QCM‑D Reveals Parallel Adsorption Pathways Governing Amyloid‑β Binding to LRP‑1

Abstract

The interaction of amyloid-β (Aβ) with the low-density lipoprotein receptor-related protein-1 (LRP-1) plays a central role in amyloid clearance, yet the molecular mechanism governing this interaction remains poorly understood. Here, quartz crystal microbalance with dissipation monitoring (QCM-D) was used to investigate the adsorption kinetics of Aβ onto immobilized LRP-1 in real time. Adsorption produced a substantial decrease in resonance frequency accompanied by an increase in dissipation, consistent with the formation of a viscoelastic protein layer. Sequential washing with phosphate-buffered saline and Hellmanex III detergent revealed two populations of adsorbed Aβ that differed markedly in their resistance to removal. To distinguish between possible kinetic mechanisms, the experimental sensorgrams were fitted using competing adsorption models. A model comprising two independent parallel adsorption pathways accurately reproduced the experimental data and yielded randomly distributed residuals, whereas a sequential adsorption–rearrangement model exhibited systematic deviations characteristic of model misspecification. Analysis across a concentration series further supported the presence of two kinetically distinguishable adsorption components. Together, these results demonstrate that Aβ binding to LRP-1 is governed by heterogeneous adsorption kinetics involving parallel interaction pathways rather than a single sequential binding mechanism, providing a more realistic framework for interpreting receptor-mediated protein adsorption.

Introduction

Alzheimer’s disease (AD) is the most common cause of dementia and is characterized by the progressive accumulation of amyloid-β (Aβ) peptides within the brain. , Although the mechanisms underlying disease progression are multifactorial, impaired clearance of Aβ from the central nervous system is now recognized as an important contributor to peptide accumulation and plaque formation. − Consequently, considerable effort has been directed toward identifying the molecular pathways responsible for regulating Aβ transport, sequestration, and elimination. Understanding these processes is essential for elucidating the molecular basis of AD and for developing therapeutic strategies aimed at restoring normal amyloid homeostasis. Among the proteins implicated in Aβ clearance, the low-density lipoprotein receptor-related protein-1 (LRP-1) has emerged as one of the principal receptors involved in maintaining brain Aβ homeostasis. LRP-1 is a large endocytic receptor expressed by neurons, astrocytes, vascular smooth muscle cells, and the endothelial cells of the blood-brain barrier, where it mediates the uptake and transport of a remarkably diverse range of ligands. − Numerous studies have demonstrated that LRP-1 facilitates the removal of Aβ from the brain parenchyma by promoting receptor-mediated endocytosis and transcytosis across the blood-brain barrier. Reduced LRP-1 expression or impaired receptor function has been associated with diminished Aβ clearance and increased amyloid burden, highlighting its central role in Alzheimer’s disease pathology. ,

Despite extensive evidence implicating LRP-1 in Aβ clearance, considerably less is known about the molecular mechanism by which Aβ initially associates with the receptor. Previous studies have established the interaction of Aβ with LRP-1 using complementary biochemical and biophysical approaches, including surface plasmon resonance (SPR) measurements with recombinant LRP-1 ligand-binding domains, radiolabeled Aβ transport and clearance studies, , fluorescence-based cellular binding and uptake assays, , and biochemical approaches to assess LRP-1-dependent Aβ association and clearance. In particular, SPR measurements have provided quantitative binding affinities for Aβ interactions with LRP-1 Clusters II and IV, whereas cellular and biochemical studies have independently demonstrated LRP-1-dependent Aβ association, internalization, trafficking, and clearance. ,, Collectively, these studies establish LRP-1 as an important Aβ-binding and clearance receptor but provide comparatively limited information regarding whether kinetically distinct Aβ populations arise through parallel binding pathways or sequential postbinding rearrangements. It remains unclear whether the characteristic biphasic binding behavior frequently observed for protein adsorption arises from a single binding event followed by a slower conformational or structural rearrangement, or whether it reflects the simultaneous operation of multiple independent adsorption pathways. Distinguishing between these mechanistic possibilities is essential because each implies fundamentally different molecular processes governing receptor recognition and surface association.

