Surface chemistry at the nanometer scale influences insulin aggregation

Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

Contents lists available at SciVerse ScienceDirect

Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb

Surface chemistry at the nanometer scale influences insulin aggregation Lalit M. Pandey a,c , Simon Le Denmat b , Didier Delabouglise a,∗ , Franz Bruckert a , Sudip K. Pattanayek c , Marianne Weidenhaupt a a b c

Laboratoire des Matériaux et du Génie Physique (LMGP), Grenoble Institute of Technology, 3 parvis Louis Néel, BP 257, 38016, Grenoble cedex 1, France Centre Interuniversitaire de MicroElectronique et Nanotechnologies (CIME Nanotech), 3 parvis Louis Néel, BP 257, 38016, Grenoble cedex 1, France Department of Chemical Engineering, IIT Delhi, New Delhi, 110016, India

a r t i c l e

i n f o

Article history: Received 17 January 2012 Received in revised form 19 April 2012 Accepted 19 May 2012 Available online 28 May 2012 Keywords: Self-assembled monolayers (SAMs) Wettability Material roughness Protein aggregation Insulin

a b s t r a c t We synthesized surfaces with different hydrophobicities and roughness by forming self-assembled monolayers (SAMs) of mixed amine and octyl silanes. Insulin aggregation kinetics in the presence of the above surfaces is characterized by a typical lag phase and growth rate. We show that the lag time but not the growth rate varies as a function of the amine fraction on the surface. The amount of adsorbed protein and the adsorption rate during the aggregation process also vary with the amine fraction on the surface and are maximal for equal parts of amine and octyl groups. For all surfaces, the growth phase starts for identical amounts of adsorbed insulin. The initial surface roughness determines the rate at which protein adsorption occurs and hence the time to accumulate enough protein to form aggregation nuclei. In addition, the surface chemistry and topography influence the morphology of aggregates adsorbed on the material surface and the secondary structures of final aggregates released in solution. © 2012 Elsevier B.V. All rights reserved.

1. Introduction The secondary structure of a protein is important for its function and involved in its aggregation behavior. The aggregation of proteins is of major concern in the fields of protein research and clinical medicine. Protein stability is essential for preparation, storage and delivery of proteins. In general, changes in the secondary structure lead to protein aggregation, in particular the formation of insoluble amyloid fibrils. These have been associated with various disorders such as Alzheimer’s disease, Parkinson’s disease, prion diseases and other amyloidoses [1–6]. To understand this phenomenon, insulin has been used as a model protein for aggregation studies [7–9]. In solution, the aggregation kinetics of insulin toward amyloid fibrils follows a sigmoid curve with three different phases: a slow nucleation phase (long lag time), a faster growth phase and a plateau phase [8,10]. It has been reported that insulin unfolds partially in certain environmental conditions and combines to form stable nuclei [7,11]. Subsequently, other insulin molecules combine to these nuclei to form fibrils during the growth phase [10,11]. The

∗ Corresponding author. Tel.: +33 4 56 52 93 56; fax: +33 4 56 52 93 01. E-mail addresses: lalit [email protected] (L.M. Pandey), [email protected] (S. Le Denmat), [email protected] (D. Delabouglise), [email protected] (F. Bruckert), [email protected] (S.K. Pattanayek), [email protected] (M. Weidenhaupt). 0927-7765/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.colsurfb.2012.05.022

secondary structure of insulin changes during the lag phase, from ␣-helix to less ordered structures (random coil and ␤-turn). These structures then form ␤-sheet structures during the growth phase [11]. Effects of various environmental factors such as pH, agitation, temperature, ionic strength, etc. on the aggregation process have been studied [10]. Insulin amyloid aggregation has been studied in presence of hydrophilic and hydrophobic surfaces [7,12–15]. A longer lag time has been observed in the presence of highly hydrophilic surfaces as compared to hydrophobic ones [7,11,16]. Also adsorption of insulin has been studied on both hydrophilic and hydrophobic surfaces [13,14]. The adsorbed mass of insulin is negligible on a plain silica surface but considerably higher on methylated silica surface [14]. Sluzky et al. [16] have correlated surface hydrophobicity with the propensity for aggregation but have not studied protein adsorption. Moreover the used surfaces in the above studies were not well characterized. Another amyloid forming polypeptide, ␤-amyloid, has been found to adsorb and aggregate on both hydrophilic mica and hydrophobic graphite surfaces [17,18]. The effects of surface morphology, hydrophobicity and protein adsorptivity on aggregation process are not fully understood. We investigated again this problem paying attention to the surface chemistry, roughness and protein adsorptivity. We prepared surfaces with different hydrophobicities using pure and mixed self-assembled monolayers of hydrophilic amine groups and hydrophobic octyl groups. We have first characterized the

70

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

wettability and topographical properties of the surfaces before studying insulin aggregation kinetics. We correlated the latter with the amount of adsorbed protein and lag time.

