Layer-by-layer assembled protein nanotubes with high DNA affinity

Abstract

In this research, protein nanotubes with high affinity for DNA molecules were prepared by alternate layer-by-layer (LbL) assembly of human serum albumin (HSA) and polyethylenimine (PEI). The well-defined hollow cylinder (PEI/HSA)₅PEI nanotubes with outer diameter 455 ± 13 nm, inner diameter 275 ± 35 nm, wall thickness 151 ± 7 nm and length 5.6 ± 0.9 μm were characterized by scanning electron microscopy (SEM), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FT-IR) and energy dispersive spectroscopy (EDS). Under optimal conditions, the maximum adsorption capacity of protein nanotubes for polyA₂₅ was 606.0 mg g⁻¹, which was far higher than the reported nanoparticles. The rates of adsorption were found to conform to the pseudo-second-order kinetics and intra-particle diffusion models with good correlation. The equilibrium adsorption data fitted well with the Langmuir isotherm model. The DNA release from DNA/(PEI/HSA)₅PEI nanotubes was associated with the ionic strength and pH in the solution.

---

1. Introduction

The adsorption or immobilization of nucleic acid on solid surfaces is essential in a wide range of research including DNA chip technology, DNA hybridization and gene delivery.^1–6 The inherent merits of nanotubes, for example, unique hollow structure, distinct inner and outer surfaces as well as two open-end terminals, make it to a promising nanomaterial for bimolecular immobilization and delivery.⁷ Inorganic nanotubes such as carbon nanotubes,⁸ silica nanotubes⁹ and magnetic nanotubes¹⁰ have been developed for gene vectors. But their biomedical applications are limited due to the lack of biocompatibility and biodegradability. It could be a good strategy to synthesize biological nanotubes. Some biological nanotubes such as lipid nanotubes¹¹ have been reported for this propose. However, protein nanotubes with DNA affinity are largely unexplored, thus is addressed in this paper.

Currently, a popular approach to prepare protein nanotubes is the combination of layer-by-layer (LbL) assembly technique with template method. Exploiting the electrostatic attraction, oppositely charged polyelectrolytes and biomacromolecules can build up multilayer into the pores of nanoporous membrane, such as anodic aluminium oxide (AAO) or track-etch polycarbonate (PC). The hollow protein nanotubes are obtained by subsequently removing the template.^12–14 This technique provides a simple and versatile means that can tune the thin films precisely on nanometer scales, and the synthesized tubular structures own tailored shapes and sizes.

Protein nanotubes were functionalized by assembling different molecules on the internal wall to realize target molecules capturing. For example, nanotubes assembled by human serum albumin (HSA) and poly-L-arginine (PLA) could bind the zinc(II) protoporphyrin IX to the HSA component in the cylindrical wall.¹⁵ The hybrid HSA/PLA nanotubes bearing single avidin layer as an internal wall captured biotin-labelled nanoparticles into the central channel.¹⁵ The nanotubes with an antibody surface interior entrapped human hepatitis B virus with size selectivity.¹⁶

One of the methods of DNA immobilization has been reported based on ionic interaction occurring between the negatively charged groups present on the DNA and positive charges covering the surface.¹⁷ Polyethylenimine (PEI) is a kind of cationic polyelectrolyte and the branched structure of PEI contains nitrogen at every third atom, resulting in a high charge density. PEI is often considered as the most prominent polymer capable of gene transfection¹⁸ and DNA immobilization.¹⁹ Therefore, we can expect that protein/PEI hybridized nanotubes with PEI on the inner and outer surfaces will provide the high efficient DNA loading.

In this article, protein nanotubes ((PEI/HSA)₅PEI) were prepared and the adsorption and release performances for oligonucleotides were investigated. Kinetic and isotherm models were applied to the experimental data for a better understanding of the binding process.

2. Experimental

2.1. Chemicals and materials

Polyethylenimine (PEI, Mw = 750 000, 50% (w/v) in water), human serum albumin (HSA, 96%–99%) and calf thymus DNA (ctDNA) were purchased from Sigma-Aldrich. The track-etch nanoporous PC membrane (no. 28420636, circle membrane diameter 25 mm, pore diameter 400 nm) was purchased from Whatman Co. Poly-adenosine (polyA₂₅) was provided by DNA-SYN Biotechnology Synthesis Lab (Beijing, China). The water was deionised (18.2 MΩ cm⁻¹) using purification Milli-Q system (Millipore, USA). All chemicals were used as received without further purification. Experiments were all carried out at 25 ± 2 °C.

