Atomic Structure of a DNA-Stabilized Ag11 Nanocluster with Four Valence Electrons

The combination of mass spectrometry and single crystal X-ray diﬀraction of HPLC-puriﬁed DNA-stabilized silver nanoclusters (DNA-AgNCs) is a powerful tool to determine the charge and structure of the encapsulated AgNC. Such information is not only relevant to design new DNA-AgNCs with tailored properties, but it is also important for bio-conjugation experiments and is essential for electronic structure calculations. Here, the eﬀorts to determine the structure of a HPLC-puriﬁed green emissive DNA-AgNC are presented. Unfortunately, the original DNA-AgNC, known to have four valence electrons, could not be crystallized. By modifying the stabilizing DNA sequence, while maintaining the original spectroscopic properties, several mutants could be successfully crystallized, and for one of them, single crystal X-ray diﬀraction data provided insight into the silver positions. While the DNA conformation is not resolved, the described approach provides valuable insight into the class of green and dual emissive DNA-AgNCs with four valence electrons. These results constitute a roadmap on how to improve crystallization and crystal quality for X-ray diﬀraction measurements.


Introduction
DNA-stabilized silver nanoclusters (DNA-AgNCs) are water-soluble emitters formed by cationic and neutral silver atoms embedded in short DNA oligomers (10-30 bases). [1]Insight into the relationship between the DNA scaffold and the stabilized AgNCs has been achieved by the availability of large screening libraries and machine-learning-assisted predictions. [2]dditionally, purification methods such as size-exclusion chromatography (SEC) [3] and high-performance liquid chromatography (HPLC), [3b,4] have been paramount in isolating atomically-precise DNA-AgNCs.This has not only facilitated detailed photophysical characterization but also size and charge determination by mass spectrometry, which demonstrated the relation between the number of neutral Ag atoms (5s 1 valence electrons) and the emission color. [5]However, single crystal X-ray diffraction measurements are still needed to gain insight into the atomic arrangement of the silver atoms, the coordinate bonds to the DNA and the overall DNA conformation.In 2019, Huard et al. reported the structure of a green-emitting (DNA) 2 Ag 8 NC, [6] while consecutive manuscripts by Cerretani et al. have described the structure of a NIR emissive Ag 16 NC and seven mutations thereof. [7]When the crystal structure of this NIR emitter was published, [7d] no mass spectrum was available and some electron density was assigned to two additional silver atoms with low occupancy (≈0.3).However, Gonzàlez-Rosell et al. recently combined mass spectrometry with a re-evaluation of the X-ray data to demonstrate the presence of two chloride ligands in the DNA-Ag 16 NC structure. [8]The electron density of a chloride ion is indeed approximately one-third of that of silver, and the experimental mass data was consistent with the theoretical isotopic distribution of (DNA) 2 [Ag 16 Cl 2 ] 8+ .These results highlight the importance of combining mass spectrometry with single crystal X-ray diffraction measurements to unravel the overall structure of this class of complex emitters.
While structure determination from crystallized DNA-AgNCs together with mass spectrometry provides a wealth of information, each individual step is neither trivial nor guaranteed to succeed.In this article, we have summarized our strategies to crystallize and solve the atomic structure of a recurring type of green emissive DNA-AgNC, the 4-valence-electron    abs and  em are, respectively, the absorption and emission ( exc = 469.5 nm) maxima measured at room temperature in 10 mM NH 4 OAc.<> is the intensity-weighted average decay time monitored at 535 nm, exciting at 469.5 nm.Q is the fluorescence quantum yield measured at room temperature ( exc = 445.4nm).a) Q determined with a relative method, using Fluorescein in 0.1 M NaOH as reference dye. [12]b) Q estimated using a single measurement for the fraction of absorbed light and a single measurement of the integrated emission.As shown for 5A-11Green-AgNC, the two methods give comparable Q values.See Supporting Information for more details.
[Ag 11 ] 7+ /[Ag 10 ] 6+ system.For this purpose, we chose the (DNA) 2 -[Ag 11 ] 7+ NC, further defined as 10Green-AgNC, previously reported by Copp et al. [5] The aim was to combine mass spectrometry results with structural information from single crystal X-ray diffraction data.Since the original DNA-AgNC could not be crystallized, the stabilizing DNA sequence was modified in several ways, while retaining the spectroscopic properties of the embedded cluster.X-ray diffraction data of one of the mutants provided insight into the arrangement of the silver atoms.
Our path to obtain well-diffracting DNA-AgNC crystals, combined with mass spectrometry and photophysical characterization, can be a roadmap for unraveling the structure/spectroscopic property relationship of other types of DNA-AgNCs.