2. Materials and methods If not otherwise stated, all chemicals were purchased from Sigma–Aldrich.

2.1. Surface modification and characterization Glass beads of 1 mm diameter were cleaned and silanized to form mono (amine or octyl) and mixed Self Assembled Monolayers (SAMs). The SAMs were formed according to previously described procedure [19]. Briefly, the glass beads were sonicated with a freshly prepared “piranha” solution (H2 SO4 and 30% H2 O2 ; v/v = 7/3) for 1 h, an ammonia solution (H2 O, 30% H2 O2 and NH3 ; v/v/v = 5/1/1) for 30 min and an HCl solution (H2 O, 30% H2 O2 and HCl; v/v/v = 6/1/1) for 30 min. They were then washed with deionized water and acetone and finally dried. The beads were immersed in 1% (v/v) silane coupling agent in anhydrous toluene for 24 h at room temperature under an inert atmosphere to form surfaces with amine, octyl or mixed SAMs. After the silanization, the beads were sonicated consecutively in three different solvents: toluene, mixture of toluene and methanol (v/v = 1:1) and methanol for 2 min each. Finally, beads were dried overnight at 37 ◦ C. The silane coupling agents were 3-aminopropyl triethoxysilane (APTES) and octyltriethoxysilane (OTES), to form amine and octyl SAMs respectively. Mixed SAMs were formed by mixing the two silanes in different ratios of OTES:APTES 1:9, 1:3, 1:1, 3:1 and 9:1. Surfaces were characterized in terms of wettability of deionized water at room temperature using a Kruss contact angle goniometer by the sessile drop method. Static contact angles were calculated by fitting a circle on the obtained drop shape. Six readings were taken on each surface made from different preparations. Images of the surface morphology were taken by Atomic Force Microscopy (AFM, Digital Instruments Multimode Nanoscope IIIA) in the tapping mode using a silicon tip (tip radius  OMCL-AC200TS-E3). n 7 nm, The roughness parameter (Ra = 1/n i=1 zi  where zi is the height of surface features) was determined using the Gwyddion software. The amplitude and extent of the surface features were used to determine the roughness factor (Rf ) according to a previously described procedure [18]. Briefly, we assumed surfaces with spherical texture with average wavelength, a and average height, Ra . Profile perimeter, P, is calculated by fitting a circle to the profile arc. Rf is defined by the relation Rf = P2 /L2 , where L2 is the scanned area and P2 is the square of the profile perimeter. Each data point is the average of three readings which were taken on samples of three different preparations.

2.2. Preparation of insulin solution Human Insulin (HI), (SIGMA I2643, recombinant, expressed in yeast, zinc content ≤1% w/w) solution was prepared in Tris–Buffered Saline, TBS (25 mM Tris–HCl, 125 mM NaCl, 2 mM MgCl2 , pH 7.4). The resulting mixture was dissolved by lowering the pH to about 3.0–3.5 with 1 M HCl, the pH was then readjusted to 7.3 ± 0.1 using 1 M NaOH. The HI concentration was determined by UV absorbance at 280 nm using an extinction coefficient of 5530 M−1 cm−1 . A final concentration of 0.5 mg mL−1 was obtained by dilution with TBS and the solution was filtered through a sterile 0.22 ␮m Millex GV low-protein binding filter.