Quartz crystal microbalance with dissipation monitoring (QCM-D) provides a powerful approach to addressing this question by simultaneously measuring changes in adsorbed mass and viscoelastic properties in real time. − Unlike equilibrium measurements that report only the final bound state, QCM-D continuously follows the entire adsorption process, allowing transient kinetic behavior, adsorption heterogeneity, and postbinding structural changes to be resolved. Furthermore, simultaneous measurement of the resonance frequency and energy dissipation enables discrimination between rigid and viscoelastic adsorbed layers, providing information that is not readily accessible using conventional biosensing techniques. When combined with appropriate kinetic modeling, QCM-D therefore offers the opportunity to distinguish between competing adsorption mechanisms and identify the processes responsible for the complex binding behavior.

In the present study, QCM-D was used to investigate the kinetics of Aβ adsorption onto immobilized LRP-1 (recombinant human LRP-1 Cluster II Fc Chimera). Competing kinetic models were evaluated to determine whether the observed adsorption behavior is more accurately described by a sequential adsorption–rearrangement mechanism or by two independent parallel adsorption pathways. Concentration-dependent measurements were further used to assess the generality of the observed kinetic behavior. By establishing the kinetic mechanism governing Aβ–LRP-1 association, this work provides new insight into the molecular basis of receptor recognition and demonstrates how real-time adsorption measurements can distinguish between mechanistically distinct pathways that are indistinguishable from equilibrium binding measurements alone.

Experimental Section

Reagents

11-Mercaptadunoic acid (11-MUA, CAS 71310-21-9), ethanolamine (CAS 141-43-5), Aβ peptide (1-42), human (CAS PP69-.25MG), phosphate-buffered saline (PBS, CAT #P2272-1L), and N-hydroxysuccinimide (NHS, CAS 6066-82-6) were obtained from Sigma-Aldrich. 1-Ethyl-3-(3-(dimethylamino)­propyl) carbodiimide hydrochloride (EDC, CAT #22981) was from Thermo Fisher Scientific. Recombinant human LRP-1 Cluster II Fc Chimera (CAT #2368-L2-050) was from Bio-Techne, R&D Systems. Hellmanex (Hellma Analytics, Müllheim, Germany), an alkaline laboratory cleaning solution, was used for sensor preparation and regeneration. It contains sodium dodecyl benzenesulfonate (SDS; CAS 25155-30-0), sodium alkylbenzenesulfonate (CAS 68411-30-3), alkylphenol ethoxylates (CAS 14409-72-4), isopropanol (CAS 67-63-0), deionized water (CAS 7732-18-5), EDTA (CAS 60-00-4), sodium hydroxide (CAS 1310-73-2), sodium carbonate (CAS 497-19-8), sodium metasilicate (CAS 6834-92-0), and proprietary antifoaming agents. All chemicals were used as received without further purification.

Preparation of Aβ Peptide 42

The purity of the peptide is reported to be 95% by HPLC. The peptide was reconstituted in DMSO and prepared in 500 nM and 1000 nM aliquots in DMSO and stored at −20 °C. Solutions in aliquots were then diluted in PBS to 100, 166, 250, and 500 pM immediately before each experiment. The oligomeric state of Aβ42 was not independently characterized immediately prior to the QCM-D measurements; the same preparation protocol and peptide lot were used for all experiments to minimize variability in the initial peptide state.

LRP-1 Cluster II Fc Chimera

The recombinant human LRP-1 Cluster II Fc Chimera used in this study represents a defined portion of the much larger full-length LRP-1 receptor. Full-length LRP-1 is an approximately 600 kDa glycoprotein that is proteolytically processed into a 515 kDa extracellular α-chain and an 85 kDa membrane-associated β-chain. In contrast, the recombinant construct contains only residues Arg786–Leu1165 of human LRP-1, corresponding to ligand-binding Cluster II, fused to residues Pro100–Lys330 of human IgG1 Fc. The construct has a predicted molecular mass of 68.1 kDa per polypeptide chain and exists as a disulfide-linked homodimer; its apparent molecular mass by reducing SDS-PAGE is approximately 90–100 kDa. Structurally, Cluster II contains eight LDL receptor class A ligand-binding repeats and one EGF-like repeat, whereas the extracellular region of full-length LRP-1 contains four ligand-binding clusters comprising 31 LDL receptor class A repeats together with additional EGF-like and β-propeller domains. , Thus, the immobilized Cluster II Fc construct preserves a major extracellular ligand-binding region of LRP-1 while lacking the remaining extracellular domains, transmembrane segment, and cytoplasmic signaling and endocytic domains of the native receptor. ,

