Depletion of amyloid‐β peptides from solution by sequestration within fibril‐seeded hydrogels

Abstract

Aggregation of amyloid‐β (Aβ) peptides in brain tissue leads to neurodegeneration in Alzheimer's disease (AD). Regardless of the kinetics or detailed mechanisms of Aβ aggregation, aggregation can only occur if Aβ concentrations exceed their local equilibrium solubility values. We propose that excess Aβ peptides can be removed from supersaturated solutions, including solutions in biological fluids, by the addition of hydrogels that are seeded with Aβ fibril fragments. Fibril growth within the hydrogels then sequesters excess peptides until equilibrium concentrations are reached. Experiments with 40‐ and 42‐residue Aβ peptides (Aβ40 and Aβ42) in phosphate buffer at 24°C and in filtered fetal bovine serum at 37°C, using crosslinked polyacrylamide hydrogels, demonstrate the validity of this concept. Aβ sequestration in fibril‐seeded hydrogels (or other porous media) may prove to be a useful technique in experiments with animal models of AD and may represent a possible approach to preventing or slowing the progression of AD in humans.

Introduction

Neurodegeneration in Alzheimer's disease (AD) is generally believed to be a direct or indirect consequence of the aggregation of amyloid‐β (Aβ) peptides within brain tissue.1, 2 A variety of Aβ aggregates have been characterized in various levels of detail, including transient or metastable oligomers comprised of ∼12 or fewer Aβ monomers,3, 4 larger oligomers and protofibrils,5, 6, 7, 8, 9, 10, 11 and thermodynamically stable amyloid fibrils,12, 13, 14, 15, 16, 17, 18 which are themselves polymorphic.19, 20, 21, 22, 23 A variety of toxicity mechanisms for Aβ aggregates have been proposed, including interference with synaptic function,24, 25, 26, 27 stimulation of neuroinflammation,28, 29 disruption of cell membranes,30, 31, 32 interaction with specific cell‐surface receptors,33, 34 oxidative stress associated with bound metal ions,35, 36, 37 interference with the blood–brain barrier or vasculature,38 and other mechanisms.39, 40 Although there is considerable uncertainty about the identities of the most important Aβ aggregates and their toxicity mechanisms, it seems possible that complete prevention of Aβ aggregation within brain tissue would prevent AD. Moreover, prevention of additional Aβ aggregation after initial diagnosis may impede the progression of AD.41, 42, 43, 44

According to basic principles of physical chemistry, regardless of the kinetics or detailed molecular mechanism of aggregation, Aβ aggregation can occur only if the concentrations of Aβ peptides exceed their intrinsic solubility values at equilibrium, that is, if solutions are supersaturated. Solubility values depend on temperature, pH, ionic strength, and solvent composition (see below). The fact that aggregation occurs in brain tissue implies that Aβ concentrations do exceed their solubility limits, at least locally. Methods for depleting excess Aβ peptides from solution are therefore of interest as potential preventive or therapeutic approaches for AD. Here we describe and demonstrate such a method, based on the incorporation of excess Aβ into fibrils within hydrogels that are deliberately seeded with Aβ fibril fragments.

Results

Strategy for depletion of Aβ peptides from solution

Our method for depleting excess Aβ peptides from solution is shown schematically in Figure 1. After preparing long 40‐ or 42‐residue Aβ (Aβ40 or Aβ42) fibrils from either synthetic or recombinant peptides [Fig. 2(A)], we break the fibrils by sonication, resulting in fibril fragments with lengths in the 50–200 nm range, called “seeds” [Fig. 2(B)]. Seeds are incorporated within crosslinked polymer hydrogels by mixing the seeds with the solution of precursor monomers before initiation of the polymerization and crosslinking reactions. Hydrogel rods are cast within Teflon molds with appropriate dimensions, extruded and washed, then cut into sections with a razor blade [Fig. 2(C)]. When seeded hydrogel sections are added to an aqueous solution that contains Aβ peptides, the peptides in solution diffuse into the hydrogels. As long as the Aβ peptide concentration exceeds its solubility limit, fibril seeds grow by addition of peptides to their ends. Fibril growth within the hydrogels continues until the peptide concentration in solution drops to its solubility limit, at which point the net rate of addition of peptides to the fibril seeds equals the net rate of release of peptides from the seeds.45

Figure 1.

Schematic representation of the experimental strategy. Hydrogel pieces (yellow blocks) that contain Aβ fibril fragments (red bars) are placed in a solution containing an excess concentration of Aβ molecules (red dots). Aβ molecules diffuse into the hydrogels, where they add to the growing ends of fibrils. Fibril elongation proceeds until the concentration of soluble Aβ is reduced to its equilibrium value, after which no additional net aggregation of Aβ occurs.

Figure 2.

(A) TEM images of Aβ40 and Aβ42 fibrils, negatively stained with uranyl acetate. (B) TEM images of fibril seeds, prepared by sonication of Aβ40 and Aβ42 fibrils. (C) Photograph of polyacrylamide hydrogel rods and sections.

In principle, any hydrogel or similar porous medium could be used, provided that (i) the polymerization and crosslinking reactions (or other processes that produce the medium) do not destroy the seeds or cause chemical modifications that prevent addition of Aβ peptides to their ends; (ii) the internal porosity and external dimensions of the medium are such that soluble Aβ peptides, which could be either monomeric or oligomeric, diffuse into the medium on an appropriate time scale (typically several hours or less); (iii) the medium does not allow seeds to diffuse out and does not block fibril growth; (iv) the medium retains its integrity while being immersed in the relevant fluid for extended periods of time. For use within living organisms, absence of toxicity and other harmful biological responses is also an obvious condition.

