Amino-acid conjugated protein–Au nanoclusters with tuneable fluorescence properties

Au-based protein nanoclusters (PNCs) represent an emerging class of fluorescence probes that are inherently biocompatible and combine the functionality of proteins and optical properties of Au nanoclusters. Here we report on a methodology to create conjugated Au PNCs using amino acid coupling strategies from a series of common laboratory proteins. We discover that the host protein and the specific conjugation chemistry has a profound influence on the resulting fluorescence properties. Synchrotron analyses showcase local Au NC aggeration upon PNC conjugation, which causes local environment changes to invoke differences in fluorscence properties. The observed aggeration does not give rise to plasmonic properties nor signifigant fluorescence quenching, strongly indicating the PNCs are still in a near-native cluster state. Our methodology and findings here could open new pathways for tuning PNC fluorescence properties in a rational fashion, having a potential impact in host of biomedical and sensing applications.


Introduction
Metal nanoclusters (NCs) are engineered at dimensions approaching the Fermi wavelength of electrons, allowing for the formation of discrete energy states that result in interesting fluorescence properties that differ to those observed in larger metal nanoparticles with plasmonic properties [1][2][3][4]. While avoiding the toxic components of traditional fluorescent quantum dots (QDs), metal NC have demonstrated high quantum yield, tuneable photonic emission, highly efficient two-photon absorption, and tailorable hybridization pathways. For biomedical applications in particular, the tailoring of NC properties and biocompatibility through biomediated synthesis approaches has proven instrumental [5][6][7][8][9]. Not surprisingly, protein-stabilized nanoclusters (PNC) have emerged as versatile hybrid platforms that can leverage functional properties of both the NC and protein components [10][11][12][13][14].
In traditional metal NCs, ligand parameterization and corresponding electronic structure is very well established [15][16][17]. In PNCs, however, identification of the stabilization region and specific ligand interactions of NCs within a protein scaffold remains elusive and represents a major deficiency towards controllable PNC synthesis and optimized functionality. If cluster location and protein interaction could be better understood, one can learn how NCs are stabilized and what reaction mechanisms allow for growth of the NC to their targeted size, electromagnetic properties, and responsive functionality. With this knowledge, the rational identification and engineering of new proteins could be accomplished to grow NCs with pre-determined properties. As Au-thiol interactions available through metal-binding residues within the protein present likely target regions for NC stabilization, recent studies have targeted these through both simulation and designer protein approaches [18]. While such studies provide secondary support for likely stabilization regions, more direct characterization is required to definitively identify NC-ligand interactions in a protein host.
Direct examination of the cluster region with a protein structure requires the utilization of x-ray and crystallographic techniques. For the latter, highly crystalline materials are required which have yet to be achieved with PNC hybrids. One avenue to create highly-ordered PNC structures is through their incorporation into metal-organic framework (MOF) scaffolds, however, solvent incompatibility has proven a challenging issue for the integration of NCs into ordered MOFs [19,20]. One method to overcome this barrier involves the precipitation of nanoclusters with a metal ion solution, such as Zn 2+ , to form aggregates that are amenable to methanol dissolution. With proper precipitation, the NCs have been demonstrated compatible for processing into zeolitic imidazole framework (ZIF-8) MOFs [20,21]. While PNCs have, to date, not been reported in composite form with MOF structures, proteins incorporation into ZIF-8 has been shown [22]. In addition to ordering through MOFs, researchers have recently begun to examine large-pore protein crystals (LPC) as scaffolds for NC incorporation. With LPC pore diameters up to 8 nm and aqueous processing conditions, glutathione capped Au25 nanoclusters can be readily adsorbed into the patterned crystal pores [23]. The ability to utilize purely organic frameworks reduces the processing requirements in addition to providing a platform to enable precise patterning of the embedded materials within the LPC.
More recently, a new methodology to directly conjugate PNCs and form crystal-like structures has been developed [24]. By synthesizing PNCs in the presence of select amino acids (AA), protein conjugation enabled the formation of supramolecular structures. Interestingly, through simple variations in pH and AA composition, Ding et al demonstrated the ability to controllably tune NC fluorescence emission wavelength across the visible spectrum for BSA-stabilized gold nanoclusters (AuNCs) conjugates [24]. As this approach enables formation crystal-like structures composed nearly entirely of the PNC of interest with limited additional materials, subsequent analysis of the gold-ligand stoichiometry can be more readily studied without extraneous contaminating material signals. Additionally, it is hypothesized that the crystal-like structure could generate PNCs of consistent orientation in a close-packed cluster; enabling more precise x-ray analysis to resolve protein-Au NC interacations.