2.2. Preparation of protein nanotubes

(PEI/HSA)₅PEI nanotubes were fabricated in PC nanopores by the revised LbL assembly according to previous method.¹⁵ Briefly, PC membrane was placed into a filter. 10 mL 1 mg mL⁻¹ PEI was firstly flowed through at the speed of 0.25 mL min⁻¹ using a syringe pump. Then the loosely adsorbed PEI was washed by 10 mL pH 7.4 PB at the rate of 1 mL min⁻¹. Following this, the membrane was dried under a stream of N₂. In next step, 2 mg mL⁻¹ HSA dissolved in pH 7.4 PB solution was injected (10 mL, 0.5 mL min⁻¹) through the membrane with assembled PEI layer. After injection of pH 7.4 PB solution (10 mL, 1 mL min⁻¹) to remove the loosely bound HSA, the membrane was again N₂ dried. Between each deposition steps, a cotton swab with water was used to eliminate the adsorption of PEI and HSA on the top and bottom of PC surfaces. After repeating 5.5-cycle injection, the membrane was dried in vacuum for 12 h (0.09 Mpa) and immersed in DMF subsequently. The PC was quickly dissolved and the (PEI/HSA)₅PEI nanotubes were precipitated. The liberated nanotubes were washed 3–5 times with DMF and the supernatants were discarded. Finally, the lyophilized (PEI/HSA)₅PEI nanotubes powder was yielded by freeze-dried the dispersion under vacuum (−60 °C, <20 Pa).

2.3. Characterization of protein nanotubes

For TEM observation, the dispersion of (PEI/HSA)₅PEI stained with 3% phosphotungstic acid was observed on a H600 electron microscope (Hitachi, Japan) at 75 kV. The SEM and EDS images were obtained using a JSM-6390 electron microscope (JEOL, Japan) by analyzing the lyophilized sample that sputter-coated with gold using an ion sputter (E1010, Hitachi). FT-IR spectra were recorded by using a TENSOR 27 FT-IR spectrophotometer (Bruker, German). UV-vis spectroscopy measurements were conducted using infinite M200 PRO (TECAN, Switzerland) to evaluate the adsorption and desorption experiments.

2.4. Adsorption experiments

The adsorption property of (PEI/HSA)₅PEI for DNA was studied. PolyA₂₅ was selected as typical DNA. A stock solution of 1 mg mL⁻¹ was prepared and diluted to the required concentrations at pH 7.4. In each experiment, 200 μL polyA₂₅ solution was added into a 1.5 mL centrifuge tube with 0.1 mg (PEI/HSA)₅PEI nanotubes. The mixtures were stirred continuously using an orbital shaker. At regular intervals, polyA₂₅ concentration in supernatant was determined by infinite M200 PRO at 260 nm after centrifugation with 6000 rpm. The amount of polyA₂₅ adsorbed onto (PEI/HSA)₅PEI nanotubes denoted as q_e (mg g⁻¹) was calculated using the following:where C₀ and C_t is the initial and equilibrium concentration of polyA₂₅ (mg L⁻¹), respectively, V (L) is the volume of polyA₂₅ solution and m (g) is the mass of (PEI/HSA)₅PEI nanotubes.

Effect of pH ranging from 4.7 to 11 was performed using 108 mg L⁻¹ polyA₂₅. Adsorption isotherm was studied using polyA₂₅ solutions with concentrations ranging from 22 mg L⁻¹ to 500 mg L⁻¹.

2.5. Release experiments

The DNA/(PEI/HSA)₅PEI nanotubes complexes were prepared by mixing 0.1 mg (PEI/HSA)₅PEI nanotubes with 200 μL 108 mg L⁻¹ polyA₂₅ at pH 7.4 for 1 h. The complexes were collected with centrifugation of 6000 rpm and then dispersed in 200 μL of PB buffer solution at pH 7.4 and pH 5.7 in the presence of NaCl (0–1.5 M), respectively. At regular intervals, the desorbed polyA₂₅ in solution was determined by measuring the supernatant at 260 nm. The ratio of DNA release from complexes was calculated as:where C_t (mg L⁻¹) is the polyA₂₅ concentration in solution at t time, q_e (mg g⁻¹) is the pre-adsorbed amount of polyA₂₅ onto (PEI/HSA)₅PEI nanotubes, V (L) is the volume of the solution, m (g) is the mass of (PEI/HSA)₅PEI nanotubes.

