Skip to main content
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 1997 Jul 22;94(15):7942–7947. doi: 10.1073/pnas.94.15.7942

Kinetic theory of fibrillogenesis of amyloid β-protein

Aleksey Lomakin †,‡,§, David B Teplow ‡,§,, Daniel A Kirschner , George B Benedek †,
PMCID: PMC21534  PMID: 9223292

Abstract

Prior quasielastic light scattering (QLS) studies of fibrillogenesis of synthetic amyloid β-protein (Aβ)-(1–40) at low pH have suggested a kinetic model in which: (i) fibrillogenesis requires a nucleation step; (ii) nuclei are produced by Aβ micelles in addition to seeds initially present; and (iii) fibril elongation occurs by irreversible binding of Aβ monomers to the fibril ends. Here we present the full mathematical formulation of this model. We describe the temporal evolution of the concentrations of Aβ monomers and micelles as well as the concentration and size distribution of fibrils. This formulation enables deduction of the fundamental parameters of the model—e.g., the nucleation and elongation rate constants kn and ke—from the time dependency of the apparent diffusion coefficient measured by QLS. The theory accurately represents the experimental observations for Aβ concentrations both below and above c*, the critical concentration for Aβ micelle formation. We suggest that the method of QLS in combination with this theory can serve as a powerful tool for understanding the molecular factors that control Aβ plaque formation.

Keywords: Alzheimer disease, light scattering


A seminal pathogenetic event in Alzheimer disease (AD) is the formation of fibrous amyloid plaques in the brain parenchyma and vasculature (1). The primary protein component of plaques is the amyloid β-protein (Aβ) (2). In the plasma and cerebrospinal fluid, amyloid β-protein exists primarily as a soluble peptide 40 or 42 residues long (3). However, in the senile plaque, Aβ exists in the form of amyloid fibers (4). The conversion of soluble Aβ into insoluble fibers produces structures with neurotoxic activity (5). This observation, coupled with accumulating genetic evidence which links increased production of fibrillogenic forms of Aβ with familial AD (6), makes inhibiting fibrillogenesis an attractive therapeutic strategy.

To properly target inhibitors, the structural stages and the kinetics of Aβ fibrillogenesis must be determined. In the past, turbidity and thioflavin T binding have been used to quantify levels of Aβ aggregation and amyloidogenesis, respectively (79). However, both methods detect only large polymeric structures and provide no information on either nucleation or elongation rate, the two key parameters controlling the kinetics of Aβ fibrillogenesis. Recently, however, static and dynamic light scattering, sophisticated optical methods capable of monitoring fibril length and structure, have been applied to the Aβ fibrillogenesis problem (1012).

In a previous communication (12) we reported an experimental study of the temporal evolution of Aβ fibrils using quasielastic light scattering (QLS). That study introduced an in vitro model system which permits quantitative monitoring of the kinetics of Aβ fibrillogenesis and enables the determination of the numerical values of the nucleation rate and elongation rate of the fibrils. Knowledge of these rate constants is vital in the effort to understand the molecular factors that control fibril creation and growth. This understanding can serve as an invaluable guide in a search for the reagents capable of affecting Aβ fibrillization.

It is the purpose of this paper to present the equations for the temporal evolution of fibril length distribution. These equations provide a quantitative explanation of all the experimental findings reported in our previous work (12). We shall demonstrate how the biologically relevant parameters of Aβ fibrillogenesis can be deduced from QLS measurements. The present mathematical analysis, taken together with our in vitro model system and the light scattering method, provides a powerful tool for the investigation of the molecular factors that control the fibrillogenesis of Aβ.

Kinetic Theory of Fibrillogenesis

A Model of Aβ Fibrillogenesis.

The most striking finding of our experiments (12) is that the fibrillogenesis pathway differs depending upon whether the total protein concentration, C, is above or below a certain critical concentration, c*. For C > c*, the initial rate of elongation is essentially independent of C and when Cc* the final size of fibrils also becomes independent of C. On the other hand, for C < c*, the initial rate of elongation is proportional to C, while the final size of fibrils significantly exceeds that found for C > c* and appears to increase as the concentration C approaches zero. To explain these findings we have introduced in ref. 12 the kinetic scheme illustrated in Fig. 1.

Figure 1.