When the Aβ peptide concentration in solution exceeds its solubility limit, growth of fibril seeds within the hydrogels competes with spontaneous nucleation of new fibrils outside the hydrogels. The spontaneous nucleation rate increases with increasing peptide concentration. Therefore, the success of experiments described below depends on the initial peptide concentration being above its solubility limit, but not enormously so.

Experiments in phosphate buffer at 24°C

Experiments were performed with crosslinked polyacrylamide hydrogels, which were cast within Teflon molds as rods with 2.25 mm diameters and 20 mm lengths. Acrylamide concentrations were 7.33% w/v or 6.10% w/v for hydrogels that were crosslinked with either N,N′‐methylene‐bis‐acrylamide (MBA, 0.23% w/v) or bis‐acryloylcystamine (BAC, 0.43% w/v), respectively. These conditions result in nominal pore radii that are approximately 85 nm or 45 nm, respectively.46 Hydrogel rods were cut into sections with 1.5–3.0 mm lengths. Seeded hydrogels contained fibril fragments with concentrations (by Aβ monomer) of approximately 100 or 400 μM, that is, 0.43 or 1.7 mg/mL.

Figure 3(A) shows the results of experiments in which 10–12 MBA‐crosslinked hydrogel sections (100 μL total volume) were placed in vials containing 2.0 mL of 10 μM Aβ40 in 10 mM sodium phosphate buffer, pH 7.4, at room temperature. Aβ40 concentrations in solution were monitored by optical absorption at 220 nm, using analytical high‐performance liquid chromatography (HPLC), a reverse‐phase C18 column, and 20 μL aliquots from the solutions for each time point (Fig. S1, Supporting Information). When gels with 400 μM Aβ40 seed concentrations (based on Aβ40 monomers) were used, the soluble Aβ40 concentration decreased by a factor of 30 (to ∼0.3 μM) in 51 h. As described above, we attribute the depletion of soluble Aβ40 to diffusion of Aβ40 into the hydrogels and subsequent addition to the fibril seeds. In principle, this process should continue until the soluble Aβ40 concentration decreases to the value that corresponds to thermodynamic equilibrium with fibrils, which has been shown previously to be about 0.4 μM under the conditions of these experiments.45

Figure 3.

(A) Dependence of Aβ40 concentration in solution on incubation time with seeded and unseeded hydrogels. The initial Aβ40 concentration in solution was 10 μM. (B) Dependence of Aβ42 concentration in solution on incubation time with seeded and unseeded hydrogels. The initial Aβ42 concentration in solution was 5 μM. Solutions were 10 mM sodium phosphate, pH 7.4, 24°C. Hydrogels were MBA‐crosslinked polyacrylamide. Aβ40 concentrations were measured by optical absorption as described in the text. Data points and error bars are means and standard deviations from measurements on three separate solutions. Initial values are normalized to 100.

When gels with 100 μM Aβ40 seed concentrations were used, the soluble Aβ40 concentration decreased more slowly. Fitting the data in Figure 3(A) with exponential decays (Table 1), we obtain rates of depletion of Aβ40 from solution equal to 0.068 ± 0.004 h⁻¹ for data with 400 μM seeds and 0.034 ± 0.005 h⁻¹ for data with 100 μM seeds. The fact that the rate of depletion decreases by an approximate factor of two when the seed concentration decreases by a factor of four indicates that the rate of depletion is not limited only by the density of seeds. Further experiments would be required to characterize the dependence of the rate of depletion on seed concentration more fully.

Table 1.

Rates of Depletion of Aβ From Solution, Obtained by Least‐Squares Fits of Data in Figures 3A, 3B, 5, and 6B With the Form [Aβ] = C₀ + C₁ × exp(–kt)

Peptide in solution	Peptide in seeds	Buffer	C 0 (%)	C 1 (%)	Depletion rate k (h−1)	rms residual (%)
10 μM Aβ40	100 μM Aβ40	Phosphate, 24°C	−10 ± 8	107 ± 7	0.034 ± 0.005	1.9
10 μM Aβ40	400 μM Aβ40	Phosphate, 24°C	0.1 ± 1.8	100 ± 2	0.068 ± 0.004	1.1
5 μM Aβ42	400 μM Aβ42	Phosphate, 24°C	13 ± 2	77 ± 4	0.114 ± 0.018	4.3
10 μM Aβ40	400 μM Aβ42	Phosphate, 24°C	28 ± 19	68 ± 18	0.020 ± 0.009	3.2
5 μM Aβ42	400 μM Aβ40	Phosphate, 24°C	8 ± 2	87 ± 3	0.086 ± 0.009	3.0
50 nM Aβ40	400 μM Aβ40	Filtered FBS, 37°C	12 ± 1	88 ± 2	0.065 ± 0.004	0.7
5 μM Aβ40	140 μM Aβ40 (microspheres)	Phosphate, 24°C	10 ± 3	86 ± 9	0.26 ± 0.06	3.0

Values of C ₀, C ₁, and the root‐mean‐squared (rms) residual are given as percents of the initial Aβ peptide concentration in solution.

When unseeded hydrogels were used, the soluble Aβ40 concentration remained above 7.0 μM until at least 78 h. In experiments with unseeded hydrogels, an initial drop to about 9.5 μM is attributable to the 5% increase in available volume when hydrogels were added to the Aβ40 solution. The subsequent gradual decrease in soluble Aβ40 concentration is attributable to spontaneous aggregation, either in solution or within hydrogels.