The adsorption and release measurements were conducted in triplicate and reported as the mean values ± standard deviation (SD).

3. Results and discussion

3.1. Characterization of (PEI/HSA)₅PEI nanotubes and DNA/(PEI/HSA)₅PEI nanotubes complexes

SEM images of (PEI/HSA)₅PEI nanotubes (Fig. 1A) displayed the well-defined hollow cylinders with outer diameter of 455 ± 13 nm, inner diameter of 275 ± 35 nm and wall thickness of 151 ± 7 nm. The outer diameter was dependent on the PC template used, thus the nanotube size was controlled by the pore size. The (PEI/HSA)₅PEI nanotubes could be considered as the eleven-layered cylinder model, in which each HSA layer had a single-protein thickness. The HSA dimension was assumed to be 8 nm from the single crystal structure^20,21 and thus the thickness of PEI layer was calculated as 18.5 nm. This value was between the reported values for polyelectrolyte layers fabricated into porous template.¹⁵ From the TEM image in Fig. 1B, a good tube structure was observed. The length was estimated to be 5.6 ± 0.9 μm, close to the PC template thickness (6 μm). Fig. 1C showed the FT-IR spectra of the fabricated nanotubes and HSA. In spectrum of HSA, peaks at 1650 cm⁻¹ and 1535 cm⁻¹ were the characteristic adsorptions of the C=O stretching vibration and the in-phase bending of N–H bond coupled with stretching of C–N bond in amide, respectively. The presence of peaks at 3320 cm⁻¹ and 2942–2820 cm⁻¹ was contributed to N–H stretching and C–H stretching vibrations.²² Above results confirmed that the HSA layers were successfully assembled. Fig. 1D exhibited the EDS spectra of nanotubes. The main compositions of carbon, nitrogen, oxygen and sulfur were detected, indicating that the expected distributions of PEI and HSA existed.

Fig. 1 (A) SEM and (B) TEM images of (PEI/HSA)₅PEI nanotubes, (C) FT-IR spectra of (PEI/HSA)₅PEI nanotubes and HSA, (D) EDS spectra of (PEI/HSA)₅PEI nanotubes, the inset is the elemental percentage.

The nanotubes with interior and exterior surfaces had cationic charges that could strongly bind the negatively charged DNA through electrostatic interaction. Fig. 2 showed the TEM images of DNA/(PEI/HSA)₅PEI nanotubes complexes. Compared with polyA₂₅ (Fig. 2A), the detail structure was clearly demonstrated by using the ctDNA. Indicated by black arrow in Fig. 2B, the adsorbed ctDNA chains were visible.

Fig. 2 TEM images of polyA₂₅ (A) and ctDNA (B) complex with (PEI/HSA)₅PEI nanotubes in pH7.4 PB solution. Black arrow indicates the adsorbed ctDNA chains onto nanotube.

3.2. Effect of pH

Since electrostatic interaction plays a key role in determining the binding between DNA and nanotubes, it is necessary to study the pH effect on the adsorption. From Fig. 3, it was found that polyA₂₅ binding decreased slightly as the pH increased. For example, the binding reduced from 94.8% to 87% by raising the pH from 4.7 to 11. The (PEI/HSA)₅PEI nanotubes surfaces are terminated by primary, secondary and tertiary amino groups of PEI and the pK_a values of these groups are around 9, 8 and 6–7, correspondingly.^23,24 For polyA₂₅, it always maintains negative charges in the pH tested for the phosphate group.²⁵ At pH below 9, amino groups were protonated to give a highly positively charged surface. As pH above 10, deprotonation made the decrease of adsorbed polyanion DNA.

Fig. 3 Effect of pH on polyA₂₅ adsorption onto (PEI/HSA)₅PEI nanotubes. ((PEI/HSA)₅PEI nanotubes dose = 0.1 mg, C₀ = 108 mg L⁻¹, t = 1 h).