Figure 1

Schematic representation of the kinetic model for Aβ fibrillogenesis. Fibrillization of Aβ protein is nucleation dependent. Two pathways of fibril nucleation are proposed. One is fibril nucleation on seeds. The second is nucleation within micelles, whose presence is postulated provided that the peptide concentration exceeds the critical micellar concentration c*. Micelles are in fast equilibrium with free monomers at concentration c*. Nuclei are spontaneously formed out of micelles with rate constant kn. Fibrils grow by binding monomers to fibril ends with the rate proportional to the concentration of free monomers. The corresponding rate constant is ke.

Two pathways of fibrillogenesis operate. These pathways differ in the mechanism of nucleation. For C < c*, fibrils grow on “seeds” initially present in the solution. For C > c*, we have proposed the existence of micelles, which are in rapid equilibrium with Aβ monomers. These micelles provide domains of high local protein concentration in which fibril nuclei form. Elongation proceeds by attachment of free Aβ monomers at the fibril ends. In our model, the quantity c* is identified as the critical micellar concentration of Aβ monomers. Thus, for C > c*, the monomer–micelle equilibrium maintains a nearly constant concentration of free Aβ monomers. This explains the observation that the initial elongation rate is essentially independent of the total Aβ concentration. In addition, the mechanism of micelle-facilitated nucleation explains the weak dependence of the final fibril size on concentration for Cc*, as will become clear from the forthcoming analysis.

In our scheme, nuclei are produced continuously while micelles are present. Thus, at any moment, there will be a distribution of fibrils having different aggregation numbers. From a knowledge of this distribution, all the macroscopic properties of the solution can be derived.

Kinetic Equations for the Distribution of Fibril Aggregation Numbers.

To quantify the kinetic scheme shown in Fig. 1, we introduce the time-dependent state variables M, Np, and c, which are, respectively, the number concentration of micelles, p-mer fibrils, and free Aβ monomers. We further introduce two structural parameters: m0 and n0. These are, respectively, the numbers of monomers in a micelle and in a nucleus. The time evolution of the system is determined by the numerical values of two kinetic parameters, kn and ke. The nucleation rate kn is the average number of nuclei produced by a single micelle per unit time. The number of monomers in a nucleus may in principle be distributed about the number n0. However, this initial uncertainty is small in comparison with the large final fibril aggregation number. Thus, in the interest of simplicity, we shall assume that all nuclei consist of n0 monomers. The elongation constant ke is the coefficient of proportionality between the number of monomers attached per unit time to each fibril and the concentration c of free Aβ monomers in solution.

The time derivative of the concentration of p-mer fibrils, p, is given by

graphic file with name M1.gif 1

The first term on the right hand side of Eq. 1 describes the creation of the fibrils of size p from those of size p − 1 by monomer binding. The rate of such binding is proportional to the concentration c of monomers. The second term accounts for the reduction in p-mer concentration due to conversion into (p + 1)-mers. The third term describes the generation of n0-mer nuclei from micelles. Eq. 1 assumes unidirectional fibril growth: monomer binding is irreversible and breaking or merging of fibrils is neglected. Preexisting seeds shall be accounted as an initial condition Np(t = 0).

The two state variables, c and M, are related according to the principles of thermodynamic equilibrium between monomers and micelles (13). Though generally there is a distribution of micelles with aggregation numbers above the minimal number m0, we will limit ourselves to a very simple two-state model for micelle formation. In this model all micelles have the same aggregation number m0, and it can be shown that

graphic file with name M2.gif 2

Here the critical micellar concentration is connected to the free energy of micellization, Δμ0, by (14)

graphic file with name M3.gif

where is the molar volume of water. Both M and c are related to the Nps by the conservation of mass condition:

graphic file with name M4.gif 3

Here N(1) ☰ ∑ pNp is the total amount of the protein in fibrillar form.

Eqs. 13 constitute a complete set of coupled nonlinear kinetic equations for the variables c, M, and each of the Np. These equations may be solved and the temporal evolution of the entire fibril distribution Np(t) may be determined provided that the initial distribution Np(0) and the total concentration C are known. One can then calculate experimentally observed quantities, such as the scattered light intensity or the distribution of hydrodynamic radii or diffusion coefficients.

Considerable mathematical simplification and physical insight can be achieved by examining the equations governing the temporal evolution of the moments of the fibril distribution. The kth moment N(k) of the distribution Np is defined as

graphic file with name M5.gif 4

Clearly, the zeroth moment N(0) is the total number concentration of fibrils of all sizes and the first moment N(1) is the total number concentration of proteins found in all fibrils. Note also that the total intensity of the scattered light is proportional to the second moment N(2) of the fibril distribution, provided that the fibril sizes are small compared with the wavelength of light.

