PEGylation of silver nanoparticles by physisorption of cyclic poly(ethylene glycol) for enhanced dispersion stability, antimicrobial activity, and cytotoxicity

Silver nanoparticles (AgNPs) are practically valuable in biological applications. However, no steady PEGylation has been established, which is essential for internal use in humans or animals. In this study, cyclic PEG (c-PEG) without any chemical inhomogeneity is physisorbed onto AgNPs to successfully PEGylate and drastically enhance the dispersion stability against physiological conditions, white light, and high temperature. In contrast, linear HO–PEG–OH and MeO–PEG–OMe do not confer stability to AgNPs, and HS–PEG–OMe, which is often used for gold nanoparticles, sulfidates the surface to considerably degrade the properties. TEM shows an essentially intact nanostructure of c-PEG-physisorbed AgNPs even after heating at 95 °C, while complete disturbance is observed for other AgNPs. Molecular weight- and concentration-dependent stabilization by c-PEG is investigated, and DLS and ζ-potential measurements prove the formation of a c-PEG layer on the surface of AgNPs. Furthermore, c-PEG-physisorbed AgNPs exhibit persistent antimicrobial activity and cytotoxicity.


Introduction
The emerging importance of nanoscience cannot be overemphasized, and the eld of nanotechnology has drawn special worldwide attention especially in the cases of noble metals such as gold and silver. 1 The distinct properties of silver nanoparticles (AgNPs) have led to their broad applications in medical imaging, 2 drug delivery, 3 cell electrodes, 4 biosensors, 5 cancer diagnosis and treatment, 6 and cytotoxic agents 7 as well as antimicrobial agents against a broad spectrum of Gram-negative and Gram-positive bacteria. 8 The nano-environment of AgNPs has notable effects on their response/activity in many applications. However, unlike gold nanoparticles (AuNPs), AgNPs are not a stable material and susceptible to light, dissolved electrolytes, and various chemical species, and transformations occur leading to aggregation, dissolution, change in structure, activity loss, etc. 9 In diverse elds of nanoscience, the instability of AgNPs oen hinders their applications and commercialization. Although several capping agents have been explored for AgNPs, transformations, dissolution and agglomerations in various environments still remain a signicant issue.
The use of stabilizers such as cetyltrimethylammonium bromide (CTAB), 10,11 sodium dodecyl sulfate (SDS), 10,12 and other surfactants has been well reviewed with limitations of non-biocompatibility and instability. The utmost crucial factors of stability and biocompatibility have been desirable and sought aer in the application of nanoparticles. In this regard, poly(ethylene glycol) (PEG), a non-ionic polymer with a exible structure, is the most commonly used biocompatible polymer accepted by the United States Food and Drug Association and has wide applications including food, commodities, and drugs as well as uses in agriculture and manufacturing industries. [13][14][15][16] In order to use PEG as a stabilizer for metal nanoparticles, especially AuNPs, thiol-functionalized PEG (HS-PEG-OMe) is oen employed through the chemisorption between the sulfur atom and metal surface. 17 However, the use of HS-PEG-OMe for AgNPs results in the formation of a silver sulde (Ag 2 S) layer on the surface drastically disturbing the nanoparticles' properties 18 and leading to dissolution, 19,20 inhibiting PEGylation of AgNPs by the thiol chemisorption. A few reports show that AgNO 3 is

Synthesis of c-PEG
Cyclization of HO-PEG-OH was carried out by the previously reported method. 52 Thus, a solution of vacuum dried HO-PEG-OH (5.0 g) and tosyl chloride (190 mg) in 100 mL of dry THF was added over 144 h to a dispersion of powdered potassium hydroxide (3.3 g) in 100 mL of a mixture of dry THF and nheptane (75/25 v/v) at 40 C using a syringe pump in a dry nitrogen atmosphere. Additional 24 h of stirring was allowed for complete cyclization. The reaction mixture was ltered, and the solvent was removed under reduced pressure. Chloroform was added to the ltrate, washed with brine followed by deionized water. The organic phase was dried with magnesium sulfate, and the solvent was removed under reduced pressure. Silica gel column chromatography was carried out with a mixture of chloroform/acetone (9/1 v/v) to elute polymeric products that underwent intermolecular reactions, followed by a mixture of chloroform/methanol (9/1 v/v) to elute the crude containing c-PEG. The crude was dissolved in dichloromethane, and nheptane was slowly added until the solution turned cloudy. The cloudy solution was heated to 40 C and cooled to 25 C with two resultant layers. The upper clear layer containing a relatively large proportion of c-PEG was collected, and this procedure was repeated several times to obtain pure c-PEG as a white solid. The yields of c-PEG 2k , c-PEG 3k , and c-PEG 9k were 303 (6.1%), 247 (4.9%), and 145 mg (2.9%), respectively. Concerning the very low isolated yields, the purity was prioritized over the yield, resulting in a major loss during the isolation.

