Potassium-Induced Emission Enhancement of Bovine Serum Albumin-Stabilized Red-Emitting Au Nanoclusters: Mechanism and Application to Blood Plasma

Aggregation-induced emission (AIE) enhancement is attractive for bioimaging as it offers higher quantum yields (QYs) from uorophores via modulation of their immediate environment. Fluorophores with high QYs are essential probes for investigating the spatiotemporal distribution of physiologically important metal ions such sodium and potassium. Potassium ions are vital for normal function of living organisms. The research reported here evaluates the emission intensity of the bovine serum albumin-stabilized red-emitting gold nanoclusters (BSA-rAuNCs) upon uptake of K + . The integrated sphere method was used to determine the absolute QYs and a custom-built uorescence setup recorded the emission spectra of BSA-rAuNCs. Enhancement of emission intensity was observed upon increasing K + concentration, within the physiological concentration range of 0 - 150 mM. The emission enhancement was correlated with the particle size and charge analysis. Aggregation of BSA-rAuNCs was found to be responsible for the observed emission enhancement. The QY of BSA-rAuNCs in bovine blood plasma was found to be four times lower than corresponding QY in water.


Introduction
The small size and tunable emission properties of gold nanoclusters (AuNCs) have triggered research interest in the potential applications of AuNCs in bioimaging, biosensing and cancer therapy [1][2][3]. The emission quantum yield (QY) and stability of AuNCs depends on the nature of stabilizing molecules and the properties of the solvents used with them [4]. Thiols have often been used to stabilize AuNCs due to the strong a nity of sulphur for gold. Stabilization with dendrimer molecules resulted in much higher QYs, but an increased ensemble size and low compatibility with aqueous solutions have been a major obstacle to in vivo imaging using highly emissive AuNCs [5]. Protein scaffolds offer another biocompatible way to stabilize AuNCs [6]. Among various proteins, albumins have been widely used due to their known ion-and drug-binding properties. Recently, our group showed that bovine serum albumin (BSA) can hold several Au 8 NCs. BSA-encapsulated blue-emitting Au 8 NCs were found to exhibit twophoton absorption which could be advantageous for in-vivo imaging [7]. The interaction of BSA-Au 8 NCs with lipid membranes provided insights into the conformation changes that BSA-Au undergoes when in contact with lipid membranes [8]. Several research groups demonstrated applications of conjugated BSA-AuNCs in uorescent contrast imaging of living cells, including cancer [9][10][11]. From the nanomaterial safety standpoint the renal clearance and toxicity of red-emitting AuNCs have also been evaluated [12].
Albumins act as transport proteins which bind to various ligands and ions thus, in principle, BSA-AuNCs could be used to mark the distribution of metal ions inside cells, and their movement across the cell membrane. Potassium ions are important for normal cell functioning so being able to visualize their uptake and release is very useful for investigating cell metabolism. However, few potassium-speci c uorescent probes have been developed so far. Benzofuran isopthalate (PBFI) was the rst commercially available probe for imaging spatiotemporal potassium distribution within cells [13]. However, it suffered from low selectivity in the presence of sodium ions and poor intracellular loading [14,15]. Consequently, uorophores conjugated with ion-size speci c ionophores have been synthesized, improving selectivity and cell membrane permeability through modi cation with acetoxymethyl (AM) esters. Asante potassium green (APG) probes are commercially available for intracellular potassium imaging. However, a drawback of APG-AM probes is their low QY and requirement for esterase enzymes to be present for uorescence to occur. These probes also require a long incubation time in order to obtain a discernible uorescent signal. The QY of the APG-2AM probe in water is negligible. In addition it was shown that uorescence reached saturation at a K + concentration of 80 mM, thus the probe is not able to respond to K + changes at physiological intracellular potassium concentrations, which is 140-150 mM [15]. Recently, genetically encoded uorescence resonant energy transfer (FRET) probes were bioengineered and used for imaging intracellular potassium ion concentration with high selectivity and uorescence intensity [16]. The di culty in using these bioengineered probes lies in the time required for their synthesis and their high cost. The reported high a nity of FRET probes for potassium ions is also far from physiological intracellular potassium concentration.
Aggregation-induced emission (AIE) is a photophysical process that has been used in recent years to turn non-or weakly-luminescent probes into highly emissive probes [17]. For example, binding of several uorescent probes with albumins yielded a huge enhancement of uorescence QYs [17,18]. This enhancement was attributed to restricted intramolecular motions (rotations and vibrations) in the proteinuorescence probe adduct. Consequently, uorogens with AIE have emerged as a new class of uorescent probes for studying spatiotemporal ion distribution. Albumins were found to provide the necessary environment for many uorescent probes to enhance their QYs.
The lack of small and cost-effective uorescent probes with high QY for potassium imaging prompted us to evaluate the emission QY of BSA-stabilized red-emitting AuNCs (BSA-rAuNCs) in the presence of potassium ions. The emission intensity was evaluated in response to potassium ions within physiological concentration range (1.5 mM -150 mM). The observed emission QY enhancement was correlated with the particle size and charge analysis. Based on these studies an enhancement mechanism has been proposed. Synthesis of BSA-rAuNCs BSA-rAuNCs were prepared according to the previously reported procedure [19]. In brief, aqueous solution of HAuCl 4 (10 mM, 10 ml) and BSA (50 mg/ml, 10 ml) were mixed under vigorous stirring conditions.