Kinetic Equations for the Moments of the Fibril Distribution.

Multiplying all Eq. 1 by pk and summing, we obtain a set of recursive equations that define the temporal evolution of the kth moment:

graphic file with name M6.gif 5

In this sequence of equations the first two are of particular interest. For the zeroth moment Eq. 5 gives

graphic file with name M7.gif 6

Here we see that the total number concentration of fibrils changes solely as a result of the process of nucleation from the micelles. For the first moment Eq. 5 gives

graphic file with name M8.gif 7

Here we see that the total number concentration of proteins in fibrillar form changes as a result of both the creation of nuclei (first term on the right-hand side) and the binding of monomers to the ends of each fibril.

Eqs. 6 and 7 together with Eqs. 2 and 3 form a closed set of equations for the four state variables N(0), N(1), M, and c, which can actually be solved analytically under the condition that m0 ≫ 1. Furthermore, these two moments provide a clear characterization of the principal features of fibril distribution. In particular, the ratio of these moments gives the average fibril aggregation number

graphic file with name M9.gif 8

Kinetic Evolution of N(0), N(0), M, and .

Let us denote the total number concentration of protein in nonfibrillar form as X:

graphic file with name M10.gif 9

Here, if we express M as a function of c according to Eq. 2 we obtain

graphic file with name M11.gif 10

We shall assume that m0 ≫ 1. Analysis of Eq. 10 shows that in this simple two-state model of micellization either M or c is essentially constant in each of two domains of X, in particular

graphic file with name M12.gif
graphic file with name M13.gif 11

In each concentration domain the relevant Eq. 11 together with Eqs. 3, 6, and 7 forms a complete set of linear equations for variables N(0), N(1), M, and c, which can be solved analytically.

Consider first the case X < c* (regime I)—i.e., low initial concentration of nonfibrillar protein; in this case, M ≅ 0, hence according to Eq. 6, N(0) remains equal to N0, the number of fibrils and heterogeneous nuclei (“seeds”) preexisting in the solution at t = 0. Eq. 7 then reduces simply to

graphic file with name M14.gif 12

In this regime N(1) and c are connected according to Eq. 3, by N(1) + c = C. Thus Eq. 12 shows that the concentration c of free monomers decreases exponentially with the time constant (keN0)−1 from its initial value c0, whereas the amount of protein in fibrillar form grows according to

graphic file with name M15.gif 13

Because N(0) remains constant and equal to N0, the mean aggregation number = N(1)/N(0) exponentially approaches its final value f = C/N0 as described in Eq. 13.

Let as now consider regime II, the case of high initial concentration, X > c*. Here the concentration of monomers remains essentially constant, c = c*, micelles are present, M > 0, and new fibrils are nucleated from micelles. Eliminating N(1) and M in Eqs. 3, 6, and 7 in favor of N(0) gives the following second-order differential equation:

graphic file with name M16.gif 14

with γ = n0kn/m0 and ω02 = knkec*/m0. The initial conditions at t = 0 are

graphic file with name M17.gif 15

where NM is defined as

graphic file with name M18.gif 16

As we shall see, NM is the total number concentration of fibrils that would ultimately emerge from micelles in the absence of seeds. Therefore, for Cc*, when most fibrils are nucleated from micelles and the role of seeds is negligible, NM is related to the final mean aggregation number f by C = fNM. In addition, if Cc*, then at the start of fibrillogenesis Cm0M(0). Our experiments show dramatic growth of fibrils, i.e., fn0. Hence, m0M(0) ≫ n0NM, and we see from Eq. 16 that γ/ω0 ≪ 1. Thus we may neglect the second term in Eq. 14, and the solution for N(0)(t), consistent with the initial conditions as given by Eq. 15, may be expressed in the form

graphic file with name M19.gif 17

where

graphic file with name M20.gif 18

and

graphic file with name M21.gif 19

Substituting Eq. 17 into Eqs. 6 and 3, we readily find that

graphic file with name M22.gif 20

Here N(1)(0) is the number concentration of protein in the seeds at t = 0. At time T the micelles become exhausted—i.e., M(T) = 0, as can be seen from Eqs. 6 and 17. After this time T no further nucleation occurs and regime II terminates with the final number concentration of fibrils being N(0)(T) = NT. Thus, according to Eq. 18, when N0 = 0, NM is indeed the total number concentration of fibrils in the absence of seeds.

