Environ. Sci. Technol. 2007, 41, 2324-2330
Adsorption of Pathogenic Prion Protein to Quartz Sand X I N M A , †,‡ C R A I G H . B E N S O N , ‡ DEBBIE MCKENZIE,§ JUDD M. AIKEN,§ AND J O E L A . P E D E R S E N * ,†,‡,| Department of Soil Science, University of Wisconsin, Madison, Wisconsin 53706, Department of Civil and Environmental Engineering, University of Wisconsin, Madison, Wisconsin 53706, Department of Comparative Biosciences, University of Wisconsin, Madison, Wisconsin 53706, and Molecular and Environmental Toxicology Center, University of Wisconsin, Madison, Wisconsin 53706
Management responses to prion diseases of cattle, deer, and elk create a significant need for safe and effective disposal of infected carcasses and other materials. Furthermore, soil may contribute to the horizontal transmission of sheep scrapie and cervid chronic wasting disease by serving as an environmental reservoir for the infectious agent. As an initial step toward understanding prion mobility in porous materials such as soil and landfilled waste, the influence of pH and ionic strength (I) on pathogenic prion protein (PrPSc) properties (viz. aggregation state and ζ-potential) and adsorption to quartz sand was investigated. The apparent average isoelectric point of PrPSc aggregates was 4.6. PrPSc aggregate size was largest between pH 4 and 6, and increased with increasing I at pH 7. Adsorption to quartz sand was maximal near the apparent isoelectric point of PrPSc aggregates and decreased as pH either declined or increased. PrPSc adsorption increased as suspension I increased, and reached an apparent plateau at I ∼ 0.1 M. While trends with pH and I in PrPSc attachment to quartz surfaces were consistent with predictions based on Born-DLVO theory, non-DLVO forces appeared to contribute to adsorption at pH 7 and 9 (I ) 10 mM). Our findings suggest that disposal strategies that elevate pH (e.g., burial in lime or fly ash), may increase PrPSc mobility. Similarly, PrPSc mobility may increase as a landfill ages, due to increases in pH and decreases in I of the leachate.
Introduction Transmissible spongiform encephalopathies (TSEs), or prion diseases, are fatal, neurodegenerative disorders affecting a variety of mammalian species and include bovine spongiform encephalopathy (“mad cow” disease), chronic wasting disease (CWD) of deer, elk and moose, sheep scrapie, and Creutzfeldt-Jakob disease in humans (1). These proteinmisfolding diseases are characterized by spongiform degeneration of the brain, accumulation of abnormal prion * Corresponding author phone: (608) 263-4971; fax: (608) 2652595; e-mail:
[email protected]. † Department of Soil Science. ‡ Department of Civil and Environmental Engineering. § Department of Comparative Biosciences. | Molecular and Environmental Toxicology Center. 2324
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
protein, personality and memory changes, loss of coordination, and inevitably death (1). No cure exists for these diseases. The infectious agents in these diseases, referred to as prions (from proteinaceous-infectious particles), are apparently devoid of nucleic acid, and are composed primarily, if not exclusively, of a misfolded form of the prion protein, a normally benign cell-surface glycoprotein designated PrPC (1). The disease-associated form of the prion protein (PrPSc) is identical to PrPC in amino acid sequence and covalent post-translational modifications (1). PrPC and PrPSc differ only in their conformation (i.e., folding): PrPC has a high R-helix content (42%) and negligible β-sheet character (3%), while PrPSc is rich in β-sheet (43%) and exhibits diminished R-helix content (30%) relative to PrPC (2). Conversion of PrPC to PrPSc represents the central event in prion disease propagation and confers distinct biophysical properties to the protein including dramatically increased protease resistance, marked detergent insolubility, and a propensity to aggregate (1). Prions exhibit extraordinary resistance to a variety of conditions that inactivate conventional pathogens including exposure to ionizing, ultraviolet, and microwave radiation; protease treatment; contact with