Reagents and materials
After stirring at 750 rpm for 10 minutes at 37°C, NaOH solution (1 M) was added to adjust pH to 12. Then, the reaction was allowed to proceed at 37°C for 12 h at 700 rpm. After cooling, the resultant brown solution was dialyzed against ultrapure water using cellulose membrane (MWCO = 30 kDa), and stored in the dark at 4°C. In the experiments, 10 µl of AuNCs solution was mixed with 1990 µl of KCl solutions within the physiological concentration range.
Luminescence studies of BSA-rAuNCs Luminescence spectra were acquired using a commercial spectro uorometer (FP-6500 JASCO), and a custom-built uorescence set-up. Emission spectra recorded at the FP-6500 were taken using a scan speed of 200 nm/min. A rectangular suprasil quartz cell (10 mm path, FP-1003 JASCO) was used as the sample container.
The custom-built uorescence set-up consisted of a green laser diode (532 nm, 0.06 W/cm 2 , JMS-30RB, Light Vision Technologies) used as the excitation source. Iris diaphragms were used for optical alignment and for regulating the intensity of the incident beam. Band-pass lters (Semrock, LP03-532RE) were used to pass beam wavelengths within a 532 nm range and to reject (attenuate) frequencies outside that range. Neutral-density (ND) lters were used to reduce laser beam intensity. A rectangular suprasil quartz cell (10 mm path, FP-1003 JASCO) was used as the sample container. The cuvette was mounted in the black cuvette holder (10 × 10 mm, Newport) to maintain the same geometrical conditions during measurements with different samples. Emission spectra were acquired with a spectrograph (Acton SP-150, 300g/mm gratings) and liquid N 2 -cooled CCD camera (Princeton Instruments, Spec10:256E) having dark current of 3 electrons/scan. Wavelength calibration before each measurement was done using a neon lamp. The overall spectral resolution of the system is 0.3 nm. For each measurement, the sample was continuously stirred during spectrum acquisition, to avoid thermal gradients.