Synthesis of MeO-PEG-OMe
Methylation of HO-PEG-OH was carried out according to the previous method. 52 Typically, in a dry nitrogen atmosphere, chlorobenzene (20 mL) was added to nely powdered potassium hydroxide (2.3 g), and the mixture was stirred at 25 C. Iodomethane (0.35 g) was added to the mixture, followed by slow addition of HO-PEG 2k -OH (5.0 g) dissolved in chlorobenzene (50 mL) over 25 min. The mixture was stirred for 24 h. Filtration was carried out, and the ltrate was reduced to a small volume under reduced pressure, followed by washing with distilled water and deionized water. The organic phase was dried with magnesium sulfate and concentrated under reduced pressure. The residue was applied to a silica gel column with a mixture of chloroform/acetone (9/1 v/v), and the product was eluted with a mixture of chloroform/methanol (9/1 v/v). The solvent was removed and vacuum dried to obtain dimethylated poly(ethylene glycol) (MeO-PEG 2k -OMe) (3.6 g, 72%) as a white solid.

NMR
1 H NMR (400 MHz) and 13 C NMR (100 MHz) were recorded on a JEOL JNM-ESC400 instrument at room temperature at a polymer concentration of 20 mg mL À1 . Deuterated chloroform was used as a solvent.

SEC
Size exclusion chromatography measurements were performed on a Shodex GPC-101 gel permeation chromatography system (Shodex DU-2130 dual pump, Shodex RI-71 reective index detector, and Shodex ERC-3125SN degasser) equipped with a Shodex KF-G guard column (4.6 mm Â 10 mm; pore size, 8 mm) and two Shodex KF-804L columns (8 mm Â 300 mm) in series. THF was used as an eluent at a ow rate of 1.0 mL min À1 . Calibration was performed with PEG standard samples.

Recycling preparative SEC
A Japan Analytical Industry LC-908 recycling preparative HPLC system (Hitachi L-7110 pump and JAI RI detector RI-5) was used. JAIGEL-2H and 3H columns and a pre-column were connected in series. Chloroform was used as a solvent, and the ow rate was set at 3.5 mL min À1 .

UV-Vis spectroscopy
UV-Vis absorption spectra were recorded using a JASCO Ubset V-670 spectrophotometer at 25 C in a micro quartz cuvette (M25-UV2, GL Science Inc., Japan) with a path length of 10 mm. Deionized water was used as a blank. Spectra were acquired at a wavelength range of 300-800 nm. Optical density at 600 nm (OD 600 ) in the antimicrobial activity experiment was determined by the intensity of an incubated specimen at 600 nm with that of the medium subtracted.

DLS and z-potential
DLS and z-potential measurements were carried out using a Zetasizer Nano ZS instrument (He-Ne laser, 633 nm, max 4 mW, Malvern Panalytical Ltd). A micro quartz cuvette (ZEN2112, Hellma Analytics) and Zetasizer nano cell (DTS1060, Malvern Instruments, Ltd) were used. Measurements were carried out at 25 C with a 120 s equilibration time. A cumulant analysis performed using the inbuilt soware of the instrument was used to determine the z-average size.

c-PEG's molecular weight-dependent stability
An aqueous dispersion of AgNPs 10 (0.54 mL) was added to c-PEG 2k , c-PEG 3k , or c-PEG 9k (0.15 mg) in a 1.5 mL Eppendorf tube, and the mixture was vortexed for 1 min to form AgNPs 10 /c-PEG 2k , AgNPs 10 /c-PEG 3k , or AgNPs 10 /c-PEG 9k , respectively. Subsequently, the tenfold-concentrated PBS solution (0.06 mL) was added to the cuvette, and the resulting mixture was 0.6 mL with pH 7.4 and a NaCl concentration of 150 mM with a PEG concentration of 0.25 wt%. A time-course UV-Vis measurement was performed for 1000 min.