2.3. Aggregation studies 200 mg of glass beads coated with different SAMs were placed in a 0.5 ml borosilicate glass tube (24715, Supelco Analytical) which was filled with the insulin solution and sealed with Parafilm. The tubes were agitated at 60 rpm at 37 ◦ C on PTR-60 Grant Bio vertical shaker. At different time intervals, tubes were removed and the insulin samples were analyzed as follows: fluorescence spectroscopy was used to quantify the fibril formation, FTIR spectrophotometry was used to determine the change in protein secondary structure, the Bicinchoninic Acid (BCA) assay was used to determine the adsorbed amount of protein on beads and AFM was used to image the surface morphology of the adsorbed/aggregated protein. All the experiments were repeated at least three times. Fluorescence measurements were performed after adding 20 ␮M of Thioflavin T (ThT), which has excitation/emission wavelength of 450/482 nm after binding to amyloid fibrils. These measurements were performed on a Tecan Infinite M1000 multimode microplate reader (TECAN USA) with 5 nm excitation and emission slits. The aggregation kinetics proceeds in three phases: a lag phase for which the signal is not statistically different from the baseline (mean value ± standard deviation), a linear growth phase and a plateau. Experimentally, the lag time is determined by the intercept between the linear growth phase and the x-axis. The nucleation rate is the inverse of the lag time and the slope of the linear growth phase gives the aggregate growth rate (h−1 ), which was normalized to the maximum value of the signal. These parameters were calculated on individual kinetics corresponding to different samples, and the given statistics display the average and the standard deviation for each parameter. The beads that had been incubated with insulin were washed three times with TBS solution. The adsorbed insulin on beads was fully desorbed in 100 ␮l of 5% Sodium Dodecyl Sulfate solution (SDS) by agitating at 37 ◦ C for 1 h and then quantified by the BCA assay. The adsorption kinetics were fitted with the exponential function m(t) = m0 exp(kt), where m(t) represents the evolution of adsorbed protein mass (mg m−2 ) in time, m0 the initial adsorbed protein mass (mg m−2 ) and k the adsorption rate (h−1 ). The conformation of native or aggregated HI was analyzed by FTIR (VERTEX V70 FTIR spectrophotometer). HI solutions were prepared as before but in D2 O buffer (pD 6.90). HI was incubated with coated beads to induce aggregation. The solutions were placed between two CaF2 windows, separated by a 0.2 mm Teflon spacer and FTIR spectra were recorded in the transmission mode. All the measurements were repeated three times. The processing of the spectra was done using the OPUS 6.5 software. The characteristic peaks of the amide-I band in the 1600–1700 cm−1 range were further analyzed. Peak positions were determined from the second derivative of FTIR spectra. Peaks were fitted with a Gaussian shape. AFM was used to characterize surface morphology of aggregated insulin on the material surfaces. The beads that had been incubated with insulin were washed three times with TBS solution, then washed with de-ionized water and dried at room temperature before observation using a silicon tip as described above for surface characterization.

3. Results 3.1. Characterization of SAM-coated surfaces Water contact angles are 10◦ , 62 ± 1◦ and 102 ± 1◦ on clean glass (hydroxyl surface), amine and octyl SAMs, respectively. Contact angles on surfaces coated with mixed SAMs are intermediate between those of amine and octyl SAMs (Fig. 1, see also [19]). Experimental data are very close to calculated values of contact angle

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

Fig. 1. Characterization of mixed silane coated glass surfaces. Mixed APTES-OTES surfaces were prepared as described in Section 2. The contact angle (closed squares) and the roughness (open circles) are represented as a function of the amine fraction. Each data point is the average of three readings taken on samples of three different preparations.

using the Cassie equation, cos  = fA cos  A + fO cos  O [20], where, f is the fraction of each constituent and subscript A refers to amine and O to octyl. This confirms that silanes are immobilized on the surface in proportion to the mixing ratios of coupling agents in solution. AFM imaging was used to assess the roughness and topography of surfaces coated with SAMs and measured roughness values, Ra and Rf are presented in Table 1 and on Fig. 1. The surface roughness, Ra increases for SAMs with increasing amine fraction until equal amine and octyl parts are used for SAM coatings. For SAMs with higher amine fractions, roughness decreases until the value of pure amine SAM is obtained (0.63 nm). The roughness factor, Rf also follows a similar trend as Ra . The maximum value of Rf is observed for mixed 1:1 SAM (1.0335 nm). AFM images of surfaces with different SAMs reflect their topography as shown in Fig. 2. Untreated glass bead surfaces are flat but exhibit defects in the nanometer range. Surface coating with pure SAMs increases the roughness slightly but mixtures of amine and octyl SAMs result in surfaces with irregular topographies. Thus mixing of SAMs with different chemical groups result in surfaces with different chemical energies, topographies and roughness that can be used to study the effect of surface characteristics on protein aggregation. 3.2. Kinetics of insulin aggregation Insulin aggregation kinetics was monitored in the presence of glass beads (200 mg) whose surface was modified by the previously described coatings. The aggregation level was determined by measuring HI-bound ThT fluorescence. An increase in ThT fluorescence reflects the formation of amyloid fibrils. Nucleation and growth rates are deduced from the ThT fluorescence data as described in Table 1 Average roughness of beads surfaces before and after insulin adsorption. Beads surface