LRP-1 Film Formation via EDC/NHS Coupling

A gold QSX 301 Q-Sense sensor was cleaned by sonication in a 2% Hellmanex solution for 15 min to remove organic contaminants. The sensor was then rinsed by alternating washes of Milli-Q ultrapure water (resistivity 18.2 MΩ·cm) and anhydrous ethanol, each for 15 s, to eliminate residual detergent. After drying under a gentle stream of nitrogen gas, the sensor surface was further treated with UV–ozone for 15 min to ensure complete removal of organic residues and to increase surface hydrophilicity. Following cleaning, the sensor was functionalized by immersing it in 10 mM 11-mercaptoundecanoic acid (11-MUA) dissolved in methanol overnight at 4 °C to form a self-assembled monolayer. The carboxyl groups were then activated by incubation in 5 mM EDC/NHS prepared in Milli-Q water for 2 h at 4 °C. Subsequently, the activated surface was incubated with 25 μg/mL of LRP-1 Cluster II Fc Chimera in PBS for 2 h at 4 °C to allow covalent coupling. Finally, unreacted sites were blocked by incubating the sensor in ethanolamine–HCl buffer for 2 h to prevent nonspecific binding during QCM-D measurements. Between each incubation, the sensor was washed with the following solvents (in order as steps are described): anhydrous ethanol, PBS, Milli-Q, and finally Milli-Q following ethanolamine incubation.

QCM-D Experiments

An E4 QCM-D (Nanoscience Instruments) was used to measure changes in frequency and dissipation. The seventh harmonic frequency was used for consistency. Each QCM-D crystal has an active area of 0.2 cm² and operates at a fundamental frequency of 4.95 MHz ± 50 kHz. All QCM-D measurements were performed at 37 °C using PBS (pH 7.2) as the running buffer. PBS was initially flowed over the sensor at 0.1 mL/min to establish a baseline, which is defined as a change in frequency of less than 1 Hz over a period of 5 min. Following baselining, Aβ was flowed through at 0.1 mL/min at 100 pM. Once a plateau was achieved, PBS was flowed through to rinse the surface and wash off any loosely bound Aβ. This was followed by a perfusion of 2% Hellmanex and PBS to regenerate the LRP-1 surface. The perfusion of each was stopped when a plateau was observed. The above process of perfusing Aβ and washing off the loosely bound peptide was repeated for three other concentrations of Aβ on the same LRP-1 surface: 166, 250, and 500 pM. All experiments were repeated thrice.

Data Fitting

All QCM-D Aβ binding transients were expressed as the magnitude of the frequency shift (|ΔF|), which is proportional to the adsorbed mass under the Sauerbrey approximation. Although the Sauerbrey relation is strictly valid only for thin, rigid, and uniformly coupled films, adsorbed protein layers may exhibit viscoelastic contributions that introduce systematic errors in the absolute mass estimate. , In the present study, however, the Sauerbrey transformation is used primarily as a linear scaling of the measured frequency shift rather than as a quantitative determination of adsorbed mass. Consequently, any systematic error in the mass conversion affects only the absolute amplitude of the response and does not alter the temporal profile of the binding transient or the kinetic parameters extracted from fitting the association curves. Nonlinear regression analyses were performed using GraphPad Prism (version 11, GraphPad Software, San Diego, CA). Time-dependent binding transients were fitted using user-defined equations describing the proposed kinetic models. Equilibrium binding data were analyzed using Prism’s built-in two-site binding model to obtain approximate dissociation constants for the two binding populations. All fits were performed by least-squares nonlinear regression. Three independent experiments were analyzed separately using the same fitting procedure, and the reported kinetic parameters correspond to the mean ± standard deviation of the fitted values.