c-PEG's concentration-dependent stability
An aqueous dispersion of AgNPs 10 (0.54 mL) was added to c-PEG 9k (0.3, 1.5, 3.0, or 7.5 mg) in a 1.5 mL Eppendorf tube, the mixture was vortexed for 1 min to form AgNPs 10 /c-PEG 9k . Subsequently, the tenfold-concentrated PBS solution (0.06 mL) was added to the cuvette, and the resulting mixture was 0.6 mL with pH 7.4 and a NaCl concentration of 150 mM with a PEG concentration of 0.05, 0.25, 0.40, or 1.25 wt%, respectively. A time-course UV-Vis measurement was performed for 1000 min.
2.14 AgNPs' size-dependent stability c-PEG 9k (1.5 mg) was added to an aqueous dispersion of AgNPs 10 , AgNPs 20 , or AgNPs 30 (0.54 mL), and the mixture was vortexed for 1 min. Subsequently, the tenfold-concentrated PBS solution (0.06 mL) was added to the cuvette, and the resulting mixture was 0.6 mL with pH 7.4 and a NaCl concentration of 150 mM with a PEG concentration of 0.25 wt%. A time-course UV-Vis measurement was performed for 1000 min.

Stability against white light
An aqueous dispersion of AgNPs 10 (2.1 mL) was two times diluted with deionized water (2.1 mL) and added to HO-PEG 9k -OH, MeO-PEG 9k -OMe, HS-PEG 9k -OMe, or c-PEG 9k (10.5 mg) in a 50 mL Falcon tube. The mixture was vortexed for 1 min to form AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, or AgNPs 10 /c-PEG 9k , respectively, where the PEG concentration was 0.25 wt%. AgNPs 10 /No PEG was prepared by diluting AgNPs (2.1 mL) with deionized water (2.1 mL) and vortexed for 1 min. The mixtures were kept under 860-990 lux light intensity using a white light emitting tube at 25 C for 35 d. 0.60 mL of the mixtures was withdrawn from the Falcon tubes immediately aer mixing (day 0) and subsequently at a 7 day interval for an absorption measurement. The measured samples were not returned to the Falcon tubes.

TEM
A few drops from the above AgNPs 10 /No PEG, AgNPs 10 /HS-PEG 9k -OMe, or AgNPs 10 /c-PEG 9k heated at 95 C for 4 h were placed on a carbon coated Formvar TEM grid and air blown with a blower. Measurements were performed with a Japan Electron Optics Laboratory JEM-2010 operated at 200 kV.

Antimicrobial activity
E. coli was grown in a Muller Hinton Broth (MHB) medium containing ampicillin (100 mg mL À1 ) at 37 C for 24 h and standardized using 0.5 McFarland standard (10 8 CFU mL À1 ). A tenfold-concentrated PBS solution (300 mL, pH 7.4, NaCl 1500 mM) was added to AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, or AgNPs 10 /c-PEG 9k (2.7 mL) with a PEG concentration of 0.25 wt% and incubated for 24 h. The resulting mixture was centrifuged at 3000 rpm for 20 min, and the supernatant or dispersion was ultraltered to reduce the volume to 100 mL and mixed with 10 mL of 10 5 CFU mL À1 E. coli in a 1.5 mL Eppendorf tube. The mixture was added to a test tube containing an MHB medium (2.9 mL) and incubated at 37 C and 200 rpm for 24 h. UV-Vis absorption spectra were recorded.