3.3. Adsorption kinetics

The time-dependent adsorption kinetics of the (PEI/HSA)₅PEI nanotubes for polyA₂₅ was investigated. As shown in Fig. 4, the amount of adsorbed polyA₂₅ increased rapidly in initial time, and then increased gradually until reached the maximum. For lower concentration of 22 mg L⁻¹, the time to reach the equilibrium was about 30 min. while for concentrations of 58–108 mg L⁻¹, 50 min was required. For larger concentrations of 163–228 mg L⁻¹, a longer time of 300 min was necessary. The effect of initial polyA₂₅ concentration on adsorption was also observed. A rise of concentration from 22 mg L⁻¹ to 228 mg L⁻¹ caused an increase in the amount of equilibrium adsorption from 42.1 mg g⁻¹ to 392.0 mg g⁻¹, which was ascribed to more availability of polyA₂₅ to the active sites on nanotubes at higher concentration.

Fig. 4 Effect of adsorption time on polyA₂₅ adsorption onto (PEI/HSA)₅PEI nanotubes ((PEI/HSA)₅PEI nanotubes dose = 0.1 mg, C₀ = 22–228 mg L⁻¹, pH = 7.4 ± 0.2).

In order to understand the mechanism of adsorption kinetics, three types of kinetic model, the pseudo-first-order, pseudo-second-order and intra-particle diffusion models, were used to analyze the experimental dates above.

The pseudo-first-order and pseudo-second-order kinetics^26,27 are expressed as:

ln(q_e − q_t) = ln q_e − k₁t (3)

where k₁ (1/min) and k₂ (g/(mg min)) are the rate constants of pseudo-first-order and pseudo-second-order equations, respectively. q_e (mg g⁻¹) and q_t (mg g⁻¹) are the amount of adsorbed polyA₂₅ at equilibrium and t time, respectively.

As shown, Fig. 5A and B were the fit of the pseudo-first-order and pseudo-second-order models, respectively. The obtained kinetic parameters were listed in Table 1. It was found the pseudo-second-order equation fitted better than the pseudo-first-order model by comparing the correlation coefficients. The calculated values q_e,cal (i.e. 42.4, 111.1, 196.5, 325.7, 408.2 mg g⁻¹) from the former were close to the experimental values q_e,exp (i.e. 42.1, 106.1, 195.1, 315.0, 392.0 mg g⁻¹), suggesting that the pseudo-second-order kinetics more suitably described the adsorption of polyA₂₅.

Fig. 5 (A) The pseudo-first-order kinetic model, (B) the pseudo-second-order kinetic model, (C) the intra-particle diffusion model for polyA₂₅ adsorption onto (PEI/HSA)₅PEI nanotubes.

Table 1 Comparison of pseudo-first-order, pseudo-second-order kinetic and intra-particle diffusion models parameters, calculated values q_e,cal and experimental values q_e,exp for different initial concentrations of polyA₂₅ adsorption onto (PEI/HSA)₅PEI nanotubes

Model	Parameter	C0 (mg L^−1)
		22	58	108	163	228
	qe,exp (mg g^−1)	42.1	106.1	195.1	315.0	392.0
Pseudo-first-order	k1 (min^−1)	0.0384	0.0721	0.0652	0.0354	0.0238
	qe,cal (mg g^−1)	13.1	52.5	70.8	273.2	361.0
	R^2	0.7967	0.9149	0.8688	0.9746	0.9576
Pseudo-second-order	k2 (g/(mg min))	0.0147	0.0021	0.0015	0.0003	0.0002
	qe,cal (mg g^−1)	42.4	111.1	196.5	325.7	408.2
	R^2	0.9999	0.9997	0.9996	0.9993	0.9987
Intra-particle diffusion	ki1 (mg/(g min^0.5))	14.6	26.4	45.7	46.2	59.8
	R^2	0.9958	0.9866	0.9707	0.9753	0.9986
	ki2 (mg/(g min^0.5))	1.44	9.60	21.1	18.2	19.4
	R^2	0.9032	0.9292	0.9754	0.9502	0.9316
	ki3 (mg/(g min^0.5))	0.0204	0.0344	0.0399	1.11	2.89
	R^2	0.9382	0.9795	0.9363	0.9191	0.9503

---

The adsorption process was further analyzed using the intra-particle diffusion model²⁸ expressed as:

q_t = k_it^0.5 (5)

where k_i (mg/(g min^0.5)) is the rate constant of intra-particle diffusion. The value of k_i can be calculated from the slope of plot q_t versus t^0.5.