Plain Glass Pure amine Mixed 1:9 Mixed 1:3 Mixed 1:1 Mixed 3:1 Mixed 9:1 Pure octyl

Before adsorption

After adsorption

Average Height, Ra (nm)

Roughness factor, Rf

Average Height, Ra (nm)

0.18 ± 0.1 0.63 ± 0.1 1.10 ± 0.1 1.59 ± 0.2 1.61 ± 0.2 1.06 ± 0.1 0.85 ± 0.1 0.83 ± 0.1

1.0001 1.0036 1.0136 1.0177 1.0335 1.0151 1.0112 1.0056

0.46 ± 0.4 3.72 ± 1.0 3.94 ± 1.3 2.73 ± 0.8 8.17 ± 2.5 4.82 ± 1.5 3.34 ± 1.0 1.74 ± 0.4

Borosilicate beads were surface-modified with silanes and mixed silanes as explained in Section 2 (each data point is the average of three readings taken on samples of three different preparations).

71

Section 2. After the lag phase, which represents the time to form stable aggregate nuclei on the surface, the ThT fluorescence increases sharply (growth phase) and reaches a plateau value (Fig. 3A). In the absence of beads the lag time last more than 2 weeks. Uncoated (hydroxyl) beads show a minimal nucleation rate of about 0.03 h−1 (data not shown). The nucleation rate for pure octyl-coated beads is 0.07 h−1 and increases as the fraction of amine increases (Fig. 3B). Interestingly, mixed 1:1 SAM-coated beads exhibit the fastest nucleation rate (0.44 h−1 ) which then slightly decreases as the amine fraction is raised from 0.5 to 1.0 (pure amine, 0.38 h−1 ). Mixing equal amounts (100 mg) of amine-coated beads and octylcoated beads resulted in an intermediate nucleation rate of 0.21 h−1 (Fig. 3B). The difference in the nucleation rate obtained on 1:1 SAMcoated beads and pure amine-coated beads is small (0.06 h−1 ) but significant. The growth rates of insulin aggregation on different SAMs are given in Fig. 3C. They are very similar for all the surfaces (about 0.6 h−1 ) except for octyl-coated beads (0.25 h−1 ). The amount of insulin adsorbed on surfaces coated with different SAMs increases with time and reaches a maximum at the end of the growth phase (Fig. 4A and B). Once aggregation is complete, the adsorbed amount decreases during the stationary phase (the plateau) because the insulin pool is depleted and aggregates detach from the surface. On octyl-coated beads about 17 mg m−2 of insulin accumulated (Fig. 4B). Interestingly, the adsorbed protein mass increases as the amine fraction of mixed SAM coatings increases with a maximum for mixed 1:1 SAM coatings (60 mg m−2 ) (Fig. 4A). On pure amine-coated beads about 30 mg m−2 of insulin is adsorbed. The insulin adsorption curves can be fitted with an exponential function (see Section 2) that allows us to determine the adsorption rate as a function of the amine fraction of the SAM coatings (Fig. 4C). Adsorption rate shows a hyperbolic dependence on the amine fraction of the SAM coating with a maximum value for mixed 1:1 SAMs (0.6 h−1 ). The adsorption rate for an equal mixture of amine-coated beads and octyl-coated beads is 0.25 h−1 . The variation of the surface chemistry of the SAM coatings has a significant influence on the protein adsorption rate which is about 6 times faster for mixed 1:1 SAM than for octyl coated beads. A correlation with the surface roughness can also be drawn (Fig. 5). 3.3. Secondary structure of native and aggregated insulin Changes in secondary structures of insulin take place during aggregation. FTIR spectra of native and aggregated insulin in the presence of different SAMs are shown in Fig. 6. The peak positions of ␣-helix, ␤-sheet and ␤-turn components in the amide-I region were taken from previous reports [11,21]. Native HI in the solution contains about 84% ␣-helix, 9% ␤-sheet and 7% ␤-turn components in accordance to Ref. [11]. In the aggregated state, the peak around 1650 cm−1 corresponding to ␣-helix of native insulin disappears and conversely ␤-sheet (1631 cm−1 ) and ␤-turn (1663 cm−1 ) peaks become predominant (Fig. 6B–D) [11,21]. This ␣-helix to ␤-sheet conformational change has been documented earlier for insulin aggregation at low pH [11]. Interestingly the chemistry of the surface coating influences the distribution of secondary structures in final insulin aggregates. The peak corresponding to the ␤-turn component is less prominent on amine SAM-coated surfaces than on octyl or mixed SAM-coated surfaces while the amount of ␤-sheet varies in the order: amine SAM > mixed SAM > octyl SAM. 3.4. Surface morphology of insulin aggregates on beads AFM images of beads coated with different SAMs at the end of HI aggregation are shown in Fig. 7. Table 1 summarizes roughness before and after protein adsorption. On plain glass beads, the surface remains very smooth with a Ra of 0.46 ± 0.2 nm. For all the other surfaces, protein adsorption induces a large increase of