Modification of 10Green-AgNC for Crystallization
The original mass spectrometry data of 10Green-AgNC was published in 2014 by Copp et al. [5] The green emissive 10Green-AgNC is composed by two 10-base DNA stands, 5′-TCCACGAGAA-3′, wrapped around eleven silver atoms, of which four are neutral and the remaining seven are cationic ([Ag 11 ] 7+ ). [5]Recently, a similar 4-valence-electron [Ag 11 ] 7+ NC has been described by Petty et al. to be formed when using 5′-CCCCAACCCCT-3′ as the stabilizing scaffold.The closely related [Ag 10 ] 6+ core has also been reported several times by Petty and co-workers. [9]Hence, the combination of structural information with knowledge of the charge from mass spectrometry would allow one to compute the electronic structure [8,10] of what seems to be a recurring type of green emissive AgNC. Figure 1 shows that 10Green-AgNC has an absorption and emission maximum at 472 and 535 nm, respectively.The intensityweighted average decay time, <>, was determined to be 0.65 ns and the fluorescence quantum yield (Q) was estimated to be 0.11 (Table 1).
Our first attempt to crystallize 10Green-AgNC consisted of screening a large number of commonly used crystallization conditions (see Supporting Information), utilizing the hanging-drop vapor diffusion method.Even though this approach was successful for a series of NIR emissive DNA-Ag 16 NCs, [7d] it did not yield any crystals for 10Green-AgNC.Based on our previous work, [7c,d] we modified the original sequence by adding one adenosine at the 5′-end, as we noticed for DNA-Ag 16 NC that the terminal adenosine did not affect the photophysical properties but could promote or alter the crystal packing interactions.
The additional base was attached to the 5′-and not the 3′terminus because the 3′-end of the DNA strand was already rich in adenosines.Moreover, the introduction of the 5′ overhang to enhance crystal packing interactions has been routinely used in RNA crystallography. [11]The new DNA sequence, 5′-ATCCACGAGAA-3′, indeed stabilized a nearly identical green emitter (further referred to as 11Green-AgNC), as illustrated by the spectra in Figure 1 and the data in Table 1.Both 10Green-AgNC and 11Green-AgNC were HPLC purified prior to crystallization attempts (the chromatograms are reported in Figures S1 and S2, Supporting Information).The addition of the extra adenosine at the 5′-terminus enabled the crystallization of 11Green-AgNC (Figures S29 and S30, Supporting Information).
While the first obstacle was overcome, the quality of the collected single crystal X-ray diffraction data was too poor to attempt structure determination (Table S1, Supporting Information).The next step was to further extend the 3′-and 5′-ends with even more adenosines and/or thymidines, since the latter are known to be the least favorable to coordinate silver atoms.Figure 2 and Table 1 show that the seven new HPLC-purified mutants displayed no significant alterations of the photophysical properties in solution.Minor concurrent changes in the fluorescence decay time and fluorescence quantum yield are most likely due to mutantspecific changes in the non-radiative decay rate.
For this series, the best X-ray diffraction results were achieved for 5A-11Green-AgNC, when using a 2 Å X-ray beam.The 5A-11Green-AgNC crystals diffracted with a resolution of 2.6 Å (Table S1, Supporting Information), which allowed us to narrow down the space group to two different possibilities: P6 2 22 or P6 4 22 (hexagonal) with unit cell dimensions of a = b = 28.7 Å and c = 193.7 Å.However, the phase determination using the singlewavelength anomalous dispersion (SAD) method failed.While it is unclear why this approach did not work, it could be that the silver atoms encapsulated in the DNA scaffold are disordered.
Crystals were also obtained for 3T-11Green-AgNC and 3A-11Green-AgNC, but the quality of the diffraction data was again poor (Table S1, Supporting Information).Since the SAD method using silver was unsuccessful, we tried to incorporate bromine in the structure and use the corresponding anomalous signal to solve the phase.For 11Green-AgNC, the third base in the DNA sequence was exchanged with a brominated cytosine, where the bromine is covalently bound to the fift carbon of the pyrimidine ring.The modification is further defined as BrC3-11Green-AgNC.
The solution properties of BrC3-11Green-AgNC are once more alike (see Table 1 and Figure 2) to 10Green-AgNC.BrC3-11Green-AgNC yielded large single crystals, and good diffraction data with resolutions of 2.3 and 2.5 Å was obtained, when utilizing 2 Å and 0.915 Å X-ray beams, respectively (Table S1, Supporting Information).The BrC3-11Green-AgNC crystal was found to be isomorphous with the 5A-11Green-AgNC crystal (space group: P6 2 22 or P6 4 22, unit cell dimensions: a = b = 28.4Å and c = 194.1 Å).Unfortunately, the anomalous signal from bromine did not help to solve the phase, and thus the atomic structure could not be determined.
Extending the DNA sequence with adenosines and thymidines to promote crystal packing interactions and/or brominating one of the bases was successful in obtaining crystals, but the quality of the diffraction data was not good enough to solve the structure.
Therefore, we decided to shorten the DNA scaffold, even though the additional nucleobases seemed to help promote crystal packing interactions.7c] Moreover, the presence of additional nucleobases meant that there were significantly more nucleotides than silver atoms.
Removing two bases from the 3′-end of 11Green-AgNC and 5A-11Green-AgNC yielded two new DNA-AgNCs, namely 11Green-2end-AgNC and 5A-11Green-2end-AgNC, respectively (see Table 1).Both modifications showed similar photophysical properties as the original green emitter (Figure 1; Figures S16  and S17, Supporting Information), and were successfully crystallized.Single crystal X-ray diffraction data of 11Green-2end-AgNC provided insight into the position of the silver atoms.Hence, a detailed spectroscopic, structural and compositional analysis of this mutant was undertaken.