Results and Discussion

QCM-D Reveals Heterogeneous Adsorption of Aβ onto Immobilized LRP-1

The overall QCM-D response observed during adsorption, washing, and sensor regeneration is shown in Figure . Introduction of amyloid-β to the LRP-1-coated sensor (Cluster II of LRP-1) resulted in an immediate decrease in resonance frequency (ΔF) accompanied by an increase in dissipation (ΔD), indicating rapid adsorption of Aβ to the immobilized receptor (Figure a). The frequency continued to decrease throughout the exposure period, reaching approximately −160 Hz, while the dissipation increased to approximately 2.8 ppm. The simultaneous decrease in frequency and increase in dissipation demonstrate that adsorption involves not only the accumulation of mass at the sensor surface but also the formation of a relatively soft and viscoelastic adsorbed layer. Following adsorption, the sensor was rinsed extensively with phosphate-buffered saline (PBS) to remove reversibly associated material (Figure b). The PBS wash produced a rapid increase in frequency together with a corresponding decrease in dissipation, indicating that a portion of the adsorbed Aβ was readily removed from the surface. However, the frequency remained substantially below its original baseline after prolonged rinsing, demonstrating that a significant fraction of the deposited material remained associated with the immobilized LRP-1. Thus, buffer washing alone was insufficient to completely regenerate the sensor surface. To remove the remaining adsorbed material, the sensor was subsequently exposed to Hellmanex III detergent solution (Figure c). This treatment produced a much larger increase in frequency, indicating efficient removal of the remaining surface-associated Aβ. Following a final PBS rinse (Figure d), both the frequency and dissipation returned to values essentially identical to those observed prior to Aβ exposure. The complete recovery of both QCM-D signals demonstrates that the regeneration protocol effectively removed the adsorbed Aβ while preserving the underlying immobilized LRP-1 layer, thereby allowing reproducible measurements to be performed on the same sensor surface. The successful regeneration of the sensor surface is consistent with previous QCM-D investigations of LRP-1–ligand interactions, where repeated regeneration with Hellmanex III and buffer permitted multiple binding cycles on the same immobilized LRP-1 surface without detectable loss of receptor activity.

1.

QCM-D monitoring of Aβ adsorption to immobilized LRP-1 during adsorption, washing, and sensor regeneration. Changes in resonance frequency (ΔF, blue) and dissipation (ΔD, red) were recorded continuously throughout the experiment. (a) Introduction of Aβ (250 pM) resulted in a rapid decrease in frequency accompanied by an increase in dissipation, indicating adsorption of Aβ and formation of a viscoelastic adsorbed layer on the immobilized LRP-1 surface. (b) Rinsing with phosphate-buffered saline (PBS) removed a readily dissociable fraction of the adsorbed material, producing a partial recovery of both ΔF and ΔD while leaving a substantial amount of Aβ associated with the surface. (c) Exposure to Hellmanex III detergent removed the remaining strongly associated Aβ, producing a further increase in frequency and decrease in dissipation. (d) A final PBS rinse restored both ΔF and ΔD to values comparable to those observed before Aβ exposure, demonstrating complete regeneration of the sensor surface while preserving the immobilized LRP-1 layer. The differential removal of Aβ by PBS and Hellmanex III provides qualitative evidence for the presence of adsorption populations with different binding stabilities, consistent with the heterogeneous adsorption behavior revealed by the kinetic analysis.

The incomplete removal of Aβ by PBS (Figure b), followed by complete removal with Hellmanex III (Figure c,d), indicates that the adsorbed layer contains populations exhibiting different stabilities on the sensor surface. A readily removable fraction is displaced by buffer alone, whereas a second population persists during extensive washing and requires detergent treatment for complete removal. This observation is consistent with the heterogeneous binding behavior revealed by the kinetic analysis presented below. Although QCM-D alone cannot uniquely identify the molecular origin of these populations, the regeneration experiments provide independent evidence that Aβ does not interact with immobilized LRP-1 through a single, kinetically homogeneous binding process.