most chemical disinfectants; boiling; autoclaving under conventional conditions; and dry heat at temperatures under 600 °C (3, 4). An environmental reservoir of prion infectivity appears to contribute to the transmission of CWD and scrapie (5). Several lines of evidence support the concept that soil comprises a component of this reservoir (5). Prions can persist in soil for g 3 y (6) and can be introduced into soil when infected carcasses decompose and through shedding in saliva and, presumably, feces (7). The presence on the landscape of decomposed infected carcasses or residual excreta from infected animals is sufficient to transmit CWD to mule deer (8). Oral transmission of CWD via saliva from infected deer has been demonstrated (7), and shedding of prions in urine from animals with chronic kidney inflammation has been reported (9). Since herbivores consume soil both deliberately and incidentally (10, 11), ingestion of prion-contaminated soil may contribute to the natural CWD and scrapie transmission (5, 12). Montmorillonite-associated prions were recently demonstrated to retain infectivity despite extremely avid adsorption of PrPSc to the clay particles (12). Mechanisms governing prion interaction with soil particles remain poorly understood. Greater insight into these mechanisms will improve our ability to predict prion bioavailability and mobility in natural environments and engineered systems (e.g., landfills). For example, the accessibility of prions to grazing animals and other soil-ingesting species is expected to depend in part on retention of the infectious agent near the soil surface, while the risk posed by landfilling infectious materials depends on prion mobility in the waste, soils used for daily cover, and granular materials used for leachate collection (5). As an initial step toward understanding prion interaction with soil and landfill materials, batch sorption experiments were conducted to assess pH and ionic strength (I) influences on PrPSc adsorption to quartz sand. Experimental results were interpreted using the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory of colloid stability (13, 14).
Materials and Methods Chemicals. Analytical grade 3-N-morpholinopropanesulfonic acid (MOPS), Tris-(hydroxymethyl)aminomethane HCl (Tris), and sodium acetate were obtained from Fisher Scientific (Hampton, NH). 10.1021/es062122i CCC: $37.00
2007 American Chemical Society Published on Web 03/06/2007
PrPSc Source. Samples enriched in PrPSc were prepared from brains of hamsters clinically infected with the Hyper strain of hamster-adapted transmissible mink encephalopathy using the Bolton et al. (15) procedure modified by excluding proteinase K digestion (16). The resulting preparation (in 0.01 M Tris pH 7.4, 0.133 M NaCl) corresponded to 4 g brain‚mL-1 (see Supporting Information for protein gel and immunoblot of prion preparation) and contained ∼109 infectious units‚g-1 as determined by end-point titration bioassay. Characterization of PrPSc. Electrophoretic mobilities of PrPSc aggregates were measured using a ZetaSizer Nano ZS instrument (Malvern Instruments, Worcestershire, UK) over a range of pH and I conditions. Prion-enriched samples (2 µL) were diluted to the equivalent of 8 mg brain‚mL-1 in 750-µL solutions of desired pH and I, equilibrated for g10 min, sonicated for 3 min, and transferred to folded capillary cells for electrophoretic mobility measurement. For I >0.05 M, the Fast Field Reversal mode was used. Instrument performance was verified using manufacturer-provided polymer microsphere transfer standards. Changes in apparent PrPSc aggregate size as a function of pH and I were examined by dynamic light scattering (DLS; ZetaSizer Nano ZS). After diluting PrPSc preparation to the equivalent of 32 mg brain‚mL-1 in solutions of desired pH and I, the suspension was equilibrated for g10 min, sonicated for 3 min, and transferred to a glass cuvette. Three measurements (10 runs per measurement) were acquired from each of triplicate samples for each solution condition. The autocorrelation function was analyzed by the method of cumulants to obtain the moments of the PrPSc aggregate