For t > T regime I applies. N(0) remains equal to NT and N(1) changes according to Eq. 13 with the starting monomer concentration c0 being c*, i.e.,

graphic file with name M23.gif 21

We may summarize the evolution of the fibril average aggregation number in the case of high initial concentration, X > c*, by the following expressions. In the domain 0 < t < T

graphic file with name M24.gif 22a

where 0 = N(1)(0)/N0 is the mean number of proteins in a seed. Furthermore, in the domain t > T we obtain (t) by dividing Eq. 21 by N(0)(T) = NT, hence

graphic file with name M25.gif 22b

According to Eqs. 22b and 18 the final average aggregation number of fibrils is given by

graphic file with name M26.gif 23

We may now make a connection with the qualitative analysis of fibril growth presented previously (12). In that analysis, we assumed Xc* and neglected seeds entirely (N0 = 0). In that case according to Eq. 23 f = C/NM. From the definition of NM in Eq. 16 we then find f = n0ω0/γ = (m0kec*/kn)1/2, as was deduced previously. Also we see from Eq. 19 that when seeds are absent T = π/2ω0 = (π/2)(m0/knkec*)1/2 regardless of the initial protein concentration. This result coincides with our previous (12) qualitative estimation of T apart from the factor of π/2.

QLS Spectroscopy as a Quantitative Assay for Fibrillogenesis

Relationship Between Aggregation Number and Hydrodynamic Radius for a Single Fibril.

In our experiments we monitor fibril growth using the method of QLS (15, 16). To utilize the theory for the distribution of fibril lengths presented above we must first establish a connection between the dimensions of an individual fibril and its diffusivity as measured by QLS. The diffusion of a fibril is more complex than that of a sphere. For long fibrils whose length L ≥ 1/q, where q is the scattering vector, the temporal autocorrelation function of the scattered light no longer can be represented as a single exponential even for a monodisperse solution (17). Indeed, no explicit analytical expression is available for the autocorrelation function even in the simple case of rigid rods. Nevertheless, it is possible to calculate and measure the first cumulant, Inline graphicD̄q2, and thus determine the mean diffusion coefficient . To calculate the first cumulant we only need to know the form of the diffusion equation for the scattering particles (18). In this work we use the tensor of diffusion coefficients for a cylinder as given by de la Torre and Bloomfield (19) to calculate the first cumulant for a fibril of length L = p/λ and diameter d. Here λ is the number of monomers per unit length of the fibril. We take λ = 1.6 monomers per nm (10, 12).

It is conventional to express the first cumulant in terms of the effective hydrodynamic radius, RH, according to the Stokes–Einstein relation,

graphic file with name M28.gif 24

Here η is solvent viscosity, T is the absolute temperature, and RH is the radius of a sphere whose diffusion coefficient is the same as the mean diffusion coefficient of the rigid rod in question. In Fig. 2 we show RH as a function of p = λL for several relevant values of d and q. As is seen from this figure, the dependence of RH on p is insensitive to the variations in the fibril diameter or wave vector q within the range of interest. In our calculations we neglected fibril flexibility, which could be characterized by a persistence length of several hundred nanometers (11). This leads to an underestimation of the length of those fibrils that are comparable to, or longer than, the persistence length. This effect could be accounted for by some rescaling of the RH(p) curve in Fig. 2. It is clear, however, that such rescaling will not impair our ability to assess the effect of variation in solution conditions or in the protein structure on the kinetic constants ke and kn. Therefore, for the sake of simplicity we shall model Aβ fibrils as rigid rods and use Fig. 2 as the calibration curve for hydrodynamic radius of a p-mer fibril.

Figure 2.

Figure 2

Calibration curve relating the hydrodynamic radius RH of a monodisperse solution of rigid rods as a function of the rod length L, or the number of Aβ monomers p. The results are shown for three different rod diameters, d, and for scattering vectors q = 11.8/λ0, corresponding to 90° scattering in aqueous solution for light having wavelength λ0 of 633, 514, and 488 nm.

Connection Between Observed Mean Hydrodynamic Radius H(t) and the Predictions of the Kinetic Theory of Fibrillogenesis.