11Green-2end-AgNC
As briefly mentioned in the previous section, spectroscopic characterization revealed that 11Green-2end-AgNC has the same steady-state properties as the original 10Green-AgNC, but with a small drop in the fluorescence decay time and hence quantum yield (Figure 1 and Table 1).The mass spectrum of 11Green-2end-AgNC (Figure 3) shows a molecular ion peak centered at 1647.79 m/z with a z = 4 − charge state.The experimental mass data is consistent with the theoretical isotopic distribution of a compound comprising two DNA strands and 11 silver atoms with an overall charge of 7+.This confirms that the silver nanocluster in 11Green-2end-AgNC is compositionally the same as the 4-valence-electron (DNA) 2 -[Ag 11 ] 7+ NC (10Green-AgNC) reported by Copp et al. [5] While the spectroscopic data indicates that all DNA modifications formed alike emitters, mass spectrometry measurements were also performed on two other mutants to further prove the similarities.Both the 11Green-AgNC and the 5A-11Green-AgNC gave molecular ion peaks corresponding to a Unlike 11Green-AgNC and 5A-11Green-AgNC, 11Green-2end-AgNC crystals diffracted with a resolution of 2.0 Å, using a 2 Å X-ray beam (Table S1, Supporting Information).The space group of the crystal was found to be P1 (triclinic), with asymmetric units formed by two DNA-AgNCs.In contrast to DNA-Ag 16 NC, the geometry of this AgNC looks less defined.Figure 4A shows the electron density of 11 silver atoms belonging to one of the (DNA) 2 [Ag 11 ] 7+ systems in the crystallographic unit.
The silver atoms are arranged in an overall rod-like shape (Figure 4A,B), where six of them resemble a tetragon pair with a shared bond and edge [13] (yellow dashed lines, "boat") and the other five assume a square planar arrangement with a single atom extension (red and blue dashed lines, "lounge chair").The Ag-Ag distances within the "boat" subsection seem in line with previously reported distances, ranging from 2.5 to 3.4 Å. [6,7d] This indicates that some Ag-Ag distances are below the bulk silver distance (2.88 Å) [14] and some approach the Van der Waals distance (3.44 Å). [15] With regard to the "lounge chair", the distances connecting the central atom in the square planar structure appear to be unrealistically short (2.1-2.3Å).It is worth noticing that the silver atoms in the "lounge chair" section of the second DNA-AgNC in the asymmetric unit (Figure S26, Supporting Information) are arranged in a square pyramidal geometry instead, and the interatomic distances display more realistic values of 2.8-3.0Å, with the exception of very short (2.4 Å) and very long distances (3.9 and 4.8 Å) at the base of the pyramid.
We thus speculate that the real arrangement of the silver atoms lies somewhere in between both (Figure 4D; Figure S26, Supporting Information) and it might indicate large flexibility in this part of the AgNC.Unfortunately, the "boat" section of the second DNA-AgNC in the asymmetric unit (Figure S26, Supporting Information) could not be fully resolved, meaning that only 9 silver positions in total were assigned with certainty.It is reasonable to assume that this second DNA-AgNC should also contain 11 silver atoms, given the single emissive species found in the solution and the single mass peak corresponding to (DNA) 2 -[Ag 11 ] 7+ NC.The difficulty in pinpointing the two silver atoms in the "boat" part of the second DNA-AgNC, along with the range of too small and too large distances for the "lounge chair" section, indicate that [Ag 11 ] 7+ might be intrinsically very dynamic.This also implies that obtaining a meaningful snapshot of the interatomic distances could be hard to achieve.Finally, the two sections ("boat" and "lounge chair") interact with each other (green dashed lines in Figure 4B) with interatomic distances ranging between 3.1 and 3.4 Å.
The structure presented in Figure 4 shows similarities with the geometrically optimized model of Ag 10 NC@hpC 12 reported by Gupta et al. [16] This computed 10-silver nanocluster embedded in a C12 hairpin is also formed by two parts: a similar distorted square planar section with a single extending atom ("lounge chair" part) and a pseudo-trapezoidal Ag 5 section that shows a resemblance to our "boat" arrangement.
While some rough outlines of electron density corresponding to the DNA bases could be observed, the quality was not good enough to build a DNA model or assign nucleotide residues.Nevertheless, thanks to mass spectrometry data, we can confidently state that each [Ag 11 ] 7+ cluster is embedded in two DNA strands.
Next, we investigated the spectroscopic properties of 11Green-2end-AgNC in the crystalline state.Figure 5A,B shows brightfield and wide-field fluorescence images of 11Green-2end-AgNC crystals.Interestingly, emission spectra of individual crystals, recorded with a home-built confocal microscope (Figure 5C), [17] display two emission bands.Besides the green emission peak, which is slightly red-shifted with respect to the solution data, a second band centered at around 675 nm is present.This indicates that, while only showing a nanosecond-lived fluorescence band in solution (Table 1), 11Green-2end-AgNC behaves as a dual emitter in the crystalline state.Consistent with previously reported dual emissive DNA-AgNCs, [9a,f,g,18] the additional redshifted band of 11Green-2end-AgNC is characterized by a microsecond decay time.More precisely, the green emissive state has a lifetime of 0.50 ns (Figure S27 Emission spectra and decay times of 11Green-AgNC and 5A-11Green-AgNC crystals were additionally acquired to confirm the homogeneity of the spectroscopic properties among different mutants (Figures S30-S33, Supporting Information).While the emission spectra of 11Green-AgNC crystals were very alike to those of 11Green-2end-AgNC, the ratio of the red-to-green band was significantly higher for the 5A-11Green- AgNC crystals.These crystals were however 2 years old, hence the crystallization buffer had completely evaporated.The addition of 40% PEG 3350 to the 5A-11Green-AgNC crystals made the photophysical features more in line with those measured for the freshly prepared 11Green-2end-AgNC (3 weeks old) and 11Green-AgNC crystals (7 days old), even though the mean fluorescence decay time remained longer (Figure S34, Supporting Information).This clearly indicates that the hydration level plays also a critical role in the observed red-to-green band ratio.To further test this idea and exclude age-related causes, the reversibility of these changes was checked by measuring a droplet of 11Green-2end-AgNC in 10 mM NH 4 OAc on the microscope and recording a series of emission spectra before and after the evaporation of the solvent.As shown in Figure S35 (Supporting Information), when the droplet dries out, the emission spectra display two bands, with the second 675 nm band even more pronounced than in the crystalline state.Upon rehydration, 11Green-2end-AgNC again behaves as a green emitter with a single emission band.The three cycles illustrate that the effect is fully reversible and the long-lived emission band is only present in the solid state.Additionally, given the much less intense red band in the hydrated crystalline state, it is also fair to assume that the determined structure is closer to that in solution compared to the dehydrated state.To the best of our knowledge, this is the first time that hydration-dependent emission properties have been demonstrated for DNA-AgNCs in the solid state, while this effect is commonly observed for zeolite-stabilized AgNCs. [19]