size distribution. This method is appropriate for particle suspensions with narrow to intermediate polydispersities (e.g., 0.20.4). The average polydispersity of PrPSc aggregates under the solution conditions employed was 0.4. The intensityaverage (Z-average) hydrodynamic radius (rh,z) was calculated from measured diffusivities using the Stokes-Einstein equation. The rh,z is a single, intensity-averaged value representing the entire size distribution. Quartz Sand Preparation and Characterization. IOTA quartz sand (Unimin Corporation, New Canaan, CT) was fractionated by wet sedimentation/flotation to obtain the 0.18-0.25 mm particle size fraction (17) (98% between 0.12 and 0.25 mm, sieve analysis, ASTM D 422). Metal and organic contaminants were removed from fractionated sand by 24-h soaking in 12 N HCl, rinsing with distilled deionized water (ddH2O), and baking overnight at 800 °C (17, 18). Cleaned sand was stored in a vacuum desiccator and rehydrated by boiling for g1 h in ddH2O prior to use. X-ray diffraction (Scintag PAD V diffractometer, Cupertino, CA) and X-ray fluorescence analyses (Bruker AXS Model 3400, Madison, WI) indicated that the cleaned sand contained 99.5% quartz and 0.5% bytownite plagioclase feldspar (average formula: Ca0.87Na0.13Al2Si2O8) by mass. The sand had a 2.62 ( 0.11 mmol(+)‚kg-1 cation exchange capacity (compulsive exchange with Ba2+, ref 19), a 0.20 m2‚g-1 specific surface area (Kr adsorption, Brunauer-Emmet-Teller method, Micrometrics Analytical Services, Norcross, GA), and a 2.65 specific gravity (ASTM D 854). Batch Sorption Experiments. PrPSc adsorption to quartz sand particles was examined as a function of pH, I, and prion protein concentration in batch sorption experiments. Solution pH was maintained using 0.01 M acetate (pH 3-5), MOPS (pH 6-7), or Tris (pH 8-9); the desired I was achieved by NaCl addition. Quartz sand (100 mg) was equilibrated with 6 mL of solutions of desired pH and I in 50-mL fluorinated ethylene polypropylene (Teflon FEP) tubes for >15 h. An aliquot of prion preparation [20 µL (equivalent to 0.08 g brain) for pH and I experiments; 5-50 µL (equivalent to 0.02-0.2 g brain)
for isotherm experiments] was sonicated for 3 min at 750 W (CV33 probe, Sonics and Materials GE750, Newtown, CT), then added to the sand. The mixture was then gently shaken (∼14 rpm) at 23-25 °C for 4 h. Previous experiments (12) indicated that adsorption was complete within this time. All experiments were conducted in triplicate. To separate unbound from adsorbed PrPSc, the PrPSc-sand suspension was settled under gravity through a 2-mL sucrose cushion (0.75 M) followed by 7-min centrifugation at 1500g through a 0.5-mL sucrose cushion (0.75 M). Supernatants from both separations were combined. The solutions used in adsorption experiments were used to prepare the sucrose cushions (12). The sedimented sand was washed four times with the same solution as used in the adsorption experiment, and the first three wash solutions were combined with the supernatant. The final wash solution was analyzed separately (results combined with those of the supernatant during data analysis). After 3-min sonication, 100-µL aliquots from each supernatant were transferred into LoBind microcentrifuge tubes. A 100-µL aliquot of 10 M urea (in 0.01 M Tris HCl, pH 7.4) was added to each supernatant and final wash sample to achieve a 5 M urea concentration. PrPSc in these samples was denatured by 10-min heating at 100 °C in 5 M urea. Adsorbed PrPSc was desorbed from the sand using two sequential extractions. In the first step, PrPSc was extracted and denatured with 10 M urea at 100 °C for 10 min. Urea extracts were analyzed by enzyme-linked immunosorbent assay (ELISA). The sand was then re-extracted with 10% sodium dodecylsulfate (SDS) in 0.1 M Tris pH 8.0, 7.5 mM EDTA, 0.1 M DTT, and 30% glycerol at 100 °C for 10 min (12). SDS extracts were analyzed by immunoblotting. Enzyme-Linked Immunosorbent Assay. PrPSc in the supernatant, sedimented sand wash, and primary extract of the sedimented sand was measured using a double-antibody sandwich ELISA (SPI-bio, Massy Cedex, France) distributed by Cayman Chemicals (Ann Arbor, MI). Following the manufacturer’s instructions, samples were diluted to e0.5 M urea and analyzed with the colorimetric ELISA. Absorbance at 410 nm was measured with a Dynatech model MRX microplate reader (Chantilly, VA). PrPSc was quantified against 8-point calibration curves prepared by serial dilution of the starting PrPSc preparation in the same solutions as used in adsorption experiments. Samples with absorbances outside the linear range of the standard curve were diluted and reanalyzed. Preliminary experiments using immunoblot analysis indicated that capture and detection antibody epitopes were not lost on detachment from the quartz grains, and that the pH and I of the solutions used in the adsorption experiments did not affect quantitation by ELISA. Immunoblot Analysis. Immunoblot analysis was conducted as described previously (12). Briefly, samples were fractionated on 4-20% pre-cast polyacrylamide gels (BioRad, Hercules, CA) under reducing conditions, transferred to PVDF membranes, and immunoblotted with anti-PrP monoclonal antibody 3F4. Horseradish peroxidase-conjugated goat anti-mouse immunoglobulin G (Bio-Rad) was used for detection. Born-DLVO Calculations. The total energy of PrPSc interaction with quartz surfaces (ΦT) was calculated by summing the electrostatic double layer (ΦEDL), van der Waals (ΦVDW), and Born (ΦBorn) contributions (20):
ΦT ) ΦEDL + ΦVDW + ΦBorn
(1)
Electrostatic double layer interaction energies were calculated for constant potential surfaces by modeling the PrPSc-quartz interaction with sphere-plate geometry (21): VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2325
[
{
ΦEDL ) π0rrp 2ψpψcln
]
1 + exp(- κh) + (ψp2 + ψc2)ln 1 - exp(- κh) [1 - exp( -2κh)]
}
(2)
where 0 is the permittivity of free space, r is the relative permittivity of water, rp is the PrPSc aggregate radius, ψp and ψc are the PrPSc aggregate and the quartz particle surface potentials, κ is the inverse Debye length, and h is the separation distance between the PrPSc aggregate and the quartz surface. Potential energy calculations used rh,z for PrPSc aggregate size. While rh,z are larger than hydrodynamic radii (rh) based on number distributions, calculating rh from number distributions requires numerous assumptions about the aggregate size distribution that may not be valid. The ζ-potentials of PrPSc aggregates and quartz surfaces were used as estimates for ψp and ψc. The magnitude of the κ depends solely on the properties of the liquid and was calculated for the aqueous 1:1 electrolyte solutions at 25 °C using the following equation (22):
κ)
0.304
(3)
x[NaCl]
The retarded van der Waals interaction energy was calculated using the following expression (23):
Arp 14h ΦVDW ) 1+ 6h λ
[
-1
]
(4)
where A denotes the Hamaker constant. The characteristic wavelength of the dielectric, λ, is typically assumed to be 100 nm (24). The Born repulsion energy was calculated using the following expression (25):
ΦBorn )
[
]
AσB6 8rp + h 6rp - h + 7560 (2r + h)7 h7 p
(5)
where the Born radius (σB) was taken to be 0.5 nm (21). This strong, very short-range repulsive force arises from the overlap of electron clouds of atoms and determines the smallest separation distance between two molecules (22). The Hamaker constant (App) for PrPSc (p) is not currently available. Bovine fibrin was used as a surrogate. Like PrPSc, bovine fibrin is hydrophobic, tends to aggregate in aqueous suspension (25), and has similar R-helix and β-sheet content [∼30% and ∼40% (26) as compared to ∼30% and ∼43% for PrPSc (2)]. An estimate of App (7.47 × 10-20 J) was obtained from the Lifshitz-van der Waals component of the total surface tension (γpLW ) 40.2 mJ‚m-2) of bovine fibrin using the following(25):
App ) 24πlo2γLW p
(6)
where lo denotes the minimum equilibration distance (1.568 ( 0.093 Å, ref 25). The Hamaker constant (Apwq) for PrPSc interacting with quartz (q) across water (w) was calculated as 5.42 × 10-21 J using the following combining relation (22):
Apwq ≈ (xApp - xAww)(xAqq - xAww)
(7)
where Aww and Aqq are the Hamaker constants for water (4.62 × 10-20 J, ref 25) and crystalline SiO2 (9.47 × 10-20 J, ref 27).
