (9,10).
To mimic Tau fibrillization in vivo, polyanionic cofactors such as heparin have been used to induce Tau filament formation in vitro (1,(11)(12)(13)(14). It has been shown that Tau and heparan sulfate (HS) 2 coexist in nerve cells with overt neurofibrillary lesion (12). The molecular basis for such an inducement process remains unknown. However, the discovery that other polyanions such as RNA (15) or arachidonic acid (16) can also cause Tau filament formation indicates that Tau aggregation is rather influenced by electrostatics than by the precise nature of the negatively charged polymers (17). The kinetics for inducer-mediated Tau filament formation can be characterized by a lag period, followed by a period of exponential growth and an asymptotic approach to equilibrium (13, 18 -20). The filamentous structures induced by heparin are structurally similar to those found in Alzheimer disease (1), and thus heparin-induced Tau aggregation serves as a good model for pathological Tau fibrillization.
In the present study, we quantitatively characterized the interaction between recombinant human Tau fragment Tau 244 -372 and heparin as well as heparin-induced fibril formation of Tau 244 -372 by using several biophysical methods, such as static light scattering, isothermal titration calorimetry (ITC), turbidity assays, and transmission electron microscopy. Our data demonstrated that at physiological pH and temperature, human Tau 244 -372 did form a 1:1 complex with heparin. Heparin shows a strong binding affinity to Tau 244 -372 in the presence of dithiothreitol (DTT), triggering Tau fibrillization under such reducing conditions.

EXPERIMENTAL PROCEDURES
Materials-Heparin (average molecular mass of 7 kDa) was obtained from Sigma-Aldrich. DTT was obtained from Ameresco (Solon, OH). All other chemicals used were made in China and were of analytical grade. All of the reagent solutions were prepared in 10 mM HEPES buffer (pH 7.4) containing 100 mM NaCl.
Plasmids and Proteins-The cDNA encoding human Tau fragment Tau 244 -372 was amplified using the plasmid for human Tau40 (kindly provided by Dr. Michel Goedert) as a template. The PCR-amplified Tau 244 -372 was subcloned into pRK172 vector. Recombinant Tau 244 -372 was expressed in Escherichia coli and purified to homogeneity by SP Sepharose chromatography as described (21). Purified Tau protein was analyzed by SDS-PAGE with one band. The concentration of human Tau fragment was determined according to its absorbance at 214 nm with a standard calibration curve drawn by bovine serum albumin.
Turbidity Assays-Samples of 160 M Tau 244 -372 were incubated at 37°C in 10 mM HEPES buffer (pH 7.4) containing heparin at different concentrations (20 -640 M) and 100 mM NaCl in the presence and absence of 1 mM DTT or in 10 mM HEPES buffer (pH 7.4) containing 160 M heparin and NaCl at different concentrations (0 -500 mM) in the presence of 1 mM DTT. The solutions with a volume of 250 l were placed into a well of a 96-well plate followed by monitoring the turbidity at 350 nm using a SpectraMax 190 microplate reader (Molecular Devices, Sunnyvale, CA). Each sample was run in triplicate or quadruplicate. The profile of absorbance versus time was analyzed as follows. A seventh order polynomial was fitted to all of the data {t, A 350 }from replicates at each heparin concentration. The maximum slope of the best fit polynomial (dP/dt) max , the value of t corresponding to maximum slope of the polynomial t*, and the value of the polynomial at this time P(t*) were evaluated. A straight line with slope (dP/dt) max was drawn through {t*, P(t*)}, and the value of t at which this line crosses the t axis (P ϭ 0) is designated the lag time.
Transmission Electron Microscopy-The formation of filaments by human Tau fragment was confirmed by electron microscopy of negatively stained samples. Sample aliquots of 10 l were placed on copper grids and left at room temperature for 1-2 min, rinsed with H 2 O twice, and then stained with 2% (w/v) uranyl acetate for another 1-2 min. The stained samples were examined using an H-8100 transmission electron microscope (Hitachi, Tokyo, Japan) operating at 100 kV or an FEI Morgagni transmission electron microscope operating at 80 kV.