Three regions were seen from the plots of q_t versus t^0.5 at different initial concentrations of 22–228 mg L⁻¹, which represented the three consecutive mass transport steps in adsorption (Fig. 5C). The rates of adsorption at corresponding stages expressed as k_i1, k_i2, k_i3 could be derived from the slopes of linear portions. The first stage also the fastest (k_i1) was attributed to large amounts of polyA₂₅ diffusion to the external surface of nanotubes. The second portion (k_i2) was the stage of polyA₂₅ migrating into the pores. With concentration of polyA₂₅ decreasing in solution, the intra-particle diffusion gradually slowed down until reached the equilibrium stage (k_i3). As listed in Table 1, the rate was k_i1 > k_i2 > k_i3, and the values increased as initial concentration increased, since the intra-particle diffusion model was based on Fick's Law. k_i3 was larger at higher initial concentration and thus longer equilibrium time was needed, which well interpreted the experimental results of equilibrium time needed.

The correlation of rate parameter k_i1 and initial concentration²⁸ was represented as following, and the plot of ln k_i1 versus ln C₀ was shown in Fig. 6.

The values of constant were listed in Table 2.

Fig. 6 Plot of ln k_i1 against ln C₀ for polyA₂₅.

Table 2 Values of rate parameter constants in equation

Material	Parameter	Parameter value
PolyA25	An	2.40
	Bn	0.59
	R^2	0.9653

---

3.4. Adsorption isotherm

The Langmuir and Freundlich isotherm equations are the typical models to predict the adsorption type of adsorbate on an adsorbent. The Langmuir isotherm equation is based on monolayer adsorption on a homogeneous surface of adsorbent²⁹ and the linear form is expressed as:where C_e (mg L⁻¹) is the equilibrium concentration of polyA₂₅, q_e (mg g⁻¹) is the amount of adsorbed polyA₂₅ at equilibrium, b (L mg⁻¹) and Q⁰ (mg g⁻¹) are Langmuir constants related to adsorption energy and adsorption capacity, respectively. The values of Q⁰ and b can be calculated from the slope and intercept of plot C_e/q_e versus C_e.

A dimensionless equilibrium parameter (R_L),²⁹ which is represented as:

where b (L mg⁻¹) is Langmuir constant, C₀ (mg L⁻¹) is the highest polyA₂₅ concentration. The value of R_L means the type of isotherm to be either unfavorable (R_L > 1), linear (R_L = 1), favorable (R_L < 1) or irreversible (R_L = 0).

The Freundlich isotherm model is regarded as adsorption on a heterogeneous surface³⁰ and expressed as:

where K_F (mg g⁻¹) and n are the Freundlich constants related to adsorption capacity and adsorption intensity, respectively. The values of 1/n and K_F can be obtained from the slope and intercept of plot ln q_e versus ln C_e.

The isotherm curve of polyA₂₅ adsorption on (PEI/HSA)₅PEI nanotubes was presented in Fig. 7 and the fitting models by Langmuir and Freundlich isotherm equations were shown in Fig. 8. The derived parameters were summarized in Table 3.

Fig. 7 PolyA₂₅ adsorption isotherm of (PEI/HSA)₅PEI nanotubes. ((PEI/HSA)₅PEI nanotubes dose = 0.1 mg, C₀ = 22–500 mg L⁻¹, t = 8 h, pH = 7.4 ± 0.2).

Fig. 8 Fitting by (A) Langmuir model and (B) Freundlich model for polyA₂₅ adsorption onto (PEI/HSA)₅PEI nanotubes.

