Sodium citrate monobasic, sodium chloride, Trizma base, glycine, hydrochloric acid, 70% nitric acid, HSA, and glycated HSA (gHSA) were from Sigma-Aldrich (St. Louis, MO). The degree of glycation of gHSA was 3 mol hexose (as fructosamine) per mol albumin and the purity was 94% according to the manufacturer’s certificate of analysis. BioPure citrate stabilized AgNPs were purchased from nanoComposix (San Diego, CA). AgNPs had a nominal diameter of 40 nm and were supplied at a concentration of 4.8 nM in 2 mM citrate solution.

Unless otherwise noted, all solutions were prepared in Millipore water (18.2 MΩ·cm at 25°C) and all samples were prepared in a buffer consisting of 5 mM citrate—5 mM NaCl buffer at pH 6.5 (herein, citrate buffer). The pH of the buffer was adjusted through dropwise addition of 0.1 M and 1.0 M NaOH. Stock solutions of HSA and gHSA were prepared to a concentration of 5 μM in Millipore water. Stocks were aliquoted and frozen at −20°C for later use.

Human hepatoma (HepG2) cells were obtained from ATCC (Manassas, VA); Dulbecco’s modified Eagle medium (DMEM) from Mediatech (Manassas, VA); fetal bovine serum and penicillin—streptomycin from Invitrogen (Carlsbad, CA); sodium pyruvate and MEM non-essential amino acids from Life Technologies (Carlsbad, CA); trypsin/EDTA from CellGro (Manassas, VA); and WST cell proliferation assay kit from Dojindo Molecular Technologies (Rockville, MD).

The various techniques used across this study have different sample concentration requirements, so the concentration of AgNPs, HSA, and gHSA were varied to improve the signal-to-noise for each experiment. Specifically, for dynamic light scattering (DLS), zeta potential, UV-vis spectroscopy, and dissolution studies the concentration ratio was maintained as approximately 7×10⁴:1 protein:AgNPs, which was achieved by incubating 24 pM. AgNPs with 1.5 μM HSA or gHSA. For cell viability studies the concentration ratio was also approximately 7 × 10⁴:1 protein:AgNPs, which was achieved by incubating 10 pM. AgNPs with 0.70 μM protein or by incubating 100 pM. AgNPs with 7.0 μM protein. Langmuir adsorption isotherms were performed using a concentration ratio of 2 × 10⁴:1 protein:AgNPs, which was achieved by incubating 24 pM. AgNPs with varying concentrations of protein (25–500 nM). The decrease in the protein:AgNP ratio was necessary to achieve optimal binding conditions for quantitative analysis. For CE, the AgNP concentration was increased to improve detection sensitivity while the protein concentration was decreased to maintain the separation efficiency, resulting in a lower molar ratio than used in the other experiments. CE studies were performed using a concentration ratio of 6 × 10²:1 protein:AgNPs, which was achieved by incubating 24 pM. AgNPs with varying concentrations of protein (50–150 nM). Finally, for circular dichroism (CD) studies, the protein concentration was increased to improve detection sensitivity while the AgNP concentration was decreased to minimize interference due to light scattering from the NPs, which also resulted in a lower molar ratio than used in the other experiments. CD studies were performed using a concentration ratio of 3 × 10²:1 protein:AgNPs, which was achieved by incubating 3.0 nM AgNPs with 0.94 μM protein.

AgNPs were characterized using dynamic light scattering (DLS) to measure the hydrodynamic diameter, polydispersity index (PDI) and zeta potential. Measurements were recorded with a Malvern Zetasizer Nano-ZS instrument (Malvern, PA). All samples were prepared with a citrate buffer that was twice filtered with a 0.2 μm nylon syringe filter. Triplicate samples of AgNPs were prepared to a final concentration of 24 pM and contained 0 or 1.5 μM HSA or gHSA. Samples were incubated in the dark at room temperature for 24 or 72 h prior to analysis. DLS measurements were recorded using 173° backscatter mode after temperature equilibration for 2 min at 25°C. For each replicate sample, 11 sub-runs were recorded per measurement and 5 measurements were recorded and averaged. Reported values represent the average and standard deviation of three preparative replicates of AgNPs, AgNPs-HSA, and AgNPs-gHSA. Zeta potential measurements were recorded in the same way, but the number of sub-runs per measurement was set to automatic and constrained between 10–100. A Pd dip cell was used to measure zeta potential and values were determined using the Smoluchowski equation.