Isothermal Titration Calorimetry-ITC experiments on the interaction of heparin with Tau 244 -372 in the absence of DTT were carried out at 37.0°C using a VP-ITC titration calorimetry (MicroCal, Northampton, MA). Freshly purified wild-type Tau 244 -372 and heparin were thoroughly dialyzed against 10 mM HEPES buffer (pH 7.4) containing 100 mM NaCl. A solution of 40 M Tau protein was loaded into the sample cell (1.43 ml), and a solution of 500 M heparin was placed in the injection syringe (280 l). A titration experiment consisted of 13 consecutive injections of 20-l volume and 40-s duration each, with a 4-min interval between injections, and the 14th injection of 20-l volume and 40-s duration, with an 8-h interval. Dilution heat of heparin was measured by injecting heparin solution into buffer alone and was subtracted from the experimental curves prior to data analysis. The resulting data (injections 1-13) were fitted to a single set of identical sites model using the MicroCal ORIGIN software supplied with the instrument, and the standard molar enthalpy change for the binding, ⌬ b H m 0 , the binding constant, K b , and the binding stoichiometry, n, were thus obtained. The standard molar free energy change, ⌬ b G m 0 , and the standard molar entropy change, ⌬ b S m 0 , for the binding reaction were calculated by the fundamental equations of thermodynamics (22).
Calorimetric data of injection 14 was integrated of the heat power peak of the reaction. Composition Gradient Static Light Scattering-Tau 244 -372 and heparin were thoroughly dialyzed against 10 mM HEPES buffer (pH 7.4) containing 100 mM NaCl and 1 mM DTT. Immediately before light scattering measurement, both protein and heparin solutions were filtered through a 0.02-m filter (Anotop 10 Plus, Whatman) and centrifuged at 5000 rpm and 4°C for 30 min to remove microscopic bubbles. The concentration of Tau 244 -372 in solution A (w A,load ) and the concentration of heparin in solution B (w B,load ), determined in an Optilab rEX differential refractometer (Wyatt, Santa Barbara, CA), were 0.55 and 0.42 g/liter, respectively. The specific refractive increment of Tau (A in equations below) is dñ/dw A ϭ 0.185 cm 3 /g (23). The specific refractive increment of heparin (B in equations below) was determined experimentally. A calibration curve of refractive index versus heparin concentration was constructed, and the best fit value of the slope dñ/dw B ϭ 0.126 cm 3 /g is in agreement with the published values (23,24). A composition gradient was created using the Calypso system (Wyatt, Santa Barbara, CA), in which the fraction of solution A (f A ) is decreased from 1 to 0 in steps of 0.08, and the total concentrations of Tau (A) and heparin (B) in each of these mixtures is given by the following equations.
These solution mixtures were pumped in sequence into parallel flow cells in which the static light scattering intensity and refractive index of each mixture were measured simultaneously in a DAWN-EOS light scattering photometer (Wyatt) and an Optilab rEX differential refractometer (Wyatt). The raw data obtained from each instrument were respectively converted into values of the normalized scattering intensity R/K opt and f A as described (25,26).
The equilibrium constant for formation of a 1:1 complex of Tau (A) and heparin (B) is defined as follows, where c AB , c A , and c B , respectively, denote the molar concen-

Heparin-Tau Interaction
trations at equilibrium of the complex, free A, and free B. The conservation of mass is given by the following, where w A,tot and w B,tot , respectively, denote the total w/v concentrations of Tau and heparin, and M A and M B , respectively, denote the molar masses of Tau and heparin. Given the values of w A,tot , w B,tot , M A , M B , and K AB , Equations 5-7 may be solved analytically to yield the equilibrium values of c A , c B , and c AB . The normalized scattering intensity is given by the following equations (27).