72

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

Fig. 2. AFM images of mixed silane coated glass surfaces. Mixed APTES-OTES surfaces were prepared as described in Section 2. The surface morphology was visualized by AFM and representative images are shown of each surface treatment. In each image, the average level has been adjusted to the red color, and the range of z values is 12–14 nm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

roughness, which confirms the accumulation of large amounts of insulin. Two zones can be distinguished when inspecting surface morphology: a flat background and prominent aggregates of about 10 nm. The insets of mixed 1:3 and mixed 3:1 focus on the background and reveal that the surface is smoother than the initial coated surfaces (compare to Fig. 2). Insulin adsorption therefore masks the surface topography at the nanometer scale. The

morphology of aggregates varies with surface functional groups. The height of the aggregates on the surfaces with amine and octyl SAMs lies in the range of 5–10 nm while, on the surfaces with mixed SAMs, it ranges from 10 to 20 nm. The diameter of the aggregates on all the surfaces is 50–100 nm (this value is overestimated because the finite radius of the AFM tip adds on the diameter of the object). The aggregates have an elongated morphology (100–200 nm) on

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

Fig. 3. Kinetics of insulin aggregation on mixed silane-coated glass surfaces. Mixed APTES-OTES borosilicate beads were prepared as described in Section 2. They were then incubated with 0.5 mg mL−1 insulin for the indicated time in the presence of Thioflavin T. The thioflavin T fluorescence kinetics as then analyzed as explained in Section 2 to determine the nucleation and the growth rates. A: Thoflavin T fluorescence kinetics. Open circles: mixed O:A 1:1 surfaces, closed circles: amine surfaces, crosses: the sample contained an equal mixture of amine- and octyl-coated beads, closed squares: octyl surfaces. B: nucleation rate of insulin aggregation. Closed squares: nucleation rate as a function of the amine fraction. Open circle: the sample contained an equal mixture of amine- and octyl-coated beads. C: growth rate of insulin aggregation. Closed squares: nucleation rate as a function of the amine fraction. Open circle: the sample contained an equal mixture of amine- and octyl-coated beads. Data points are the mean of four experiments.

the surfaces containing amines up to a mixed 1:3 SAM. On surfaces containing more than half of octyl mixed with amine in the SAM, the morphology of the aggregates is more spherical. 4. Discussion We prepared glass beads with surface coatings made of amine, octyl and mixed SAMs and used them to study material surface-induced aggregation of insulin. In our experiments, insulin aggregation follows a characteristic sigmoid shape with lag phase, exponential growth phase and stationary phase as previously

