Impacts of Papain and Neuraminidase Enzyme ... - ACS Publications

May 5, 2009 - Electrohydrodynamics and IgG-Mediated Agglutination of Type A Red ... On the basis of these electrohydrodynamic characteristics, the ove...
0 downloads 0 Views 1MB Size
pubs.acs.org/Langmuir © 2009 American Chemical Society

Impacts of Papain and Neuraminidase Enzyme Treatment on Electrohydrodynamics and IgG-Mediated Agglutination of Type A Red Blood Cells )

Atsushi Hyono,† Fabien Gaboriaud,‡ Toshio Mazda,§ Youichi Takata,† Hiroyuki Ohshima,† and Jer^ome F. L. Duval*,

)

† Faculty of Pharmaceutical Science, Tokyo University of Science, 2641 Yamazaki, Noda, Chiba 278-8510, e, Japan, ‡Laboratoire de Chimie Physique et Microbiologie pour l’Environnement (LCPME), Nancy Universit CNRS, 405 rue de Vandoeuvre, F-54600 Villers-l es-Nancy, France, §Japanese Red Cross Central Blood Institute, 2-1-67 Tatsumi, Koto-ku, Tokyo 135-8521, Japan, and Laboratoire Environnement et Min eralurgie, Nancy Universit e, CNRS, 15 avenue du Charmois, B.P. 40, 54501 Vandoeuvre-l es-Nancy, Cedex, France

Received January 8, 2009. Revised Manuscript Received April 15, 2009 The stability of native and enzyme-treated human red blood cells of type A (Rh D positive) against agglutination is investigated under conditions where it is mediated by immunoglobuline G (IgG) anti-D antibody binding. The propensity of cells to agglutinate is related to their interphasic (electrokinetic) properties. These properties significantly depend on the concentration of proteolytic papain enzyme and protease-free neuraminidase enzyme that the cells are exposed to. The analysis is based on the interpretation of electrophoretic data of cells by means of the numerical theory for the electrokinetics of soft (bio)particles. A significant reduction of the hydrodynamic permeability of the external soft glycoprotein layer of the cells is reported under the action of papain. This reflects a significant decrease in soft surface layer thickness and a loss in cell surface integrity/rigidity, as confirmed by nanomechanical AFM analysis. Neuraminidase action leads to an important decrease in the interphase charge density by removing sialic acids from the cell soft surface layer. This is accompanied by hydrodynamic softness modulations less significant than those observed for papain-treated cells. On the basis of these electrohydrodynamic characteristics, the overall interaction potential profiles between two native cells and two enzyme-treated cells are derived as a function of the soft surface layer thickness :: in the Debye-Huckel limit that is valid for cell suspensions under physiological conditions (∼0.16 M). The thermodynamic computation of cell suspension stability against IgG-mediated agglutination then reveals that a decrease in the cell surface layer thickness is more favorable than a decrease in interphase charge density for inducing agglutination. This is experimentally confirmed by agglutination data collected for papain- and neuraminidasetreated cells.

1. Introduction Under certain conditions, an individual’s serum may contain antibodies that can specifically bind to antigens located on the red blood cell (RBC) surface membrane, a so-called immune response.1,2 Clinical tests, such as Coombs tests,3 are routinely undertaken to screen for atypical antibodies possibly present in blood plasma with the goals of the preparation of blood for transfusion, the detection of antibodies in blood plasma of pregnant women as part of antenatal care, and the detection of antibodies for the diagnosis of immune-mediated hemolytic anemia. Unfortunately, these tests are not efficient in detecting all antibodies upon simple mixing of blood and serum. As an example, immunoglobulin G (IgG) antibodies fail to agglutinate RBCs suspended in saline whereas immunoglobulin M (IgM) antibodies may lead to the agglutination of cells under similar conditions. (See Figure 1 and ref 4 and references therein.) To improve the degree of antibody detection of the aforementioned *To whom correspondence should be addressed. E-mail: jerome.duval@ ensg.inpl-nancy.fr. Tel: 00 33 3 83 59 62 63. Fax: 00 33 3 83 59 62 55. (1) Mollison, P. L.; Engelfriet, C. P.; Contreras, M. In Blood Transfusion in Clinical Medicine, 10th ed.; Blackwell Science: Oxford, U.K., 1997. (2) Geifman-Holtzman, O.; Wojtowycz, M.; Kosmas, E.; Artal, R. Obstet. Gynecol. 1997, 89, 272. (3) Coombs, R. R. A.; Mourant, A. E.; Race, R. R. Brit. J. Exp. Path. 1945, 26, 255. (4) Pollack, W.; Hager, H. J.; Peckel, R.; Toren, D. A.; Singher, H. O. Transfusion 1965, 5, 158.

Langmuir 2009, 25(18), 10873–10885

tests, professionals commonly resort to enzyme-based methodologies. Native RBCs are then “treated” by selective enzymes prior to antibody detection. Upon action of the enzymes, interfacial properties of RBCs are modified with, as a consequence, possible agglutination of the cells by antibodies under conditions where untreated cells would not have agglutinated. Hemagglutination via IgG antibody molecules occurs when the (native- or enzyme-treated) blood cell surface membranes, where the anchoring antigen D sites are located (Figure 1), are separated by ∼14 nm.1,4,5 This distance corresponds to the geometrical separation of binding entities within the IgG molecule.6 Agglutination of RBCs by IgG may then be observed if the overall interaction between (charged) RBCs allows for their sufficient proximity so that IgG anti-D antibody connection possibly occurs. The difference invoked above between the respective capacities of IgG and IgM antibodies for agglutinating native RBCs may be explained by realizing that the separation distance between binding sites in IgM molecule is around 30 nm.7 IgM then has a better agglutinability capacity than IgG because RBCs (5) Issit, P. D.; Anstee, D. J. The Immune Response, Production of Antibodies, Antigen-Antibody Reactions. In Applied Blood Group Serology, 4th ed.; Issit, P. D., Anstee, D. J., Eds.; Montgomery Scientific Publications: Durham, NC, 1998; Chapter 2. (6) Raghupathy Sarma, V.; Silverton, E. W.; Davies, D. R.; Terry, W. D. J. Biol. Chem. 1971, 246, 3753. (7) Frazer, K.; Capra, J. D. Immunoglobulins: Structure and Function. In Fundamental Immunology, 4th ed.;Paul, W. E., Ed.; Raven Press: New York, 1998.

Published on Web 05/05/2009

DOI: 10.1021/la900087c

10873

Article

Figure 1. Schematic illustration of the conditions required for IgG- and IgM-mediated red blood cell agglutination.

do not have to cross any repulsive electrostatic interaction energy barrier for IgM agglutination to occur. The purpose of this work is to identify the effects of a proteolytic enzyme (papain) and protease-free neuraminidase enzyme on RBC surface properties and thereby understand their impact on the propensity of blood cells to agglutinate in the presence of IgG antibody. Even though the processes underlying the effect of enzymes on RBC surface properties in relation to hemagglutination have been discussed in various theoretical and experimental studies,4,8-10 we are forced to recognize that the literature is still full of controversy. Attempts have been reported for computing the DLVO interaction potential between native or enzyme-treated red blood cells so as to appreciate their agglutination with IgG anti-D antibody.8,9 In these studies, the interaction energy is based on RBC surface properties expressed in terms of the zeta potential or electrokinetic potential, as measured by electrophoresis. Van Oss and Absolom8,9 concluded their work by stating that agglutination with IgG anti-D antibodies after proteolytic enzyme treatment was essentially the result of a decrease in the RBC zeta potential, thereby decreasing the interparticular repulsive interaction energy. However, Stratton et al.10 suggested that the reduction of steric hindrance between neighboring cells could be more critical than the decrease in their respective zeta potential in governing the agglutination process. This conclusion was qualitatively supported by the observation that RBC treatment with neuraminidase enzyme leads to the agglutination of RBCs by IgG anti-D antibodies that is less significant than that obtained with papain-treated RBCs. One serious shortcoming in the aforementioned studies is that the reasoning is based on the concepts of zeta potential and surface charge for describing the electrosurface properties and electrostatic interaction of/between red blood cells. Red blood cells exhibit at their outer periphery a permeable charged glycoprotein layer of 3-15 nm11 thickness that confers upon them the typical characteristics of soft permeable particles. Following the definition given by Ohshima,12 soft particles indeed generally consist of a hard core of radius denoted as a, which is stricto sensu impermeable to fluid flow, and an adsorbed polyelectrolyte-type layer, such as a protein layer, characterized by a 3D spatial distribution of hydrodynamically stagnant, ionogenic groups. It has long been recognized that the electrophoretic behavior of soft particles deviates substantially from that of hard (i.e., rigid) (8) Van Oss, C. J.; Absolom, D. R. Vox Sang. 1983, 44, 183. (9) Van Oss, C. J.; Absolom, D. R. Vox Sang. 1984, 47, 250. (10) Stratton, F.; Rawlinson, V. I.; Gunson, H. H.; Phillips, P. K. Vox Sang. 1973, 24, 273. (11) Levine, S.; Levine, M.; Sharp, K. A.; Brooks, D. E. Biophys. J. 1983, 42, 127. (12) Ohshima, H. Adv. Colloid Interface Sci. 1995, 62, 189. (13) Duval, J. F. L. Electrophoresis of Soft Colloids: Basic Principles and Applications. In Environmental Colloids and Particles: Behaviour, Separation and Characterisation; Wilkinson, K. J., Lead, J., Eds.; IUPAC Series on Analytical and Physical Chemistry of Environmental Systems; John Wiley & Sons: Chichester, England, 2007; Vol. 10, Chapter 7.