Conclusion
We described a series of mutants of 10Green-AgNC, which was known to be a green emissive (DNA) 2 -[Ag 11 ] 7+ NC with four valence electrons.Modifications of the DNA sequence of the original emitter were introduced to promote crystallization without major changes in the emissive properties.Addition of adenosines and/or thymidines at the 3′-and 5′-ends did not affect the photophysical properties, but enabled crystallization.Bromination of the DNA strand also preserved the spectroscopic characteristics of the original green emitter.Despite being able to grow welldiffracting crystals for 5A-11Green-AgNC and BrC3-11Green-AgNC, the SAD methods using the anomalous signals of silver and bromine were not successful in determining the structure.On the other hand, removing two adenosines from the 3′-end of the 11-base DNA strand, 5′-ATCCACGAGAA-3′, promoted the formation of the same green emissive species with the benefit of obtaining crystals that gave insight into the silver positions.In agreement with the mass spectrum, single crystal X-ray diffraction data of 11Green-2end-AgNC unveiled the presence of 11 silver atoms in a potentially very dynamic and flexible rod-like structure.This flexibility of silver atoms within the nanocluster might also be the reason for the reversible appearance of a microsecondlived red-shifted band in the crystalline and dried states.In addition, we demonstrated for the first time that the degree of hydration plays a crucial role in the observed photophysical properties.The combination of mass spectrometry data and single crystal Xray diffraction measurements has thus provided the first insight in the structure/property relationship of HPLC-purified greenemissive DNA-stabilized [Ag 11 ] 7+ /[Ag 10 ] 6+ NC systems.