Results and Discussion Electrokinetic Properties of PrPSc Aggregates and Quartz Grains. The ζ-potential is the potential measured by elec2326
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
FIGURE 1. Intensity-averaged hydrodynamic radii of PrPSc aggregates and ζ-potentials of PrPSc aggregates and quartz particles as a function of (a) pH and (b) ionic strength. Solutions were buffered with 0.01 M acetate (pH 3-5), MOPS (pH 6-7), or Tris (pH 8-9). Desired ionic strength was achieved by addition of NaCl. In (a), ionic strength was maintained at 0.01 M; in (b) pH was maintained at 7. Mean electrophoretic mobilities of PrPSc aggregates (determined at 150 V and 25 °C from triplicate samples, 3 measurements per sample, 25 runs per measurement) and streaming potentials of the quartz sand (Paar Physica Electro Kinetic Analyzer (Anton Paar USA, Ashland VA) at the Particle Engineering Research Center, University of Florida) were converted to ζ-potentials using the Smoluchowski equation. The ζ-potential at 100 mM NaCl exhibited large variation because the change in voltage with pressure during streaming potential measurement was near zero. Data points are means of at least 3 replicate samples; error bars represent 1 standard deviation. Variations in rh,z among prion preparations were within the error of the measurement. trokinetic methods at the shear plane separating the thin layer of liquid adhering to the particle surface from the bulk solution; ζ-potentials are commonly used as surrogates for surface potentials (25). The average ζ-potential of PrPSc aggregates was slightly positive at pH e 4 and became negative as pH increased (Figure 1a). PrPSc aggregates exhibited an apparent average isoelectric point (pI) of 4.6, within the pI ranges reported previously (28, 29). The average ζ-potential at pH 7 became less negative as I increased due to electrostatic double layer compression (Figure 1b). The ζ-potentials for PrPSc aggregates remained stable over 4 h. The ζ-potentials for the sand were negative for all pH and I conditions (Figure 1a), as expected from the point of zero charge for quartz (2.9, ref 30), and were comparable to those reported previously for the same sand (18, 31). Prion Protein Aggregate Size. Transmission electron microscopy demonstrated that the prion enrichment procedure yielded the expected rod-like structures (data not shown). DLS was used to determine PrPSc aggregate hydrodynamic radii (rh) in the prion-enriched preparation. The hydrodynamic radius is equivalent to the radius of hard sphere with the same diffusion coefficient. PrPSc aggregate rh do not necessarily reflect the effective radii of individual rods as they may include contributions from aggregates of rods.
PrPSc aggregate rh,z varied depending on solution conditions (Figure 1). As pH increased from 3.0 to 4.0, PrPSc aggregate rh,z increased significantly (p ) 0.00145) (Figure 1a). PrPSc aggregate size remained relatively constant between pH 4 and 6. Near the pI, electrostatic repulsion between PrPSc molecules is at a minimum, favoring aggregation. As pH increased to 8, PrPSc aggregate size declined; further increases in pH did not change PrPSc aggregate rh,z. PrPSc aggregate rh,z increased as I rose from 4.2 to 10 mM; further increases in I did not affect rh,z (Figure 1b). Increasing I compresses the electrostatic double layer allowing closer approach of like-charged particles, favoring aggregation. The rh,z values were stable over a 4-h period. Inspection of scattering intensity distributions revealed that PrPSc aggregate size distributions were bimodal at pH 7 and 9 (I ) 10 mM) with most of the scattering intensity (87 to >99%) contributed