73

Fig. 4. Kinetics of insulin adsorption on mixed silane-coated glass surfaces. Mixed APTES-OTES borosilicate beads were prepared as described in Section 2. They were then incubated with 0.5 mg mL−1 insulin for the indicated time. The amount of adsorbed proteins was then determined as described in Section 2. The uncertainty in this measure is less than 10% (n = 3) and was omitted for clarity. In order to determine the initially adsorbed mass and adsorption rate, the adsorption kinetics were fitted with a single exponential (see Section 2) on the following ranges: 0–4 h (mixed O:A 1:1, mixed O:A 1:9, mixed O:A 1:3), 0–5 h (amine, mixed O:A 3:1, mixed O:A 9:1), 0–20 h (octyl), 0–7 h (equal mixture of amine and octyl beads). A and B: insulin adsorption kinetics. A: open squares: mixed O:A 1:1 surfaces, closed squares: mixed O:A 1:9 surfaces, open triangles: amine surfaces, closed triangles: mixed O:A 9:1 surfaces, open circles: equal mixture of amine- and octyl-coated beads, B: octyl surfaces. C: initial adsorbed mass (closed squares) and adsorption rates (open circles) as a function of amine fraction. Data points are the mean of three experiments.

reported [7,15]. The nucleation rate reflects the lag phase during which protein accumulates on the material surface and once a critical density is reached, aggregation starts. The rapid aggregate growth phase reaches a plateau when the soluble insulin pool is depleted. Upon further incubation, aggregates continue to detach from the surface. Our results confirm that insulin aggregation takes place on the bead surface, where protein adsorption triggers conformational changes that henceforth drive aggregation. Aggregation of insulin occurs on both hydrophilic and hydrophobic surfaces but with different nucleation rates (Fig. 3B).

74

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

Fig. 5. Insulin adsorption rate as a function of the surface roughness. The values are taken from Fig. 1 (initial roughness) and Fig. 4C (insulin adsorption rate). Open circle: samples with amine, octyl and mixed SAMs. Data points are the mean of three experiments.

Pure octyl SAMs present slower nucleation rates than pure amine SAMs. Different nucleation rates refer to different mechanisms of insulin adsorption on these surfaces. Indeed, the surface-exposed functional groups of the different SAMs influence the adsorption behavior. On hydrophilic amine SAM-coated beads adsorption of insulin occurs mainly due to electrostatic interactions. Amine groups are positively charged at pH 7.4 (pKa 9.8 [22]) and insulin carries a net negative charge (pI 5.6 [23]). Thus, adsorption is favored by strong electrostatic attractions. On the other hand,

insulin adsorbs through hydrophobic groups on octyl SAM exposing hydrophilic groups toward the solution. This conformation results in longer lag times on surfaces coated with octyl SAMs probably because the formation of aggregation nuclei is less efficient in this case [7,10,13]. Thus, depending on the physico–chemical mechanisms driving adsorption, the time to establish aggregation-prone nuclei on the surface varies. Combining hydrophobic interactions inducing slower nucleation rates with electrostatic attraction by creating mixed SAMs with various amine fractions on the same beads accelerates the nucleation rate up to 8 times. Interestingly the 1:1 SAM coating shows the quickest nucleation rate, slightly but significantly higher than obtained with a pure amine SAM. The combination of mixed 1:1 SAM coatings on the same beads triggers significantly faster nucleation rates than when an equal mixture of pure amine and octyl SAMs coated beads are used (Fig. 3B). This shows that, beyond the physico–chemical nature of the functional groups, other parameters such as surface topography and nanometer roughness also influence aggregation kinetics. Our data show that the amine fraction in the SAM coating influences both the initial surface roughness (Fig. 1) and the protein adsorption rate (Fig. 4C). For a mixed 1:1 SAM surface roughness is the highest and so is the protein adsorption rate. Surfaces with mixed SAMs are rougher than surfaces with mono SAMs (Fig. 1), probably due to spatial exclusion between hydrophilic amine and hydrophobic octyl groups. This increased roughness might provide multiple anchoring points for the adsorption of insulin and as a consequence increase the adsorption speed and the amount of adsorbed proteins (Fig. 4). Consequently, mixed SAMs accelerate insulin adsorption and aggregation by combined phenomena

Fig. 6. FTIR spectra of insulin aggregates. Plain and mixed APTES-OTES borosilicate beads were prepared as described in Section 2. They were then incubated with 0.5 mg mL−1 insulin at 37 ◦ C until full insulin aggregation had occurred. The insulin aggregate suspension was separated from the beads and a FTIR spectrum (dotted line) was recorded in the range 1580–1720 cm−1 . The spectra were decomposed (thin lines) in ␣-helical (1650 cm−1 ), ␤-sheet (1632 cm−1 ) and ␤-turn (1665 cm−1 ) peaks (see Section 2). The resulting adjusted spectra are very close to the recorded ones and are omitted for clarity. A: FTIR spectrum of the initial insulin solution. B: insulin FTIR spectrum after aggregation on amine-coated surfaces. C: insulin FTIR spectrum after aggregation on octyl-coated surfaces. D: insulin FTIR spectrum after aggregation on mixed O:A 1:1 coated surfaces.