In the above equations, dñ/dw i , M i , and c i , respectively, denote the specific refractive increment, molecular weight, and molar concentration of the ith scattering species, ñ 0 is the refractive index of buffer, 0 is the wavelength in vacuum of incident light (here 690 nm), and N A is Avogadro's number. To facilitate modeling of scattering in mixtures containing scattering species of significantly different refractive increment, we scale the above equations as follows. We define a new optical constant, where dñ/dw protein is set equal to the average for polypeptides, 0.185 cm 3 /g (23). Then Equation 3 may be rewritten as follows.
2 is thus a measure of the scattering power of a given substance relative to that of protein. The specific refractive increments of Tau (A) and heparin (B) are as follows: dñ/dw A ϭ 0.185 cm 3 /g (23); dñ/dw B ϭ 0.126 cm 3 /g. The refractive increment of the 1:1 complex AB is a mass average (25).
The experimentally determined dependence of R/K opt upon f A was modeled using Equations 3-7 to calculate the equilibrium values of c A , c B , and c AB and then using Equation 11 with the values of G A , G B , and G AB given above to calculate R/K opt .
The values of M A , M B , and log 10 K AB were allowed to vary to achieve a best fit via nonlinear least squares.

Interaction between Tau and Heparin under Reducing
Conditions-The experimentally measured dependence of R/K opt on f A is plotted in Fig. 1. The dependence calculated using Equation 13 with the best fit parameter values given below is plotted together with the data. Here we quantitatively characterize the interaction between recombinant human Tau fragment Tau 244 -372 and heparin in the presence of DTT using static light scattering. The best fit parameter values with 95% confidence limits of estimate were: M A ϭ 14,100 Ϯ 700 g mol Ϫ1 ; M B ϭ 7200 Ϯ 800 g mol Ϫ1 ; log K AB ϭ 6.5 (Ϫ1, ϩ indeterminate). Our data clearly indicated that at physiological pH, heparin and human Tau 244 -372 form a tight 1:1 complex with an equilibrium association constant exceeding 10 6 M Ϫ1 under reducing conditions.
Interaction between Tau and Heparin under Oxidative Conditions-ITC provides a direct route to the complete thermodynamic characterization of noncovalent, equilibrium interactions (28,29), and DTT concentrations as low as 1 mM can cause severe base-line artifacts caused by background oxidation during the titration. Therefore ITC was used to measure the binding affinity of heparin to Tau protein in the absence of DTT. ITC profiles for the binding of heparin 7K to wild-type Tau 244 -372 at 37.0°C in 10 mM HEPES buffer (pH 7.4) containing 100 mM NaCl are shown in Fig. 2. Fig. 2A shows the raw ITC curve resulting from the injections of heparin into a solution of wild-type Tau 244 -372 . The titration curve shows that heparin binding to Tau 244 -372 is exothermic, resulting in negative peaks in the plots of power versus time.  obtained by fitting the data to a single set of identical sites model, indicating that one heparin bound to one Tau 244 -372 with a moderate affinity in the absence of DTT. The formation of the heparin-Tau complex was driven by a large favorable entropy increase with a small favorable enthalpy increase under such oxidative conditions. Our data clearly indicated that at physiological pH, heparin and human Tau 244 -372 formed a 1:1 complex with an equilibrium association constant of approximately 10 5 M Ϫ1 under oxidative conditions.