Kinetic Model Discrimination Supports Parallel Adsorption Pathways

When a biomolecule adsorbs to an immobilized receptor, the observed binding kinetics may arise from more than one underlying molecular mechanism. One possibility is sequential adsorption, in which the analyte first forms an initial surface complex that subsequently undergoes a slower conformational or orientational rearrangement to produce the final bound state. Such behavior is frequently observed when binding induces structural changes in either the ligand or receptor, or when the initially adsorbed complex undergoes surface relaxation. This process can be described by,

$$\mathrm{L}\mathrm{R}\mathrm{P}+\mathrm{A}\beta \underset{k_{-1}}{\overset{k_1}{\rightleftharpoons }}[\mathrm{L}\mathrm{R}\mathrm{P}-\mathrm{A}\beta ]\underset{k_{-2}}{\overset{k_2}{\rightleftharpoons }}{[\mathrm{L}\mathrm{R}\mathrm{P}-\mathrm{A}\beta ]}_{\mathrm{final}}$$

And the kinetic model can be described by,

$$|\mathrm{\Delta }F|(t)={|\mathrm{\Delta }F|}_{\max }({1-e}^{{-k}_{\mathrm{o}\mathrm{b}\mathrm{s}}t})$$ 1

For a given Aβ concentration ([Aβ]₀), the observed rate constant (k _obs) is,

$$k_{\mathrm{o}\mathrm{b}\mathrm{s}}=k_{-2}+k_2\frac{{[\mathrm{A}\beta ]}_0}{{[\mathrm{A}\beta ]}_0+K_{\mathrm{d},1}}$$ 2

where K _d,1 is the binding constant of the first step (K _d,1 = k _–1/k ₁). Alternatively, adsorption may occur through parallel binding pathways, whereby two independent populations of binding events proceed simultaneously, each characterized by distinct kinetic and thermodynamic properties. This process can be described by,

$$\begin{array}{l}\mathrm{L}\mathrm{R}\mathrm{P}+\mathrm{A}\beta \underset{k_{-1}}{\overset{k_1}{\rightleftharpoons }}{[\mathrm{L}\mathrm{R}\mathrm{P}-\mathrm{A}\beta ]}_1\\ \mathrm{L}\mathrm{R}\mathrm{P}+\mathrm{A}\beta \underset{k_{-2}}{\overset{k_2}{\rightleftharpoons }}{[\mathrm{L}\mathrm{R}\mathrm{P}-\mathrm{A}\beta ]}_2\end{array}$$ 3

The corresponding kinetic model for this parallel process is,

$$[\mathrm{\Delta }F](t)=|\mathrm{\Delta }F{|}_{\max ,1}(1-e^{-k_{\mathrm{o}\mathrm{b}\mathrm{s},1}t})+|\mathrm{\Delta }F{|}_{\max ,2}(1-e^{-k_{\mathrm{o}\mathrm{b}\mathrm{s},2}t})$$ 4

In this model, the two observed rate constants are,

$$\begin{array}{l}{k}_{\mathrm{o}\mathrm{b}\mathrm{s},1}=k_{-1}+k_1{[\mathrm{A}\beta ]}_0\\ k_{\mathrm{o}\mathrm{b}\mathrm{s},2}=k_{-2}+k_2{[\mathrm{A}\beta ]}_0\end{array}$$ 5

In all expressions above, the |ΔF|_max values are the equilibrium QCM frequency shifts at a given Aβ concentration as t → ∞. Both above models assume pseudo-first-order conditions such that the analyte ([Aβ]₀) concentration remains effectively constant during the experiment.

In heterogeneous biological systems, parallel processes may reflect receptor-specific interactions occurring alongside weaker nonspecific adsorption or binding to multiple classes of receptor sites. Because both mechanisms can generate biphasic adsorption kinetics, distinguishing between them requires quantitative comparison of competing kinetic models rather than visual inspection of the sensorgrams alone. Given the structural complexity of Aβ and the heterogeneous nature of the immobilized LRP-1 surface, both sequential adsorption–rearrangement and parallel binding mechanisms represent physically plausible descriptions of the observed QCM-D response.