UV-vis experiments were performed using a Cary UV-vis spectrophotometer (Agilent Technologies, Inc.). AgNPs were prepared to a final concentration of 24 pM. in citrate buffer and were titrated with HSA or gHSA in the concentration range 0–500 nM. All samples were prepared in triplicate and incubated for 24 h in the dark at room temperature. All samples were analyzed in a semi-micro quartz cuvette with a 1 cm pathlength. Absorbance spectra were recorded from 400—450 nm at 60 nm·min⁻¹. The shift in the localized surface plasmon resonance (LSPR) band was used to determine the association constant, K _a, for the formation of the AgNP-protein complex according to (Boulos et al., 2013; Dennison et al., 2017; Boehmler et al., 2020): ∆ λ ∆ λ max = K a C p 1 + K a C p (1) where Δλ is the shift in the LSPR band relative to the sample without protein, Δλ _max is the maximum shift in the LSPR band, and C _p is the protein concentration. Qualitative absorbance spectra of AgNPs were also recorded using the same concentration ratio of protein:AgNPs used in other studies. Specifically, 24 pM. AgNPs were incubated for 24 h in the dark at room temperature in buffer alone or in the presence of 1.5 μM HSA or gHSA. Absorbance spectra were recorded from 350 to 500 nm using a scan rate of 300 nm·min⁻¹.

CE was carried out on a P/ACE MDQ Plus capillary electrophoresis system from AB SCIEX. An uncoated fused silica capillary was used with an inner diameter of 50 μm. The total length of the capillary was 60.2 cm and the length to the detector was 50.0 cm. The separation buffer consisted of 5 mM Tris—500 mM glycine with a pH of 7.8. Each day the capillary was flushed successively for 10 min with Millipore water, 10 min with 0.1 M NaOH, 10 min with Millipore water, and 20 min with separation buffer. Between each run, the capillary was flushed for 1 min with Millipore water and 2 min with separation buffer. The separation buffer and the solutions used for rinsing the capillary were filtered with a 0.2 μm nylon syringe filter prior to use. Triplicate samples of AgNPs were prepared to a final concentration of 240 pM. in the separation buffer and contained 0, 50, 100, or 150 nM HSA or gHSA. All samples were incubated in the dark at room temperature for at least 30 min prior to analysis. Samples were injected hydrodynamically at 4 psi for 5 s (≈35 nL injection volume) and a 25 kV separation voltage was applied (applied field ≈ 415 V/cm). Each replicate sample was subjected to triplicate CE analysis to obtain reliable parameters for quantitative analysis, as described below.

Non-equilibrium capillary electrophoresis of equilibrium mixtures (NECEEM) was used to characterize AgNP-protein complex formation (Berezovski and Krylov, 2002; Berezovski et al., 2003; Krylov and Berezovski, 2003; Okhonin et al., 2004; Riley et al., 2018). In accordance with NECEEM theory, electropherograms were integrated to determine the areas under the unbound AgNP peak (A ₁), the AgNP-protein complex peak (A ₂), and in the region of dissociation (A ₃; Supplementary Figure S3). These peak areas were used to calculate the ratio of unbound to bound AgNPs, according to: R = A 1 A 2 + A 3 (2)

Then, using the calculated value of R and the initial concentrations of protein and AgNPs, [P]₀ and [AgNPs]₀, respectively, the dissociation constant was calculated, according to: K d = P 0 1 + R − A g N P s 0 1 + 1 R (3)

To determine the rate constant for AgNP-protein complex dissociation, k _off, CE electropherograms were fit in region A ₃ according to: I t = I t 0 e k o f f t A g N P − P t A g N P − t A g N P − P t − t 0 (4) where I _t is the absorbance intensity at some point in time t and I _t0 is the intensity of the absorbance signal at time t ₀. Time t ₀ represents the beginning of the exponential decay between the unbound AgNP peak at time t _AgNP and the AgNP-protein complex at time t _AgNP-P (Supplementary Figure S3). Using the calculated values of K _d and k _off, the rate constant for AgNP-protein complex association, k _on, was calculated according to: k o n = k o f f K d (5)