Figure 3(B) shows results from experiments with MBA‐crosslinked hydrogels that were seeded with Aβ42 fibrils (400 μM concentration, 100 μL total hydrogel volume), using initial soluble Aβ42 concentrations of 5 μM and 2.0 mL solution volumes. In the presence of seeded gels, but not in the presence of unseeded gels, the soluble Aβ42 concentration decreased by a factor of 10 (to ∼0.5 μM) in 50 h.

To confirm that fibril growth within the seeded hydrogels accounts for depletion of Aβ40 from solution, we performed thioflavin T (ThT) fluorescence and transmission electron microscopy (TEM) experiments with BAC‐crosslinked hydrogels, which can be dissolved by treatment with concentrated 2‐mercaptoethanol solution.47 A 1.0 mL volume of 10 μM Aβ40 solution was added to each vial containing 100 μL of hydrogels, seeded with 10 μM Aβ40 fibril fragments (0.043 mg/mL). After incubation for various time periods, the hydrogel sections were rinsed and dissolved. Concentrated ThT stock solution was added to produce a final ThT concentration of 60 μM and fluorescence emission spectra were recorded with excitation at 423 nm. As shown in Figure 4(A,B), the fluorescence at 495 nm increased monotonically with incubation time, indicating an increasing total mass of Aβ40 fibrils within the hydrogels.48 As shown in Figure 4(C), TEM images of the dissolved hydrogels showed fibrils with lengths that greatly exceed the initial lengths of the fibril seeds.

Figure 4.

Fluorescence spectra of Aβ40‐seeded hydrogels after incubation in Aβ40 solutions for the indicated time periods. Hydrogels were BAC‐crosslinked polyacrylamide, initially containing 10 μM Aβ40 fibril seeds. Solutions were 10 mM sodium phosphate, pH 7.4, 24°C, with initial Aβ40 concentrations of 10 μM. Hydrogels were dissolved before addition of ThT for fluorescence measurements. (B) Dependence of the ThT fluorescence peak area on incubation time. (C) TEM images of Aβ40‐seeded hydrogels after incubation in Aβ40 solutions for the indicated time periods before dissolution and preparation of TEM grids, which were negatively stained with uranyl acetate.

TEM images in Figure 4(C) show large variations in fibril lengths after incubation with Aβ40 solution. These variations may indicate that fibril seeds closest to the surface of a hydrogel grow most rapidly as they “capture” peptides from the solution, while seeds farther from the surface grow more slowly because the local concentration of soluble Aβ40 remains lower. Alternatively, growth of certain seeds may be inhibited by interactions with polyacrylamide chains within the hydrogels. Although micron‐long fibrils could be found on the TEM grid after 1 h of incubation, the density of long fibrils increased at longer incubation times.

It is worth noting that sequestration of Aβ peptides from a 2.0 mL volume of solution with an initial Aβ concentration of 10 μM into a 100 μL volume of hydrogel with a seed concentration of 400 μM (1.7 mg/mL) requires that the seed length increase by ∼50% on average (because the solution initially contains 20 nmol of Aβ molecules and the hydrogel initially contains 40 nmol of Aβ molecules). Thus, under the conditions of experiments in Figure 3, the average increase in seed length is rather small. Under the conditions of experiments in Figure 4(C) (where the solution initially contains 10 nmol of Aβ molecules and the hydrogel initially contains 1 nmol of Aβ molecules) fibril seeds are expected to increase in length on average by an approximate factor of 10.

Cross‐seeding between Aβ40 and Aβ42

Figure 5 shows results from “cross‐seeding” experiments with MBA‐crosslinked hydrogels. When hydrogels seeded with Aβ42 fibril fragments were added to Aβ40 solutions, a gradual decrease in the soluble Aβ40 concentration was observed. However, when hydrogels seeded with Aβ40 fibril fragments were added to Aβ42 solutions, the soluble Aβ42 concentration decreased from 5 μM to about 0.3 μM within 50 h. Cross‐seeding experiments in free solution (i.e., without hydrogels) also indicated efficient seeding of Aβ42 fibril formation by Aβ40 fibril fragments, and less efficient seeding of Aβ40 fibril formation by Aβ42 fibril fragments (Figs. S2 and S3, Supporting Information). Discrepancies with a previous report by Cukalevski et al., in which efficient seeding of Aβ42 by Aβ40 fibrils was not observed,49 may be due to the absence of deliberate fragmentation of the Aβ40 fibrils in the experiments by Cukalevski et al. Other groups have reported no detectable seeding of Aβ40 by Aβ42 fibrils,14, 50 possibly due to differences in cross‐seeding efficiency for different Aβ42 fibril polymorphs, or due to other differences in experimental conditions. Cross‐seeding between Aβ40 and Aβ42 in both directions has been observed in some experiments.51

Figure 5.

Dependence of Aβ40 or Aβ42 concentration in solution on incubation time with “cross‐seeded” hydrogels. Solutions were 10 mM sodium phosphate, pH 7.4, 24°C. Hydrogels were MBA‐crosslinked polyacrylamide. Initial concentrations in solution were 10 and 5 μM for Aβ40 and Aβ42, respectively. Aβ40 concentrations were measured by optical absorption as described in the text. Data points and error bars are means and standard deviations from measurements on three separate solutions. Initial values are normalized to 100.

Depletion of Aβ42 from solution by Aβ40 seeds may occur by addition of Aβ42 molecules to the ends of the Aβ40 fibril fragments, resulting in individual fibrils that contain Aβ40 and Aβ42 in separate segments. Alternatively, Aβ40 seeds may provide surfaces for heterogeneous nucleation of pure Aβ42 fibrils.52 Our data do not distinguish between these possibilities.