To distinguish between possible adsorption mechanisms, the experimental binding curve was fitted using two competing kinetic models using eqs and (Figure ). The latter assumes that Aβ binds to LRP-1 through two independent, parallel adsorption pathways, each characterized by its own maximum binding capacity and observed rate constant. The former represents a sequential adsorption mechanism in which an initially formed surface complex undergoes a slower postadsorption rearrangement before reaching the final bound state. As shown in Figure a, the parallel binding model closely reproduces the complete experimental sensorgram over the entire time course, whereas the sequential adsorption–rearrangement model systematically deviates from the data, particularly during the transition between the initial rapid adsorption phase and the subsequent slower approach to equilibrium. These differences become more apparent in the residual analysis (Figure b). The residuals obtained from the parallel binding model are randomly distributed about zero throughout the experiment, indicating that the model adequately captures the observed adsorption kinetics. In contrast, the sequential adsorption model produces pronounced time-dependent residuals, demonstrating systematic deviations between the model and the experimental data that are characteristic of model misspecification rather than experimental noise. Collectively, these results indicate that Aβ adsorption to LRP-1 is more consistent with two kinetically distinct binding pathways than with a single adsorption event followed by a conformational or structural rearrangement. While the kinetic analysis alone does not establish the molecular identities of these two binding processes, the results are consistent with the coexistence of a high-affinity, receptor-specific interaction together with a weaker, nonspecific adsorption pathway that proceeds independently in parallel. The sequential adsorption–rearrangement model fails to account for the initial adsorption kinetics, producing systematic residuals characteristic of model misspecification. In contrast, the parallel-process model yields randomly distributed residuals, indicating that the observed kinetics are better explained by two kinetically distinct adsorption pathways.

2.

Comparison of parallel and sequential kinetic models for Aβ adsorption to immobilized LRP-1. (a) Experimental QCM-D binding data (red symbols) were fitted using two competing kinetic models. The parallel binding model (black curve) assumes that Aβ (100 pM) adsorbs through two independent binding pathways, each characterized by its own maximum binding capacity and observed rate constant. The two-step binding model (blue curve) assumes that adsorption is followed by a slower post-adsorption rearrangement of the surface-bound complex. The parallel model accurately reproduces the experimental sensorgram over the entire time course, whereas the sequential adsorption–rearrangement model exhibits systematic deviations from the data. (b) Residual plots for the parallel binding model (i) and the two-step binding model (ii) during initial 100 s of deposition. Residuals from the parallel model are randomly distributed about zero, indicating an adequate description of the adsorption kinetics. In contrast, the two-step model produces pronounced time-dependent residuals, demonstrating systematic model misspecification. Collectively, these results indicate that Aβ adsorption to LRP-1 is more consistent with two kinetically distinct binding pathways than with a single adsorption event followed by conformational rearrangement.

It should be emphasized that the sequential adsorption–rearrangement model considered here represents a simplified kinetic description. A general coupled mechanism, A + L ⇌ AL₁ ⇌ AL₂, can exhibit more than one kinetic relaxation depending on the forward and reverse rate constants of the individual steps and therefore need not reduce to a single-exponential response. Accordingly, the superior agreement obtained with the two-component parallel model demonstrates that it provides a better description of the present QCM-D data than the simplified sequential model tested here, but does not constitute proof that all possible sequential adsorption–rearrangement mechanisms can be excluded.

To further evaluate whether the parallel-binding behavior was retained across concentrations, Aβ adsorption to immobilized LRP-1 was examined over a concentration series, and each sensorgram was fitted using the same two-process kinetic model (Figure a). In all cases, the experimental traces were better described by the parallel adsorption model (eq ) than by a single sequential adsorption–rearrangement process (eq ), indicating that the two-component behavior was not unique to a single concentration. Equilibrium dissociation constants (K _d) are not reported because the Aβ concentration range examined (100–500 pM) does not sufficiently span the binding isotherm or approach saturation. Consequently, K _d values derived from these data would be poorly constrained and potentially misleading. We therefore restrict our quantitative interpretation to the kinetic parameters and adsorption behavior directly supported by the QCM-D measurements. The persistence of two kinetically distinguishable components across the entire concentration range nevertheless provides additional support for a heterogeneous parallel adsorption mechanism over a single binding process.

3.