Samples to be analyzed by CD were prepared by diluting HSA or gHSA to a concentration of 0.94 μM in 20 mM sodium phosphate buffer at pH 7.4. Samples were analyzed in a cylindrical quartz cuvette with a 1 mm pathlength and analyzed with an Olis Rapid-Scanning monochromator. A subsequent scan was taken after the addition of AgNPs to a final concentration of 3.0 nM and for a total sample volume of 280 μL. Measurements were recorded in the wavelength range 185–260 nm and the number of increments was set to 150. Consistent with previous studies of this kind (Fleischer and Payne, 2014), spectra were acquired in millidegrees (θ_obs) and converted to mean residue ellipticity ( θ ), according to: θ = M W   θ o b s 10   l   C   n (6)

The mean residue ellipticity (units of degrees cm² dmol^-1) is a function of the observed signal in millidegrees, the average molecular weight of the protein (MW), path length (l in cm), protein concentration (C in g/L), and the total number of amino acids (n). The % α-helicity of the protein was determined according to (Adler et al., 1973): %   α   h e l i c i t y = − θ M R E − 4000 3300 − 4000 (7)

The percent α-helicity of a protein is a function of the mean residue ellipticity at 208 nm ([θ_MRE]) minus the contribution from the β-form and random coil conformations at 208 nm (4000). The observed value is compared to the mean residue ellipticity of a pure α-helix protein (33000).

HepG2 cells were maintained and grown in 100 mm tissue culture dishes (Greiner, Bio-One, Monroe, CA, United States) using DMEM supplemented with 10% fetal bovine serum, sodium pyruvate, MEM non-essential amino acids, and 1% penicillin-streptomycin, at 37°C in a 5% CO₂ humidified environment. The cells were grown to 70%–80% confluence (in approximately 7–10 days) and passaged using 0.25% trypsin/EDTA. For cell viability assays, cells (15,000 cells per well) were seeded in 96-well flat bottom plates and allowed to proliferate for 48 h. Cells were then washed with culture media alone (no AgNP or protein control), with culture media containing 0.70 or 7.0 μM HSA or gHSA (no AgNP control), or with culture media containing 10 pM or 100 pM of uncoated or protein-coated AgNPs (200 μL total volume per well). All samples were prepared in triplicate. Protein-coated AgNPs for the cell viability assays were prepared by reacting 100 pM AgNPs with 7.0 μM HSA or gHSA or 10 pM AgNPs with 0.7 μM HSA or gHSA for 10 min in DMEM. Control cells and cells exposed to the AgNP preparations were incubated for either 24 h or 72 h at 37°C in a humidified atmosphere containing 5% CO₂ . At collection, the cells were washed with media to remove the NPs and eliminate any interference due to the NPs.

Cell viability was then measured using the commercially available WST assay according to the manufacturer’s instructions. WST solution (20 μL) was added to cells at a 1:10 dilution in DMEM (200 μL total volume per well), followed by a 2-h incubation period at 37°C in a humidified atmosphere containing 5% CO₂. Absorbance was measured at 570 nm using a Tecan Infinite 200 PRO plate reader (Tecan, Switzerland). Background absorbance due to NPs was recorded using no cell controls for all the NP preparations and subtracted from the absorbance values of the experimental samples. Cell viabilities (percent) were thereafter calculated relative to controls not treated with AgNPs or proteins.

LSSV was carried out as previously reported (Hui et al., 2019; Boehmler et al., 2020). Briefly, a three-electrode setup consisting of a glassy carbon working electrode, Ag/AgCl reference electrode, and Pt wire counter electrode was used to carry out the analysis. Measurements were recorded using a BASi Epislon Eclipse potentiostat and C-3 Cell Stand from BioAnalytical Systems, Inc., (West Lafayette, IN). All solutions were sparged with N₂ (g) for 10 min and had a final dissolved oxygen concentration of approximate 8.0 mg·L⁻¹. Each day, the working electrode was polished with 0.05 μm alumina polish and conditioned using 200 cycles of cyclic voltammetry from −0.5 to 0.35 V at 0.3 V·s⁻¹. Between experiments, all electrodes were rinsed thoroughly with Millipore water, while the working electrode was placed in a 35% nitric acid solution for 30 s, rinsed thoroughly with Millipore water, and sonicated in Millipore water for 30 s. Stripping voltammetry was carried out by deposition at −0.5 V for 30 s with stirring, followed by a 5 s quiet time, and a linear sweep from −0.5 to 0.35 V at 0.1 V·s⁻¹. A same-day matrix-matched calibration curve was prepared with Ag(I) standards in the concentration range 0–240 μg·L⁻¹.