Experiments in serum at 37°C

Intrinsic solubilities of Aβ peptides (i.e., concentrations of soluble peptides at thermodynamic equilibrium, also called “critical concentrations”53) depend strongly on solvent conditions. For Aβ40 in 10 mM phosphate buffer, pH 7.4, the intrinsic solubility decreases from about 0.4 μM to about 50 nM as temperature increases from 24 to 37°C.45 Concentrations of soluble Aβ40 and Aβ42 in human cerebrospinal fluid (CSF) are reported to be about 1.4 nM and 0.1–0.2 nM, respectively.54, 55, 56 In human blood plasma, these concentrations are reported to be 0.07–0.2 nM and 0.01–0.03 nM, respectively.54, 55, 57, 58 Concentrations of soluble (in tris‐buffered saline) Aβ40 and Aβ42 in human cortical tissue are reported to be ∼0.5–4 pmol/g and 0.5–11 pmol/g, respectively59 (i.e., ∼0.4–3 nM and 0.4–8 nM, respectively). It is therefore important to test whether the strategy for depletion of Aβ peptides from solution demonstrated above is effective under biologically‐relevant conditions and whether this approach is capable of lowering Aβ peptide concentrations to the reported ranges in human tissues and fluids.

Consequently, we performed experiments with fetal bovine serum (FBS), which was filtered through 3000 MWCO membranes to remove high concentrations of high‐molecular‐weight proteins such as albumin and immunoglobulins.60 Filtration was used for two reasons: (i) the composition of filtered serum is similar to the composition of CSF,61, 62 which has relatively low concentrations of high‐molecular‐weight proteins; (ii) preliminary experiments with unfiltered FBS were unsuccessful, suggesting that high‐molecular‐weight proteins in unfiltered FBS may clog the pores of our hydrogels, impeding diffusion of Aβ peptides into the hydrogels. MBA‐crosslinked hydrogels were added to 50 nM solutions of recombinant Aβ40 in filtered FBS, containing 0.1% w/v sodium azide and 10 mM phosphate buffer at pH 7.4 and maintained at 37°C during incubation. Initial solution volumes were 1.2 mL, total hydrogel volumes were 100 μL and 80 μL aliquots were taken for quantification of Aβ40 concentrations after various incubation periods. Since soluble Aβ40 concentrations in these experiments were too low to be measured by optical absorption, we used mass spectrometry as described in “Materials and Methods” section.

Figure 6 compares the dependences of Aβ40 concentration on incubation time for filtered FBS solutions that contained either Aβ40‐seeded (400 μM seed concentration) or unseeded hydrogels. When seeded hydrogels were used, the Aβ40 concentration dropped from 50 nM to ∼7 nM within 72 h. When unseeded hydrogels were used, no significant change in Aβ40 concentration occurred over at least 72 h. These results indicate that seeded hydrogels can be used to sequester Aβ peptides from biologically relevant fluids, and that concentrations of soluble Aβ can be reduced to levels approaching those in human CSF and cortical tissue.54, 55, 56, 59

Figure 6.

Dependence of Aβ40 concentration in solution on incubation time with seeded and unseeded hydrogels. The initial Aβ40 concentration in solution was 50 nM. Solutions were filtered FBS, pH 7.4, 37°C. Hydrogels were MBA‐crosslinked polyacrylamide. Data points and error bars are means and standard deviations from measurements on three separate solutions. Initial mean values are normalized to 100.

Experiments with polyacrylamide microspheres

Experiments with animal models for AD (see Discussion) may require injectable preparations of seeded hydrogels. To this end, we prepared polyacrylamide microspheres by an emulsion polymerization method (see Materials and Methods), both with and without Aβ40 fibril seeds. Addition of allylamine to the polymerizing mixture allowed the microspheres to be fluorescently labeled with fluorescein isothiocyanate (FITC). Fluorescence microscope images [Fig. 7(A)] then indicated diameters primarily in the 20–100 μm range. To test the ability of seeded microspheres to sequester Aβ peptides from solution, 5 mg samples of lyophilized microspheres were added to 500 μL volumes of 5.0 μM Aβ40 in 10 mM sodium phosphate buffer, pH 7.4, at room temperature. Concentrations of Aβ40 in solution were monitored by optical absorption at 280 nm, using analytical HPLC as described above, taking 50 μL aliquots from the solutions for each time point. Data in Figure 7(B) show that seeded microspheres (140 μM seed concentration, based on Aβ40 monomers) caused the concentration of Aβ40 in solution to drop below 1.0 μM in less than 20 h, and to reach ∼0.4 μM (the equilibrium solubility level) in 42 h. In solutions containing unseeded microspheres or no microspheres [control in Fig. 7(B)], reductions in Aβ40 concentrations, attributable to spontaneous aggregation, were significantly less rapid.

Figure 7.

(A) Fluorescence microscope images of polyacrylamide microspheres, both with and without Aβ40 fibril seeds. Microspheres are fully hydrated and are labeled with fluorescein isothiocyanate. (B) Dependence of Aβ40 concentration in solution on incubation time with seeded and unseeded microspheres. Solutions were 10 mM sodium phosphate, pH 7.4, 24°C. Initial Aβ40 concentrations in solution were 5 μM for Aβ40. Data points and error bars are means and ranges from two replicates of each condition. “Control” is a 5 μM Aβ40 solution without microspheres, with only one replicate. Aβ40 concentrations were measured by optical absorption as described in the text. Initial values are normalized to 100, based on the control concentration at time zero. (Partial aggregation of Aβ40 in the control solution at early times may account for the higher concentrations observed in solutions containing unseeded microspheres.).