Concentration dependence of the two kinetic adsorption populations identified by parallel kinetic analysis. (a) Representative QCM-D adsorption transients for Aβ binding to immobilized LRP-1 at increasing Aβ concentrations. Solid lines represent fits to the parallel adsorption model, which resolves the total adsorption into fast and slow kinetic components with equilibrium amplitudes B_1 and B_2. (b) The equilibrium amplitudes obtained from the kinetic fits (B₁, blue circles; B₂, red squares) plotted as a function of Aβ concentration. Error bars represent the standard deviation from three independent experiments.

Discussion

The present results provide new mechanistic insight into the interaction between Aβ and LRP-1 by demonstrating that the adsorption kinetics are more consistent with parallel binding pathways than with a single adsorption event followed by a slower conformational rearrangement. The observation that a readily removable population coexists with a more strongly associated fraction (Figure ), together with the superior performance of the parallel kinetic model across multiple concentrations (Figure ), suggests that heterogeneous binding is an intrinsic property of the Aβ–LRP-1 interaction rather than an artifact of a particular experimental condition. Such behavior is biologically plausible given the structural complexity of both LRP-1 and Aβ. LRP-1 , contains multiple ligand-binding complement-type repeats capable of engaging diverse molecular partners, while Aβ is a conformationally dynamic peptide that can adopt multiple surface-associated states. Consequently, adsorption through more than one independent pathway is a realistic molecular scenario and may represent the combined contributions of receptor-specific recognition together with weaker nonspecific interactions occurring simultaneously at the sensor surface. Deane et al. reported two apparent dissociation constants for Aβ binding to LRP-1; however, these values were obtained by studying recombinant ligand-binding domains rather than the intact receptor. In the present study, only Cluster II was examined. Nevertheless, the kinetic analysis revealed two parallel adsorption pathways, demonstrating that kinetic heterogeneity exists within Cluster II itself. This finding indicates that the observed heterogeneity does not require the simultaneous presence of multiple ligand-binding clusters. Instead, Aβ appears to interact with Cluster II through at least two kinetically distinct binding modes. These may reflect binding to different sites within Cluster II, multiple binding orientations, multivalent interactions, or adsorption-induced conformational rearrangements of the receptor–ligand complex. The present kinetic data alone cannot distinguish among these possibilities.

These findings have implications for the interpretation of previous studies of Aβ binding to LRP-1. Most previous investigations have relied primarily on equilibrium binding measurements or kinetic analyses based on single-site interaction models. ,,− While these approaches have successfully established that Aβ binds LRP-1, they provide limited information regarding the complexity of the adsorption pathway itself. The present results suggest that describing the interaction using a single apparent affinity may oversimplify a system in which multiple kinetically distinguishable processes contribute to the overall binding response. Importantly, no equilibrium dissociation constants are reported here. The present analysis is deliberately restricted to the kinetically distinguishable components that are directly resolved by the QCM-D measurements. Recognizing this heterogeneity may help reconcile differences among previously reported affinity measurements obtained using different experimental techniques and surface immobilization strategies. Several limitations should also be acknowledged. QCM-D measures changes in coupled mass and viscoelastic properties at an immobilized interface and therefore cannot directly identify the molecular origin of the observed kinetic populations. Although the kinetic analyses strongly support the existence of parallel adsorption pathways, they do not by themselves establish the precise structural interactions responsible for each process. Likewise, immobilization of LRP-1 on a sensor surface provides a simplified representation of the receptor environment and does not fully reproduce the complexity of the cell membrane or receptor-mediated endocytosis occurring in vivo. Consequently, the mechanistic interpretations presented here should be viewed within the context of the experimental model system. Finally, we acknowledge that surface heterogeneity cannot be completely excluded, but we note that the observed kinetics are better explained by parallel adsorption processes at the experimental level, regardless of the molecular source of that heterogeneity.