The initial dissolution rate of AgNPs, AgNPs-HSA, and AgNPs-gHSA was measured by placing citrate buffer and the appropriate volume of protein (final concentration of 300 nM) into the electrochemical cell and sparging for 10 min. Then, the appropriate volume of AgNPs was added so that the final concentration was 4.7 pM (protein:AgNP ratio of ≈7 × 10⁴:1) and LSSV was carried out every 5 min for 2 h. The experiment was repeated in triplicate and the resulting dissolution curves (plots of [Ag(I)]_dissolved vs. time) were used to determine the dissolution rate constant, k _dissolution, according to: ln 1 − A g I t A g N P 0 = − k d i s s o l u t i o n t (8)

The percentage of dissolved silver, %AgNPs_dissolved, was determined by normalizing the [Ag(I)]_dissolved measured at the end of the dissolution kinetics experiment (t = 2 h) to the initial silver concentration placed into solution [Ag]₀, according to: %   A g N P s d i s s o l v e d = A g I d i s s o l v e d , t = 2 h A g 0 × 100 (9)

The dissolution of AgNPs was also monitored after 24 and 72 h incubation in buffer alone or buffer containing protein. Six samples of AgNPs, AgNPs-HSA, and AgNPs-gHSA were prepared in citrate buffer with final concentrations of 24 pM. AgNPs and 1.5 μM protein (protein:AgNP ratio of ≈7 × 10⁴:1) and incubated in the dark at room temperature. After 24 h, three replicates of each sample were analyzed using LSSV to determine the percentage of dissolved Ag(I) using same-day matrix matched calibration curves. After 72 h from the initial exposure, the remaining three replicates of each sample were analyzed in the same manner.

The formation of AgNP-protein complexes was confirmed qualitatively by measuring the hydrodynamic diameter, PDI, and zeta potential of AgNPs alone, in the presence of HSA, or in the presence of gHSA. Samples were analyzed at time points of 24 and 72 h following incubation to assess the stability of the AgNPs over time. After 24 h, an increase in the hydrodynamic diameter of AgNPs was observed in the presence of both HSA and gHSA relative to the no protein control (Figure 1A; Supplementary Figure S1; Supplementary Table S1). However, the magnitude of this increase differed between the two proteins, with the addition of HSA resulting in a larger increase in AgNP diameter (≈20 nm) and the addition of gHSA resulting in a more modest increase (≈7 nm). Differences in the thickness of the HSA and gHSA PCs formed on the AgNP surface could be a result of differences in the protein structure and/or orientation on the surface (vide infra). After 72 h, an increase in the diameter of AgNPs was observed suggesting slight aggregation of the particles over time. No significant differences were noted for the protein-coated AgNPs after 72 h, suggesting that the coronas was relatively stable and prevented significant aggregation of the AgNPs (Figure 1A; Supplementary Figure S1; Supplementary Table S1).

After 24 h incubation, the increase in AgNP diameter upon addition of HSA or gHSA was accompanied by either no change or a slight decrease in the PDI, providing further evidence that the formation of the AgNP-PCs provided some steric stabilization of the colloidal suspension (Figure 1B; Supplementary Table S1). Based on the decrease in PDI values, the degree of steric stabilization conferred by the proteins was more significant for gHSA than for HSA (Supplementary Table S1). After 72 h, the PDI value for all AgNPs increased, suggesting changes in the homogeneity of the samples over time and the potential for particle-particle or protein-protein aggregation. The formation of the AgNP-protein complex was further demonstrated by a decrease in the AgNP zeta potential (toward more positive potential) after 24 h incubation with HSA or gHSA (Figure 1C; Supplementary Table S1). After 72 h incubation, the zeta potential significantly decreased for the sample of AgNPs alone, further supporting the likelihood that the uncoated AgNPs had begun to aggregate. The zeta potentials for the protein-coated AgNPs showed only a subtle decrease in magnitude, further supporting the stability conferred by the protein coating. Overall, only subtle differences were observed between the AgNP-HSA and AgNP-gHSA coronas and more significant changes were observed between uncoated and protein-coated AgNPs.

