In vivo rate-determining steps of tau seed accumulation in Alzheimer's disease

Abstract

Both the replication of protein aggregates and their spreading throughout the brain are implicated in the progression of Alzheimer’s disease (AD). However, the rates of these processes are unknown and the identity of the rate-determining process in humans has therefore remained elusive. By bringing together chemical kinetics with measurements of tau seeds and aggregates across brain regions, we can quantify their replication rate in human brains. Notably, we obtain comparable rates in several different datasets, with five different methods of tau quantification, from postmortem seed amplification assays to tau PET studies in living individuals. Our results suggest that from Braak stage III onward, local replication, rather than spreading between brain regions, is the main process controlling the overall rate of accumulation of tau in neocortical regions. The number of seeds doubles only every ∼5 years. Thus, limiting local replication likely constitutes the most promising strategy to control tau accumulation during AD.

Fig. 1.. Illustration of the key processes in tau aggregate formation.

(A) Spreading denotes spatial relocation of an existing aggregate. (B) Replication, the localized production of new seeds from existing ones, is composed of subprocesses that naturally fall into two categories—those that increase seed number (multiplication) and those that increase the size of a given seed (growth). (C and D) Seed load as a function of time and distance, from the solution of Eq. 1 in the two limits. In (C), the spread is slow [D/κ = 0.00025 (unit length)²] and the system is spreading-limited; in (D), the spread is fast [D/κ = 0.025 (unit length)²] and, thus, the system is replication-limited (see Materials and Methods for an approximate conversion from these reduced units).

Fig. 2.. Data fits and effect of decreasing spread or replication.

(A) Dots show the experimentally measured distribution of tau seeds, sampled in several brain regions, at different stages of the disease from DeVos et al. (19) [mean over measurements from seven (Braak III), four (Braak IV), six (Braak V), and two (Braak VI) patients; error bars are SEM]. The regions from left to right correspond to increasing distance from the EC, where aggregates first appear. a.u., arbitrary units. (B) Solid lines are the results of fitting Eq. 1 to the data in (A), using the data at Braak stage III as a starting point. The dashed lines represent the data, assuming that the sampled regions are equidistant along the spreading path. How the seed concentrations in the different Braak stages would change is shown for a decrease of either the spreading rate (C) or the replication rate (D) by a factor of 3. The change is much more pronounced when replication is reduced, highlighting that the system is in a replication-limited regime.

Fig. 3.. Replication rate from different datasets.

To determine the replication rate κ, the solution to Eq. 1 in the replication limit was fitted to (A) the temporal evolution of seed concentration in the neocortical regions and (B) the aggregate amounts measured by AT8 staining of brain slices from the primary visual cortex (images at Braak stages 0/I, III, and VI were thresholded for AT8 response quantification, normalized by the number of cells; inset: example image at Braak stage III). (C) Seed measurements from Kaufman et al. (21). (D) Seed and ELISA measurements from Furman et al. (20). (E) Stereological counts of neurofibrillary tangles from Gómez-Isla et al. (22). (F) Longitudinal tau PET (¹⁸F-flortaucipir) measurements from Sanchez et al. (24), which determined the rate of change in tau signal over consecutive measurements approximately 2 years apart. Each data point corresponds to the rate of change plotted against the total signal in 1 of 101 individuals tested: 4 diagnosed AD, 7 with mild cognitive impairment (MCI), 27 Aβ-positive cognitively normal (CN PiB+), and 63 Aβ-negative cognitively normal (CN PiB−). These data thus measure different quantities to those shown in (A) to (E) and are hence displayed differently; however, the same model as in (A) to (E), via Eq. 5, is used to fit the data. For details of the data used, see Tables 1 and 2.

Fig. 4.. P301S mouse model is also replication-limited.

The logarithm of tau seed concentration in P301S transgenic mice, measured by fluorescence resonance energy transfer (FRET) flow cytometry, at several time points and locations throughout the brain. The data (filled and open circles) were obtained from Holmes et al. (25). The solid line is a straight line fit in logarithmic space. For the fitting, only times were taken into account for which data exist in all brain regions (filled circles) to avoid biases.

Fig. 5.. Rate comparison with other systems.

(A) Key steps of aggregation reactions illustrating how growth and multiplication combine to lead to seed replication. (B) Comparison of the rates of tau seed growth and multiplication determined in AD brains here, with the rates predicted at concentrations between 100 nM and 10 μM from in vitro experiments (28), as well as the rates for both PrP (39) and Aβ42 (40) replication in vitro at the same concentrations (for details, see Materials and Methods). Diagonal lines show the order of magnitude of the doubling time, i.e., points along the diagonal lines have the same doubling time and replication rate. (C) Time required to produce 70 billion seeds from one seed (i.e., time for 36 rounds of doubling) for comparison with disease time scales; the dashed green line shows the average time that passes between Braak stages III and VI. Only the two slowest systems, wild-type (WT) tau, are visible on a linear scale; inset shows time on a logarithmic axis.

Fig. 6.. Converting Braak stage to time.

(A) Fractions of individuals in a particular Braak stage for a given age. Data from Braak et al. (14). All individuals below Braak stage I have been grouped together into one group (not shown), whereas the remainder are as classified by Braak et al. The distributions of Braak stage I and above are fit to Gaussians, where the magnitude and the SD of the Gaussians are global parameters determined by the distributions of stages I to III, and only the midpoint is a free parameter for all stages. (B) Results of the fits; error bars are 95% confidence intervals on the mean age for each stage.