Synthesis of c-PEG and MeO-PEG-OMe
HO-PEG-OH with a molecular weight of 2, 3, and 9 kDa (HO-PEG 2k -OH, HO-PEG 3k -OH, and HO-PEG 9k -OH, respectively) was successfully cyclized by etherication. Thus, the chain ends of HO-PEG-OH were intramolecularly connected in the presence of tosyl chloride and potassium hydroxide in dilution.
Highly pure c-PEG 2k , c-PEG 3k , and c-PEG 9k were obtained aer column chromatography and repeated separation using dichloromethane and n-heptane. SEC of c-PEG showed a unimodal trace with a peak shi to the lower molecular weight region compared to the prepolymer HO-PEG-OH (Fig. S1 †). The  Paper Nanoscale Advances decrease in the hydrodynamic volume resulting from cyclization caused the shi in the apparent molecular weight. For example, M p,SEC decreased from 9640 for HO-PEG 9k -OH to 6040 for c-PEG 9k (Table 1 and Fig. S1c †). 13 C NMR spectra showed a complete disappearance of the peaks at 61.8 and 72.5 ppm from the carbon atoms adjacent to the hydroxyl end groups of HO-PEG-OH, thus conrming the elimination of the chain ends (Fig. S2 †). 1 H NMR of c-PEG also gave a single peak unlike that of HO-PEG-OH, which showed the distinguishable signals from the methylene protons close to the chain ends (Fig. S3 †). MALDI-TOF mass spectrometry of c-PEG and HO-PEG-OH further gave a striking difference in their isotopic distributions (Fig. S4 †) 44 + Ag] + with the difference arising from the elimination of a water molecule. However, a MALDI-TOF mass spectrum for PEG 9k was not obtainable due to its large molecular weight. The expected diameter of c-PEG 2k , c-PEG 3k , and c-PEG 9k was 4.5, 6.8, and 22 nm, respectively, when they form an ideal right circular conformation. Furthermore, methylation of HO-PEG-OH resulted in the successful synthesis of MeO-PEG-OMe with features similar to its precursor HO-PEG-OH in terms of the appearance of the carbon atoms adjacent to the methoxy end group (Fig. S2 †). UV-Vis spectroscopy showed no absorbance from any of these PEGs at 300-800 nm wavelength (Fig. S5 †), suggesting that the PEG samples are free of impurity and suitable for optical investigations of AgNPs.

Physisorption of c-PEG to AgNPs
According to the procedure we previously established, 43 HO-PEG-OH, MeO-PEG-OMe, HS-PEG-OMe, or c-PEG with a molecular weight of 2, 3, or 9 kDa (PEG 2k , PEG 3k , or PEG 9k , respectively), was simply mixed with an aqueous dispersion of AgNPs with a size of 10, 20, or 30 nm (AgNPs 10 , AgNPs 20 , or AgNPs 30 , respectively). The addition of HO-PEG-OH, MeO-PEG-OMe, or c-PEG to AgNPs had no effect on the surface plasmon resonance (SPR) shown in Fig. 2a. The UV-Vis absorption spectra and visual color were almost identical to those of the AgNPs dispersion without PEG, and l max remained at 398 nm. On the other hand, the addition of HS-PEG 9k -OMe abruptly reduced the absorption, broadened the peak, and deepened the yellow color of the AgNPs dispersion to yellowish brown. This change in SPR is explained as being a result of increase in the local dielectric environment upon thiol coordination to the Ag surface. 19 DLS and z-potential measurements proved the distinct formation of a c-PEG layer on the surface of AgNPs as in the case of AuNPs. 43 Thus, by DLS, AgNPs 10 /No PEG had a size of 18 nm, which on addition of HO-PEG 9k -OH and MeO-PEG 9k -OMe slightly enlarged to 24 and 20 nm, respectively ( Fig. 1 and Table  2). On the other hand, a signicant increase was seen for AgNPs 10 /HS-PEG 9k -OMe with 71 nm and AgNPs 10 /c-PEG 9k with 109 nm. Moreover, an increase in the size with increase in the molecular weight of c-PEG was evident. In the case of AgNPs 10 , complexation with c-PEG 2k , c-PEG 3k , and c-PEG 9k resulted in 35, 77, and 109 nm in size, respectively. This molecular weight dependence is consistent with the previously reported adsorption of cyclic PEG and cyclic poly(dimethylsiloxane) on silica. 45,46 When AgNPs 20 and AgNPs 30 were used with c-PEG 9k (73 and 71 nm, respectively), the increase in size was less intense compared to AgNPs 10 (109 nm). Thus, c-PEG 9k with a diameter of 22 nm in the ideal right circular conformation exhibited the strongest interaction with AgNPs 10 with a size of 18 nm.
Due to citrate anions existing at the surface of AgNPs, the zpotential of AgNPs 10 /No PEG showed a negative value of À31 mV, which on addition of non-ionic PEG reduced to À25 mV for AgNPs 10 /HO-PEG 9k -OH and À16 mV for AgNPs 10 /MeO-PEG 9k -OMe (Table 3). A neutral PEG layer on the surface was reported to shield the negative charges of citrate thus decreasing the magnitude of the z-potential. 53 A near zero potential was seen for AgNPs 10 /HS-PEG 9k -OMe and AgNPs 10 /c-PEG 9k with À2 mV. This suggests that chemisorption of HS-PEG 9k -OMe and physisorption of c-PEG to the surface of AgNPs shield the charge more efficiently. A signicant decrease in magnitude of the z- a Units are in mV.
potential by the addition of c-PEG was also previously observed in AuNPs. 43 Furthermore, an increase in molecular weight led to a reduction in the z-potential: AgNPs 10 /c-PEG 2k gave À12 mV, AgNPs 10 /c-PEG 3k was À6 mV, and a further reduction to À2 mV in AgNPs 10 /c-PEG 9k . Thus, a thicker layer formed by c-PEG with a higher molecular weight shielded the charge more effectively. The adsorption of c-PEG on AgNPs is probably an enthalpically favorable and entropically unfavorable process. Because the number of conformations of c-PEG in the unadsorbed state is limited compared to that of HO-PEG-OH and MeO-PEG-OMe, the entropic loss upon the adsorption of c-PEG is expected to be smaller. [45][46][47][48][49][50][51] On the other hand, the adsorption enthalpy would be similar for both c-PEG and the linear counterparts because they have the same chemical structure of the repeating units and the same molecular weight. Based on this, the total adsorption free energy change is likely larger in negative value for c-PEG.