The formation of AgNP-protein complexes was also confirmed quantitatively using UV-vis spectroscopy and CE to measure association and rate constants for complex formation. The 40 nm AgNPs exhibited a prominent LSPR band at approximately 420 nm that shifted to longer wavelength in the presence of HSA and gHSA (Supplementary Figure S2). Langmuir adsorption isotherms were constructed to determine the association constant, K _a , for AgNP-protein complex formation (Figure 2). The K _a values for HSA and gHSA were 2.3 (±0.2) × 10⁷ M⁻¹ and 2.4 (±0.1) × 10⁷ M⁻¹ (n = 3, R ² > 0.97), respectively, suggesting that both proteins had similar affinity for the AgNP surface (Table 1).

The formation of AgNP-protein complexes was further quantified using CE and NECEEM analysis. While NECEEM theory was initially developed to measure DNA-protein complexes, it has recently been demonstrated as an effective tool to measure the NP-PC (Riley et al., 2018). According to NECEEM theory, injection of the equilibrium mixture (containing unbound AgNPs, unbound protein, and AgNP-protein complexes) onto the capillary and application of the separation voltage will lead to a non-equilibrium condition, whereby each component of the mixture will migrate according to their unique electrophoretic mobilities. As a result, when the AgNP-protein complexes dissociate and the unbound AgNPs and protein migrate away from one another in the capillary, they are no longer proximate and cannot reestablish the AgNP-protein complex (i.e., a non-equilibrium condition) (Berezovski and Krylov, 2002; Berezovski et al., 2003; Krylov and Berezovski, 2003; Okhonin et al., 2004; Riley et al., 2018). By directly monitoring the dissociation of AgNP-protein complexes, both the dissociation constant (K _d) and the on/off rate constants (k _on/k _off) can be determined in a single experimental run (Berezovski and Krylov, 2002).

In this work, only the AgNPs generated an absorbance signal in the electropherogram because the concentration and volume of protein injected onto the capillary was too low to be detected. In the absence of protein, a single peak at approximately 3.7 min was observed and attributed to the AgNPs (Figure 3, red trace). With the addition of HSA, a new peak was observed at approximately 3.3 min and was attributed to the AgNP-HSA complex due to an observed increase in absorbance intensity of the peak with increasing protein concentration (Figure 3). Electropherograms for NECEEM analysis of AgNPs-gHSA complexes exhibited similar features (Supplementary Figure S4). By applying quantitative NECEEM analysis (Eqs 2, 3; Supplementary Figure S3), the K _d values of AgNP-HSA and AgNP-gHSA complexes were obtained and then converted to K _a values for ease of comparison. The values obtained by NECEEM were consistent with those obtained using UV-vis spectroscopy within error, confirming that HSA and gHSA have similar affinity for the AgNP surface (Table 1). NECEEM analysis of the dissociation kinetics was also carried out (Eqs 4, 5; Supplementary Figure S3), and likewise, no significant differences were noted between the k _on and k _off values determined for the AgNP-HSA and AgNP-gHSA complexes. Generally, the values obtained suggest fast adsorption and slow desorption of the proteins from the AgNP surface (Table 1).

Finally, changes to protein structure upon PC formation were probed using CD spectroscopy (Figure 4A; Table 2). A comparison of the two proteins in the absence of AgNPs revealed slight differences in the protein secondary structure, with gHSA exhibiting slightly greater α-helicity than HSA (Figure 4A; Table 2). Others have noted changes in HSA structure upon glycation, where the degree and exact nature of those changes depend upon the type of monosaccharide, concentration, and length of exposure (Nakajou et al., 2003; Qiu et al., 2021). Upon the addition of AgNPs to each protein solution, the α-helical character of the protein decreased, as measured by changes in the ellipticity at 208 and 220 nm and the calculated % α-helicity (Figure 4B; Table 2). While CD suffers from low sensitivity and an inability to distinguish free from surface-adsorbed protein, it can provide qualitative and semi-quantitative analysis of changes in protein secondary structure, which may be attributed to the formation of NP-protein complexes. Generally, changes in protein ellipticity can be taken as further evidence of the formation of the NP-PC. Our results demonstrate that the reduction of α-helicity upon interaction with AgNPs was slightly larger for HSA than for gHSA, which suggests possible differences in the surface interaction of HSA and gHSA due to protein glycation.