Table 3 Isotherm parameters for polyA₂₅ adsorption on (PEI/HSA)₅PEI nanotubes

Model	Parameter	Parameter value
Langmuir	Q^0 (mg g^−1)	606.0
	b (L mg^−1)	0.048
	R^2	0.9944
	RL	0.04
Freundlich	1/n	0.48
	KF (mg g^−1)	52.9
	R^2	0.8246

---

As seen, Langmuir model (R² = 0.9944) proved to be more fitted for describing the adsorption process, demonstrating that a monolayer adsorption occurred. The calculated maximum monolayer adsorption capacity (Q⁰) was 606.0 mg g⁻¹. The low value of b (<1) implied (PEI/HSA)₅PEI nanotubes had a high affinity to polyA₂₅. Additionally, polyA₂₅ adsorption onto nanotubes was also favorable reasoned from the value of R_L (<1).

3.5. Comparison adsorption capability with nanoparticles

Compared with spherule allophane nanoparticles (45.0 mg g⁻¹)³¹ and PEI-modified Fe₃O₄/Au nanoparticles (90.0 mg g⁻¹),³² the adsorption amount of (PEI/HSA)₅PEI nanotubes is 606.0 mg g⁻¹. This is the strongest DNA adsorption loader in all previous reports. The excellent absorptivity of (PEI/HSA)₅PEI nanotube benefits from its large surface area, unique hollow structure and sufficient DNA binding sites on both outer and inner surfaces. In addition, two open-end terminals make it possible that DNA diffuses into the tube and adsorbs on inner wall.

3.6. DNA release

The sample of DNA-loaded (PEI/HSA)₅PEI nanotubes with adsorbed amount of 195.1 mg g⁻¹ was used to investigate ionic strength induced release behaviours. As shown in Fig. 9, there was hardly any DNA released in the absence of NaCl. Once NaCl was added, the amount of DNA in solution quickly increased and gave a fast desorption kinetics. For example, by varying NaCl concentration from 0.1 M to 1.5 M, the release ratio increased from 5% to 67.1% at pH 7.4 and from 10% to 78.8% at pH 5.7 in 5 minutes, which might be in that the electrostatic interaction between polyA₂₅ and (PEI/HSA)₅PEI nanotubes was sensitive in a relatively high ionic strength.³³ At the same ionic strength, the ratio at pH 5.7 was much higher than that of at pH 7.4. This indicated that DNA release from (PEI/HSA)₅PEI nanotubes was also pH triggered, providing the necessary conditions for intracellular delivery.

Fig. 9 Kinetics of polyA₂₅ release from (PEI/HSA)₅PEI nanotubes at (A) pH 7.4 ± 0.2 and (B) pH 5.7 ± 0.2.

4. Conclusions

In this study, hybrid PEI/HSA nanotubes were successfully prepared with the maximum adsorption capability for polyA₂₅ (606.0 mg g⁻¹) at pH 7.4 and 25 °C. The adsorption kinetics and isotherm were discussed in detail and demonstrated to fit the pseudo-second-order kinetic, intra-particle diffusion and Langmuir isotherm model. DNA release from DNA/(PEI/HSA)₅PEI nanotubes complexes was mainly controlled by the ionic strength and also influenced by pH of solution. In summary, our researches would provide useful information to understand the interaction between DNA and protein nanotubes for design and development of protein nanotubes as gene vectors.

Acknowledgements

The authors gratefully acknowledge the National Science Foundation of China (21175105, 21375104 and 21327806), the Specialized Research Fund for the Doctoral Program of Higher Education of China (20126101110015), the Natural Science Foundation of Shaanxi Province of China (2014JM2050) and the Specialized Foundation of Education Bureau of Shannxi Province of China (12JK0634) for financial support.