Heparin-mediated Tau Filament Formation under Reducing
Conditions-Tau 244 -372 was incubated with various concentrations of heparin in the presence of 1 mM DTT and monitored by turbidity assays. Fig. 3 shows the dependence of turbidity changes for Tau 244 -372 filament formation at 37.0°C on the molar ratio of heparin to Tau protein. The molar ratio of heparin to Tau was 1:8, 1:6, 1:4, 1:2, 1:1, 2:1, and 4:1, respectively. The kinetic curves were analyzed as described above (Fig. 3), and the lag time and maximum slope of these curves were obtained. As shown in Figs. 1 and 3, at physiological pH, heparin and human Tau 244 -372 formed a tight 1:1 complex under reducing conditions, triggering Tau fibrillization. Further fibrillization kinetics analyses showed that the lag time appeared to be approximately invariant up to a molar ratio of 2:1 and then got longer at larger ratios of heparin/Tau in the presence of DTT (Fig. 4B). The maximum slope representing the apparent rate constant for fibril growth increased sharply with substoichiometric ratios of heparin/Tau and then decreased to some extent with ratios of Ͼ1:1 in the presence of DTT (Fig. 4A).  FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6

Heparin-mediated Tau Filament Formation under Oxidative
Conditions-To monitor the fibrillization of Tau 244 -372 in the absence of DTT, injection 14 had an 8-h interval after 13 consecutive injections. The ITC curve of the fibrillization of Tau 244 -372 was shown in Fig. 2C. The curve represented heat of the reaction as a function of time by integrating the power change (⌬P) peak of the reaction shown in the inset box, corrected by the reference. This represented the kinetics of the reaction via the enthalpy production, indicating the high accuracy and low noise of the measurement even without smoothing of the data (Fig. 2C, inset). The calorimetric data between 60 and 540 min clearly confirmed that heparin mediated Tau filament formation under oxidative conditions (Fig. 2C). Meanwhile the transmission electron microscopy image of Tau 244 -372 aggregates formed in the ITC cell showed typical long and branched fibrils, further confirming the fibrillization of Tau 244 -372 induced by heparin (Fig. 2D).

Characterization of Morphology of Human Tau Samples-
Transmission electron microscopy was used to study the morphology of human Tau samples incubated under different conditions. For wild-type Tau 244 -372, the addition of DTT had no significant effect on the morphology of Tau samples, and long and branched fibrils as well as straight filaments were observed in both samples (Fig. 5), indicating that heparin could induce Tau filament formation under reducing or oxidative conditions.

DISCUSSION
Tau monomers do not assemble spontaneously under physiological conditions over experimentally tractable time periods (20), and Tau fragments cannot assemble spontaneously to   form amyloid filaments over a period of days when the concentration is lower than 150 M (30, 31). The pathological aggregation of Tau is a nucleation-elongation reaction that involves the formation of ␤-structure containing some hexapeptide motifs of the repeat domain (21). The "driving force" for filament formation by the Tau repeat domain involves both the formation of an intermolecular disulfide bond (10) and noncovalent interactions such as hydrogen bonds, electrostatic interactions, and hydrophobic interactions (17). Then the elongation can proceed from the nuclei (10).
In the present study, Tau filaments formed within hours when the Tau fragment was incubated with heparin. Polyanionic cofactors, such as heparin and thiazine red, cause structural and conformational changes of Tau protein (17,32,33), and then the kinetic and thermodynamic barriers for Tau fibrillization can be overcome (1,34). Although heparin is not a perfect model for HS, which coexists with Tau protein in nerve cells (12), both heparin and HS induce formation of Tau filaments with very similar morphologies (35). Moreover, the addition of heparin into full-length human Tau protein induces the assembly of Tau into filaments that closely resemble those observed in Alzheimer disease brain tissue (12). Therefore, this study has implications for the putative role of endogenous HS proteoglycans in the formation of neurofibrillary tangles in Alzheimer disease. The extension of the present study of the characterization of heparin binding to human Tau fragment to the quantitative characterization of HS binding to fulllength human Tau protein should lead to a better understanding of how HS proteoglycans mediate Tau filament formation in the physiological environment in Alzheimer disease brain.