Figure 2 .
Figure 2. A) Normalized absorption and B) emission spectra of all 11Green-AgNC mutations with additional adenosines and thymidines, as well as brominated C3 (BrC3), in 10 mM NH 4 OAc at room temperature.The emission spectra were recorded exciting at 469.5 nm.The spectra are plotted with an offset for displaying purposes.The dashed lines in A) and B) correspond, respectively, to 472 and 535 nm, which are the absorption and emission maxima of 11Green-AgNC.

Figure 3 .
Figure 3. A) Mass spectrum of DNA 2 [Ag 11 ] 7+ system (11Green-2end-AgNC).The sum formula is C 174 H 220 N 72 O 100 P 16 [Ag 11 ] 7+ , which corresponds to a molecular mass of 6602.21 g mol −1 .B) Zoomed-in view of the molecular ion peak with z = 4 − charge state.The experimental isotopic distribution is reported with the corresponding Gaussian fit (blue) and the theoretical isotopic distribution (orange).The calculated average mass is 1647.79m/z.Additional zoomed-in views can be found in the Supporting Information.
, Supporting Information), similar to the solution state value, whereas the red emissive band shows an average decay time of 167 μs (Figure S28, Supporting Information).It is still an open question whether the red-shifted "phosphorescence-like" emission is due to an actual spin change to a triplet state or not.Our findings indicate that (DNA) 2 -[Ag 11 ] 7+ NC can change from being a single emitter in solution to a dual emitter in the solid state.This means that crystal packing interactions most likely affect the geometry of the [Ag 11 ] 7+ NC, and hence modify the photophysical parameters observed in solution.

Figure 4 .
Figure 4. Structure of one subunit of 11Green-2end-AgNC crystal.A) Density-modified electron density map of silver atoms obtained by the SAD phasing, contoured at 5 level.B) Nearest inter-silver distances (up to 3.4 Å) represented by colored dashed lines to highlight the subsections of the [Ag 11 ] 7+ nanocluster.Detailed views of C) the tetragon pair with shared bond and edge, and D) the square planar section with a single atom extension.All distances are given in Å.

Figure 5 .
Figure 5. A) Bright-field and B) fluorescence images ( exc = 470-495 nm,  em = 510-550 nm) of 11Green-2end-AgNC crystals grown in 10% MPD, 10 mM spermine, 10 mM Ca(NO 3 ) 2 and 50 mM MOPS at pH 7. The scale bar corresponds to 50 μm.C) Emission spectra of different crystals, recorded with a confocal microscope exciting at 458 nm.The spectra are normalized to the emission maximum and have a constant 0.2 offset for displaying purposes.

.
Overview of the steady-state and time-resolved solution data of 10Green-AgNC and derived mutants.