Knowing the hydrodynamic radius of p-mer, RH(p), we now can make a connection between the apparent hydrodynamic radius H determined from the QLS data and the distribution of fibrils Np which our theory predicts for a particular set of kinetic parameters. Hence we may deduce the kinetic parameters which correspond to the observed evolution of H. Taking into account the scattering both by micelles and by the distribution of Aβ fibrils with appropriate weighting factors, H can be related to Np by

graphic file with name M29.gif 25

Here RM is the radius of a micelle. The contribution of each p-mer fibril is weighted by its scattered intensity, which is proportional to p2α, where α(qL) is the form factor of a rod of length L = p/λ. The corresponding weighting factor for a micelle is m02. Since a micelle is small, i.e., qRM ≪ 1, we take its form factor to be unity.

We see from Eq. 25 that, to rigorously apply our kinetic model for the analysis of QLS data, the temporal evolution of the whole fibril distribution function, Np(t), must be computed numerically. However, if we neglect the polydispersity in the fibril distribution, we can take advantage of the simple analytical theory developed in the previous section. Denoting hydrodynamic radius calculated in this approximation as H, from Eq. 25 we get

graphic file with name M30.gif 26

Here M(t), N(0)(t), and (t) = N(1)(t)/N(0)(t) have already been found analytically. As the fibrils grow, the relative contribution of micelles decreases and Eq. 26 reduces simply to H = RH(). We now wish to show that, within the accuracy of our experiments, we can replace the rigorous quantity H by the approximate quantity H. For this purpose we solved numerically Eqs. 13 for Np(t). We then computed H according to Eq. 25, using the calibration curve in Fig. 2 for H(p). The results of this computation for a number of initial concentrations and for a set of kinetic parameters consistent with our experiments are shown in Fig. 3. Here the solid curves represent the time evolution of the apparent hydrodynamic radius H(t) as computed numerically. They are to be compared with the dashed curves representing the corresponding approximate H(t) found analytically.

Figure 3.

Figure 3

Temporal evolution of H(t) (dashed lines) for a polydisperse distribution of fibrils, as computed numerically, and the appropriate H(t) (solid lines) as calculated using the simple analytic theory for (t), for the following values of the total monomer concentration: curves a, C = 5c*; curves b, C = 0.5c*; and curves c, C = 0.1c*. The parameters of the model used were N0 = 0.001C, kn = 2.4 × 10−6 sec−1, ke = 90 M−1⋅sec−1, m0 = 25, c* = 0.1 mM, and n0 = 10.

Our simplified treatment of monomer–micelle equilibrium neglects the existence of some micelles for X < c*. Consequently, the simple theory underestimates the number of fibrils at low concentration. This leads to overestimation of fibril size. Nevertheless, the difference between the rigorous and the simplified analysis for realistic values of kinetic parameters is significantly less than experimental error. We therefore shall use in further discussions the simple analytical version, since it provides a clear insight into the factors that govern the fibrillogenesis process.

Significance of the Parameters of the Model and Their Deduction from Experimental Data.

The parameters introduced previously can be divided naturally into three groups. The first group consists of structural parameters. These are the linear density λ and the fibril diameter d. The second group of parameters has to do with fibril creation from micelles (c*, m0, n0, and kn) or upon preexisting seeds (N0). Finally, the parameter ke describes the rate constant for the binding of monomers to the ends of an individual fibril.

The structural parameters are needed solely to relate the fibril aggregation number p to the length L of a fibril and its observed hydrodynamic radius RH. The parameters d and λ have been estimated by other workers (10, 12). A readjustment in these parameters would result simply in a rescaling of the calibration curve in Fig. 2. The second group of parameters determines the number of fibrils and therefore the final aggregation number pf. Knowledge of these parameters facilitates the choice of conditions under which fibrillogenesis can be reproducibly controlled. The nucleation parameters also provide useful information on the intermolecular interactions between Aβ monomers in the micelle. The remaining parameter ke, which describes the rate of fibril growth, provides insight into the molecular factors that control monomer binding to each fibril end. From a practical viewpoint ke, is a valuable quantitative measure of the effectiveness of putative growth inhibitors.