Enhancement of the dispersion stability
Subsequently, the dispersion stability of AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 / HS-PEG 9k -OMe, and AgNPs 10 /c-PEG 9k with a PEG concentration of 0.25 wt% under physiological conditions was investigated. On addition of a tenfold-concentrated PBS solution (0.06 mL) to each AgNPs dispersion (0.54 mL) to form the intended physiological conditions (pH 7.4 and a NaCl concentration of 150 mM), there was an immediate color change from yellow to light brown in the case of AgNPs 10 /No PEG and AgNPs 10 /HO-PEG 9k -OH or to dark brown for AgNPs 10 /MeO-PEG 9k -OMe ( Fig. 2b and ESI Movie 1 †), which was followed by precipitation. The dark yellow color of AgNPs 10 /HS-PEG 9k -OMe remained unchanged. Remarkably, AgNPs 10 /c-PEG 9k exhibited only a slight color change and nearly retained the initial yellow color even aer 1000 min. UV-Vis spectroscopy showed only a minor decrease in the absorption of AgNPs 10 /c-PEG 9k on addition of the tenfold-concentrated PBS solution and aer 1000 min, whereas there was essentially no absorption from AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, or AgNPs 10 /MeO-PEG 9k -OMe due to precipitation (Fig. 2c). The relative absorption intensity (Rel. Abs) calculated by dividing the absorption value at l max aer 1000 min by that before the addition of PBS was 75% for AgNPs 10 /c-PEG 9k . The presence of NaCl in PBS has been well reviewed as a causative factor of aggregation of metal nanoparticles, but properly performed PEGylation can avoid aggregation. 54 Thus, c-PEG 9k protected AgNPs by physisorption to the surface thereby preventing agglomeration in the presence of increased ionic strength. In the meantime, a signicant reduction in the UV-Vis absorption spectra of AgNPs 10 /HS-PEG 9k -OMe was evident aer 1000 min (Rel. Abs ¼ 48%), which was likely caused by dissolution of AgNPs in the presence of thiol. 18 Subsequently, the stabilization effect depending on the molecular weight of c-PEG was investigated. Fig. 3a shows Rel. Abs versus time for AgNPs 10 /c-PEG 2k , AgNPs 10 /c-PEG 3k , and AgNPs 10 /c-PEG 9k in a PBS buffer solution with pH 7.4 and a NaCl concentration of 150 mM. A signicant increase in the Nanoscale Advances dispersion stability was observed with an increase in the molecular weight. AgNPs 10 /c-PEG 9k retained 74% of Rel. Abs aer 1000 min, while AgNPs 10 /c-PEG 3k retained only 39%, and AgNPs 10 /c-PEG 2k also showed a weak stabilization with Rel. Abs of 37%. Moreover, the concentration of c-PEG 9k was varied from 0.05 to 1.25 wt% (Fig. 3b). At 0.05 wt%, a continuous decrease in absorbance was seen, and Rel. Abs was 19% aer 1000 min. In the meantime, at 0.25 wt% and higher concentrations, there was a much smaller change in time course (Rel. Abs $ 76% aer 1000 min). At a c-PEG concentration of 0.05 wt%, an insufficient amount of c-PEG existed in the dispersion as the surface of AgNPs was scarcely covered. At the concentration of 0.25 wt% and higher, the amount of c-PEG was satisfactory to form a thick enough layer on the surface of AgNPs, thereby enhancing the dispersion stability. The c-PEG layer thickness was likely saturated at 0.25 wt%, and this phenomenon was previously observed for AuNPs. 