Motivated by the aforementioned study, we report on the conjugation of PNCs using a series of protein hosts and amino acid conjugation conditions. We found that the conjugated PNCs exhibit flourescence characteristics that depend on the nature of conjugation chemistry and host protein using BSA, trypsin, pepsin, and lysozyme. To study the atomic-scale origins of these phenomena, we further study the conjugated PNCs using a x-ray absorption spectroscopy (XAS) and small angle x-ray scattering (SAXS). With these methods, we observe unexpected clustering effects in the conjugated PNCs that are dependent on the protein host. These findings should help guide and refine our ability to make Au PNCs with optimized properties.

Synthesis of PNC's and AA-conjugated PNCs
The synthesis of AA-conjugated PNC (AA-PNC) was performed with slight modifications of previously reported methods [24]. In brief, hybrids were prepared with varied composition of AAs, including L-histidine (L-His) and glycine (Gly), and protein hosts to include BSA, trypsin, lysozyme, and pepsin. Stock solutions were prepared of the selected proteins (50 mg ml −1 ), L-His (31 mg ml −1 ), Gly (15 mg ml −1 ), and HAuCl 4 (16 mg ml −1 ). Samples were prepared according to table 1 by sequentially adding the protein stock (1 ml), HAuCl4 (1 ml), AA (5 ml), and DI water (3 ml) to a glass vial. The pH was adjusted as outlined in table 1 with 1 M solutions of NaOH or HCl. This procedure results in a final reaction solution of 5 mg ml −1 protein, 1.6 mg ml −1 HAuCl 4 , and 7.5 mg ml −1 Gly (or 15.5 mg ml −1 L-His depending on amino acid selection). Samples were stirred at 37 • C for 12 h. The final solution is purified through centrifugation at 10 000 g for 30 min, with collection of the pellet to target formed conjugates. The pellet was resusped in DI water and centrifugal purification repeated three times. The final pellet resuspended in 5 ml of DI water. Control protein nanoclusters were made for each protein host utilizing the same procedure outlined above without the addition of amino acids. Controls were synthesized at pH 12 and stirred for 12 h at 37 • C. Samples were purified through a 10 kDa molecular weight cut off (MWCO) centrifugal filter to remove unreacted metal ions.

Characterization
Fluorescent emission of the protein-nanocluster hybrids was measured on a Horiba Fluorolog-3 with excitation at 350 nm and emission capture range of 400-700 nm. All measurements were made in aqueous solutions in a quartz cuvette with a path length of 1 cm. XAS measurements were performed at the 12-BM beamline of the Advanced Photon Source (APS), Argonne National Laboratory (ANL). A sample from each  Sample B1  B2 B3 B4  T1  T2 T3 T4  L1  L2  L3  L4  P1  P2 P3 P4 AA L-His Gly Gly Gly L-His Gly Gly Gly L-His Gly Gly Gly L-His Gly Gly Gly pH 5.5 1.5 3.5 12.5 5.5 1.5 3.5 12.5 5.5 1.5 3.5 12.5 5.5 1.5 3. 5 12.5 protein-host with glycine-assisted AA nanocluster formation was selected from table 1 to include B3 (BSA PNC), L2 (Lysozyme PNC), T2 (Trypsin PNC), and P1 (Pepsin PNC). Solutions of Au NCs were loaded into thin-walled quartz x-ray capillaries (1 mm OD) as synthesized and studied under a fluorescence measurement geometry using a 13 channel Ge detector. XAS scans were performed from 200 eV below the Au L 3 -edge (11.919 keV) to 900 eV past the edge. Data reduction, analyzed, and modeling were performed using the Demeter XAS software package [25]. EXAFS modeling was performed using scattering contributions from the well-known crystal structures of bulk Au, Au 2 O 3 and/or experimental determined structure of thiolated Au NCs [26]. A value of 0.798 for S 0 2 was used in all fits as determined from the EXAFS modeling of a reference foil. Small angle x-ray scattering (SAXS) measurements were performed at the 12-ID-C beamline of the APS. Samples were prepared in an identical fashion as to XAS experiments probed in a transmission geometry using 12.0 keV x-rays. Data was processed onsite at 12-ID-C and further modeled to obtain pair distance distribution functions, P(r), using the program ScÅtter [27].