We now demonstrate how our theoretical analysis permits an evaluation of the above parameters using the experimental data on RH reported previously (12). We consider first the simpler case C < c*. Here no additional fibrils are nucleated during fibrillogenesis and the distribution of fibril sizes is determined primarily by the stochastic (Poisson) nature of monomer binding to fibril ends. Thus, the width of distribution Np(t) grows as Inline graphic, and whatever polydispersity may exist initially in the distribution of seed sizes can be neglected as soon as the fibril size significantly exceeds that of the largest seed. The relative width of distribution decreases in time as 1/1/2. Under these conditions the mean aggregation number (t) provides a sufficient representation of the extent of fibrillogenesis. We can deduce (t) directly from the experimentally measured mean hydrodynamic radius RH(t) by using the calibration curve in Fig. 2. The two most useful quantities to be derived from (t) are dp̄/dt at t = 0 and pf, the fibril final size. With these we can immediately find N0 and ke according to N0 = C/pf and ke = (1/C)(dp̄/dt).

We now consider the domain C > c*, where the parameters related to nucleation, namely kn and c*, play a central role. In this domain micelles are present, and the fibril distribution is the result of a continual emergence of new fibrils from micelles. As we demonstrated above, the fibril distribution, in practice, proves to be sufficiently narrow, so that H(t) can be satisfactorily represented in terms of (t) and the micellar concentration M(t) by Eq. 26. This equation accounts for the scattering from both micelles and fibrils. As the fibrils emerge and grow in size the micellar contribution to RH becomes negligible. Then, nearly linear growth of the apparent hydrodynamic radius with time is observed. Here we can again use the calibration curve in Fig. 2 to convert RH(t) into (t) and thereby determine the rate of fibril elongation, (dp̄/dt) = kec*. Since ke is known from C < c* data, this permits a deduction of the critical micellar concentration c*. Knowing kec*, the remaining parameter, kn, may be deduced from the mean final aggregation number f (Eq. 23). In the limit Cc* we expect that nucleation within micelles is the dominating source of fibrils, i.e., NMN0, and that practically all the peptide is initially in micellar form: m0M(0) ≅ C. Under these conditions f = ((kec*(m0/kn))1/2 and hence the ratio kn/m0 can be found from f. An estimation of m0 can be obtained by comparing the initial (Ii) and final (If) values of the scattered light intensity. Because Iim02M(0) = m0C and If ∼ α(f)f2N(0)(∞) = α(f)fC, it follows that m0 = α(f)f(Ii/If). We found experimentally f = 480, (Ii/If) = (1/7). Using then α(f) = 0.4 for the form factor, we may estimate that m0 = 25.

Analysis of our previously reported experimental data (12) produces the following values of the theoretical parameters: ke = 90 M−1⋅sec−1, c* = 0.1 mM, and kn = 2.4 × 10−6 sec−1. Thus, for C > c*, fibrils grow at a rate of 0.5 monomer per min, and new fibril nucleation occurs approximately once in 5 days per micelle. The predicted and the observed temporal evolution of RH(t) for two samples at C > c* is shown in Fig. 4. In this comparison we used Eq. 26, where (t) is given by Eqs. 22a and 22b. M(t) for t < T is derived by substituting Eq. 17 into Eq. 6. This figure shows that our model successfully represents the essential features of the kinetics of fibrillogenesis. Note that, in the vicinity of the knee of the RH(t) curve, the experimental points fall below the theoretical prediction. This discrepancy is most likely a consequence of the oversimplified two-state model employed to describe monomer–micelle equilibrium. In fact, the data suggest that there may be a distribution of micellar sizes around the mean aggregate number m0.

Figure 4.

Figure 4

Comparison between temporal evolution of the samples with Aβ concentration 1.16 mM (A) and 0.47 mM (B) observed experimentally in 0.1 M HCl (+) and calculated using the simple analytic theory (solid curves) with the following parameters: kn = 2.4 × 10−6 sec−1, ke = 90 M−1⋅sec−1, m0 = 25, c* = 0.1 mM, and n0 = 10. One percent of the protein was assumed to be in the form of seeds: n0N0 = 0.01C.

Conclusions

On the basis of experimental studies using QLS spectroscopy we previously proposed (12) a kinetic model for the in vitro fibrillogenesis of Aβ. According to this model: (i) fibrillogenesis requires the existence of nuclei, which are either produced from micelles of Aβ or are initially present in the solution as seeds; (ii) fibril elongation takes place by the irreversible binding of the monomeric protein to the fibril ends. In the present communication we have presented a detailed mathematical analysis of the model, giving a full description of the temporal evolution of the distribution of the fibril sizes. The theory enables a quantitative deduction of the rate constants for fibril nucleation (kn), fibril elongation (ke), and the concentration of seeds (N0). We have shown that this theory can satisfactorily describe the observed time course of fibrillogenesis for Aβ concentration both below and above c*, the critical concentration for Aβ micelle formation.