So far, our experiments were conducted with human Tau fragment, but because filaments in Alzheimer disease contain full-length Tau protein, it is important to demonstrate the behavior of full-length Tau bound by heparin. The human Tau fragment used in this study encompasses the four microtubulebinding repeats, which does exclude a number of domains that could potentially interact with heparin. The main difference is that full-length Tau forms filaments in the presence of heparin much more slowly than Tau 244 -372 induced by heparin (36). However, we did observe straight filaments formed by Tau 244 -372 induced by heparin (Fig. 5), similar to those formed by more complete Tau proteins (12) and by full-length human Tau (Fig. 6) induced by heparin. Our additional ITC experiments (Fig. 7) indicated that in the absence of DTT, heparin bound to full-length human Tau protein with a moderate affinity (2.38 ϫ 10 5 M Ϫ1 ) at the first binding site followed by one weak binding event (2.80 ϫ 10 3 M Ϫ1 ). Clearly, the binding affinity of heparin to full-length human Tau at the first binding site closely resembles that of heparin to Tau 244 -372 . Based on our results and the recently published data (17), we suggest that heparin binds to full-length human Tau mainly through its segment 244 -372, forming the 1:1 complex and triggering Tau fibrillization.
Neuronal cells normally have a reducing environment maintained by an excess of glutathione (1,8). In this study, DTT, a   FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 strong reducing agent, was used to mimic the reducing environment present in normal neuronal cells. Our data here clearly showed that at physiological pH, heparin 7K and human Tau fragment formed a tight 1:1 complex under physiological reducing conditions, triggering Tau fibrillization. High affinity binding of heparin to Tau protein under reducing conditions provides the necessary charge compensation to allow fibril growth (17). An oxidative (without DTT present plus air oxidation) experimental condition is closer to the physiological environment in Alzheimer disease brain (31,37). Our ITC data demonstrated that heparin bound to Tau fragment with a moderate binding affinity under pathological oxidative conditions, similarly triggering Tau fibrillization. Kardos and co-workers (38) have reported the first ITC study of amyloid formation, e.g. the seed-dependent extension of ␤ 2 -microglobulin amyloid fibrils under acidic conditions. Here, we report for the first time a direct thermodynamic study of the interaction between amyloidogenic protein and inducer as well as inducer-triggered fibril formation using ITC.

Heparin-Tau Interaction
Our time-dependent turbidity measurements revealed that the lag time did not vary significantly, but the apparent rate constant for the growth of fibrils increased sharply as the molar ratio of heparin to Tau increased up to 1:1 under reductive conditions, suggesting that heparin plays a regulatory role mainly in the elongation process of Tau fibrillization. When the molar ratio of heparin to Tau exceeded unity, the lag time increased, and the apparent rate constant of fibril growth decreased to some extent. The retarding effect of excess heparin is attributed to a large increase in ionic strength of the medium arising from excess unbound heparin (13,39), which was confirmed by control experiments at fixed heparin concentrations and increasing salt concentrations (supplemental Fig. S1).
Based on our kinetic and thermodynamic data and the recently reported results (17,33,34,39,40), we propose a valuable hypothesis for heparin-mediated Tau filament formation that the first step of kinetics involves binding of heparin to the natively unfolded Tau protein, forming a 1:1 complex and inducing structural and conformational changes of Tau protein, and then the nuclei are formed, which is the rate-limiting step, followed by the step of fibril elongation at physiological pH (Fig. 8). The hypothesis proposed here is different from the previous hypothesis (33) in the following aspects. First, we propose that the binding of heparin (or another polyanion) to monomeric Tau is a prerequisite for the growth of Tau fibrils.
Second, we speculate that the binding of heparin to Tau monomer induces conformational changes in Tau, as well as reducing the large net positive charge borne by Tau protein, thus reducing the activation energy required to add Tau to the end of a growing Tau fiber and accelerating fibril growth. Therefore the formation of a 1:1 complex of Tau monomer and heparin plays an important role in the inducer-mediated Tau filament formation, providing clues to understanding the pathogenesis of neurodegenerative diseases.