Notes and references

1. E. Huang, F. Zhou and L. Deng, Langmuir, 2000, 16, 3272–3280.
2. A. Steel, R. Levicky, T. Herne and M. J. Tarlov, Biophys. J., 2000, 79, 975–981.
3. A. W. Peterson, R. J. Heaton and R. M. Georgiadis, Nucleic Acids Res., 2001, 29, 5163–5168.
4. N. Jirakittiwut, N. Panyain, T. Nuanyai, T. Vilaivan and T. Praneenararat, RSC Adv., 2015, 5, 24110–24114.
5. R. Qi, S. Liu, J. Chen, H. Xiao, L. Yan, Y. Huang and X. Jing, J. Controlled Release, 2012, 159, 251–260.
6. T. M. Sun, J. Z. Du, L. F. Yan, H. Q. Mao and J. Wang, Biomaterials, 2008, 29, 4348–4355.
7. C. R. Martin and P. Kohli, Nat. Rev. Drug Discov., 2003, 2, 29–37.
8. R. Singh, D. Pantarotto, D. McCarthy, O. Chaloin, J. Hoebeke, C. D. Partidos, J.-P. Briand, M. Prato, A. Bianco and K. Kostarelos, J. Am. Chem. Soc., 2005, 127, 4388–4396.
9. C. C. Chen, Y. C. Liu, C. H. Wu, C. C. Yeh, M. T. Su and Y. C. Wu, Adv. Mater., 2005, 17, 404–407.
10. S. J. Son, J. Reichel, B. He, M. Schuchman and S. B. Lee, J. Am. Chem. Soc., 2005, 127, 7316–7317.
11. A. Karlsson, R. Karlsson, M. Karlsson, A.-S. Cans, A. Strömberg, F. Ryttsén and O. Orwar, Nature, 2001, 409, 150–152.
12. S. A. Dougherty, D. Zhang and J. Liang, Langmuir, 2009, 25, 13232–13237.
13. X. Qu, G. Lu, E. Tsuchida and T. Komatsu, Chem.–Eur. J., 2008, 14, 10303–10308.
14. D. Zhang, S. A. Dougherty and J. Liang, J. Nanopart. Res., 2011, 13, 1563–1571.
15. X. Qu and T. Komatsu, ACS Nano, 2009, 4, 563–573.
16. T. Komatsu, X. Qu, H. Ihara, M. Fujihara, H. Azuma and H. Ikeda, J. Am. Chem. Soc., 2011, 133, 3246–3248.
17. S. B. Nimse, K. Song, M. D. Sonawane, D. R. Sayyed and T. Kim, Sensors, 2014, 14, 22208–22229.
18. X. Gao, K.-S. Kim and D. Liu, AAPS J., 2007, 9, E92–E104.
19. Y. Liu, D. C. Wu, W. D. Zhang, X. Jiang, C. B. He, T. S. Chung, S. H. Goh and K. W. Leong, Angew. Chem., 2005, 44, 4782–4785.
20. S. Curry, H. Mandelkow, P. Brick and N. Franks, Nat. Struct. Biol., 1998, 5, 827–835.
21. X. M. He and D. C. Carter, Nature, 1992, 358, 209–215.
22. J. Kong and S. Yu, Acta Biochim. Biophys. Sin., 2007, 39, 549–559.
23. C. R. Dick and G. E. Ham, J. Macromol. Sci., Part A: Pure Appl.Chem., 1970, 4, 1301–1314.
24. W. Godbey, K. K. Wu and A. G. Mikos, J. Biomed. Mater. Res., 1999, 45, 268–275.
25. M. Wu, R. Kempaiah, P. J. Huang, V. Maheshwari and J. Liu, Langmuir, 2011, 27, 2731–2738.
26. D. Wang, L. Liu, X. Jiang, J. Yu, X. Chen and X. Chen, Appl. Surf. Sci., 2015, 329, 197–205.
27. H. J. Yan, J. W. Bai, X. Chen, J. Wang, H. S. Zhang, Q. Liu, M. Zhang and L. H. Liu, RSC Adv., 2013, 3, 23278–23289.
28. W. H. Cheung, Y. S. Szeto and G. McKay, Bioresour. Technol., 2007, 98, 2897–2904.
29. B. H. Hameed, A. T. Din and A. L. Ahmad, J. Hazard. Mater., 2007, 141, 819–825.
30. A. A. Ahmad, B. H. Hameed and N. Aziz, J. Hazard. Mater., 2007, 141, 70–76.
31. Y. Matsuura, F. Iyoda, S. Arakawa, B. John, M. Okamoto and H. Hayashi, Mater. Sci. Eng., C, 2013, 33, 5079–5083.
32. H. Sun, X. Zhu, L. Zhang, Y. Zhang and D. Wang, Mater. Sci. Eng., C, 2010, 30, 311–315.
33. G. S. Irmukhametova, B. J. Fraser, J. L. Keddie, G. A. Mun and V. V. Khutoryanskiy, Langmuir, 2012, 28, 299–306.

---