Discussion

Data presented above show that soluble Aβ concentrations can be reduced to levels close to equilibrium solubility values by sequestration of excess Aβ peptides within fibril‐seeded hydrogels, both for simple buffers at room temperature and for biological fluids at 37°C. In principle, seeded hydrogel pieces could be implanted in suitable cavities within the brain, such as the ventricles, to remove excess Aβ peptides from the CSF. Provided that Aβ concentrations in brain tissue are equilibrated with concentrations in the CSF, this approach may be expected to prevent the development of large excess Aβ concentrations that can lead to neurotoxic Aβ aggregation. As an alternative to direct implantation, seeded hydrogels could be contained within an external compartment that is connected to CSF‐filled cavities within the brain or spinal cord by appropriate tubing, perhaps including a pumping system to maintain fluid flow between the external compartment and the ventricular system of the brain. Experiments with transgenic mice or other animals that have been engineered to produce human Aβ peptides as models for AD can be performed to test these proposals.63, 64 If successful, seeded hydrogels would prevent or delay the onset of memory impairment, Aβ plaque development, or other AD‐like behavioral or neuropathological characteristics in the animals, compared with unseeded hydrogels.

Related approaches from other laboratories deserve mention. First, Sundaram et al. developed a crosslinked polyethylene glycol hydrogel that incorporates the “retro‐inverso” peptide ffvlk (based on the KLVFF segment of Aβ) for binding of Aβ peptides65 and reported that subcutaneous injection of this hydrogel in transgenic mice led to statistically significant reductions in total Aβ42 levels and Aβ plaque sizes in brain tissue, an increase in Aβ42 levels in serum, and an improvement in spatial memory assessed by a Y maze test.66 Compared with the approach of Sundaram et al., the fibril‐seeded hydrogels described above may have larger capacities for sequestering Aβ peptides (because each seed can increase in length by an arbitrary factor in principle, thereby sequestering an unlimited number of Aβ molecules) and may have greater specificity for Aβ peptides (because the entire peptide sequence participates in the molecular recognition process). Moreover, the binding affinities for fibril‐seeded hydrogels are essentially the same as the equilibrium solubilities of Aβ peptides, so that soluble Aβ concentrations should drop to their equilibrium levels in the presence of fibril‐seeded hydrogels, in principle regardless of the initial concentrations and fluid volumes. The same is not true of a hydrogel that incorporates a finite density of binding sites.

Another related approach is the Lixelle adsorption column, which was designed to eliminate β₂‐microglobulin (β₂m) from blood for amelioration of systemic β₂m amyloidosis in hemodialysis patients.67, 68 This column is included in a hemodialysis apparatus and employs hexadecyl alkyl chains within cellulose beads to bind β₂m as blood passes through the column. In principle, fibril‐seeded hydrogels could be used in a similar manner to deplete Aβ peptides from blood with high specificity. The effectiveness of such an approach for AD treatment would depend, among other things, on whether equilibrium is established between Aβ concentrations in blood and Aβ concentrations in brain tissue. Observations that human plasma Aβ concentrations are significantly lower than CSF Aβ concentrations54 and that depletion of Aβ from blood by proteolysis does not reduce brain Aβ levels in transgenic or wild‐type mice or in other mammals69, 70 argue against such an equilibrium. On the other hand, it has been reported that peripheral administration of compounds (such as gelsolin) that bind Aβ with micromolar affinities can significantly reduce brain Aβ levels and plaque loads44 and slow the progression of cerebral amyloid angiopathy43 in transgenic mice.

In a rather different approach, Mancini et al. designed liposomes that cross the blood‐brain barrier and bind Aβ peptides.42 Peripheral administration of these liposomes to transgenic mice was reported to prevent memory impairment, assessed by a novel object recognition test, and slow the accumulation of Aβ plaques. These effects were attributed to liposome‐mediated transport of soluble Aβ from the brain to the blood, followed by Aβ degradation. This work suggests the possibility of reducing brain Aβ levels by combining a treatment that enhances efflux of Aβ from brain tissue or CSF to the blood with an apparatus that places fibril‐seeded hydrogels in continual contact with the blood.

Finally, it should be noted that both Aβ40 and Aβ42 fibrils are polymorphic and that the thermodynamic stabilities of fibrils are polymorph‐dependent.19, 21, 22 If two different polymorphs are present in the same solution, fibrils of the less stable polymorph gradually dissolve and fibrils of the more stable polymorph gradually grow, with the overall rate being greater when the fibrils are divided into shorter lengths.45 Thus, if a fibril‐seeded hydrogel is added to a solution of fibrils (rather than a solution of peptide monomers or small oligomers), and if the seeds in the hydrogel are created from a polymorph that is more stable than the polymorph in solution, then one expects the fibrils in solution to dissolve, gradually transferring peptide molecules to the seeds in the hydrogel. In the biomedical context, this means that seeded hydrogels could in principle be used to dissolve Aβ deposits that have already formed in brain tissue, provided that the Aβ fibril polymorph in the hydrogels is thermodynamically more stable than the polymorphs in the brain tissue.