Results and discussion
Motivated by the ability to better understand NC location within a protein host, AA conjugated PNCs were synthesized using a series of proteins. While we were unable to get qaulity materials for single-crystal XRD, we did observe interesting synthesis-dependent properties that warranted further invesitation. Utilizing the methodology reported by Ding et al [24], AA conjugated PNC growth was successfully performed using four different protein models: BSA, trypsin, lysozyme, and pepsin. To the authors' knowledge, this is the first reported AA-conjugated PNCs created with trypsin, lysozyme, and pepsin protein hosts. As shown in figure  1, this approach provides tailorable fluorescent emission wavelengths that can be obtained across the visible spectrum. The control PNCs formed for each protein host yielded red emissions, which corresponds to previous studies of pH 12 synthesis of BSA [5,28], trypsin [29][30][31], pepsin [32], and lysozyme [33,34]. While pepsin has been shown to readily create smaller clusters (i.e. blue-shifted emission) through modulation of reaction pH [32], this tunability of BSA, lysozyme, and trypsin to the blue spectrum has been limited. As can be seen in figure 2, the ability to tune each protein model to the blue spectrum has been achieved through select AA integration into the synthesis process.
Although AA conjugated PNCs were succesfully formed with each protein model, fluorescent PNCs were not formed with every sample composition outlined in table 1. For non-fluorescent AA-conjugated PNCs, plasmonic responses were observed, strongly suggesting the possibility of Au NC destabilization under certain conjugation conditions. For the remaining PNCs, the fluorescent emission spectrum are shown in figure 2. In all cases, the non-conjugated NCs exhibit fluoresence at lower energy (i.e. red-shifted) wavelengths than the AA-conjugated PNCs. This suggests that the process promotes the formation of smaller cluster sizes, as Au NC emission wavelengths directly correlate to the number of Au atoms in the cluster [32].
Previous studies with BSA have demonstrated the ability for Gly and L-His to migrate into the protein during the unfolding/refolding process [24]. As the protein contains the residues required to reduce the metal salts, the additional Gly or L-His can assist in the stabilization and capping of the cluster during formation. Both amino acids, Gly and L-His, have previously shown the preferential capping of smaller metal cluster structures, shifting the emission towards the blue spectrum [35,36]. With respect to the formation of conjugated PNC structures, the presence of additional AA capping residues within the protein structure could inhibit proper protein refolding and induce aggregation. Although certain amino acids have shown the ability to inhibit protein aggregration during the unfolding/folding process, this is not the case for glycine and L-histidine [37].
In the case of BSA, lysozyme, and pepsin; conjugated PNCs were formed with both L-His and Gly approaches. Interestingly, the L-His conjugation of BSA stabilized PNCs yielded a spectral blue-shift in the cluster emission, which is supported by reported literature ( figure 2(A)); however yield a green cluster emission in the lysozyme and pepsin models along with a blue emission of the trypsin AA-PNCs (figures 2(B)-(D)). As the incorporation of L-His into the synthesis mechanism shifts the traditional spectral response of each protein model, it is likely that L-His plays a direct role in NC formation and stabilization beyond simple protein conjugation and supports direct capping/stabilization of the cluster as it is reduced/stabilized within the protein scaffold.
To better understand the effect of host protein and conjugation chemistry on the local structure of the Au clusters, XAS was used to as an element specfic probe. The x-ray absoprtion near edge spectra (XANES) for both as-synthesized Au PNC and the AA-conjugated PNCs are shown in figure 3(A). Overall, all probed PNCs exhibit a~+1-2 eV shift in E 0 from the Au reference, indicating a partial increase in oxidation state. The biggest shift in E 0 is observed for PNCs stabilized within BSA, followed by trypsin, lyzosyme, and pepsin stabilized PNCs. This shift in E 0 is expected given the small cluster size and corresponding surface atoms interactions with host protein moeties. Moreover, the XANES features observed in the as-synthesized PNCs are signifigantly broad in comparison to the bulk Au reference as expected [38,39], given the small cluster size not replicating the bulk fcc lattice of Au. Upon conjugation, all PNCs exhibit a reduced E 0 compared to their as-synthesized counterparts that is more akin to the Au reference. Additionally, XANES features better resemble that of the Au reference, albeit with an increase in white line intesity. These findings were unexpected and incidate that the protein conjugation must force the Au atoms more into a larger aggerated state, yielding XANES feature more resemblent of fcc Au. The increase in white line intensity as compared to the reference Au additionally suggest sufficient Au-ligand interactions, incidating electron transfer from the Au to the protein is still present.
After edge-step normalization, k 2 -weighting, and Fourier transform, the extended x-ray absorption fine structure spectra (EXAFS) reveal interesting local structural changes in the Au NCs ( figure 3(B)). Note the EXAFS is not corrected for phase shift, wherein peaks in the EXAFS are shifts~0.3-0.4 Å lower from actual distances. The as-synthesized PNCs all exhibit a prominate feature centered below 2.0 Å and smaller features that somewhat align with those observed in from the bulk Au. This first distance is likely Au-thiol bonding through interactions with the protein that are commonly observed in thiolated AuNC complexes with organic ligands [38]. It is important to note that the pepsin PNC has a signifigantly shifted distance for the Au-biotic interaction, perhaps indicating that diffent cluster binding motif may be present.