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

75

Fig. 7. AFM images of mixed silane coated glass surfaces after incubation with insulin. Mixed APTES-OTES surfaces were prepared as described in Section 2. They were then incubated with 0.5 mg mL−1 insulin at 37 ◦ C until full insulin aggregation had occurred. The surface morphology was visualized by AFM and representative images are shown of each surface treatment. In each image, the average level has been adjusted to the red color, and the range of z values is 50–57 nm. In order to compare with the results of Fig. 2, the regions of interest surrounded by a square in the images corresponding to Mixed O:A 1:3 and 3:1 coatings are redrawn aside. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

involving the physico–chemical nature of the surface and its roughness (Fig. 5). On all surfaces tested, the aggregation growth phase started (end of the lag time), once around 8 mg m−2 of insulin was adsorbed (compare Fig. 3A, and Fig. 4A and B), which is equivalent to 8 monomer monolayers of insulin (taking insulin monomer monolayer coverage of 1.1 mg m−2 [12]). Once this threshold amount of adsorbed insulin is attained, stable aggregation nuclei are formed to which further insulin molecules bind. This situation is

reminiscent of a surface-induced aggregation process [7,18]. Hereafter the growth rate of insulin aggregates is largely independent of the nature of the surface (Fig. 3C), which infers that the rate of detachment of aggregates from the surface is similar on all surfaces. This indicates a critical amount of protein has to accumulate on the surface before aggregation-promoting nuclei begin to form. Similarly, Giacomelli and Norde [24] showed that the interaction between the ␤-amyloid peptide (A␤) and hydrophobic Teflon surfaces proceeds in two steps. At a low surface coverage, the A␤

76

L.M. Pandey et al. / Colloids and Surfaces B: Biointerfaces 100 (2012) 69–76

peptide keeps its ␣-helical structure, but as the surface density increases, a cooperative conformational transition takes place, that results in ␤-sheet formation and subsequent aggregation. Insulin aggregation upon contact with material surfaces [15] or induced at acidic pH [8] is accompanied by a transition from mainly ␣-helical to predominantly ␤-sheet conformation. In aggregates obtained in contact with beads coated with different SAMs, varying amounts of ␣-helical, ␤-sheet and ␤-turn are observed (Fig. 6). The contents of ␤-sheet are more abundant on amine SAM as compared to octyl and mixed 1:1 SAMs, whereas ␤-turn conformations are less abundant on amine SAM than on octyl or mixed 1:1 SAMs. The surface chemistry drives the adsorption mechanism of insulin, the associated kinetics, the conformational transitions the proteins undergo and possibly the morphology of the final aggregates. Indeed, we observe that insulin aggregates form elongated structures on surfaces with amine-rich SAMs while more globular structures on octyl-rich SAMs (Fig. 7). According to its primary structure, a protein adopts different conformations when adsorbed on hydrophobic or hydrophilic surfaces. These conformations are driven by the net energy balance that accompanies the exposure of side chains either toward the surface or the solvent. Hence according to the physico–chemical properties of each side chain (hydrophobicity, charge) the adsorbed protein conformation will be different on surfaces exposing different hydrophobicities. Subsequent stacking of incoming proteins on top of the first layer of adsorbed protein thus see their final shape guided by the most stable conformation of the initially adsorbed layers. In summary, the physico–chemical nature of exposed groups as well as the roughness of surface coatings can be tuned using different ratios of amine and octyl SAMs. These mixed SAMs influence material-induced protein aggregation and represent therefore a tunable system to control this phenomenon. Surface functionalization of materials in contact with proteins is particularly important in the development of recombinant protein-based drugs or implantable devices in contact with the blood. Here, the fine control of protein adsorption using tunable surface characteristics is decisive for the successful application of the material. 5. Conclusion We have synthesized surfaces with different hydrophobicities (water contact angle 62◦ –102◦ ) by forming various SAMs. Surface morphologies, characterized by AFM, results in nano-scale smooth surfaces. Kinetics of aggregation is characteristic of a nucleation controlled mechanism. Nucleation rate depends on surface