Materials and Methods

Preparation of Aβ peptides and fibrils

Aβ peptides were synthesized using Fmoc chemistry on a Tribute peptide synthesizer (Protein Technologies). Fmoc‐Val‐Wang (0.27 meq/g, Peptides International) and H‐Ala‐HMPB‐ChemMatrix (0.41 meq/g, PCAS BioMatrix) resins were used for Aβ40 and Aβ42, respectively. Peptides were cleaved from the resins by treatment for 105 min with a standard cocktail [10 mL trifluoroacetic acid (TFA), 0.75 g phenol, 0.25 mL 1,2‐ethanedithiol, 0.5 mL thioanisole, 0.5 mL H₂O), using 7.5 mL of cleavage cocktail for half of each resin. Crude product was precipitated by adding the cleavage cocktail to 45 mL of cold t‐butyl methyl ether (TBME), followed by centrifugation. The precipitate was washed twice with 45 mL of cold TBME before drying in vacuum. For purification, 5–10 mg of crude product were dissolved in 1–2 mL of hexafluoroisopropanol, 3–4 mL of 10% acetic acid were added, and the resulting 5 mL solution was injected into a preparative HPLC system. A 19 × 300 mm Waters μBondapak C18 column was used at room temperature, with a water/acetonitrile gradient containing 0.1% TFA. Final purities were >95% for Aβ40 and >90% for Aβ42, based on liquid chromatography/mass spectrometry (LC‐MS).

For experiments in Figure 6, recombinant 15N‐labeled Aβ40 was obtained from AlexoTech and used without further purification, after verification of purity by LC‐MS. Synthetic or recombinant Aβ peptides were initially solubilized by dissolution in dimethyl sulfoxide (DMSO) to make 1.0 mM solutions, as determined by UV absorbance at 276 nm using the extinction coefficient of 1450 M⁻¹ cm⁻¹. DMSO solutions were then diluted in either phosphate buffer (10 mM sodium phosphate, pH 7.4) or filtered FBS (10 mM sodium phosphate, pH 7.4, 0.1% w/v sodium azide) to the desired concentrations immediately before experiments in Figures 3, 4, 5, 6.

For preparation of Aβ40 fibrils, a 100 μM solution of synthetic Aβ40 in phosphate buffer was seeded with fibril fragments derived from the threefold symmetric Aβ40 fibrils described by Paravastu et al.12 After several days of incubation at 24°C, the fibrils were pelleted by ultracentrifugation and resuspended in phosphate buffer to a final concentration of 800 μM. For preparation of Aβ42 fibrils, a 50 μM unseeded solution of synthetic Aβ42 in phosphate buffer was incubated at 24°C with gentle agitation for one week. After verification of fibril growth (and the absence of nonfibrillar aggregates) by TEM, the fibrils were pelleted by ultracentrifugation and resuspended in phosphate buffer to a final concentration of 800 μM. Seeds were generated by sonication for at least 10 min (Branson model 250 sonifier, 3 mm tapered horn, 10–20% duty cycle, power levels 1–2) immediately before preparation of seeded hydrogels.

Preparation of hydrogels

For MBA‐crosslinked hydrogels, acrylamide (7.33% w/v, Sigma) and MBA (0.23% w/v, Sigma) were dissolved in phosphate buffer containing the desired concentration of Aβ seeds. Ammonium persulfate (0.1% w/v, Thermo Scientific) and N,N,N′,N′‐tetramethylethylenediamine (0.1% v/v, Sigma) were then added. After vortexing for 15 s, the mixture was slowly injected into a mold, consisting of a Teflon piece into which holes with 2.25 mm diameter and 20 mm length had been drilled. After gelation for 6 h at room temperature, rod‐shaped hydrogels were extruded with a plunger and bathed overnight in 45 mL of phosphate buffer. After discarding the buffer, each hydrogel was washed three times with 10 mL of phosphate buffer, then cut into 1.5–3.0 mm sections with a razor blade.

For BAC‐crosslinked hydrogels, acrylamide (6.1% w/v) and BAC (0.43% w/v, Alfa Aesar) were dissolved in phosphate buffer containing the desired concentration of Aβ seeds. Ammonium persulfate (0.075% w/v) and N,N,N′,N′‐tetramethylethylenediamine (0.075% v/v) were then added, and hydrogel rods were prepared as described above.

To test for residual acrylamide monomer in the MBA‐crosslinked hydrogels, freshly prepared hydrogel sections, with total volumes of 100 μL and without seeds, were washed by incubation at room temperature for 24 h in 15 mL tubes containing 10 mL of 10 mM sodium phosphate buffer, with continuous end‐over‐end rotation of the tubes. The buffer was then removed for analysis. The same washing procedure was repeated four times. After the first wash, the buffer was found to contain 24.3 ± 3.4 ng/mL of acrylamide monomer (mean and standard deviation from four independent trials), using absorbance at 220 nm in HPLC chromatograms to quantify acrylamide concentrations. After the second wash, the buffer was found to contain 0.075 ± 0.011 ng/mL of acrylamide monomer. Acrylamide monomer was undetectable (less than 0.02 ng/mL) after the third and fourth washes. From these data, we conclude that residual acrylamide monomer, which is potentially neurotoxic, can be removed from the hydrogels by repeated washing. Moreover, the amount of acrylamide monomer collected in the first wash represents ∼0.005% of the monomer in the polymerization mixture, indicating that polymerization proceeds nearly to completion.