43 Furthermore, the stability in relation to the size of AgNPs was also examined. Fig. 3c shows that c-PEG 9k can stabilize AgNPs 10 , AgNPs 20 , and AgNPs 30 , which have a size of 10, 20 and 30 nm, respectively. However, stability was best conferred on AgNPs 10 /c-PEG 9k with Rel. Abs of 82% aer 1000 min compared to AgNPs 20 /c-PEG 9k (Rel. Abs ¼ 60%) and AgNPs 30 /c-PEG 9k (Rel. Abs ¼ 57%). We showed above that the DLS size and z-potential were dependent on the molecular weight and AgNPs' size (Tables 2 and 3); c-PEG 9k formed a thicker layer than c-PEG 2k and c-PEG 3k , on AgNPs 10 compared to AgNPs 20 and AgNPs 30 . The dispersion stability against the physiological conditions was in accord with the DLS size and zpotential; the thicker the PEG layer that forms on AgNPs, the better the dispersion stability is. In accordance with these results, the following experiments were mainly performed with AgNPs 10 /PEG 9k with a polymer concentration of 0.25 wt%. Moreover, divalent ionic salts have been reported to exert stronger dissolution and agglomeration effects on nanoparticles than the monovalent counterparts. [55][56][57] On account of stabilization conferred to AgNPs by c-PEG 9k against PBS with its main constituent as NaCl, we investigated the dispersion stability of AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 / MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, and AgNPs 10 /c-PEG 9k with a PEG concentration of 0.25 wt% against a 10 mM CaCl 2 solution. Similar to the case of the PBS experiment, c-PEG conferred stability to AgNPs aer 1000 min with Rel. Abs of 78% (Fig. S6 †). On the other hand, AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, and AgNPs 10 /MeO-PEG 9k -OMe precipitated with Rel. Abs $0%, while AgNPs 10 /HS-PEG 9k -OMe with a shied and broadened spectrum caused decrease in the absorption (Rel. Abs ¼ 45%). This further proved the strong dispersion stability endowed by c-PEG.
It was reported that the dissolution, aggregation, and secondary phase precipitation of AgNPs are caused by photoirradiation. 58 Thus, in production, storage, and applications, light is a well-known limiting factor which causes transformational changes of AgNPs. 59,60 Also, emphasis is always made in the safety data sheets with respect to light. Thus, we tested the stability endowed by c-PEG against photoirradiation. AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, and AgNPs 10 /c-PEG 9k with a PEG concentration of 0.25 wt% were exposed to white light at 860-990 lux (Fig. 4). AgNPs 10 /No PEG (Fig. 4a), AgNPs 10 / HO-PEG 9k -OH (Fig. 4b), and AgNPs 10 /MeO-PEG 9k -OMe (Fig. 4c) showed an initial reduction in absorbance at 398 nm, followed by the appearance of a new peak at 550 nm for AgNPs/ No PEG and 460-495 nm for AgNPs 10 /HO-PEG 9k -OH and AgNPs 10 /MeO-PEG 9k -OMe, which intensied with time. The appearance of AgNPs 10 /No PEG (black-marked tube), AgNPs 10 / HO-PEG 9k -OH (blue-marked tube), and AgNPs 10 /MeO-PEG 9k -OMe (green-marked tube) changed from light yellow to deep yellow, which intensied as the days progressed via the aggregation of AgNPs (Fig. 4f-h). The strong oscillating dipole-dipole interaction by photoirradiation reportedly causes aggregation. 61 On the other hand, no obvious change in the UV-Vis spectra in Fig. 4e suggests a superior protection of AgNPs by c-PEG 9k even aer 35 d of exposure to white light, and there was no color change as it remained yellow (red-marked tube in Fig. 4f-h). This was likely because the thick c-PEG layer on the surface inhibited the contact of AgNPs and thus prevented the aggregation. In the meantime, AgNPs 10 /HS-PEG 9k -OMe resulted in reduced absorption of AgNPs immediately aer mixing with HS-PEG 9k -OMe as seen above, 19 and subsequently gave nearly no absorption from the SPR on day 7 and later (Fig. 4d). The color changed from brownish yellow (orange-marked tube in Fig. 4f) to gray (Fig. 4g and h), which eventually turned to colorless. Moreover, it was also reported that AgNPs coated with PVP cannot withstand photoirradiation, resulting in aggregation. 61 That is to say, the unique topology of c-PEG allows for physisorption to protect AgNPs from degradative reactions taking place at the surface caused by white light exposure.