10874 DOI: 10.1021/la900087c

Hyono et al.

particles.12 (See the review in ref 13 and references therein.) Numerous studies14-24 showed that it is indeed physically irrelevant to locate a priori a shear plane (or slip plane) and define a zeta potential for soft particles. Studies where modifications of the interphasic properties of blood cells after enzymatic treatment are examined on the basis of electrophoretic mobility data analyzed according to electrokinetic theories for hard particles should be received with much caution. Recently, Duval and Ohshima24 proposed a numerical theory for the electrokinetics of soft particles where the possibility of a heterogeneous (or diffuse) distribution of charged polymer segments within the soft layer is explicitly taken into account. Successful application of this theory has been reported for elucidating the interphasic properties of humic acids,25 polysacharrides,26,27 bacteria,28-30 viruses,31 and yeasts32 over a large range of ionic strengths and pH conditions. Also, experimental work showed that this theory is useful and relevant for unraveling electrosurface phenomena such as swelling/shrinking processes and detecting associated structural modifications of the soft surface layer carried by the particle.26-29 Besides, it was recently demonstrated how this theory may be used to derive the relevant physicochemical parameters required for evaluating the electrostatic interaction forces between bacteria and a hard substrate surface to be colonized.29 In view of the above elements, the objectives of the current analysis are essentially threefold: (i) On the basis of the theory for electrokinetics of diffuse soft bioparticles,24 we present here a rigorous numerical investigation of electrophoretic mobility data measured for red blood cells of type A in the native state or subjected to enzyme treatment via exposure to papain and neuraminidase enzymes. (ii) On the basis of the electrohydrodynamic parameters characterizing the interphasic properties of native or enzyme-treated red blood cells, we theoretically derive and compute the interaction potential profiles under physiological conditions between two soft red blood cells that have been subjected or not to enzyme treatment. As argued before, in this derivation, no use is made of the concept of zeta potential, which is physically irrelevant for RBCs. (14) Ohshima, H. J. Colloid Interface Sci. 2000, 228, 190. (15) Ohshima, H. J. Colloid Interface Sci. 1994, 163, 474. (16) Ohshima, H. Colloid Polym. Sci. 2005, 283, 819. (17) Ohshima, H. J. Colloid Interface Sci. 1997, 185, 269. (18) Saville, D. A. J. Colloid Interface Sci. 2000, 222, 137. (19) Hill, R. J.; Saville, D. A.; Russel, W. B. J. Colloid Interface Sci. 2003, 258, 56. (20) Hill, R. J.; Saville, D. A.; Russel, W. B. J. Colloid Interface Sci. 2003, 263, 478. (21) Hill, R. J.; Saville, D. A. Colloids Surf., A 2005, 267, 31. (22) Lopez-Garcia, J. J.; Grosse, C.; Horno, J. J. Colloid Interface Sci. 2003, 265, 327. (23) Lopez-Garcia, J. J.; Grosse, C.; Horno, J. J. Colloid Interface Sci. 2003, 265, 341. (24) Duval, J. F. L.; Ohshima, H. Langmuir 2006, 22, 3533. (25) Duval, J. F. L.; Wilkinson, K. J.; van Leeuwen, H. P.; Buffle, J. Environ. Sci. Technol. 2005, 39, 6435. (26) Rotureau, E.; Thomas, F.; Duval, J. F. L. Langmuir 2007, 23, 8460. (27) Duval, J. F. L.; Slaveykova, V. I.; Hosse, M.; Buffle, J.; Wilkinson, K. J. Biomacromolecules 2006, 7, 2818. (28) Duval, J. F. L; Busscher, H. J.; van de Belt-Gritter, B.; van der Mei, H. C.; Norde, W. Langmuir 2005, 21, 11268. (29) Gaboriaud, F.; Gee, M. L.; Strugnell, R.; Duval, J. F. L. Langmuir 2008, 24, 10988. (30) Clements, A.; Gaboriaud, F.; Duval, J. F. L.; Farn, J. L.; Jenney, A. W.; Lithgow, T.; Wijburg, O. L. C.; Hartland, E. L.; Strugnell, R. A. PLosOne 2008, 3, e3817. (31) Langlet, J.; Gaboriaud, F.; Gantzer, C.; Duval, J. F. L. Biophys. J. 2008, 94, 3293–3312. (32) Karreman, R. J.; Dague, E.; Gaboriaud, F.; Quiles, F.; Duval, J. F. L.; Lindsey, G. G. Biochim. Biophys. Acta 2007, 1774, 131.

Langmuir 2009, 25(18), 10873–10885

Hyono et al.

(iii)

Article

Upon comparison of the theoretical results obtained in (ii) with IgG-mediated agglutination data, we provide an interpretation of the respective roles of papain and neuraminidase enzymes in promoting or not type A blood cell agglutination by IgG anti-D antibodies.

2. Materials and Methods 2.1. Enzymes and Enzyme Activity Assays. Papain was purchased from E. Merck AG (Darmstadt, Germany) and dissolved in 9 parts saline and 1 part 0.067 M sodium potassium phosphate buffer, pH 5.4, without activator. The solution was stored at -20 C until use. Proteinase activities of papain were determined using a modified version of the method of Hagihara et al.33 and were expressed in casein units per 50 μL.34,35 One casein unit is defined as the enzyme activity that at 30 C liberates for 1 min soluble digestion products (namely peptides) resulting in an increase in the absorbance at 275 nm equivalent to that produced by 1 mg of tyrosine. Neuraminidase (protease-free) isolated from Arthrobacter ureafaciens was purchased from Nacalai Tesque (Kyoto, Japan) and dissolved in 9 parts saline and 1 part 0.1 M sodium potassium phosphate buffer (PBS), pH 7.4. The solution was stored at -20 C until use. One unit of neuraminidase is the amount needed to liberate for 1 min 1μmol of N-acetylneuraminic acid from N-acetylneuraminyl-(2f3)-lactose (NAN-lactose) at pH 5, 37 C. Neuraminidase activity was given by the manufacturer and determined as described elesewhere.36 There is no equivalence between papain and neuraminidase enzyme units because the former is a protease enzyme and the latter is protease-free.

2.2. Treatment of Red Blood Cells with Enzymes Followed by Agglutination Tests. The treatment of red blood cells with enzyme was carried out by mixing the enzyme solution with an equal volume of packed red blood cells (type A, Rh D positive) that had been washed with saline. After incubation for 15 min at 37 C, the cells were washed 4 times with 10 times their volume of saline. The concentrations of neuraminidase were 0, 0.05, and 0.1 units, and concentrations of papain ranged between 0 and 5 casein units. For the agglutination test with IgG anti-D antibody, red blood cells were suspended at a concentration of 3% (v/v) in PBS, pH 7.4. Anti-D serum (25 μL) obtained from a patient with weak antiD antibodies was mixed with an equal volume of a 3% red blood cell suspension. After incubation for 15 min at 37 C, the test tube was centrifuged for 1 min at 150g. Agglutination was detected macroscopically. The IgG antibody titrations were performed using aliquots of serial master dilution of serum in saline, as detailed elsewhere.37 2.3. Electrophoretic Mobility Measurements. Electrophoretic mobility measurements were performed with a madeto-order apparatus by Central Scientific Commerce, Inc. (Tokyo, Japan). In the quartz electrophoretic cell, a constant directcurrent electric field (10 V/cm) is generated between a molybdenum electrode and a palladium-covered electrode. The red blood cell displacements were measured by the reflection of a laser beam and tracked with a charge-coupled device camera, and electrophoretic mobilities were evaluated by image-analysis software. For all experiments, the temperature was maintained at 25 C. Experiments were carried out at natural pH (∼6) and repeated twice with different cultures. The red blood cells to be (33) Hagihara, B.; Matsubara, H.; Nakai, M.; Okunuki, K. J. Biochem. 1958, 45, 185. (34) Mazda, T.; Ogasawara, K.; Nakata, K.; Shimizu, M. Vox Sang. 1987, 52, 63. (35) Ogasawara, K.; Mazda, T. Vox Sang. 1989, 57, 72. (36) Ohta, Y.; Tsukada, Y.; Sugimori, T. J. Biochem. 1989, 106, 1086. (37) Scott, M. L.; Voak, D.; Phillips, P. K.; Ann Hoppe, P.; Kochman, S. A. Vox Sang. 1994, 67, 1994.