by large aggregates; scattering intensity distributions were unimodal for pH 4 (I ) 10 mM) and for I g 100 mM (data not shown). Although derivation of hydrodynamic radii based on mass and especially number distributions from DLS data involves numerous assumptions, the mass and number distributions can be useful for making comparisons among samples. However, confidence in the rh values based on mass and number distributions is less than for rh,z. Intensity, mass, and number distributions exhibited similar trends with pH and I. Number distributions revealed that at pH 7 and 9 under low I conditions, the majority of PrPSc aggregates was substantially smaller than those responsible for most of the scattering. For the dominant mode, rh estimated on a number basis declined as pH increased (I ) 10 mM): 333 nm (pH 4) > 60 nm (pH 7) > 31 nm (pH 9). At pH 7, rh estimated from the number distribution increased from 60 nm at I ) 10 mM to 429 and 443 nm at I ) 100 and 500 mM. Effect of pH and I on PrPSc Adsorption to Quartz Particles. PrPSc adsorption to quartz sand exhibited a maximum at pH 4 (Figure 2a), near the apparent pI (4.6) of PrPSc aggregates. PrPSc adsorption declined at higher and lower pH. Similar pH-dependent behavior has been noted for the sorption of other proteins to mineral surfaces (32). At the pI, electrostatic interactions of proteins with highenergy surfaces are at a minimum while “hydrophobic” attraction is maximal (22). Declines in protein adsorption for pH values on either side of the pI have been ascribed to lateral electrostatic repulsion of like-charged protein molecules or conformational changes (unfolding) on the surface (33), increasing the area occupied by a single protein (34). The operation of different mechanisms above and below protein pI has also been proposed (32, 33). Below the pI, the positively charged protein can unfold due to strong electrostatic attraction to the negatively charged surface; above the pI, both the protein and surface carry net negative charge and electrostatic repulsion disfavors adsorption. The effect of I on PrPSc adsorption to quartz sand was examined at pH 7 because near-neutral pH values are commonly encountered in the environment. At pH 7, PrPSc adsorption to quartz particles increased as I rose from 4.2 to 10 mM (p ) 0.0447); further increases in I up to 300 mM did not enhance adsorption (Figure 2b). At this pH, both the PrPSc aggregate and the quartz surface were negatively charged. Double layer compression at higher I apparently reduced electrostatic repulsion between PrPSc and the quartz surface leading to increased absorption. Adsorption Isotherms. PrPSc adsorption to quartz sand was examined as a function of protein concentration at pH 4 and 7 (Figure 3). In the pH 7 experiments, low levels of PrPSc were observed in SDS extracts from three samples (initial PrPSc concentration, C0 g 0.02 g brain equivalent‚mL-1); for these samples, sorbed PrPSc concentrations were based on the combined results from ELISA and immunoblot analysis.
FIGURE 2. Adsorption of PrPSc to quartz sand as a function of (a) pH (I ) 0.01 M) and (b) ionic strength (pH ) 7.01 ( 0.03, 0.01 M MOPS). Initial PrPSc concentration, C0 was equivalent to 0.08 g brain per mL. Proton activity in (a) was maintained using 0.01 M acetate (pH 3-5), MOPS (pH 6-7), or Tris (pH 8-9); ionic strength was adjusted to 0.01 M with NaCl. Fraction of PrPSc adsorbed calculated from CsMs/(CsMw + CwVw), where Cs and Cw are the measured bound and unbound concentrations, Ms is adsorbent mass, and Vw is solution volume. Data points are means of triplicate measurements; error bars represent one standard deviation.