Upon conjugate formation, notable structural differences can be observed between the AA-conjugated PNCs vs their non-conjugated counterparts. Features more resemblant of fcc Au become more evident, albeit at slightly shifted distances. Qualitatvely, this indicates that protein conjugation condenses the clusters together in an aggregated cluster arrangement that is likely perturbed signifigantly from the bulk lattice due to the likely imperfectious nature of cluster assembly during protein conjugation. The majority of the PNCs also exibit a feature <~1.5 Å. This featue is shorter than the ligand-Au distances observed for the as-synthesized Au NCs, strongly indicating that protein-Au interactions are changing during AA conjugation.
The EXAFS data was then modelled as described in the experimental section with results presented in table 2. The modeled data produces nearest neighbor distances (NNDs) and coordination numbers (CNs) of the specified scattering pairs. For the as-syntheiszed Au NCs, small Au-Au CNs were achieved and are reflective of the small size of the clusters with high surface area. Au-Au NNDs are consistent with those observed in previous clusters [38,40], with a slighly longer distance of 2.94 ± 0.02 Å for NCs grown within BSA. This suggest that the structural influence of the protein may be used to alter the NC's properties using BSA. For Au-S contributions to the EXAFS, the Au-S CNs for all the as-syntheiszed NCs are~2.0 (within fitting error), suggesting strong surface coordination within the cluster and the protein. NNDs for Au-S reflect those qualtitavely observed in the EXAFS, with the Au-NC synthesized with pepsin exhibiting a slightly shorter feature. Uponc conjugation, all Au-Au CNs increase to 3.45 ± 0.61, 5.00 ± 0.51, 5.18 ± 0.13, and 8.84 ± 1.02 for PNCs implementing trypsin, pepsin, BSA, and lysozyme repsectively. Using idealized nanocrystal models, which are a rough estimation given the observed nature of cluster aggeration, these CNs coorespond to those with sizes of~0.75 nm,~1.0 nm, and~1.5 nm repsectively [41]. NNDs for Au-Au are all similar to those found in nanoscale Au as well. The shorter-scale features from Au-ligand intereactions exhibit sufficient changes upon conjugation to PNCs as well [41]. Au-S CNs are reduced by over 50% for all samples, while shorter low-Z contributions become more evident in the remaining samples. It is important to emphasized that Au nanoparticle formation is likely not occuring at an appricatable amount given that visibile fluorescence is reminescence of Au NCs and not plasmonic Au nanoparticles.
Morphological differences between as-synthesized and conjugated PNCs were probed using SAXS as shown in figure 3(C). PNCs synthesized with BSA and trypsin are shown in the black and green dotted scattering patterns respectively and showcase typical form factor scattering for metallic NCs in the immediate q-range along with some clear aggregation at lower q-range. Conversely, the AA-conjugated PNCs  exhibit markedly different SAXS profiles. The form factor scattering is largely absent, which is coupled to a highly linear region at lower q-range. This strongly suggests that protein-protein aggregation within the sample is not prevalent and that PNCs are beginning to order in solution through the conjugation process. Note that the ordering is not in a highly ordered periodic fashion to result in Bragg features, such as those observed in DNA-programmed superlattices of Au nanoparticles [42,43]. To further probe the effect of conjugation on morphology, SAXS profiles we fitted to pair distance distribution functions, P(r) (red lines, figure 3(C)), using known methods. The resulting P(r) functions are shown in figure 3(D) and showcase a large difference in size and morphology for between the as-synthesized and crystallized materials. PNCs synthesized with BSA and trypsin exhibit a main feature~20 Å, and d max values for 84 Å and 108 Å respectively. The AA-conjugated PNCs, on the other hand, yield P(r) maxima at values >60 Å for all materials and d max values >170 Å over similar q-range values. This strongly indicates and organization of PNCs into supramolecular structures caused by conjugation processes within the proteins. Taken together, the synchrotron characterization demonstrate that AA-conjugation creates an aggregation of Au NCs resembling some more similar to Au nanoparticle. However, the visibile light flourescence properties are stongly suggest non-plasmonic behavior. This finding illistrates that cluster aggeration rates and the appropriate protein host could be used to control flourescence propeties without inducing a plasmonic response, and may provide insights in the generation of new PNCs for various applications.

Conclusions
In summary, AA-conjugated PNCs were created in multiple protein models, including BSA, trypsin, pepsin, and lysozyme PNCs. Fluorescence properties were shown to be dependent on Au NC-protein interaction and the aggreation of Au NCs that ocucrs during conjugation as revealed using synchortron characterization methods. Our finding suggest that choice of host protein and the degree of NC aggeration allows for the possible tunablity of fluorescence properties, priving new pathways to create new NC fluorescence materials with user defined properties.