Langmuir 2009, 25(18), 10873–10885

Figure 2. (A) Atomic force microscopy imaging of a type A red blood cell deposited onto a glass slide. (B) Representation of the diffuse soft interphase between the membrane-anchored glycoprotein layer and outer electrolytic medium (160 mM) according to eqs 1 and 2 (a = 3 μm, δ = 5 nm). (a) R f 0, (b) R = 1 nm, (c) 2 nm, (d) 3 nm, (e) 4 nm, (f) 5 nm, (g) 6 nm, (h) 7 nm. tested were prepared in a 0.1% suspension and were obtained by mixing 1 volume of sodium potassium phosphate buffer, pH 7.4, with 9 volumes of physiological saline. The solution ionic strength (5 mM to 0.16 M in this study) was adjusted by dilution of the buffer with an isotonic sucrose aqueous solution. 2.4. AFM Measurements. Force curves were obtained at room temperature using an MFP-3D instrument (Asylum Research, Santa Barbara, CA). All force experiments were recorded under physiological conditions using microlever probes (MLCTEXMT-BF, Santa Barbara, CA) with spring constants systematically determined following the thermal calibration method (mean value equal to 22 ( 3 pN/nm). Glass slides (38  26 mm2, Menzer GmBH, Braunschweig, Germany) were washed overnight in 70% nitric acid before rinsing with distilled water. The slides were then immersed in 0.2% (w/v) polyethyleneimine solution (PEI, Sigma Chemical Co.) and left for 4 h, after which they were washed with distilled water. To immobilize the cells on the dried PEI slides, a droplet of native or enzyme-treated blood cell solution (optical density of 0.06 at 600 nm) was deposited onto the slide for a period of 1 h, and the slide was subsequently rinsed and directly transferred to the liquid cell mounted in the AFM instrument (Figure 2). The cells were precisely located using an optical image obtained from the inverted microscope with which the AFM apparatus is equipped (Supporting Information). The cantilever was then moved to the apex of the cell, and force curves were recorded. The measured force profiles were converted into force-indentation curves, and various key mechanical parameters including the blood cell spring constant, Young’s modulus, and viscoelastic energy were evaluated according to a procedure described in detail elsewhere38 and illustrated in Supporting Information.

3. Theory 3.1. Electrokinetic Model for Red Blood Cells. The electrokinetic properties of bioparticles (e.g., bacteria, yeasts, blood cells) or environmental polyelectrolytes (e.g., humic acids and succinoglycan) are governed by a complex interplay between electrostatic and hydrodynamic processes. As demonstrated in numerous experimental and theoretical studies,13,24-32 the analysis of the electrohydrodynamics of those systems cannot be performed according to theories that are strictly valid for hard colloidal objects, and the concept of the zeta potential for such systems loses its conventional meaning.13 Instead, one must necessarily resort to more advanced models where flow penetration within the particle is explicitly taken into account. (38) Gaboriaud, F.; Parcha, B. S.; Gee, M. L.; Holden, J. A.; Strugnell, R. Colloids Surf., B 2008, 62, 206.

DOI: 10.1021/la900087c

10875

Article

Hyono et al.

Recently, Duval et al. derived the governing set of electrohydrodynamic equations that account for the migration of a spherical, permeable, heterogeneous particle under the action of an externally applied electric field.24 The evaluation of the electrophoretic mobility of such a diffuse soft particle relies on the complex numerical solving of the electrostatic and transport equations for the particle and ions that are distributed around and within it. The formalism extends the analytical model originally developed by Ohshima,12 which is applicable for homogeneous particles (step-function representation of the interface) when electric double layer polarization/relaxation effects are negligible. The electrokinetic modeling of a diffuse soft particle may be done for any radial soft material density distribution f (r) obtained from either empirical, analytical, or computational considerations. In the following, in the absence, a priori, of concrete information on the local spatial distribution of the charges within the soft surface layer of thickness δ around native and treated red blood cells, we take the general following expression for f (r) as adopted earlier24 (  ) nðrÞ ω r -ða þ δ f ðrÞ ¼ ¼ 1 -tanh no 2 R

ð1Þ

with n(r) being the number density of polymer segments at radial position r (the origin is taken at the center of the core component of the particle), no being the nominal number density of polymer segments, a being the radius of the impermeable core of the particle, and the parameter R pertaining to the typical decay length of the function f (Figure 2). The ratio R/δ determines the degree of interphasic diffuseness: for R/δ f 0, f reduces to the classical step-function-like representation of soft interphase (n(r) f no) whereas for a nonzero value of R/δ the radial density of polymer segments decays continuously from bulk values deep inside the layer to zero in the electrolyte solution. Dimensionless parameter ω is determined in such a way that the total number of polymer segments remains constant upon variation of R and/or δ. It is therefore given by the expression ω ¼

½ða þ δÞ3 -a3  = 3

Z

¥

f ðrÞr2 dr

ð2Þ

0

Equations 1 and 2 typically illustrate the heterogeneity of a soft surface layer as a result of osmotic swelling taking place upon the decrease in electrolyte concentration.39,40 The spatial dependence for the segment density across the interphase, as imposed by eq 1 and illustrated in Figure 2, has been experimentally measured for a number of swollen systems using light-scattering techniques or neutron reflectivity.39 Such swelling occurs upon the condition of a constant number of charges in the layer (eq 2) with an increase in the surface layer thickness as a result of layer extension. It can be verified that the layer thickness may be expressed as δ + 2.3R because the position r = a + δ + 2.3R corresponds to that where n(r) is only a few percent of the value deep inside the surface layer.24 For the sake of simplicity, the red blood cells are assimilated to spheres of core radius a ≈ 3 μm11 with a coated glycoprotein layer of thickness δ supported by a lipid bilayer of thickness d. (See section 3.3 for a discussion on values adopted for δ and d.) From an electrostatic point of view, this geometry corresponds to that compatible with the thin electric double layer limit because for (39) Karim, A.; Satija, S. K.; Douglas, J. F.; Ankner, J. F.; Fetters, L. J. Phys. Rev. Lett. 1994, 73, 3407. (40) Muller, F.; Romet-Lemonne, G.; Delsanti, M.; Mays, J. W.; Daillant, J.; Guenoun, P. J. Phys.: Condens. Matter 2005, 17, S3355.

10876 DOI: 10.1021/la900087c

Figure 3. Description of the nomenclature used for computing the local electrostatic potential ψ(x) between two red blood cells with charged glycocalyx and a protein soft surface layer of thickness δ, with a lipid bilayer of thickness d and separated by a distance H + 2δ . A typical representation of the resulting potential ψ(x) is given, and the location of antigen D of the red blood cells is specified. We take the x axis perpendicular to the membranes with the origin at the core surface of the left membrane.

ionic strength values of practical interest here, the condition κa . 1 is always verified (κ is the reciprocal Debye layer thickness). As such, the assimilation of the biconcave disk blood cells to spheres with large radius a is not a matter of concern (within electrokinetic analysis) because we are essentially investigating systems for which electrokinetic theory applied to flat-plate geometry particles is acceptable, as we shall demonstrate below (section 4.3). To verify this assumption, we have cautiously examined the electrokinetic data commented on in section 4.1 and, more specifically, the movies displaying the trajectories of red blood cells under the action of the applied field. At low ionic strengths (20 mM and below) (i.e., in the ionic strength regime where the above assumption may be the most questionable), we were able to clearly identify both discocytes (biconcave disk blood cells) and spherocytes (spherical red blood cells). Their respective trajectories and electrophoretic mobilities were then determined separately. The results displayed in Supporting Information unambiguously show that within experimental error the mobilities of spherocyte and discocyte blood cells are identical, thus supporting our approach that consists of assimilating blood cells to spherical particles. For a given polymer segment density profile f, the electrophoretic mobility, μ, is obtained upon the numerical resolution of the set of coupled Navier-Stokes, nonlinear Poisson-Boltzmann, and continuity-governing equations as detailed in ref 24. The friction exerted by the flow on the soft surface layer is modeled along the lines set by Debye-Bueche,41 who assimilate the polymer segments to so-called resistance centers. For further details, the reader is referred to ref 24. Basically, the electrokinetic theory involves the following key parameters: (i) the nominal volume charge density across the soft surface layer, denoted as Fo; (ii) the nominal softness parameter λo with (λo)-1 being the typical flow penetration length within the soft surface layer; (iii) geometrical dimensions a and δ; and (iv) the heterogeneity parameter R/δ that reflects the deviation of the radial distribution of polymer segments from the homogeneous step-function-like profile. 3.2. Overall Gibbs Energy of Interaction between Two Red Blood Cells. Consider two identical soft red blood cell particles that are composed of an impermeable (hard) core and a ion-permeable (soft) surface layer of thickness δ as depicted in Figure 3. The particles are immersed in a 1:1 electrolyte of bulk (41) Debye, P.; Bueche, A. J. Chem. Phys. 1948, 16, 573.

Langmuir 2009, 25(18), 10873–10885

Hyono et al.

Article

concentration c¥ and are separated by a distance denoted as H. The overall Gibbs energy of interaction, denoted as V (in J), between the red blood cells is taken as the sum of the electrostatic (Uel), van der Waals (UvdW), and steric contributions (Uster) according to VðHÞ ¼ fUvdW ðHÞ þ Uel ðHÞ þ Uster ðHÞgS

ð3Þ

where S is the surface area of cells that is involved in the interaction. Below, the respective contributions Uel, UvdW, and Uster are detailed for flat disk-assimilated cells. In such geometry, the surface area S is, according to Van Oss et al.,8,9 given by S = πa2/3, with a ≈ 3 μm being the cell radius. In Supporting Information, the expressions for Uel and UvdW are given for the interaction between torus-shape-assimilated blood cells. As demonstrated in Supporting Information, the consideration of such a geometry that possibly better represents the reality of blood cells does not qualitatively alter the conclusions of our analysis. 3.2.1. Electrostatic Component of the Gibbs Energy of Interaction between Two Red Blood Cells. In line with the nomenclature introduced in section 3.1, the soft surface layers carry a nominal volume charge density denoted as Fo. For the sake of generalization, the polymer segment distribution follows a radial profile f (r). Providing that charges are uniformly distributed along single polymer chains, the local charge density within the layer, denoted as Ffix(r), is expressed by Ffix ðrÞ ¼ Fo f ðrÞ

ð4Þ

with the limit Ffix(r) f Fo for R/δ f 0. The tacit assumption underlying the validity of eq 4 is that the functional groups responsible for the overall charge of red blood cells are completely ionized so that the quantity Fo does not depend on the local electrostatic potential ψ (or on ionic strength) at the interphase formed with the outer electrolyte.24 Under the pH conditions of our electrokinetic experiments (∼6), the assumption of complete ionization of polymer segments responsible for the cell charge is fully justified because (i) the charge predominantly originates from the dissociation of carboxyl groups of sialic acids42 (see comment below) and (ii) the pK of these carboxyl groups is known to be close to 2.6.42 Because the extent of ionization of the carboxyl groups is given by 1/(1 + 10pK - pH exp[-Fψ/RT])43 with F being the Faraday constant, R being the gas constant, and T being the temperature, it is easily verified that complete ionization is reached under the conditions of interest here. In section 4, the properties of cell stability against IgGmediated agglutination will be quantitatively analyzed on the basis of the theory reported in the current section. The analysis will be carried out under physiological conditions (160 mM) where the inequalities κa . 1 and κδ . 1 are well satisfied. For this reason, we neglect here curvature effects and consider the situation of electrostatic interaction between 1D soft surface layers (Figure 3) of thickness δ supported by a hard substrate of thickness d. For RBCs, the soft surface layer is made of charged glycoprotein layers (δ ≈ 3-15 nm11), and the hard supporting substrate is the presumably uncharged lipid bilayer11 of thickness d ≈ 4 nm.44 Neglect of the charge on the lipid bilayer (charge due to the presence of phosphate groups) is legitimate because, as mentioned above, previous studies on blood cells have clearly (42) Eylar, E. H.; Madoff, M. A.; Brody, O. V.; Oncley, J. L. J. Biol. Chem. 1962, 237, 1992. (43) Voigt, A.; Donath, E.; Kretzschmar, G. Colloids Surf., A 1990, 47, 23. (44) Heinrich, V.; Ritchie, K.; Mohandas, N.; Evans, E. Biophys. J. 2001, 81, 1452.

Langmuir 2009, 25(18), 10873–10885

demonstrated that their overall negative charge at pH ∼6 is primarily due to the dissociation of carboxyl groups of sialic acids distributed throughout the glycoprotein layer rather than the phosphate groups of the phospholipid bilayer.42 More quantitatively, the number of charges issued from sialic acid dissociation in human erythrocytes at pH ∼6 is about 4 to 5 times larger than that which originates from the dissociation of functional groups located at the lipid bilayer.42 The electrostatic contribution to the Gibbs interaction energy (per unit surface area) between two parallel planar surfaces, denoted as Uel, is obtained by integration of the disjoining pressure, Πel, according to Z Uel ðHÞ ¼ -

H ¥

Pel ðH 0 Þ dH 0

ð5Þ

where H0 is a dummy integration variable. Πel is the amount by which the normal component of the pressure tensor exceeds the :: outer pressure. Within the Debye-Huckel approximation (valid for red blood cells at 160 mM), Πel is defined by Pel ðHÞ ¼ c¥ RTfym ðHÞg2

ð6Þ

where ym is the potential at the midposition xm = κ (δ + H/2). Solving the Poisson-Boltzmann equation within the Debye:: Huckel approximation under conditions of uncharged core surface and absence of overlap between soft surface layers, we obtain ym ¼ -β

cosh2 ðxm Þ sinhðxm Þ

Z

¥

f ðxÞ coshðxÞ dx þ

0

Z β sinhðxm Þ

xm

f ðxÞ coshðxÞ dx ð7Þ

0

where β = -Fo/(2c¥F) and x = κ(r - a). In the limit R/δ f 0, we have ym = -β sinh(κδ)/sinh(xm), and Uel (in J m-2) may then be explicitly evaluated according to the relationship (   ) 2c¥ RTβ2 KH -1 sinh2 ðKδÞ coth Kδ þ Uel ðHÞ ¼ 2 K

ð8Þ

which is in line with the expression given by Ohshima et al.45 3.2.2. van der Waals Component of the Gibbs Energy of Interaction between Two Red Blood Cells. The overall van der Waals interaction energy (in J m-2) between two flat plates having adsorbed layers may be expressed as a function of the separation distance H according to the general relationship given by Vincent46 Uvdw ðHÞ ¼ fFo ðHÞ þ Fi ðHÞ -2Foi ðHÞgAPL þ Ff ðHÞALB þ 2fFof ðHÞ þ Fif ðHÞgALB 1=2 APL 1=2 ð9Þ where APL and ALB are the Hamaker constants for the adsorbed protein layer and for the lipid bilayer, respectively. The terms Fj=o,i,oi,f,of,if (H ) in eq 9 are functions of the separation distance H, as detailed in ref 46. 3.2.3. Steric Interaction between Two Red Blood Cells. As discussed in ref 47, the key parameters governing the strength of (45) Ohshima, H.; Makino, K.; Kondo, T. J. Colloid Interface Sci. 1987, 116, 196. (46) Vincent, B. J. Colloid Interface Sci. 1973, 42, 270. (47) Rijnaarts, H. H. M.; Norde, W.; Lyklema, J.; Zehnder, A. J. B. Colloids Surf., B 1999, 14, 179.

DOI: 10.1021/la900087c

10877

Article

steric interactions between polymer segments are the segment density distribution and the Flory-Huggins interaction parameters χij that reflect the interaction between pairs of groups i and j across the solvent. According to recent theories,48 classical DLVO forces (electrostatics and van der Waals contributions included) may not be simply added to the steric contributions because the charge carried by the segments affects the segment density distribution, which in turn impacts the magnitude of χij. In view of this major difficulty, we shall consider here the picture of an infinitely steep repulsive steric interaction energy Uster(H) in the neighborhood of x = κδ (step-function-like dependence on separation distance). This procedure, adopted in ref 47 within the context of the adhesion of bacteria onto macroscopically flat surfaces, considerably simplifies the analysis by resorting to a single parameter, the steric thickness δ. Despite this simplicity, this formulation is strongly in line with theory and experiments on steric interactions that show that that the typical decay length for Uster(H ) is several orders of magnitude lower than that for UvdW(H) and Uel(H). 3.3. Assignment of the Parameters for the Analyses of the Electrokinetic and Stability Features of Red Blood Cells. We find it important to discuss here the numerical values of parameters APL, ALB, δ, and d that will be used for the quantitative investigation of the electrohydrodynamic properties and stability of native and enzyme-treated red blood cells. Concerning the membrane lipid bilayer, a thickness of d ≈ 4 nm is a good estimate as judged from values provided in the literature.44 The Hamaker constant ALB for lipid/water systems is very well documented with published data spanning a relatively small range [e.g., (3.4-6.8)  10-21 J,49 (3.7-4)  10-21 J,50 (4.31 ( 0.44)  10-21 J, (3.73 ( 0.35)  10-21 J,51 and (3.6 ( 0.8)  10-21 J 52]. On the basis of these data, we shall adopt, where appropriate, the value ALB ≈ 4  10-21 J (∼1 kBT at 298 K). Regarding the RBC soft surface layer that corresponds to the glycoprotein layer, the wide range of reported thickness δ reflects the various analytical methods used and the associated experimental and interpretative difficulties. From an examination of ghost cell surface ultrastructure, Hillier et al.53 suggested a value for δ of about 3 nm. However, Weinstein54 argued that Hoffman results could be influenced by the collapse of glycoprotein layer molecules as a result of surface tension forces of evaporating water introduced during the airdrying step. More recently, Linss et al.55 argued, on the basis of electron microscopy data, that the membrane-anchored glycoprotein layer could extend the bilayer foundation of the membrane to the order of ∼10 nm. In this work, we shall consider δ values in the range of ∼3 to 15 nm and specifically address the impact of the magnitude of δ on the electrokinetic analysis and on the interpretation of interaction potential profiles in relation to IgG agglutination data. Because of the complexity of the composition of the glycoprotein layer and the ongoing controversy regarding its extension, the Hamaker constant pertaining to this surface layer remains very difficult to address with accuracy. Fortunately, as correctly mentioned by Lund et al. in their formulation of a mesoscopic model for protein-protein (48) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M. H. M.; Cosgrove, T.; Vincent, B. In Polymer at Interfaces; Chapman and Hall: London, 1993; Chapter 11. (49) Gingell, D.; Parsegian, V. A. J. Theor. Biol. 1972, 36, 41. (50) Brooks, D. E.; Levine, Y. K.; Requena, J.; Haydon, D. A. Proc. R. Soc. London, Ser. A. 1975, 347, 179. (51) Requena, J.; Haydon, D. A. Proc. R. Soc. London, Ser. A. 1975, 347, 161. (52) Ohshima, H.; Inoko, Y.; Mitsui, T. J. Colloid Interface Sci. 1982, 86, 57. (53) Hillier, J.; Hoffman, J. F. J. Cell. Comp. Physiol. 1953, 42, 203. (54) Weinstein, R. S. The morphology of adult red cells. In The Red Blood Cell.; Surgenor, D. M., Ed.; Academic Press: New York, 1974. (55) Linss, W.; Pilgrim, C.; Feuerstein, H. Acta Histochem. 1991, 91, 101.

10878 DOI: 10.1021/la900087c

Hyono et al.

interactions in solutions,56 Hamaker constants of proteins in water are not subject to large variations with values generally in the range of (3-10)kBT. Implementing such value for APL within the interaction modeling described above reveals that under physiological conditions, untreated red blood cells should strongly aggregate. This denotes an overestimation of the van der Waals component connected to the presence of the glycoprotein layer. The authors verified that the removal of this unrealistic aggregation under physiological conditions goes along with APL , ALB or APL f 0. This latter equality comes to state that the number density of molecules in the soft surface glycoprotein layer is too small to provide a significant van der Waals force as compared to that which originates from the presence of lipid bilayer. As revealed by electrokinetic results discussed below, the charge density carried by the soft layer is around 50 mM (expressed in concentration of moles of charges). Considering that the molecules responsible for the charge represent 10% of the total number of molecules in the soft surface layer, the total number density of molecules is then about 0.5 M, which is 1% of the number of water molecules. Because Hamaker constants are proportional to number density squared, the difference in Hamaker constants for the aqueous medium and for the protein layer (this difference defines APL) is about 0.01%, which is in line with APL f 0.

4. Results 4.1. Electrokinetics. In Figure 4A, we report the electrophoretic mobility μ (expressed in dimensionless form) as a function of ionic strength for untreated RBCs and for RBCs treated with papain and neuraminidase at different concentration levels (indicated). Under all conditions, μ is negative, thus denoting a negative charge carried by the glycoprotein layer. As discussed in section 3.2.1, at the pH of the experiments, this charge predominantly originates from the complete dissociation of the carboxyl groups of sialic acid within the soft surface layer.42,57 Overall, the ionic-strength dependence for the electrophoretic mobility μ is typical of that obtained for soft permeable particles.13,24-32 Upon increasing ionic strength, μ decreases in magnitude as the result of screening of the cell charge by the ions present in the electrolyte medium. For sufficiently large ionic strengths, μ asymptotically reaches a finite nonzero constant value, which is the characteristic signature of the presence of a permeable charged surface layer (the glycoprotein layer) in the peripheral zone of the blood cells. The mobility plateau value is given here by12,24 μjc¥ f¥ f

Fo coshðλo δÞ -1 ηλ2o coshðλo δÞ

ð10Þ

Upon closer inspection, it is observed that for a fixed ionic strength, mobilities of enzyme-treated cells are lower in magnitude as compared to that for native cells. For a given enzyme, the larger the concentration of enzyme to which blood cells were exposed, the more significant the aforementioned mobility decrease. These variations of the cell mobility reflect the changes in electrohydrodynamic properties of blood cells after enzymatic treatment (i.e., the changes in charge density Fo, hydrodynamic softness λo, and possibly thickness δ pertaining to the soft surface layer; see section 4.3). 4.2. Cell Agglutination Mediated by IgG Anti-D Antibodies. Before quantitatively analyzing the electrokinetic data :: (56) Lund, M.; Jonsson, B. Biophys. J. 2003, 85, 2940. (57) Kawahata, S.; Ohshima, H.; Muramatsu, N.; Kondo, T. J. Colloid Interface Sci. 1990, 138, 182.

Langmuir 2009, 25(18), 10873–10885

Hyono et al.

Article

Figure 4. (A) Dimensionless electrophoretic mobility as a function of medium ionic strength for native, papain-treated, and neuraminidasetreated red blood cells (RBC). The concentration (in units U, see details in the main text) of enzyme used for the treatment of the red blood cells is indicated. The quantity εo is the dielectric permittivity of vacuum, εr is the relative dielectric permittivity of water, η is the dynamic viscosity of water, kB is the Boltzmann constant, and e is the elementary charge. (B) Agglutination data (anti-D titer) for type A red blood cells as a function of concentration of neuraminidase and papain. Points at 0.001 enzyme unit refer to untreated cells. See the text for further details.

along the lines set forth in section 3.1, we now comment on the agglutination properties of native and enzyme-treated blood cells so as to underline the intimate relationship between electrosurface properties of RBCs and hemagglutination in the presence of IgG anti-D antibodies. The results of agglutination tests under physiological conditions are displayed in Figure 4B. In this Figure, the y scale refers to the agglutination titer for the anti-D IgG antibody, that is, the degree of maximum dilution of the stock solution of IgG that lets red blood cells agglutinate. For a given enzyme concentration, titer 0 means that cell agglutination is not detected when IgG antibody concentration was that determined by the initial operation that consists of mixing 25 μL of anti-D serum (obtained from a patient with weak anti-D antibodies) with an equal volume of 3% red blood cells previously exposed to enzyme. Titer X (with 0 < X e 64 under the conditions of Figure 4B) means that cell agglutination is detected for an IgG solution with a concentration X times (at most) lower than that of the initial IgG solution. The higher the titer, the easier cell agglutination. In light of Figure 4B, it is found that papain treatment of blood cells considerably favors their agglutination. A dramatic increase in the anti-D titer is observed for RBCs exposed to papain concentrations ranging from 0.1 to 5.0 units. The titer increases with the enzyme units, tending to a maximum value of ca. 64 for papain (reached for cells treated with 5 papain units) whereas in the case of neuraminidase-treated cells the plateau value for the titer is about 4 (reached for cells treated with 1 neuraminidase unit). This reveals that cell exposure to neuraminidase enzyme does not favor the agglutination of the cells as strongly as exposure to papain. To examine the nature of the action of the enzymes on blood cells and to attempt an explanation of the apparently distinct surface modification that they undergo, we now quantitatively interpret the electrohydrodynamic profiles of native and enzyme-treated cells depicted in Figure 4A. 4.3. Quantitative Analysis of Electrokinetic Data. As a first step, the numerical analysis of the data is carried out by considering an homogeneous soft interphase between the glycoprotein layer and the outer electrolyte medium. We now tackle a segment density distribution, within the surface layer, characterized by R/δ f 0 (section 3.1). Note that such a step-function-like Langmuir 2009, 25(18), 10873–10885

interphase representation is the most adequate for sufficiently large ionic strength in the medium (i.e., when the intramolecular repulsion of charges embedded in the soft surface layer is screened to such an extent that swelling of the glycoprotein layer and accompanying interphasic heterogeneity is nearly absent or insignificant13,24). With such a representation, the remaining unknown parameters are Fo and λo, taking for the thickness δ values in the range of 3 to 15 nm (section 3.3). Using a leastsquares fitting method, we were able to determine for a given δ the unique searched set of electrohydrodynamic parameters Fo and λo that reproduces the data at sufficiently large ionic strength levels (typically above 30 mM). The reliability of the fitting procedure is addressed in the Supporting Information, where a typical dependence of the theoretical mobility on Fo and λo is given for the sake of example. Illustrative merging between experimental data and numerical theory is given in Figure 5A for untreated red blood cells. For the sake of comparison, the mobility values predicted from the flat-plate Ohshima expression12 are also depicted. It is found that the latter are in excellent agreement with those obtained from rigorous numerical theory over a large range of ionic strengths and that significant discrepancies start to appear for electrolyte concentrations lower than ca. 10 mM. Above this concentration, mobilities are essentially independent of δ as demonstrated in Figure 5A. In this concentration range, the excellent agreement between numerical theory and flat-plate theory by Ohshima denotes that double layer relaxation by the applied electric field is insignificant. This is so because (i) the local electrostatic potentials within the soft surface layer are all small and (ii) the inequality a . δ goes along with maintaining a nearly spherical electric double layer around blood cells. Only for ionic strengths lower than 10 mM does the double layer relaxation starts to play a significant role, with, as a result, an increased inaccuracy of Ohshima’s expression if compared to the exact numerical prediction. In doing so, mobility becomes dependent on the soft surface layer thickness (Figure 5A) because increasing the latter comes to increase the local electrostatic potentials and increase therewith double layer relaxation, thus giving rise to a reduction in the magnitude of the mobility. Also, this decrease in mobility upon increasing δ is concomitantly governed by an DOI: 10.1021/la900087c

10879

Article

Hyono et al.

Figure 5. (A) Experimental electrophoretic mobility (in dimensionless form) of native red blood cells (points) and theoretical analysis as a function of ionic strength using the analytical expression by Ohshima for homogeneous soft particles12 (indicated) and using the numerical scheme developed by Duval et al.24 for diffuse soft bioparticles in the limit R f 0. The predictions are reported for three soft surface layer thickness δ = 5, 10, and 15 nm of the peripheral glycoprotein layer (indicated). (B) Quantitative interpretation of the electrokinetics of native red blood cells with the numerical scheme developed by Duval et al.24 for diffuse (swollen) soft bioparticles in the limits (a) R f 0 and (b) R = 1 nm, (c) 2 nm, (d) 3 nm, and (e) 4 nm (δ = 5 nm, Fo = -50 mM, 1/λo = 1.17 nm). See the text for further details.

increase in the drag exerted by the polyelectrolyte layer on the electroosmotic flow. The reader is referred to ref 24 for further information on the limits of Ohshima’s model and on its comparison with exact numerical prediction. On the basis of Figure 5A, the electrophoretic mobility is essentially independent of the thickness δ of the glycoprotein soft surface layer within the range of electrolyte concentrations adopted for the experiments (5-160 mM). This feature was systematically verified not only for untreated cells (Figure 5A) but also for cells exposed to papain and neuraminidase at different concentration levels (Figure 6). For ionic strengths lower than 30 mM, the numerical electrokinetic theory or equivalent analytical Ohshima predictions, both carried out for homogeneous interphases (R/δ f 0), do not satisfactorily merge with experimental data. At such ionic strengths, the approximation of the homogeneous interphase is very poor because the ongoing osmotic swelling process24 for the glycoprotein layer is inevitably accompanied by a diffuse or gradual decay of the segment density distribution from the core surface membrane to the outer electrolyte. This is confirmed in Figure 5B where electrokinetic data collected for electrolyte concentrations below 30 mM may be reconstructed upon an appropriate increase in the decay length R for the segment density distribution. As extensively discussed elsewhere,24 the decrease (in magnitude) in mobility for a given c¥ upon increase of the interphasic diffuseness (Figure 5B) is explained by an increase;at the outer edge of the soft surface layer (Figure 2) where electroosmotic flow is the strongest;of the friction exerted by the polymer segments on electroosmotic flow. Also, as mentioned by Levine in his pioneering analysis of the electrokinetics of red blood cells,11 “expansion of the polypeptide chains can be assumed at low ionic strength [...] with a tighter structure at physiological ionic strengths than under low ionic strengths conditions”. Our results confirm this expectation and are further in line with experimental observations of swelling of RBCs under osmotic stress conditions.58 On a qualitative level, the conclusion that the decay length R increases with decreasing ionic strength remains when adopting a different mathematical form for the segment density distribution as

(58) Greenwalt, J.; Rugg, N.; Dumaswala, U. J. Transfusion 1997, 37, 269.

10880 DOI: 10.1021/la900087c

compared to that given in eq 1. Providing this distribution verifies the necessary condition given by eq 2. The analysis discussed above for the case of untreated blood cells (Figure 5) has been systematically performed for RBCs treated with 0.1, 0.5, and 5 units of papain (Figure 6A-C) and 0.05 and 0.1 units of neuraminidase (Figure 6D,E). As for native RBCs, experimental data at ionic strengths larger than ∼30-40 mM are consistent with the homogeneous representation of the soft interphase whereas swelling and accompanied interphasic diffuseness come into play for lower ionic strength levels in solution. In Figure 7, relevant electrohydrodynamic parameters Fo and λo obtained from electrokinetic analysis are collected as a function of the concentration and nature of enzymes. For untreated cells, we find that Fo ≈ -50 mM and 1/λo ≈ 1.2 nm. As argued by Levine et al.,11 the mass M of sialic acid per cell is 17.7 ( 1.5 fg/ cell, the surface area Scell of a cell is 145-163 μm2, and the thickness δ of the surface layer is 3-15 nm (section 3.3). On the basis of this, the expected charge density, estimated according to Fo = MF/(ScellMwδ) with Mw = 309.3 g mol-1, which is the molecular weight of sialic acid, is in the range of -28 to -80 mM, which is in agreement with our findings. The hydrodynamic softness parameter λo is further related to the number density of polymer segments no (introduced in eq 1) and the polymer segment radius ao according to λo = (6noπao/η)1/2, with η being the dynamic viscosity of the medium.12,24 Considering the values of no and ao provided in ref 11, it is determined that λo-1 lies in the range of ∼0.6 to ∼1.9 nm, which is again in agreement with the outcome of our electrokinetic study for untreated cells. From Figure 7A, we observe that the typical flow penetration length 1/λo within the glycoprotein layer decreases upon increasing the enzyme concentration before leveling off at ∼0.05 units of neuraminidase and ∼2 units of papain. The extent of this decrease is significantly different for cells exposed to papain and neuraminidase. For the former, 1/λo decreases from 1.2 to 0.74 nm whereas for the latter it decreases from 1.2 to 0.91 nm. The charge density Fo carried by the glycoprotein layer continuously decreases (in magnitude) with neuraminidase concentration whereas this decrease in Fo is strongly attenuated for papain at a concentration of ca. 1 to 2 Langmuir 2009, 25(18), 10873–10885

Hyono et al.

Article

Figure 6. Experimental (points) and theoretical (plain line: analytical expression by Ohshima; dashed lines: predictions based on theory for diffuse soft bioparticles24) electrophoretic mobilities for red blood cells treated with 0.1, 0.5, and 5 units of papain (panels A-C, respectively) and with 0.05 and 0.1 units of neuraminidase (panels D and E, respectively). The meaning of the curves marked (a), (b), (c), and (d) is the same as that specified for Figure 5B. Electrohydrodynamic parameters Fo and 1/λo obtained by least mean squares analysis of the experimental data are reported in Figure 7.

Figure 7. Dependence of (A) the characteristic flow penetration length 1/λo within the soft glycoprotein layer and (B) charge density Fo on papain and neuraminidase concentrations. The dashed lines are only guides for the eye. Points at 0.01 enzyme unit refer to untreated cells. See the text for further details.

units (Figure 7B). In doing so, the decrease in Fo over the examined papain and neuraminidase concentration ranges represents 40 and 70%, respectively, of the value obtained for untreated cells. Langmuir 2009, 25(18), 10873–10885

4.4. AFM Analysis of the Nanomechanical Properties of Papain-Treated Red Blood Cells. Following the electrokinetic analysis described in the preceding section, we performed an AFM investigation of the nanomechanical properties of papainDOI: 10.1021/la900087c

10881

Article

Hyono et al.

modulus, however, appear and are connected to the different tip geometries used. The ∼35 and ∼40% decreases in E upon increase in papain concentration in the range of 0-5 units are the nanomechanical signatures of the damage done to the blood cell interphase when exposed to papain. These modifications of cell and soft surface layer generate strong variations (increase) in the dissipative viscoelastic energy with increasing enzyme concentration. An explanation for this fact is not easily given because the viscoelasticity of soft polymeric and biological material is intrinsically related to their molecular rearrangement (relaxation) after the retraction of the AFM tip. Enzyme concentration-dependent viscoelastic properties as depicted in Figure 8C surely demonstrate major structural modifications of cell surfaces driven by the enzymes. It is possible that the associated cell interphasic heterogeneities lead to stronger dissipation as a result of significant internal reorganization of chains upon decreasing the load force in the retraction part of the indentation profiles.

5. Discussion

Figure 8. Nanomechanical parameters derived from AFM nanoindentation analysis carried out on red blood cells treated with 0 (native cells), 0.5, and 5 units of papain . (A) Young’s modulus, (B) red blood cell spring constant, and (C) viscoelastic energy.

treated cells. In view of the significant reduction of 1/λo with increasing enzyme concentration, these cells are likely to undergo the most remarkable surface status change (if compared to neuraminidase-treated RBCs). Results are depicted in Figure 8A-C where the dependence of Young’s modulus E, the cell spring constant kcell, and the viscoelastic energy Aviscoeiastic on papain concentration is indicated. Red blood cells are very soft biocolloids (from a mechanical point of view) because 0.5 μm indentation within the cell may be reached for load forces as low as 1 nN. (See illustrative raw indentation data in Supporting Information.) For the sake of comparison, characteristic nanomechanical indentations for Shewanella putrefaciens bacterial cells generally do not exceed 0.3 μm for load forces as large as 10 nN.29,59 As a consequence, the magnitudes of the blood cell Young’s modulus and spring constant are low with values of about 250 Pa and 7 mN m-1, respectively, as compared to 0.2 MPa and 35 mN m-1 for Shewanella putrefaciens at pH 4.59 We verified that the raw force curves and resulting cell spring constant, kcell, obtained in this study are in quantitative agreement with previous measurements60 on native RBCs under conditions similar to those adopted here. Some discrepancies in the Young’s (59) Gaboriaud, F.; Dague, E.; Bailet, S.; Jorand, F.; Duval, J. F. L.; Thomas, F. Colloids Surf., B 2006, 52, 108. (60) Bremmell, K. E.; Evans, A.; Prestidge, C. A. Colloids Surf., B 2006, 50, 43.

10882 DOI: 10.1021/la900087c

5.1. Action of Papain and Neuraminidase on Red Blood Cell Interphasic Properties. The electrokinetic results depicted in Figure 7A reveal that hydrodynamic softness is seriously reduced for cells treated with papain whereas charge density within the peripheral glycoprotein layer is mainly affected for cells exposed to neuraminidase (Figure 7B). Papain is a proteolytic enzyme able to break peptide bonds.61 The significant reduction of 1/λo upon increasing papain concentration reflects a change in the status of the peripheral protein layer. A reduction of 1/λo corresponds to a decrease in the number density no of polymer segments and/or a decrease in the polymer segment radius ao. The break in peptide bonds leads to the removal of proteins from the cell surface, in line with a reduction of the degree of flow penetration within the glycoprotein layer. The removal of proteins is likely accompanied by a decrease in charge density Fo because the action of papain targets charged and uncharged parts of protein segments. The decrease in both the Young’s modulus E and the cell spring constant kcell with enzyme concentration (Figure 8) reflects a loss of rigidity for the cell peripheral region that includes the permeable glycoprotein layer, in line with the removal of proteins from the surface of red blood cells. Such protein removal necessarily impacts the integrity of cell surface membranes, which then offers resistance to compression by the AFM tip that is significantly reduced as compared to that when the soft surface layer is intact. Neuraminidase is a glycosyl hydrolase enzyme able to cleave terminal sialic acid residues from substrates such as glycoprotein.62 The reduction of Fo (in magnitude) upon increasing neuraminidase concentration agrees with this mode of action. It has been reported that cell exposure to sufficiently large neuraminidase concentration could result in a 90-95% reduction of the electrophoretic mobility of untreated cells.42 Because the electrophoretic mobility of a red blood cell particle necessarily goes along with the details of electroosmotic flow around and within the cell particle,13 the variation in charge density for neuraminidase-treated cells is necessarily accompanied by a reduction in magnitude of the hydrodynamic flow field and therewith of 1/λo. It is noteworthy that the electrokinetic patterns depicted in Figure 6C,E, which pertain to 5 units of papain and 0.1 unit of neuraminidase-treated cells, respectively, point out a very distinct (61) Komatsu, K.; Tsukuda, K.; Hosoya, J.; Satoh, S. Exp. Neurol. 1986, 93, 642. (62) Henrissat, B.; Davies, G. Structure 1995, 3, 853.

Langmuir 2009, 25(18), 10873–10885

Hyono et al.

Article

Figure 9. (A) Typical dependence of the overall Gibbs energy of interaction for native red blood cells on separation distance H + 2δ and soft surface layer thickness δ (indicated). The parameters used for the computation were derived from electrokinetic analysis (electrostatic contribution, eq 8), and those pertaining to the attractive van der Waals component (eq 9) are detailed in the text (section 3.2). The coordinates (h*, V*) of the secondary minimum are explicitly indicated. (B) Details of the electrostatic, van der Waals, and steric contributions to the overall interaction energy for native red blood cells taking δ = 4 nm.

response in terms of the extent of swelling or equivalently interphasic diffuseness. Indeed, upon decreasing ionic strength, the increase in R is more pronounced for cells exposed to papain than for those treated with neuraminidase. Besides, swelling sets in at larger ionic strengths for papain-treated cells than for neuraminidase-treated cells. Recalling that osmotic swelling processes for isolated colloids are inherently governed by an interfacial pressure drop that depends on the charge density squared, Fo2 (see the term β2 involved in eq 8), the swelling features reflected in Figure 6C,E are consistent with the strongest reduction (in magnitude) of Fo as determined for neuraminidase-treated cells. According to the results displayed in Figure 4B, cell agglutination mediated by the IgG anti-D antibody is more important for papain-treated cells than for neuraminidase-treated cells. In our previous study,63,64 similar conclusions were obtained in the case of IgG-mediated agglutination of papain and neuraminidasetreated type O red blood cells, though the agglutination and electrokinetics of cells were measured less systematically over a restricted range of enzyme unit concentrations. In addition, the analysis of electrokinetic data using the approximate mobility expression by Ohshima did not allow for a quantitative interpretation of deviations at low ionic strengths. Still, despite these shortcomings, the fact that type O and type A human blood cell agglutination is impacted in a similar way by papain and neuraminidase enzyme treatment reveals the nonspecific nature of the mechanisms underlying the effect of papain and neuraminidase on cell interphasic properties. By confronting hemagglutination data with quantitative electrohydrodynamic information obtained for native and enzymetreated cells, it appears that (i) papain promotes agglutination primarily because it leads to a deterioration of the soft surface layer with some reduction in thickness and (ii) neuraminidase primarily increases agglutination via a decrease in Fo. Also, it suggests that a decrease in soft surface layer thickness δ at constant charge likely leads to an important reduction in (63) Hyono, A.; Mazda, T.; Okazaki, H.; Tadokoro, K.; Ohshima, H. Vox Sang. 2008, 95, 131. (64) Hyono, A.; Mazda, T.; Duval, J. F. L.; Gaboriaud, F.; Okazaki, H.; Kenji, K.; Ohshima, H. Vox Sang. 2008, 95, 180.