It is also known that temperature is another important factor for the aggregation of AgNPs, 61 and stability against temperature would offer opportunities in various applications such as photothermal therapy. 62,63 Thus, we further tested the stabilization by c-PEG against heating. Keeping AgNPs 10 /No PEG (Fig. 5a), AgNPs 10 /HO-PEG 9k -OH (Fig. 5b), AgNPs 10 /MeO-PEG 9k -OMe (Fig. 5c), or AgNPs 10 /c-PEG 9k (Fig. 5e) at 4 and 37 C for 4 h gave no change in the absorption spectra. On the other hand, AgNPs 10 /HS-PEG 9k -OMe (Fig. 5d) showed a red shi and reduced absorption intensity at 4 C as in the cases of the stability tests against PBS and CaCl 2 . 18 Further decrease in the absorption spectra of AgNPs 10 /HS-PEG 9k -OMe was seen when kept at 37 C with a brownish appearance (orange-marked tube in Fig. 5g). When the temperature was raised to 95 C, AgNPs 10 / No PEG, AgNPs 10 /HO-PEG 9k -OH, and AgNPs 10 /MeO-PEG 9k -OMe aer 4 h resulted in Rel. Abs of 42%, 76%, and 72%, respectively, while that of AgNPs 10 /HS-PEG 9k -OMe was only 5% (Fig. 5a-d). In contrast, AgNPs 10 /c-PEG 9k showed an insignicant change with a Rel. Abs of 98% under the same conditions (Fig. 5e). Aer heating at 95 C for 4 h, AgNPs 10 /No PEG and AgNPs 10 /HS-PEG 9k -OMe turned colorless, AgNPs 10 /HO-PEG 9k -OH and AgNPs 10 /MeO-PEG 9k -OMe gave a slight faint yellow color, whereas AgNPs 10 /c-PEG 9k remained in the original yellow color (Fig. 5h). Moreover, commercial AgNPs 80 /HS-PEG 5k -OMe was also heated at 95 C for 4 h, and there was no conferment of stability to AgNPs as reduction in absorption intensity with disappearance of the yellow color of AgNPs was seen (Fig. S7 †). This was in tandem with the above result for AgNPs 10 /HS-PEG 9k -OMe and further proved the inability of HS-PEG to stabilize AgNPs.
TEM measurement of AgNPs 10 /No PEG, AgNPs 10 /HS-PEG 9k -OMe, and AgNPs 10 /c-PEG 9k kept at 95 C for 4 h explained the effect of heating (Fig. 6). TEM photographs of AgNPs 10 /No PEG showed aggregated AgNPs likely caused by the dipole-dipole interaction enhanced at the high temperature (Fig. 6d). 61 AgNPs 10 /HS-PEG 9k -OMe aer heating drastically also changed its form; particles with reduced size (#5 nm) along with stainlike objects with a few hundred nanometer in size were observed. The coordination of thiol to the AgNPs surface leads to an Ag 2 S layer. 18,19 At high temperature, dissociation of Ag 2 S from AgNPs was stimulated to reduce the median particle size shown in Fig. 6e. Moreover, the large stain-like objects were likely Ag 2 S aggregated upon drying. What needs to be emphasized here is that AgNPs 10 /c-PEG 9k was intact aer heating with no signicant change in size, and the particles were still well dispersed (Fig. 6f). Because the nanoparticles were separated from each other by the c-PEG layer formed on the surface, aggregation of AgNPs did not take place to retain the original size and shape. Concerning this, the nanoparticles of AgNPs 10 /c-PEG 9k observed in Fig. 6c and f were separated from each other likely by the c-PEG layer, compared to those of AgNPs/No PEG in contact with each other as shown in Fig. 6a. These experiments demonstrate how AgNPs/No PEG and AgNPs/HS-PEG-OMe are degraded at high temperature, and c-PEG serves as an effective protection for AgNPs.