FIGURE 3. Isotherms for PrPSc adsorption to quartz sand at pH 4 and 7. Solution pH was buffered to 4.04 or 7.07 with 0.01 M acetate or MOPS. Solid lines correspond to least-squares fits of the experimental data to a linear isotherm. For pH 4, Kd ) 8.8 ( 1.6 L‚kg-1 (p < 0.0001), R 2 ) 0.65; for pH 7, Kd ) 3.0 ( 0.4 L‚kg-1 (p < 0.0001), R 2 ) 0.82. Dashed lines are least-squares fits of the experimental data to a linearized form of the Freundlich isotherm (obscured in the case of the pH 7 data). Freundlich equation parameters: for pH 4, log KF ) 1.52 ( 0.46 (p ) 0.0044), n ) 0.88 ( 0.13 (p < 0.0001), R 2 ) 0.74; pH 7, log KF ) 0.10 ( 0.49 (p ) 0.85), n ) 1.10 ( 0.14 (p < 0.0001), R 2 ) 0.80. Note KF value for pH 7 was not significant. The lack of an asymptotic approach to a plateau in the Cs vs Cw plots suggests that the adsorption capacity of the quartz sand was not attained in these experiments. The presence of PrPSc in SDS extracts at concentrations below the limit of immunoblotting detection cannot be excluded, but would be expected to decline as C0 decreased. To evaluate VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2327
the effect of hypothetical and nondetectable PrPSc in SDS extracts, the maximum error in the sorbed PrPSc concentration was computed assuming that the concentration of nondetectable PrPSc in the extract equaled the limit of immunoblotting detection (∼1.2 × 10-4 g brain equivalent). With a single exception, this calculation indicated that nondetectable PrPSc introduces a maximum error in the sorbed concentrations of 0.9 to 9.7%. The exception was a maximum 29.0% error in samples with C0 ) 0.02 g brain in the pH 7 experiments. Actual error in this and the other samples with no detectable PrPSc in the SDS extracts was likely smaller. Experimental data were fit to a linear isotherm, Cs ) Kd‚ Cw, where Cs and Cw are the concentrations sorbed to the quartz particles and remaining in suspension as measured by ELISA (and immunoblotting for the three highest PrPSc concentrations in the pH 7 experiment), and Kd is the distribution coefficient (L‚kg-1). Least-squares fits to the linear model had R 2 ) 0.65 (pH 4) and 0.82 (pH 7). Scatter in the data is attributed to heterogeneity in the PrPSc population, which is inherent structurally (e.g., degree of glycosylation of monomeric PrPSc, glycan composition), conformationally (e.g., both proteinase K-sensitive and -resistant PrPSc are present; ref 37) and in aggregation state. As expected from observed trend in PrPSc adsorption with pH (Figure 2), Kd was larger at pH 4 (near the average apparent pI of the PrPSc aggregates; Kd ) 8.8 ( 1.6 L‚kg-1) than at pH 7 (PrPSc aggregates carrying net negative charge; Kd ) 3.0 ( 0.4 L‚kg-1). In general, least-squares fits of the experimental data to the Freundlich, Langmuir, Langmuir-Freundlich, Temkin, and Redlich-Peterson equations either resulted in nonsignificant model parameters (p > 0.05) or failed to increase R 2 (data not shown). The sole exception was a fit at pH 4 to the linearized form of the Freundlich equation: log Cs ) n‚log Cw + log KF, where KF is the adsorption capacity at a specific Cw, and n provides a measure of the distribution of adsorption energies. The adsorption isotherm at pH 4 exhibited marked nonlinearity (n ) 0.88 ( 0.13, p < 0.001), suggesting a distribution of energies of interaction between PrPSc aggregates and quartz surfaces. Interaction Energy Profiles. At pH 7 (I ) 10-500 mM) and 9 (I ) 10 mM), the sand and PrPSc aggregates carried net negative charge (Figure 1). Thus, electrostatic repulsion should inhibit PrPSc attachment to the quartz surfaces. Nevertheless, PrPSc adsorption was observed under unfavorable electrostatic conditions. To obtain further insight into mechanisms controlling PrPSc adsorption to the quartz grains, interaction energies for PrPSc aggregates and quartz surfaces were calculated as a function of separation distance by summing electrostatic double layer and van der Waals interaction energies (DLVO theory) and the Born repulsion energy (20). Total Born-DLVO interaction energy profiles indicated favorable conditions for PrPSc attachment to quartz surfaces at pH 4 (I ) 10 mM) (Figure 4); no secondary minimum (Φ2°min) and no energy barrier (Φmax) to adsorption in the primary minimum (Φ1°min) were predicted (Table 1). At pH 7 and 9 (I ) 10 mM), substantial Φmax (110 kBT and 217 kBT) and shallow Φ2°min values (0.40 kBT and 0.19 kBT) were predicted (Table 1). At pH 7, increasing I from 10 to 100 mM results in disappearance of Φmax and deepening of Φ1°min allowing adsorption in the primary minimum. While the predicted trends with pH and I appear consistent with our experimental results (Figure 2), the shallow Φ2°min predicted for pH 7 and 9 (I ) 10 mM) appear insufficient for attachment (i.e., Φ2°min