This work enables the following advances. First, by combining the experimental method of QLS with the present kinetic theory we obtain a powerful assay method for the quantitative determination of the fundamental kinetic coefficients kn and ke, which control fibrillogenesis. Second, the discovery and quantitation of nucleation within micelles gives us means to control the concentration of nuclei that initiate fibrillogenesis. These two advances now place on a solid quantitative footing further examination of the various biochemical and physiologic factors that affect the formation of Aβ fibrils. This knowledge can be highly valuable for the discovery and optimization of pharmacological agents that can inhibit Aβ plaque formation.

Acknowledgments

We thank Ms. Margaret Condron for expert technical assistance. This work was supported in part by the Foundation for Neurologic Diseases (D.B.T.) and by National Institutes of Health Grants 1-PO1-AG-14366 (D.B.T., G.B.B.) and 5-R37-EYO5127 (G.B.B.).

ABBREVIATIONS

amyloid β-protein

QLS

quasielastic light scattering

References

  • 1.Selkoe D J. J Neuropath Exp Neurol. 1994;53:438–447. doi: 10.1097/00005072-199409000-00003. [DOI] [PubMed] [Google Scholar]
  • 2.Glenner G G, Wong C W. Biochem Biophys Res Commun. 1984;120:885–890. doi: 10.1016/s0006-291x(84)80190-4. [DOI] [PubMed] [Google Scholar]
  • 3.Seubert P, Vigo-Pelfrey C, Esch F, Lee M, Dovey H, Davis D, Sinha S, Schlossmacher M G, Whaley J, Swindlehurst C, McCormack R, Wolfert R, Selkoe D J, Lieberburg I, Schenk D. Nature (London) 1992;359:325–327. doi: 10.1038/359325a0. [DOI] [PubMed] [Google Scholar]
  • 4.Narang H K. J Neuropath Exp Neurol. 1980;39:621–631. doi: 10.1097/00005072-198011000-00001. [DOI] [PubMed] [Google Scholar]
  • 5.Yankner B A. Neuron. 1996;16:921–932. doi: 10.1016/s0896-6273(00)80115-4. [DOI] [PubMed] [Google Scholar]
  • 6.Selkoe D J. Science. 1997;275:630–631. doi: 10.1126/science.275.5300.630. [DOI] [PubMed] [Google Scholar]
  • 7.Naiki H, Nakakuki K. Lab Invest. 1996;74:374–383. [PubMed] [Google Scholar]
  • 8.LeVine H., III Protein Sci. 1993;2:404–410. doi: 10.1002/pro.5560020312. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Evans K C, Berger E P, Cho C G, Weisgraber K H, Lansbury P T. Proc Natl Acad Sci USA. 1995;92:763–767. doi: 10.1073/pnas.92.3.763. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Shen C-L, Fitzgerald M C, Murphy R M. Biophys J. 1994;67:1238–1246. doi: 10.1016/S0006-3495(94)80593-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Shen C L, Murphy R M. Biophys J. 1995;69:640–651. doi: 10.1016/S0006-3495(95)79940-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Lomakin A, Chung D S, Benedek G B, Kirschner D A, Teplow D B. Proc Natl Acad Sci USA. 1996;93:1125–1129. doi: 10.1073/pnas.93.3.1125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Tanford C. The Hydrophobic Effect. New York: Wiley; 1980. [Google Scholar]
  • 14.Missel P J, Mazer N A, Benedek G B, Young C Y. J Phys Chem. 1980;84:1044–1057. [Google Scholar]
  • 15.Chu B. Laser Light Scattering: Basic Principles and Practice. Boston: Academic; 1991. [Google Scholar]
  • 16.Clark N A, Lunacek J H, Benedek G B. Am J Phys. 1970;38:575–585. [Google Scholar]
  • 17.Maeda T, Fujime S. Macromolecules. 1984;17:1157–1167. [Google Scholar]
  • 18.Akcasu A Z. Polymer. 1981;22:1169–1176. [Google Scholar]
  • 19.de la Torre J G, Bloomfield V A. Q Rev Biophys. 1981;14:81–139. doi: 10.1017/s0033583500002080. [DOI] [PubMed] [Google Scholar]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES