Hierarchical Networks of Casein Proteins: An Elasticity Study Based

Langmuir 2004, 20, 5079-5090

5079

Hierarchical Networks of Casein Proteins: An Elasticity Study Based on Atomic Force Microscopy V. I. Uricanu,* M. H. G. Duits, and J. Mellema Physics of Complex Fluids Group, University of Twente, Faculty of Science and Technology, associated with the J.M. Burgerscentrum for Fluid Mechanics, and Institute of Mechanics, Processes and ControlsTwente (IMPACT), Postbus 217, 7500 AE Enschede, Netherlands Received December 16, 2003. In Final Form: February 13, 2004 2D- and 3D-atomic force microscopy (AFM) experiments were performed on single casein micelles (CM) in native state, submerged in liquid, using a home-built AFM instrument. The micelles were immobilized via carbodiimide chemistry to a self-assembled monolayer supported on gold-coated slides. Off-line data analysis allowed the extraction of both surface topography and elastic properties. Relative Young moduli (E*) were derived from force-vs-indentation curves, using the Hertz theory. The obtained E* values were found to increase with CM diameter, following a straight line dependence. The data showed that temperature, via its influence on both the protein-protein interactions and the composition of the micelle, has a clear effect on the mechanical properties of the CMs: higher temperatures and lower serum casein concentrations result in stiffer micelles. For pH e5.6, effecting calcium phosphate release from the micelles by decreasing the pH does not have a large effect on CM stiffness. On decrease of the pH below 5.0, particulate gels and multilayers were obtained. Their measured elasticity (expressed by an equivalent G′AFM) agrees remarkably well with the storage moduli as measured with a conventional rheometer. Compared to single micelles, gels from nonheated CM suspensions are about 3 orders of magnitude softer. The “softness” of these gels (measured under compression or shear) therefore must come from the microscopic and/or mesoscopic links rather than the micelles themselves.

1. Introduction In cases where rheology is used as a probe of the sample microstructure, the implicit assumption is that the material is homogeneous between the macroscopic scale and the length scale of interest. Many systems, however, show levels of organization at intermediate length scales. For these cases, it is desirable to be able to do microrheology on specific, i.e., selected, microscopic features. Atomic force microscopy (AFM) proves to be a useful tool in accessing the structure-property relationships at different length and force scales. We have made use of this AFM versatility to study the structure and elastic properties of units as are present in acidified milk gels. These are particulate gels, built up from casein micelles (CM). In turn, each CM is an association colloid, consisting of a large number (∼104) of casein macro-molecular chains. One of the unresolved issues in the dairy industry is where the elasticity of acidified milk gels originates from. Both the micelles themselves, and the intermicellar links can contribute to the overall elastic properties of the gel, with each of the contributions depending on chemical composition and thermal/chemical treatment. The main focus of the present paper is the elastic modulus of individual casein micelles. Let us first consider some molecular properties of casein micelles. Depending on the animal source, the caseins’ molecular structure (and, with it, the micellar composition and/or size) can be different.1 Still, a similar mixture of hydrophobic and electrostatic forces holds the proteins together and mediates their mutual interactions as well as protein association with inorganic ions. Besides proteins, also calcium phosphate is found in the micelles as a combination of micellar calcium (bound directly to the * To whom correspondence should be addressed. E-mail: [email protected]. (1) Ginger, M. G.; Grigor, M. R. Comp. Biochem. Physiol. 1999, B 124, 133.

caseins via their phosphorylated residues) and colloidal calcium (associated with the micelle but not bound directly to the casein polypeptide).2,3 Small amounts of magnesium, sodium, potassium, and citrate are also incorporated in the supramolecular micellar construction. The principal components of a micelle are obviously the casein molecules. A typical bovine micelle contains four distinct types of caseins: R S1-, R S2-, β-, and κ-caseins. The last one is located mainly in the outer hairy layer of the more or less spherical protein arrangement and has a major contribution for micellar stability. In the micelle interior, the proteins are not packed in dense regions and the micelles hold also relatively high water amounts (around 2 mL per gram protein). In native conditions, the micelles are in equilibrium with soluble casein molecules and dissolved salts in the serum. Also the fact that there is a micellar size distribution should be appreciated, and the possible dependence of micellar properties on size considered. For bovine casein micelles, Dalgleish et al.4 made a detailed analysis of the composition versus micelle diameter. These authors reconfirmed that smaller micelles have a higher proportion of κ-caseins. Meanwhile, the only correlations found between the CM size and the content of the constitutive polypeptides were that (i) the inner micellar structure contains an invariable percentage of R-caseins (≈ 47%) and (ii) the β-casein concentration increases linearly with CM’s radius. Similar results were obtained by Pierre and co-workers5 on goat milk casein micelles. These last authors advanced the idea that: “Micelles of different sizes might be different steps in building of the casein micelle structure. Large micelles, having the highest polymeri(2) Farrell, H. M. J.; Thomson, M. P. In Calcium binding proteins; Thomson, M. P., Ed.; CRC Press: Boca Raton, FL, 1988; pp 117-137. (3) Holt, C. Adv. Protein Chem. 1992, 43, 63. (4) Dalgleish, D. G.; Horne, D. S.; Law, A. J. R. Biochim. Biophys. Acta 1989, 991 (3), 383. (5) Pierre, A.; Michel, F.; Zahoute, L. Int. Diary J. 1999, 9, 179.

10.1021/la0363736 CCC: $27.50 © 2004 American Chemical Society Published on Web 05/13/2004

5080

Langmuir, Vol. 20, No. 12, 2004

zation state, would correspond to the most complex micelle structure while smaller particles could be micelles at intermediary stages of building”. To effect aggregation between the micelles, different methods can be used: adding a nonsolvent, or an enzyme that cleaves κ-casein chains,6,7 or lowering the pH. The latter method is used in the present paper and will therefore be elaborated below. On decrease of the pH (from physiological values of 6.7-6.9 toward 5.2-4.6), the net negative charge of the individual casein molecules and micelles is diminished. As the zeta potential approaches zero, the steric stabilization layer of the CMs collapses and the system begins to aggregate. Gelation in CM dispersions appears as result of van der Waals forces dominance over the steric interactions. Lowering the pH also solubilizes the mineral calcium phosphate in the micelle and, depending on the temperature, changes the protein composition of the casein micelles, via release of β- and κ-casein into the serum phase of the milk.8-13 A more detailed description of the whole gelation process can be found in ref 14. Because the micelles seem to maintain their global organization (even when the environmental conditions are altered by pH reduction11,12), the CMs were considered to behave in the gelation process as sticky spheres.15,16 The model developed (i.e., the “adhesive sphere” model) describes properties of CM dispersions up to the onset of flocculation, based on the concept of a sensitive steric stabilization layer surrounding a protein-rich core. Destabilization during renneting or alteration of the solvent quality reduces the ability of this polymeric brush to oppose the attraction operating between the casein spheres. After the gel point is reached (i.e., the CMs have assembled into a network which immobilizes the aqueous phase), the casein gel properties can still evolve over time, in relation to changes in the internal micellar integrity. To describe this, Horne17,18 adopted the dual-binding model. In this model, the building of the micellar assembly takes place via two distinct types of bonds: (a) cross-links between the hydrophobic regions of the caseins and (b) cross-links between protein phosphoseryl clusters and positively charged calcium phosphate microcrystals or nanoclusters. When the pH is lowered, in the micellar interior, the (b)-type bonds are broken due to dissolution of the calcium phosphate nanoclusters. If the temperature is lowered, increased solubilization of individual caseins, particularly of β- and κ-casein, weakens the hydrophobic interactions. Both pH and temperature treatments can be applied during milk processing. Hence, the CM and gel properties may vary according with the sequence and rate of the pH- and/or T-modifications.14,19 On the basis of such insights at the molecular level, an appreciable number of rheological studies have been (6) Walstra, P.; Jenness, R. Diary Chemistry and Physics; John Wiley and Sons: New York, 1984; Chapter 13. (7) de Kruif, C. G. Int. Diary J. 1999, 9, 183. (8) Walstra, P. In Cheese: Chemistry, Physics and Microbiology, 2nd ed.; Fox, P. F., Ed.; Chapmann & Hall: New York, 1993; pp 141-191. (9) Roefs, S. P. F. M.; Walstra, P.; Dagleish, D. G.; Horne, D. S. Neth. Milk Dairy J. 1985, 39, 119. (10) Roefs, S. P. F. M. Ph.D. Thesis, Wageningen University, The Netherlands, 1986: (a) pp 41-42, (b) pp 31, 36, 55-56, (c) p 38. (11) Dalgleish, D. G.; Law, A. J. R. J. Dairy Res. 1988, 55, 529. (12) Dalgleish, D. G.; Law, A. J. R. J. Dairy Res. 1989, 56, 727. (13) Lucey, J. A.; Singh, H. Food Res. Int. 1998, 30 (7), 529. (14) Vasbinder, A. J.; Rollema, H. S.; Bot, A.; de Kruif, C. G. J. Dairy Sci. 2003, 86, 1556. (15) de Kruif, C. G.; Jeurnink, T. J. M.; Zoon, P. Neth. Milk Diary J. 1992, 46, 123. (16) de Kruif, C. G. J. Colloid Interface Sci. 1997, 185, 19. (17) Horne, D. S. Colloids Surf., A 2003, 213, 255. (18) Lucey, J. A.; Johnson, M. E.; Horne, D. S. J. Diary Sci. 2003, 86, 2725.

Uricanu et al.

performed at the macroscopic level. The viscoelastic behavior of the milk protein gels has been studied mostly for “model milk” in the sense that certain components were removed, or deliberately added in excess amounts, allowing elimination, respective, studying of their influence. The rheological properties of acidified milk gels were determined using dynamic rheology at relatively low frequencies (0.001-10 Hz).10,20 For nonheated skim milk gelled at or close to room temperature (using the slow acidulant: glucono-δ-lactone (GDL)), rather small storage moduli values (G′ below 10-20 Pa) were obtained.20-22 In some of these experiments, the milk samples contained not only CM but also whey proteins: R-lactalbumin and β-lactoglobulin. The native whey proteins actually “act as an inert filler in acid milk gels”20 and mixing the CM with larger (than physiological) amounts of whey proteins leads to a decrease in G′. Acidification at low temperatures, followed by rheological measurements over an extended temperature range (between 2 and 25 °C) was performed by Roefs.10 The effects of mild heating (up to 30 or 40 °C) during acidification were also investigated.10,13 As the gelation temperature rises, the gels become more firm. An even stronger effect on the rheological properties is observed if the samples are heat-treated above 70-80 °C, prior to acidification. In such cases, additional interactions between the CM and the whey proteins22 increase the G′ up to values close to (but still below) 1 kPa.13,20 For very high temperatures (in the range 100-140 °C), Schreiber made some interesting findings.24 This author studied the strength of “whey protein free” (i.e., made only from casein micelles) gels by renneting CM dispersions after heat treatment. As the heat load applied increased, the equilibrium between calcium diffusible and colloidal phases is shifted toward the latter. The heat-induced changes in the calcium distribution resulted in a decreased firmness of the gels. Working with highly concentrated CM dispersions (1825% protein) at a pH much above the gelation point, Zhong25 measured the storage and loss moduli during cooling (from 80 °C down to 5 °C) at different rates. All the samples showed a significant sensitivity on the cooling rate and had high G′ values (around 5 kPa for the 25% CM dispersion). Even higher G′ values (up to 40 kPa) were detected (applying the same treatment) on a commercial processed cheese. Recently, Smyth and co-workers26 used a complementary technique (i.e., high-frequency ultrasonic measurements) for analysis of the shear and dilatational viscoelasticity of casein gels in the frequency range 5-25 MHz. The magnitudes of both shear moduli and the storage (G′ ) 30 kPa, at 7 MHz) and the loss moduli (G′′ ) 100 kPa, at 7 MHz) of the gel are several orders higher than those determined by dynamic rheology at low frequency (0.1 Hz), indicating different contributions of the gel network to the changes in the viscoelastic parameters at low and high frequencies. The dilatational storage moduli (19) de Kruif, C. G.; Holt, C. Advanced Dairy Chemistry, 3rd ed.; Fox, P. F., McSweeney, P. L. H., Eds.; Kluwer Academic: New York, 2003; Vol. 1, Chapter 5. (20) Lucey, J. A.; Munro, P. A.; Singh, H. Int. Dairy J. 1999, 9, 275. (21) Horne, D. S. Int. Dairy J. 1999, 9, 261. (22) Vasbinder, A. J.; Alting, A. C.; Visschers, R. W.; de Kruif, C. G. Int. Dairy J. 2003, 13, 29. (23) Vasbinder, A. J.; Alting, A. C.; de Kruif, C. G. Colloids Surf., B 2003, 31, 115. (24) Schreiber, R. Int. Diary J. 2001, 11, 553. (25) Zhong, Q. Proceedings of RheoFuture, the International Forum for Material Characterization, Karlsruhe, 2002. (26) Smyth, C.; Kudryashov, E. D.; Buckin, V. Colloids Surf., A 2001, 183-185, 517.

Networks of Casein Proteins

were found to be of the same order of magnitude as the shear moduli of the gel. Despite all the work done on the rheological behavior of casein particle gels, there are no reported data on the elastic properties of individual casein micelles. The small size of these supramolecular arrangements limits the range of experimental techniques that can be used. We chose to address the issue, taking advantage of the atomic force microscopy capabilities: high spatial and force resolution. We have performed contact mode experiments to localize and characterize (i.e., classify) casein micelles and their assemblies. Subsequently, force-distance experiments on single micelles were done to obtain elastic properties. Studying biological specimens with AFM requires several conditions which must be satisfied during sample preparation/attachment to a surface. Extreme care should be taken in using solutions of highest possible purity, since contamination may affect both imaging and force measurements. In addition, special attention should be paid to strongly attaching the molecules to an immobile substrate. Good attachment is a must in this technique since the scanning probe may shift a weakly bound sample around, thus precluding stable imaging or force registration. While the attachment should be firm enough to avoid undesired motions, it should still allow a certain freedom. For “single molecules force spectroscopy” (like in studies of receptor/ligand interactions),27 the molecules (while having one end immobilized by the attachment) must be still able to change their conformation during biological activity or interaction with the other molecular partner. A too strong attachment along considerable portions of the molecular surface may lead to undesirable denaturing effects and should be avoided. The most used alternative for protein binding makes use of self-assembled monolayers (SAMs) and covalent coupling (between the free end of the SAMs molecules and the protein) using a linker.28,29 Our approach was to use carbodiimide chemistry (see section 2) for CM binding to an initially carboxylterminated SAM. Further, the conditions for stable CM attachment were determined as a function of pH for two temperatures: 4 and 20 °C. In our elasticity measurements, compressions were done at 20 or 23 °C onto/into single micelles or aggregates (i.e., gels or multilayers). Hertz theory30 was used to obtain (apparent) relative Young moduli (E*) from the forceindentation curves. The obtained results will be discussed in correlation with the existing data on CM gels and cheese. 2. Experimental Section Preparation of Milk Protein Samples. Whey protein free reconstituted skim milk (WPF) was prepared by dissolving 8.8 g of WPF powder (prepared by ultra- and microfiltration, NIZO Food Research, Ede, The Netherlands) in 91.2 g of distilled water. After dissolution at room temperature, this CM-containing solution was stirred at 45 °C for 1 h. After the solution was cooled to 20 °C, 0.02% sodium azide was added (to prevent bacterial growth) and the WPF suspension was kept overnight at 4 °C before use. For experiments at 20-23 °C, the suspension was re-equilibrated at these temperatures for 2 h, prior to acidification. Its initial pH was 6.7. The samples were acidified by addition of (27) Zlatanova, J.; Lindsay, S. M.; Leuba, S. H. Prog. Biophys. Mol. Biol. 2000, 74, 37. (28) Lahiri, J.; Isaacs, L.; Tien, J.; Whitesides, G. M. Anal. Chem. 1999, 71 (4), 777. (29) Ostuni, E.; Yan, L.; Whitesides, G. M. Colloids Surf., B 1999, 15, 3. (30) For a didactic presentation of the Hertz mechanical contact theory, the reader can consult: (a) Johnson, K. L. Contact Mechanics; Cambridge University Press: Cambridge, London, 1985. (b) Wang, Y. C.; Lakes, R. Int. J. Solids Struct. 2002, 39, 4825.

Langmuir, Vol. 20, No. 12, 2004 5081 glucono-δ-lactone (GDL, Sigma) and incubated for 24 h at 4 °C (for cold-WPF) or 20-23 °C (for WPF). The next step was to perform the micelles attachment onto a functionalized substrate. Functionalized Substrate. Glass slides were gold sputtered (i.e., conditions chosen as to give a surface roughness below 10 nm). Next, the gold-covered slides were immersed in an ethanolic (2 mM) solution of 11-mercapto-1-dodecanoic acid (HS-[CH2]11COOH, Aldrich). The alkanethiol is deposited onto the gold (via chemisorption) and forms self-assembled monolayers (SAMs). After 12 h, the SAM-supporting substrates were extensively rinsed with ethanol and distilled water and immersed for 15 min in a freshly prepared aqueous mixture of NHS (0.1 M) and EDC (0.4 M). Protein attachment was performed via carbodiimide chemistry. N-Hydroxysuccinimide (NHS) and N-ethyl-N′-(dimethylaminopropyl)carbodiimide (EDC) react with the carboxylic acid groups on the SAM surface to form an active ester; subsequently, this active ester can react with alkylamines of the caseins located in the superficial layer of the micelles. The latter reaction was allowed to progress for 1 h at the conditioning temperature (i.e., 4 °C for cold-WPF and 20 °C for WPF). To preserve as much as possible the native state of the micelles during the AFM experiments, the samples were further washed carefully with serum. In this way, only the attached micelles remain on the substrate and the proteins are kept continuously in the hydrated state. The serum was prepared by complete CM removal (ultracentrifugation) from the (respective) acidified WPF suspensions. The whole procedure adopted ensures that, for every single sample, the CMs are kept at constant pH during sample preparation and AFM experiments. Before the AFM imaging/ compression measurements, the CM-coated substrates were allowed to equilibrate under serum at controlled constant temperature (Tmeasurements ) 20 or 23 °C). At fixed time intervals (1 or 5 h), investigations were done at the same temperature. Atomic Force Microscopy. A home-built instrument, based on the stand-alone principle,31 was used to perform the atomic force microscopy studies. The instrument is designed to keep the sample immobile and to allow the xyz piezoscanner to move the cantilever. For experiments on single micelles, the same cantilever (silicon nitride, Park Scientific Instruments), with a nominal spring constant 0.01 N m-1, was used in all measurements. All samples were supported on the functionalized substrate (which also served as the bottom of our custom-made liquid cells) and studied under serum. To observe the topography (height profile) of the casein micelles, the microscope was operated in the constant force imaging mode at a setpoint control force of less than 500 pN. The interactions (between the AFM tip and the CMs) as well as the deformations of the micelles were measured using the force-distance mode. In these measurements, the cantilever was ramped up and down at a frequency of 1 Hz. Prior to, and after each experiment on the milk proteins, the deflection sensitivity of the cantilever was determined (checked) by recording a force-distance curve at a different lateral position on the substrate, where the glass-gold substrate was not covered by soft (protein) material. In all cases, in the noncontact region, no interaction was measured between the silicone nitride tip and the substrate under serum. These experiments proved that no protein contamination had affected the tip; hence, the local geometry while indenting into the micelles is given by the shape of the tip. When the soft samples are measured, the local elastic behavior can be extracted from a selected part of the contact line (as explained further on in section 3). The data to be reported are representative results for the casein micelles and gels studied, based on a large number of recordings for each of the samples, covering different lateral positions on the same sample but also repetitions of scans “on the same spot”.

3. Results and Discussion (a) Conditions for Reliable AFM Experiments. Atomic force microscopy is a powerful technique in performing different types of experiments on, sometimes, (31) van der Werf, K. O.; Putman, C. A. J.; de Grooth, B. G.; Segerink, F. B.; Schipper, E. H.; van Hulst, N. F.; Greve, J. Rev. Sci. Instrum. 1993, 64 (10), 2892.

5082

Langmuir, Vol. 20, No. 12, 2004

Uricanu et al.

Figure 1. AFM images acquired in the soft contact mode (Fcontact ) 500 pN), after protein removal from the functionalized substrate by an initial scanning with higher force: Finitial ) 8 nN on the sample region shown in (b). Both images (a) 5 × 5 µm2 and (b) 2.5 × 2.5 µm2 are recordings of the error signal, and the brighter spots are casein micelles. The gold substrate remains visible (middle of (a)) after protein/SAM shaving. The sample was prepared at 20 °C from a WPF suspension (12 wt % CMs) with pH ) 5.2.

very small objects. The most used application of AFM is in imaging/detection of height differences. For hard samples, the real height can be extracted with good accuracy. As the samples under investigation are softer, the AFM probe compresses the material and only semiquantitative heights can be obtained. If the scanning force is too high, damage can be induced, either in the internal structure (i.e., tip penetration in the sample, with irreversible breakage of material’s integrity) or by cutting the sample-substrate linkages. For our samples, the used force regime was adjusted accordingly. Soft contact (i.e., scanning laterally in constant-force mode), with set-point forces below 500 pN, was found to leave the single CMs or their assemblies unaltered. Successive imaging proved that the protein bonding to the substrate is strong enough and the CM are not removed by the AFM sampling. One can ask: how do we imagine the local contact between the CM and the flat substrate? Is the attachment operative for a “single point” or on a small region? The latter situation is more plausible; the polypeptide chains protruding into the serum phase have much more entropic freedom than the molecules trapped deep inside the CM. The relatively high density of the amino groups along the partially pendant macromolecules together with their flexibility should provide an enhanced probability for covalent binding with the SAM. In this respect, the resultant attachment is supposed to function via multiple casein-SAM linkages. The cooperative effect of such bonding manifests itself in a direct manner when the CMs are subjected to repeated imaging via AFM. Only scanning with forces higher than 7 nN removes the attached micelles. The threshold value (i.e., 7 nN) is in the range of measured forces (by Amro et al.32) inducing the “shaving” of SAMs from gold (via cutting of the Au-S bonds). In our experiments, contact forces higher than 7 nN create a protein-free region on the substrate (Figure 1). The pH range for successful CM’s covalent binding (to the substrate) is limited by the net surface charges. Earlier studies28 showed that: “the pH of the coupling buffer, the concentration of protein, the ionic strength of the solution and the molecular weight of the protein”, all are parameters which influence the attachment efficiency and the final amount of protein linked to a given SAM. Compared to single molecules, the attachment of a casein micelle is much more complex. Local electrostatic charges, the ability for protein orientation/ folding in accessible (to bonding) (32) Amro, N. A.; Xu, S.; Liu, G. Y. Langmuir 2000, 16, 3006.

Figure 2. Error signal AFM images for CM covalently linked to the substrate and scanned at T ) 20 °C. The substrate is visible in all cases. The scan area was the same: 5 × 5 µm2. The coverage density depends on the sample pH (indicated for every case).

conformations and the total residence time of the CM close to the functionalized substrate can result in differences in micellar attachment to the SAM. Our investigations showed that at pH above 5.6, despite long time intervals (.1 h) allowed for the chemical reaction, only few CM are occasionally coupled to the substrate. Also, the size of the bounded micelles at pH >5.6 was sometimes too low to be used in elasticity measurements (see details on the discussion related to the 3D-AFM measurements). At room temperature (20 °C) and pH below 6, electrokinetic measurements done on freely dispersed CMs8 showed that the average zeta potential drops drastically down to almost zero (around pH ) 5.0) and has a subsequent rise as the pH moves to even lower values. The change in pH is accompanied also by calcium and inorganic phosphate dissolution/diffusion into the serum phase (Figure 4). For the WPF dispersion studied by us, the above-mentioned changes are well reflected in the recorded AFM images (Figure 2). From pH ) 5.6 to 5.2, a smaller electrostatic charge is responsible for a net increase in the coverage (bonding) density. The CMs flocculate (gel) at pH ) 5.0. To benefit (for our studies) from this naturally occurring process, we measured the elasticity (via 3D AFM, see section b4 in Results) in three situations: (a) undisturbed whole-gel sample; (b) multilayers (prepared by CM sedimentation after vigorous stirring/dispersion of the gel); (c) single micelles, bound

Networks of Casein Proteins

Langmuir, Vol. 20, No. 12, 2004 5083

Figure 4. The proportion of calcium (Ca) and inorganic phosphate (Pin) inside casein micelles, and the electrokinetic potential (ζ), as a function of pH. Reprinted with permission from ref 8. Copyright 1993, Chapman & Hall. Table 1. Calculated Coverage Density after Protein Bonding on the Substrate-Supported SAM (based on the AFM height images)a coverage density (%) pH 5.6 pH 5.4 pH 5.2 cold-WPF WPF: T ) 20 °C

5 7

28 36

22 70

pH 5.0

pH 4.8

34 29 100 (gel) 100 (gel)

a Note that the 100% covered samples were identified as such without AFM characterization.

Figure 3. Error signal AFM images of milk proteins acidified at T ) 4 °C and imaged at T ) 20 °C. The scan area was 5 × 5 µm2 (in the pH range 5.6-5.0) and 10 × 10 µm2 (for pH ) 4.8).

to the functionalized substrate from a highly diluted (with the respective serum) b sample. Only the single micelle case gives reproducible images while scanning in the soft contact mode.

If the acidification proceeds for 24 h at 4 °C, no macroscopic gelation is observed down to pH ) 4.8. Only for pH ) 4.6, the CMs are arrested in a gel structure. Even if gelation was not obviously observed at the macroscopic level, it seems like there is more or less advanced aggregation for pH ) 5.2 and below (Figure 3). The aggregates’ dimensions grow with pH lowering. The particle-particle interactions inside these aggregates are high enough as to suppress micelle removal from the aggregates during washing with serum. The AFM error signal images in Figures 2 and 3 can give an estimate of the coverage density (see Table 1). Since, for the aggregated systems, there is no one-to-one correspondence between the coverage density and the relative size of the protein structures, the AFM recordings must be analyzed also from a tridimensional perspective (like in Figure 5). Cross-analysis (relative height versus width) data indicate a transition: from single micelles linked to the substrate, to a mixture of single CM and aggregates. Representative examples are given in Figure 6. With pH decrease at 4 °C, the attached aggregates assume different configurations; more three-dimensional-like structures are seen at 4.8 as compared with pH ) 5.2 where part of the aggregates are “surface associations”. (b) 3D AFM Experiments. b1. Typical Force vs Distance Curves. Even if the AFM tip radius at the apex is very small (R ) 50 nm), it approaches the values for the micellar radius (between 50 and 150 nm). Together with the spherical geometry of the CMs, it makes a real challenge to interpret/process the results of the force-distance experiments. Performing a large number of 3D AFM measurements on single CMs, with a high spatial resolution, showed that the registered curves can be more or less divided into classes, depending on the relative position of the tip axis with respect to the micelle center. The main patterns are the ones given in Figure 7. If the CM is subjected to a diametral load (case a1), the compression curve indicates a relatively “tough” material behavior (curve in Figure 7-b1). For vertical scans done at horizontal locations further away from the CM center, an apparent softening (hence smaller slope) is observed

5084

Langmuir, Vol. 20, No. 12, 2004

Uricanu et al.

Figure 5. Tridimensional perspective of the protein structures deposited on the hard substrate from cold-WPF at pH ) 5.4 (a) and pH ) 5.2 (b). At pH ) 5.4, only single CMs are attached to the substrate. For low pH ()5.2), around 15% from the counted entities are aggregates. Note that contrary to Figures 2 and 3, these are now height images.

Figure 6. Results of cross-section analysis based on the AFM images recorded on CMs covalently linked to the SAM at pH ) 5.2 and temperature: 4 °C (a1-a2) and 20 °C (b1-b2). The number of the CMs belonging to the most numerous population was taken as reference in the calculations of normalized counts. The calculated height (upper graphs) for the single CMs are considered (as a first approximation) as being equal to the micelle diameter (D). Relative height versus width (i.e., size along the scanning direction) for single micelles (open symbols) or CM aggregates (closed symbols) are plotted in the lower graphs. For aggregates, an appreciable scatter in the data is observed, compared to the rather smooth correlation (full line) for single CMs.

(Figure 7-b2). For compressions close to the micelle periphery (Figure 7-a3), a metastable regime is observed and slip effects are obvious in the recorded curves (Figure 7-b3), after which the tip touches the hard substrate. The last regime is a frictional one, when, during the vertical up-down movements, the tip barely senses the micelle (see Figure 7-a4,b4). On a given CM, the a1-a2 regimes give well-reproducible data: approach and retract curves were found to coincide within the same scan, as well as on repeating the scan. Meanwhile, the metastability is a “unique event”: the location of the sliding points (or zones) changes per measurement. Also the correspondence between approach and retract scans was lost for the a3 regime. The imbalance in the local pressures developed inside the casein association generates these nonequilibrium events. In these particular circumstances, we restricted our data processing only to force-distance curves acquired in the a1-a2 conditions. The size of the bound CMs represents in itself a limitation of the analysis range; for small micelles (diam-

eter