To test for leakage of fibril seeds from the hydrogels during incubation, freshly prepared hydrogel sections, with a total volume of 100 μL and with 400 μM Aβ40 seeds, were washed twice with 10 mL of 10 mM sodium phosphate buffer for 2 h, then incubated at room temperature for 45 h in a 1.5 mL tube containing 1.0 mL of buffer, with continuous end‐over‐end rotation of the tube. After incubation, a 0.5 mL aliquot of the buffer was taken from the 1.5 mL tube and lyophilized. The lyophilized material (which would contain Aβ40 fibril fragments if leakage from the hydrogels occurred) was then dissolved with 60 μL of formic acid, mixed with 60 μL of acetonitrile, and analyzed by HPLC. Based on absorbance at 220 nm, the formic acid/acetonitrile solution contained ∼0.15 μM Aβ40, corresponding to a concentration of ∼0.04 μM Aβ40 in the buffer before lyophilization. We attribute this low level of Aβ40 to gradual dissolution of seeds within the hydrogel sections, which is expected to occur until the Aβ40 concentration reaches its equilibrium solubility level (∼0.4 μM at 24°C). We find no evidence for leakage of fibril seeds from the hydrogels.

Incubation of hydrogel‐containing Aβ solutions

For experiments in Figures 3 and 5, 100 μL total volumes of hydrogel sections were placed in flat‐bottom glass vials with 8.0 mL capacities. A 2.0 mL volume of either 10 μM Aβ40 or 5 μM Aβ42 was added. During quiescent incubation at 24°C, 20 μL aliquots were taken at various time points for analytical HPLC measurements. Three vials were prepared for each experimental condition.

For experiments in Figure 6, FBS (Gibco) was first filtered through a centrifugal filter device (Amicon Ultra‐15 3K, Millipore). Sodium azide (0.1% w/v) and phosphate buffer powder (10 mM) were added, and the resulting solution was stored at −20°C prior to use. Hydrogels were prepared as described above, but with the additional precautions that the phosphate buffer, gel mold, razor blade, plunger, and glass vials were autoclaved at 115°C for 15 min before use. Glass vials with 2.0 mL capacities were used. The 1.2 mL volumes of 50 nM recombinant, 15N‐labeled Aβ40 were added to vials that contained 100 μL total volumes of hydrogel sections, and 80 μL aliquots were taken at various time points during quiescent incubation at 37°C. Three vials were prepared for each experimental condition. Each aliquot was lyophilized, then dissolved in 30 μL of 96% formic acid, with alternating sonication and vortexing for 4–5 min. Prior to LC‐MS measurements, 70 μL of degassed water containing 100 nM of melittin (see below) was added to each aliquot.

Analytical HPLC measurements

Soluble Aβ concentrations for experiments in Figures 3 and 5 were determined from ultraviolet absorbance measurements with a Thermo Dionex Ultimate 3000 HPLC system, using a 3.9 × 150 mm² Waters μBondapak C18 column at 40°C with 1.0 mL/min flow rates and a linear solvent gradient from 10% acetonitrile/95% water/0.01% TFA to 60% acetonitrile/40% water/0.01% TFA over 12 min after 3 min equilibration at 5% acetonitrile/95% water/0.01% TFA. Eight microliters of a 1:1:1 mixture of acetic acid, DMSO, and acetonitrile were added to each 20 μL aliquot from these experiments, and the resulting samples were placed in the HPLC autosampler, which was maintained at 10°C. Injection volumes were 20 μL. Absorbance peaks from Aβ40 or Aβ42 at 220 nm (Fig. S1, Supporting Information) were integrated with Thermo Chromeleon software, which also controlled the HPLC system.

LC–MS measurements

Soluble A40 concentrations for experiments in Figure 6 were determined by mass spectrometry, using the Thermo Dionex Ultimate 3000 HPLC system and Thermo MSQ Plus single quadrupole mass spectrometer. An Acclaim PepMap 300 C18 column (Thermo Fisher Scientific) was used at 40°C, with 0.2 mL/min flow rates and a linear solvent gradient from 5% acetonitrile/95% water/0.01% TFA to 60% acetonitrile/40% water/0.01% TFA over 20 min after 2 min equilibration at 5% acetonitrile/95% water/0.01% TFA. Injection volumes were 80 μL from each 100 μL aliquot prepared as described above. Elution times and mass‐to‐charge ratio (m/z) spectra of ¹⁵N‐labeled Aβ40 and melittin (added to each aliquot as an internal m/z intensity reference) were first determined using standard samples. Selected ion monitoring (SIM) was then used to only monitor the three most intense m/z peaks for Aβ40 (877.2 Da, 1096.0 Da, and 1461.1 Da, eluting at ∼16.2 min) and the two most intense peaks for melittin (712.3 Da and 949.6 Da, eluting at ∼19.3 min). Aβ40 concentrations were determined from peak areas in total ion count (TIC) chromatograms, after normalizing to a constant melittin peak area [Fig. S4(A), Supporting Information].

A separate set of SIM measurements on Aβ40/melittin solutions with a range of Aβ40 concentrations and constant melittin concentrations indicated that the normalized Aβ40 peak areas in TIC chromatograms varied approximately linearly with Aβ40 concentration from 2.5 to 50 nM under the conditions described above [Fig. S4(B), Supporting Information].

ThT fluorescence

For experiments in Figure 4, 100 μL total volumes of BAC‐crosslinked hydrogels were placed in flat‐bottom glass vials with 2.0 mL capacity. Two milliliters of 10 μM Aβ40 in phosphate buffer were added to each vial. After quiescent incubation at 24°C for various time periods, hydrogels were removed from the vials, washed three times with 1.0 mL of phosphate buffer, then immersed overnight in 200 μL of neat 2‐mercaptoethanol solution to dissolve the disulfide crosslinks in the hydrogels. One vial was prepared for each time point. For ThT fluorescence measurements, ∼20 μL of 540 μM ThT solution were mixed with the dissolved hydrogel, to produce a final ThT concentration of 60 μM. Fluorescence emission spectra were recorded at room temperature with a StellarNet Black‐Comet‐TEC fluorescence spectrometer, using excitation at 423 nm from a monochromatic light‐emitting diode and using a Hellma 105.250‐QS cuvette.