Results And Discussion
Emission spectra of BSA-rAuNCs Figure 1 shows a photoexcitation spectrum acquired on a dialyzed sample with pH = 9.6 (left spectrum). The spectrum shows a main peak at 527 nm and a shoulder at 500 nm. On the right side of the graph, emission spectra upon excitation at 500 and 527 nm are shown. At both excitation wavelengths, a broad peak is observed with the maximum at 670 nm for 500 nm excitation and a peak at 678 nm for 527 nm. Such excitation wavelength-dependent shift in the emission spectra was observed for BSA-rAuNCs as well as for BSA-AgNCs [22,23]. Some authors attributed the shift to the presence of NCs with different size, and some to the presence of Au(I) oligometric structures.
Emission origin of BSA-rAuNCs A synthesis protocol, proposed by Xie. et al. is widely used to prepare BSA-rAuNCs. In this protocol, HAuCl 4 is mixed with BSA while the pH of the mixture is raised to 12 [19]. For a long time, the red emission of solutions prepared using the protocol has been attributed to the presence of Au 25 NCs in BSA scaffold. Recently, however, the origin of the red-emission has been disputed. It was earlier proposed that the Au 3+ ions of the gold precursor are rst reduced to Au (0) by the hydroxyl group of tyrosine residues of BSA, since required pH was close to the pKa of tyrosine (10.46) [19]. Dixon et al. investigated the occurrence of the red emission in the pH range from neutral to basic [24]. Their studies revealed that the onset of the red emission occurs at pH = 9.7, which is below pKa of tyrosine. In another paper, the same authors have debated assignment of the red emission to Au 25 NCs, nucleated at the single-site [25]. This assignment to Au 25 NCs was based on results obtained using the matrix-assisted laser desorption ionization time-of-ight (MALDI-TOF) technique [19]. The method can only reveal the mass increase of BSA upon the uptake of Au ions and provides information on the total number of Au atoms but not on the cluster size or their distribution inside protein. Indeed, our studies revealed a linear dependence on the m/z with HAuCl 4 concentration, and a time-dependent m/z increase [7,26]. Since crystal structure of BSA-rAuNCs is unknown, the MALDI results cannot be de nitely used to assign the origin of the red emission to Au 25 NCs.
These earlier studies prompted renewed investigation aimed at identi cation of reducing residues and the origin of the red emission. Recently, a thorough systematic study was published by Hsu et al. who investigated the NCs capping moieties of BSA [27]. A detailed analysis of mass spectra under enzyme and disul de bond digestion revealed that the red-emitting core fraction contains Au and cysteine residues. The disul de pairs C75-C91/C90-C101 in domain IA, C315-C360/C359-C368 in domain IIB, and C513-C558/C557-C566 in domain IIIB were identi ed as capping sites of red-emitting AuNCs. By mapping 97% of the BSA sequence, multiple aminoacid oxidation sites have been identi ed. BSA-rAuNCs were extensively oxidized at multiple methionine and cysteine sites and moderately oxidized at histidine sites.
A small percentage of tyrosine residues were found to be oxidized. Based on these studies it has been proposed that methionines and cysteines are responsible for the reduction of Au ions. BSA has only one free cysteine residue. The other 17 cysteines form disul de bridges. The primary nucleation sites are likely the free cysteine and cysteines freed from their disul de bond via hydrolysis in alkaline solutions. It has been shown that uncoordinated Au atoms are e cient in cleaving disul de bonds [28]. At pH 12, disul de bonds cleavage along with the BSA conformation change from N-to A-form favors nucleation [29]. Nucleation process involves formation of S-Au-S staple motifs. Such motifs are well known in the chemistry of thiols with Au [30]. According to molecular dynamics (MD) simulation, Au clusters grow close to cysteine sites across all three domains of BSA [31]. The domains IIB and IA were reported to accommodate large clusters composed of more than 12 gold atoms. The widely observed excitation wavelength-dependent shift in the emission spectra is in line with the presence of AuNCs of different sizes across different regions of BSA [32].
The majority of X-ray photoelectron spectroscopy (XPS) studies for the red-emitting BSA-AuNCs show In order to understand the origin of the emission, Chevrier et al. used enzyme digestion to show a correlation between intensity of the red emission and rigidity of BSA. [37]. The rigidity is attributed to interlocking Au(I)-SR moieties that also enhance formation of Au(I)-Au(I) bonds. High rigidity resulted in strong emission. The effect is well-known to enhance the luminescence of Au complexes composed of aurophilic oligomers [38, 39]. Such oligomeric structures exhibit a microsecond-long photoluminescence decay component, observed also in this study (Fig. S1&S2) [34]. In order to understand the observed emission enhancement of the samples, the size and charge of particles present in the solutions were evaluated. Table 1 summarizes the results and the results of QY measurements for a given KCl concentration. The observed size of prepared BSA-rAuNCs is around 9.6 nm, consistent with the previous dynamic light scattering (DLS) and small-angle X-ray scattering (SAXS) results. At pH 9.5, (pH after dialysis) the zeta potential is −37mV and QY is 13%. Aggregation-induced emission enhancement of BSA-rAuNCs Au(I)-thiolate complexes exhibit strong luminescence emission by the mechanism of aggregationinduced emission [34,36]. The origin of strong emission has been attributed to the aurophilic interactions [39]. According to the EXAFS spectra analysis, luminescence intensity of glutathione-stabilized AuNCs was proportional to the degree of Au(I)-Au(I) bonds [36]. Recent transmission electron miscopy (TEM) images of BSA-rAuNCs, after a prolonged incubation, show the presence of long chains with triangularshaped side chains [41]. It is likely that the backbone chains in the images correspond to Au(I)-Au(I) linear chains and the side chains are Au(I)-(SR) residues. Francos et al. found that addition of sodium and potassium into solutions containing Au-SG NCs enhances their emission via formation of inter-cluster electrostatic linkages that promote formation of aurophilic interactions [42].
In view of the emerging new facts about Au-SG NCs, the enhancement of the emission of BSA-rAuNCs in the presence of KCl observed in this work can be explained as follows. At low KCl concentrations, BSA molecules containing various amount of Au atoms are far apart and protected by water molecules forming a hydration shield (HS) of some thickness (Fig. 5a). Upon increasing KCl concentration, the positive ions bind to carboxylate groups of aminoacids, with Cl − ions also binding to Au atoms. At high KCl concentrations, BSA molecules start to aggregate due to the salting-out effect which causes thinning of hydration layer on protein.
In the aggregated protein assemblies, the interprotein aurophilic gold bridges form that restrict BSA rotational freedom resulting in the enhanced luminescence (Fig. 5b).
Emission of BSA-rAuNCs in bovine blood plasma Potential application of AuNCs in cell imaging have been demonstrated in recent years thus emission intensity of BSA-rAuNCs was also investigated upon mixing with bovine blood plasma [11,43]. The emission spectrum acquired with the custom-built uorescence set-up is shown in Fig. 6 along with the spectrum obtained in water.
In blood plasma, the maximum wavelength of emission is shifted towards longer wavelength, and intensity is lower than in water. At 532 nm excitation the QY was around 3.33% but was decreasing with time, suggesting changes of cluster size and its chemical morphology due to interaction with blood plasma proteins.

Conclusions
The emission intensity of the red-emitting BSA-AuNCs was investigated as function of KCl within concentration range encountered under physiological conditions (0-150 mM). Increasing of potassium ion concentration resulted in the enhancement of light emission. This increase was attributed to intramolecular aggregation via Au(I)-Au(I) bond formation. The absolute QY of BSA-rAuNCs measured with the integrated sphere was 13%. When NCs are mixed with the bovine plasma, the QY is only 3%. This small QY is insu cient for in-vivo imaging. Thus, application of the NCs in bioimaging should be reconsidered in view of new facts that are emerging about BSA-rAuNCs aggregation, reactivity and poor renal clearance [12].

Con icts of interest
There are no con icts to declare.