Studying the effect of protein coating on cell viability is important because NPs used in the body will inevitably be coated with a PC when they come in contact with biological fluids (e.g., blood) or culture medium. Such a protein coating has been shown to influence targeting abilities, cellular uptake, and immunotoxicity of NPs (Monteiro-Riviere et al., 2013; Nierenberg et al., 2018). Thus, we hypothesized that the HSA and gHSA PCs would each alter cell toxicity, but that there would be no significant differences between their impact since biophysical analyses suggested that the coronas had remarkably similar properties under the chosen experimental conditions. To evaluate this hypothesis, we investigated the role of protein coatings on the toxicity of AgNPs to human cells. We chose liver hepatocellular carcinoma (HepG2) cells as our cell model, as they are widely used in the literature to determine toxicity. The liver has also shown to be a target organ for several NPs, making liver cells biologically relevant for toxicity studies (Balasubramanian et al., 2010; Lankveld et al., 2010; Khosravi-Katuli et al., 2017; Wu and Tang, 2018).

HepG2 cells were incubated with varying concentrations of AgNPs under three conditions: AgNPs exposed to HSA, gHSA, and no protein (uncoated AgNPs). Control studies included incubation of HepG2 cells in culture media alone or with culture media containing different concentrations of HSA or gHSA. After 24 or 72 h incubation, a WST assay was performed to evaluate cell viability. As expected, relative to cells incubated in cell culture media alone, no toxicity was observed for cells exposed to just HSA or gHSA with no AgNPs (Supplementary Table S2). The toxicity due to the addition of AgNPs was concentration dependent with higher levels of toxicity observed at higher AgNP concentrations. Interestingly, the results revealed increased toxicity for both HSA-coated and gHSA-coated AgNPs relative to uncoated AgNPs (Figure 5; Supplementary Table S3).

As a first step towards investigating the mechanism by which the HSA and gHSA PCs decrease HepG2 cell viability, LSSV was used to measure AgNP dissolution in the absence and presence of both proteins. For all LSSV analyses, same-day matrix-matched calibration curves were obtained, which was important to account for any impact of the proteins on the electrochemical signal. All calibration curves had sub-micromolar LODs (typically around 0.10–0.30 μM) under all experimental conditions evaluated (Supplementary Tables S4, S5) suggesting that the protein did not unduly effect LSSV analyses.

First, the initial rate of dissolution was measured for AgNPs alone or in the presence of HSA or gHSA over the first 2 h following incubation. The dissolution rate constant, k _dissolution, of AgNPs alone was 1.6 × 10⁻⁴ min⁻¹ indicating rapid dissolution of AgNPs after dilution in buffer (Supplementary Figure S5; Supplementary Table S4). In contrast, AgNPs added to buffer containing either HSA or gHSA exhibited slow dissolution and a nearly constant concentration of Ag(I) over the 2 h analysis period, so dissolution rate constants were unable to be measured (Supplementary Figure S5; Supplementary Table S4). At the conclusion of the 2 h AgNP dissolution experiment, the faster dissolution kinetics of AgNPs in the no protein condition led to a 2-fold increase in the percentage of dissolved Ag(I) relative to AgNPs in the presence of HSA and gHSA (Supplementary Figure S5; Supplementary Table S4).

Analysis of AgNP dissolution behaviors was extended by measuring the percentage of dissolved Ag(I) after 24 and 72 h incubation in buffer alone, or in buffer containing HSA and gHSA. These time points were chosen to exactly mimic the conditions used for the cell viability assay. Under all experimental conditions a slight increase in the percentage of dissolved Ag(I) was observed after 72 h incubation relative to 24 h (Figure 6; Supplementary Table S5). This increase was most significant for the AgNPs incubated in buffer only, consistent with kinetic dissolution experiments that showed the AgNPs alone undergo much faster dissolution. Relative to the AgNPs alone, AgNPs incubated with HSA or gHSA exhibited a statistically significant decrease in the percentage of dissolved Ag(I) at both 24 and 72 h following incubation (Figure 6; Supplementary Table S5). Specifically, the dissolution of AgNPs in the presence of protein was ≈5-fold lower than the no protein control after 24 h and ≈6-fold lower after 72 h.