The differential removal of Aβ by PBS and Hellmanex demonstrates the presence of populations with distinct resistance to washing, but this behavior does not uniquely identify their molecular origin. Differences in ligand orientation, penetration depth within the interfacial layer, nonspecific adsorption, viscoelastic coupling, or heterogeneity in the immobilized LRP-1 layer could all contribute to the observed response. Changes in the ΔF/ΔD relationship before and after rinsing provide a qualitative measure of whether the retained material exhibits altered mechanical coupling or rigidity at the sensor interface. In a representative experiment, the adsorbed layer exhibited |ΔF/ΔD| = 56.0 Hz per 10^–6 dissipation unit before rinsing (ΔF = −157.0 Hz, ΔD = 2.80 × 10^–6), increasing to 190.8 after the PBS rinse (ΔF = −109.7 Hz, ΔD = 0.58 × 10^–6). Treating the two populations as contributing additively to the measured response, the PBS-removable fraction corresponds to ΔD/|ΔF| = 47.2 × 10^–9 Hz^–1 compared with 5.2 × 10^–9 Hz^–1 for the retained fraction, indicating that the material surviving the rinse is substantially more rigidly coupled to the sensor surface. However, this parameter should be interpreted as supporting evidence for distinct interfacial states rather than as a unique molecular assignment.

An important limitation of the present study is that QCM-D resolves kinetically distinct interfacial populations but does not directly establish their molecular identities. Thus, although the kinetic analysis strongly favors parallel adsorption pathways over the sequential binding–rearrangement mechanism tested here, alternative molecular origins of the two populations cannot be excluded. For example, Aβ can exist as a distribution of monomeric and oligomeric species, and different solution-state species could exhibit distinct association and dissociation kinetics toward LRP-1. In addition, immobilization of the LRP-1 Cluster II Fc Chimera may introduce heterogeneity in receptor orientation, accessibility, or conformation at the sensor interface, thereby generating binding environments with different apparent kinetic stabilities. Such interfacial heterogeneity is inherently difficult to distinguish from ligand heterogeneity using QCM-D alone. Consequently, the parallel kinetic model should be interpreted as describing two experimentally distinguishable adsorption populations rather than uniquely assigning two molecular binding mechanisms. Further studies combining QCM-D with independent characterization of Aβ oligomeric state and approaches capable of probing the orientation and conformational distribution of immobilized LRP-1 would be required to establish the molecular origin of these populations.

Future studies should focus on directly correlating the experimentally observed kinetic populations to specific molecular interactions. Computational approaches, including molecular dynamics simulations, together with complementary biophysical techniques and studies employing individual LRP-1 ligand-binding domains or site-directed mutagenesis, could help identify the structural origins of the parallel adsorption pathways identified here. More broadly, the analytical framework presented in this work provides a general strategy for distinguishing between sequential and parallel adsorption mechanisms in complex biomolecular systems. Applying similar kinetic analyses to other receptor–ligand interactions may reveal that adsorption heterogeneity is more widespread than is currently appreciated, leading to a more complete understanding of protein recognition at biological interfaces.

Conclusions

Quartz crystal microbalance with dissipation monitoring was used to investigate the adsorption kinetics of amyloid-β onto immobilized LRP-1 and to distinguish between competing mechanistic models describing the interaction. Sequential washing experiments revealed that the adsorbed Aβ layer comprises populations exhibiting different binding stabilities, with a readily dissociable fraction removed by buffer alone and a more strongly associated fraction requiring detergent treatment for complete removal. Kinetic model comparison demonstrated that the experimental adsorption behavior is more accurately described by two independent parallel adsorption pathways than by a sequential adsorption–rearrangement mechanism. This conclusion was further supported by residual analysis and by concentration-dependent measurements, which consistently reproduced the heterogeneous kinetic behavior across the concentration range investigated. Equilibrium dissociation constants are not reported because the concentration range examined does not adequately sample the binding isotherm; the analysis is therefore restricted to the kinetic behavior directly supported by the measurements. Collectively, these findings demonstrate that Aβ adsorption to immobilized LRP-1 is intrinsically heterogeneous and proceeds through kinetically distinct parallel pathways rather than a single sequential binding process. This work establishes QCM-D combined with kinetic model discrimination as a powerful approach for resolving complex protein adsorption mechanisms and provides new insight into the molecular processes underlying Aβ recognition by LRP-1. The analytical framework presented here should be broadly applicable to other receptor–ligand systems in which multiple adsorption pathways contribute to the observed binding kinetics.

†.

G.L. and E.T. made an equal contribution to this work.

The authors declare no competing financial interest.