Langmuir 2009, 25(18), 10873–10885

interparticular steric repulsion energy, which promotes the approach of neighboring cells at separation distances compatible with the binding of cell antigen D by IgG. In the light of the above results, such a scenario is more favorable than a decrease in repulsive electrostatic interaction energy linked to that of Fo. These features are quantitatively confirmed in the next section following the theoretical formalism presented in section 3.2. 5.2. Quantitative Analysis of Red Blood Cell Stability against IgG-Mediated Agglutination. Given the charge densities evaluated by electrokinetic analysis for native and enzymetreated red blood cells (Figure 7B), the overall Gibbs energy of interaction V between cells may be computed under physiological conditions (160 mM) where R f 0 by means of eqs 3, 8, and 9. A typical illustration of the dependence of such interaction Gibbs energy with separation distance H + 2δ (Figure 3) and soft surface layer thickness δ is provided in Figure 9A for untreated blood cells (Fo = -50 mM). Results are shown for δ values in agreement with soft surface layer thickness as reported in the literature (section 3.3). The most remarkable feature is the presence of a secondary minimum V* of hundreds of kBT located at a separation distance denoted as h*. The presence of this secondary minimum results from the respective magnitudes of the electrostatic, van der Waals, and steric components, as illustrated in Figure 9B. The associated repulsive barrier easily ensures the stability of native blood cells against aggregation. Upon a decrease in the soft surface layer thickness δ, h* is reduced; accordingly, the secondary minimum becomes more pronounced. On the basis of the interaction energy profiles as illustrated in Figure 9, the dependence of h* and V* on δ may be explicitly determined. The results for native blood cells and papain- and neuraminidase-treated cells are collected in Figure 10A,B. The quasi-linear dependence of h* on δ (Figure 10A) is in part imposed by the assumption of infinitely steep repulsive steric interaction in the neighborhood of the outer soft surface layer (section 3.2.3). For a given δ, upon decreasing Fo (i.e., when increasing the concentration of enzyme, Figure 7B), h* decreases because cells get closer following a reduction in electrostatic repulsion. For a given Fo, h* decreases with decreasing δ because cells get closer following a reduction in the steric repulsion barrier. DOI: 10.1021/la900087c

10883

Article

Hyono et al.

Figure 10. Dependence of the (A) location h* and (B) associated interaction energy V* for the secondary minimum on enzyme concentration (expressed in units defined in section 2.1) for papain- and neuraminidase-treated red blood cells. The dashed lines in panels A and B reflect the necessary condition for IgG-mediated agglutination to occur (separation distance between red blood cells e14 nm, see text for details) for 0.1 unit of neuraminidase-treated cells and 5 units of papain-treated cells. In panel B, symbols X and Y indicate the possible course of the soft surface layer reduction upon increasing concentration of papain and neuraminidase enzymes, respectively. See the text for further details. The symbol O pertains to untreated red blood cells that do not agglutinate in the presence of IgG. In panels A and B, the curves pertaining to untreated (control) red blood cells and 0.1 unit of papain-treated cells are identical because the corresponding volume charge densities Fo are similar (Figure 7B).

The characteristic separation distance between two antigen D-binding sites of IgG leading to the agglutination of two blood cells is 14 nm (Figure 1), recalling that the D antigen is located on the hydrophobic polypeptide that extends very little above the lipid bilayer.5 We can transpose this feature by considering that the agglutination of cells by IgG may occur when the secondary minimum of magnitude V* is positioned at a cell-cell separation distance h* that verifies h* e 14 nm. This condition is satisfied at δ e 5.6 nm for 0.1 unit of neuraminidase-treated cells or δ e 5.1 nm for 5 units of papain-treated cells as shown in Figure 10A. From these values of δ, we can estimate the range of secondary minima in line with the occurrence of hemagglutination (Figure 10B). The values given above for the surface layer thickness δ in line with a cell-cell separation distance of 14 nm at most should be viewed as averages of the distribution around the cell of the glycoprotein surface layer thickness. Indeed, it is expected that the reduction of the glycoprotein layer thickness upon enzyme action is not uniform around the cell. However, we have no experimental method for measuring this layer thickness distribution around the cell as a function of the concentration of the enzyme to which it was exposed. The treatment of blood cells with the enzymes of interest here may lead to either a decrease in the cell volume charge density Fo and/or a decrease in the thickness of the glycoprotein layer δ. A comparison of the dependence of the IgG antibody dilution factor (Figure 4B) and of the charge density Fo (Figure 7B) on papain concentration indicates that the significant increase in cell agglutination with increasing enzyme concentration cannot be governed by the variation of Fo that remains roughly constant in the range of 0.5-5 unit of papain. This result is entirely consistent with the dependence of the secondary minimum V* on δ obtained for papain-treated cells (Figure 10B). Indeed, for a given surface layer thickness δ compatible with cell agglutination, the magnitude of the minima V* for 0.5 and 5 units of papaintreated cells differs at most by 10% . This is not in line with the 10884 DOI: 10.1021/la900087c

300% increase in the titer in the range of 0.5-5 units of papain enzyme (Figure 4B). Consequently, cell agglutination upon papain treatment is mostly governed by a reduction in the steric layer thickness δ rather than by a decrease in interparticular electrostatic repulsion or, equivalently, a decrease in Fo. In Figure 9B, the cross symbols (X) illustrate a possible course of decrease of δ upon increase of papain concentration. This decrease in δ is made consistent with the agglutination results of Figure 4B. Following similar reasoning for neuraminidase-treated cells, the poor agglutination of these as compared to that observed for papain-treated cells may be viewed as the result of a decrease in Fo with a mild reduction of δ. This mechanism, which is thermodynamically less favorable for promoting agglutination (Figure 10B) than that commented for papain-treated cells, is schematically pictured in Figure 10B where positioning of the symbols (Y) is consistent with the agglutination data of Figure 4B. As a final comment, we stress that all conclusions drawn from the analysis of the stability features of cells against IgG-mediated agglutination (analysis based on flat disk-like geometry assumed for the cells, section 3.2) are the same as those obtained if reasoning with torus geometry that could be more realistic for representing the biconcave disk form of the blood cells. The proof of this is given in Supporting Information, where the counterparts of Figure 10A,B are provided for the torus-geometry-based evaluation of blood cell interaction.

6. Conclusions In this study, we report a quantitative analysis of the electrohydrodynamic properties of blood cells of type A (Rh D positive) either in the native state or subjected to treatment after exposure to papain and neuraminidase enzymes at various concentrations. It is found that the mode of action for these two enzymes is significantly different. Whereas neuraminidase action leads predominantly to a decrease in the density of charges carried by the Langmuir 2009, 25(18), 10873–10885

Hyono et al.

permeable blood cells (removal of sialic acids within the soft peripheral glycoprotein layer), papain induces a significant reduction of the hydrodynamic flow permeation within the cells. This latter feature is explained by a reduction of the glycoprotein soft surface layer thickness following the breaking of peptide chains, in agreement with nanomechanical AFM analysis. An examination of agglutination data for native and enzyme-treated blood cells reveals that cell agglutination is promoted to a larger extent by a decrease in steric layer thickness than by a decrease in charge density. This is supported by the theoretical evaluation of the stability features of native and enzyme-treated cells against IgG-mediated agglutination. The stability diagrams indeed demonstrate that cell agglutination by IgG is better promoted by a reduction of the surface layer thickness at constant charge than by a reduction of the charge at constant surface layer thickness. The former case corresponds more appropriately to the situation encountered for papain-treated cells whereas the latter better describes the situation of neuraminidase-treated cells,

Langmuir 2009, 25(18), 10873–10885

Article

as judged from the variation of the electrokinetic parameters with enzyme concentration. Acknowledgment. J.F.L.D. thanks the Conseil Regional de Lorraine for the financial support that partially funded the 6 month stay (March-August 2008) of A.H. at the Laboratoire Environnement et Mineralurgie (LEM, UMR CNRS-INPL 7569), Vandoeuvre-les-Nancy, France. Supporting Information Available: Detailed steps for the analysis of the nanomechanical indentation force curves by atomic force microscopy and additional information on the evaluation of the Gibbs interaction energy on the basis of torus geometry for red blood cells. Also, the reliability of the quantitative interpretation of the electrokinetic data is addressed. This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la900087c

10885