chemistry as well as surface morphology and nanometer roughness. Aggregation growth phase start after formation of stable nuclei, which corresponds to adsorbed amount of about 8.0 mg m−2 on all the surfaces. Adsorption rate follows a similar trend as surface roughness; while growth rate, which also reflects the detachment rate of aggregates from surface, is independent of surface chemistry. Surfaces with different chemistry induce aggregates of different secondary structures and so different morphologies. Acknowledgments Special thanks to the European Union for allocating an Erasmus Mundus fellowship to Lalit M. Pandey. We also thank Laurent Nault and Thomas Ballet (LMGP) for fruitful discussions, Carmen Jimenez and Catherine Picart (LMGP) for technical help. References [1] D.J. Selkoe, Ann. N.Y. Acad. Sci. 924 (2000) 17. [2] B.A. Yankner, Neuron 16 (1996) 921. [3] C. Guela, C.K. Wu, D. Saroff, A. Lorenzo, M. Yuan, B.A. Yankner, Nat. Med. 4 (1998) 827. [4] M.F. Perutz, Curr. Opin. Struct. Biol. 6 (1996) 848. [5] S.B. Prusiner, Trends Biochem. Sci. 21 (1996) 482. [6] F. Rousseau, J. Schymkowitz1, L. Serrano, Curr. Opin. Struct. Biol. 16 (2006) 118. [7] V. Sluzky, J.A. Tamada, A.M. Klibanovt, R. Langer, Proc. Natl. Acad. Sci. U.S.A. 88 (1991) 9377. [8] J.L. Whittingham, D.J. Scott, K. Chance, A. Wilson, J. Finch, J. Brange, G.G. Dodson, J. Mol. Biol. 318 (2002) 479. [9] M.I. Ivanova, S.A. Sievers, M.R. Sawaya, J.S. Wall, D. Eisenberg, Proc. Natl. Acad. Sci. U.S.A. 106 (2009) 18990. [10] L. Nielsen, R. Khurana, A. Coats, S. Frokjaer, J. Brange, S. Vyas, V.N. Uversky, A.L. Fink, Biochemistry 40 (2001) 6036. [11] M. Bouchard, J. Zurdo, E.J. Nettleton, C.M. Dobson, C.V. Robinson, Protein Sci. 9 (2000) 1960. [12] S.H. Mollmann, J.T. Bukrinsky, S. Frokjaer, U. Elofsson, J. Colloid Interface Sci. 286 (2005) 28. [13] P. Nilsson, T. Nylander, S. Havelund, J. Colloid Interface Sci. 144 (1991) 145. [14] J. Buijs, C.C. Vera, E. Ayala, E. Steensma, P. Håkansson, S. Oscarsson, Anal. Chem. 71 (1999) 3219. [15] T. Ballet, F. Bruckert, P. Mangiagalli, C. Bureau, L. Boulangé, L. Nault, T. Perret, M. Weidenhaupt, Biochemistry 51 (2012) 2172. [16] V. Sluzky, A.M. Klibanovt, R. Langer, Biotechnol. Bioeng. 40 (1992) 895. [17] T. Kowalewski, D.M. Holtzman, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 3688. [18] M. Zhu, P.O. Souillac, C. Ionescu-Zanetti, S.A. Carter, A.L. Fink, J. Biol. Chem. 277 (2002) 50914. [19] L.M. Pandey, S.K. Pattanayek, Appl. Surf. Sci. 257 (2011) 4731. [20] A.B.D. Cassie, Faraday Discuss. 3 (1948) 11. [21] J. Kong, S. Yu, Acta Biochim. Biophys. Sin. 39 (2007) 549. [22] L.A. Belyakova, A.M. Varvarin, N.V. Roik, Appl. Surf. Sci. 253 (2006) 784. [23] C. Tanford, J. Epstein, J. Am. Chem. Soc. 76 (1954) 2163. [24] C.E. Giacomelli, W. Norde, Biomacromolecules 4 (2003) 1719.