Transmission electron microscopy

TEM images were obtained with an FEI Morgagni transmission electron microscope, operating at 80 kV and equipped with an AMT Advantage HR CCD camera. For images in Figures 2, S2, and S3, solutions of fibrils or fibril seeds were diluted by factors of 5–10 in deionized water. Ten microliter aliquots were adsorbed to glow‐discharged carbon films on lacey‐carbon‐coated copper mesh grids for ∼60 s. Grids were then blotted, rinsed with 10 μL of deionized water, blotted, stained with 10 μL of 3% w/v uranyl acetate for 15 s, blotted, and dried in air.

Dissolved hydrogels from ThT measurements were used for images in Figure 4(C). Approximate 5 μL aliquots of dissolved hydrogel solutions were adsorbed to glow‐discharged grids, which were then rinsed multiple times with 10 μL aliquots of deionized water, until the viscous hydrogel solutions were mostly removed. Grids were then stained with 10 μL of 3% w/v uranyl acetate for 15 s, blotted, and dried in air.

Preparation and testing of polyacrylamide microspheres

Polyacrylamide microspheres were prepared with the apparatus in Figure S5, Supporting Information. For unseeded microspheres, a 1.0 mL syringe (Hamilton model 1001 TLL SYR) was loaded with 200 μL of cyclohexane, followed by 500 μL of a solution consisting of 487 μL of 15% w/v acrylamide and 0.5% w/v MBA in sodium phosphate buffer (10 mM, pH 7.4, degassed), 10 μL of TEMED, and 3 μL of allyl amine. A 0.1 mL syringe (Hamilton model 1710 TLL SYR) was loaded with 70 μL of 20% w/v APS in sodium phosphate buffer. The contents of the two syringes were mixed at a total flow rate of 1.1 mL/min, using a syringe pump (KD Scientific model 80230) and the “T mixer” arrangement shown in Fig. S5, Supporting Information, which was constructed from two polyether ether ketone tubes (20 cm lengths, 250 μm inner diameters), a stainless‐steel T junction, and a brass exit tube (9 cm length) into which was soldered a copper tube with 100 μm inner diameter. The end of the exit tube was immersed in a 2.0 mL volume of mineral oil, which was mixed with Tween 80 and Span 80 surfactants (50 μL each) and which was stirred rapidly. Activation of the syringe pump resulted in injection of the acrylamide mixture into the mineral oil mixture, producing a white “water‐in‐oil” emulsion.

For seeded microspheres, Aβ40 fibrils were grown from recombinant peptide (rPeptide, Watkinsville, Georgia) at a concentration of 70 μM, pelleted at 280,000g for 60 min, resuspended in deionized water, and pelleted again for 60 min. The fibril pellet was then resuspended in phosphate buffer containing 15% w/v acrylamide and 0.5% w/v MBA to an Aβ40 concentration of 140 μM (assuming complete fibrillation of Aβ40 and negligible loss during centrifugation steps). Fibrils were fragmented by sonication as described above before addition of allyl amine and TEMED and loading into the 1.0 mL syringe.

Cyclohexane was included in the 1.0 mL syringe (adjacent to the plunger) so that cyclohexane would wash the acrylamide solution out of the T mixer at the end of each injection, thereby minimizing polymerization within the T mixer. In experiments with bulk solutions, addition of 100 μL of 20% w/v APS to 1.0 mL of an acrylamide/MBA/TEMED solution with the composition given above resulted in gelation within 10 s. We therefore believe that droplets in the emulsion created by rapid mixing and injection into mineral oil polymerize to form polyacrylamide microspheres within 10 s or less.

Examination of the emulsions by optical microscopy (Olympus model BX50 with SPOT model 1.5.0 camera) indicated microsphere diameters primarily in the 20–100 μm range. Microspheres were extracted from mineral oil and washed by addition of deionized water (∼10 mL), vortexing, and pelleting (18,000g, 60 min, swinging bucket rotor). Washing was performed three times. After addition of fluorescein isothiocyanate dye (dissolved in DMSO) to an aliquot of the washed, hydrated microspheres, the microspheres became fluorescent due to reaction of the dye with amine groups of allyl amine monomers that were copolymerized with acrylamide monomers. For images in Figure 7(A), a suspension of microspheres on a glass slide was illuminated with blue light (Nightsea SFA‐RB source, 450 nm wavelength) and imaged by using only the red and green channels of the microscope's camera.

Lyophilization of washed microspheres resulted in a white powder. From the mass of this powder, we estimate the yield of polyacrylamide microspheres to be ∼40%, relative to the initial mass of acrylamide. Upon addition of aqueous buffer, lyophilized microspheres returned to their original spherical shapes and diameters within several seconds, as verified by optical microscopy.

For data in Figure 7(B), 50 μL aliquots of Aβ40 solutions were taken at each time point and immediately mixed with 40 μL of 1:1 v/v acetonitrile/formic acid. HPLC measurements used 80 μL injections, an Acclaim PepMap 300 C18 column (Thermo Fisher Scientific), 0.2 mL/min flow rates, and a linear solvent gradient from 5% acetonitrile/95% water/0.01% TFA to 60% acetonitrile/40% water/0.01% TFA over 20 min after 2 min equilibration at 5% acetonitrile/95% water/0.01% TFA. Aβ40 concentrations were measured as the areas of the appropriate chromatogram peak, detected at 280 nm. The identity of the chromatogram peak was confirmed by mass spectrometry.

Supporting information

Supporting Information