Biological applications
Antimicrobial activity is inherent and one of the most important properties of AgNPs, and that against Gram-negative Escherichia coli (JM 109) in a Muller Hinton Broth (MHB) medium was evaluated. Thus, AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, and AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, and AgNPs 10 /c-PEG in PBS (pH 7.4, NaCl 150 mM) were added to an E. coli-containing medium. Upon addition of AgNPs 10 /c-PEG to E. coli, immediate change in the color to brownish was observed, while the other mixtures did not cause the change. Aer 24 h of incubation, the E. coli-containing medium with AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, and AgNPs 10 /MeO-PEG 9k -OMe became cloudy, suggesting the growth of E. coli (Fig. 7a). For AgNPs 10 /HS-PEG 9k -OMe, a gradual disappearance of the transparent yellow color of AgNPs to form a cloudy solution was evident, which also suggests loss of antimicrobial activity. However, the medium with AgNPs 10 /c-PEG 9k was transparent and remained brownish in color even aer 24 h, showing inhibited growth of E. coli. Furthermore, UV-Vis spectra were recorded to show transparency of the AgNPs 10 /c-PEG 9k specimen with a clearly observable SPR absorption peak, while the other specimens were turbid with a considerably increased baseline through scattering by grown E. coli (Fig. 7b). The growth of E. coli was quantied by optical density at 600 nm (OD 600 ): AgNPs 10 /No PEG, 0.79; AgNPs 10 /HO-PEG 9k -OH, 0.63; AgNPs 10 /MeO-PEG 9k -OMe, 0.82; AgNPs 10 /HS-PEG 9k -OMe, 0.86; AgNPs 10 /c-PEG 9k , 0.08. This revealed that the antimicrobial efficacy was preserved in AgNPs 10 /c-PEG 9k but lost in all other specimens. Because HO-PEG 9k -OH and MeO-PEG 9k -OMe could not maintain the dispersibility of AgNPs in PBS, the antimicrobial potency was quenched before the addition to E.
coli. In the case of AgNPs 10 /HS-PEG 9k -OMe, no precipitate formed, but the suldation of AgNPs nullied the antimicrobial efficacy. In contrast, physisorption of c-PEG exhibited an improved dispersion stability of AgNPs and evidently retained the antimicrobial efficacy.
Following the various dispersion stability and antimicrobial activity experiments, further evaluation of the biological properties of AgNPs/c-PEG via cytotoxicity and scratch assay experiments using HeLa cells were performed. Fig. S8 † shows the  result of cytotoxicity of AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, and AgNPs 10 /c-PEG 9k aer incubation with HeLa cells in a DMEM medium for 24 h. AgNPs 10 /c-PEG 9k had the lowest cell viability of 79% which was statistically signicant (p < 0.05) compared to AgNPs 10 /No PEG with a viability of 85% with sextuplicate experiments. This suggests that c-PEG helps in the dispersion of AgNPs in the medium and preserves the cytotoxicity. On the other hand, no cytotoxicity was seen in AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, and AgNPs 10 /HS-PEG 9k -OMe likely due to precipitation under the conditions.
In addition, migration and recovery of HeLa cells were evaluated by a cell scratch assay. A conuent monolayer was scratched on AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, AgNPs 10 /HS-PEG 9k -OMe, and AgNPs 10 /c-PEG 9k aer 2 h of incubation (Fig. S9 †). The scratches in AgNPs 10 /No PEG, AgNPs 10 /HO-PEG 9k -OH, AgNPs 10 /MeO-PEG 9k -OMe, and AgNPs 10 /HS-PEG 9k -OMe specimens were recovered to some extent aer 22 h. However, most of the HeLa cells in AgNPs 10 /c-PEG 9k were stripped off from the plates upon scratching, and basically no recovery was observed. This could be explained by AgNPs 10 /c-PEG 9k leading to cell death in the large area, thereby inhibiting adhesion of the cells. These results further conrmed the cytotoxicity through the enhanced dispersion stability of AgNPs by c-PEG.

Conclusions
Our research has shown the rst steady PEGylation method for AgNPs conferred by physisorption of c-PEG, which cannot be attained with HS-PEG-OMe due to the formation of silver sulde. Physisorption of c-PEG provided outstanding dispersion stability to AgNPs, against physiological conditions, white light, and high temperature, whereas HO-PEG-OH or MeO-PEG-OMe did not provide such dispersion stability. This method further exhibited persistent antimicrobial activity and cytotoxicity, which are two of the most important properties of AgNPs. Coupled with the excellent biocompatibility of PEG and the simple physisorption method, c-PEG would pave the way and broaden the uses of AgNPs especially in the biological and medicinal elds. Moreover, as we previously proved that the physisorption of c-PEG can also enhance the dispersion stability of AuNPs, 43 the present method has great potential for application in a wide variety of metal nanoparticles.

Conflicts of interest
The authors declare no competing interest.