Fluorescence-tunable Ag-DNA biosensor with tailored cytotoxicity for live-cell applications

DNA-stabilized silver clusters (Ag-DNA) show excellent promise as a multi-functional nanoagent for molecular investigations in living cells. The unique properties of these fluorescent nanomaterials allow for intracellular optical sensors with tunable cytotoxicity based on simple modifications of the DNA sequences. Three Ag-DNA nanoagent designs are investigated, exhibiting optical responses to the intracellular environments and sensing-capability of ions, functional inside living cells. Their sequence-dependent fluorescence responses inside living cells include (1) a strong splitting of the fluorescence peak for a DNA hairpin construct, (2) an excitation and emission shift of up to 120 nm for a single-stranded DNA construct, and (3) a sequence robust in fluorescence properties. Additionally, the cytotoxicity of these Ag-DNA constructs is tunable, ranging from highly cytotoxic to biocompatible Ag-DNA, independent of their optical sensing capability. Thus, Ag-DNA represents a versatile live-cell nanoagent addressable towards anti-cancer, patient-specific and anti-bacterial applications.

determined by the variant sizes and aspect ratios of the noble metal nanoclusters 36 . Similar to larger noble metal nanoparticles, larger silver rods correspond to a longer wavelength resonance than smaller particles [38][39][40] , with the DNA template also playing a role in the fluorescence variance of the emitters.
Using these dependencies, we designed Ag-DNA templates to function as an optical sensor in living cells' interior environments (Fig. 1b-d) (for DNA sequences please see Table S1). At the same time, by utilizing the endogenous silver core, we tuned Ag-DNA from cytotoxic to completely biocompatible ( Fig. 1b-d), creating an intracellular toxicity-tunable optical sensor. Here, we demonstrate the functionality of the designed nanoagents in a live-cell environment.

Results and Discussion
Ag-DNA nanoagent design. The design of DNA sequences is the result of optimizing fluorescence and/ or sensor properties of Ag-DNA within the cell by varying the base sequences for enhanced performance. The strong interaction of cytosine and guanine bases with Ag(I)ions allows effective stabilization of a silver cluster. Furthermore, the addition of adenine and thymine bases throughout the DNA strand increase the yields of fluorescent species and fluorescence quantum yields. By varying the base sequences, the resulting different sizes and shapes of Ag-DNA structures define their particular functionality as nanoagents.
First, we demonstrate a robust and bright Ag-DNA structure, applicable as a fluorescent marker inside living cells. The DNA template, consisting of 28 single bases, is designed to yield an Ag-DNA construct (Ag-28b in Fig. 1b) with strong and stable fluorescence properties upon insertion into living cells. Thus, in this DNA-design, the encapsulated silver cluster preserves its fluorescence properties when inside living cells, not reacting to the complex intracellular environment.
While the relatively long single strand wrapping around the silver cluster seems to protect the cluster enough to maintain a strong fluorescence, any exposed parts of the silver surface could still interact with cellular molecules, altering cellular processes.
Small changes in the DNA template can have a significant effect on the functionality of the Ag-DNA nanomaterials. Here, we demonstrate that by a simple modification of Ag-28b, we produced an Ag-DNA construct which shows optical sensor functionality inside living cells: shortening the 28b sequence by nine bases produces the Ag-19b construct (Fig. 1c), which exhibits an optical response to the intracellular environment. By using the shorter sequence, we reduce the shielding of cellular molecules from the silver surface as compared to Ag-28b, resulting in a higher level of cytotoxicity and yielding a stronger sensitivity to the change in conditions inside living cells.To decrease the toxicity level to a biocompatible Ag-DNA for an optical biosensor sensitive to intracellular ions, we designed an Ag-hairpin (Ag-HP) construct (Fig. 1d). As the molecular dynamics of single-and double-stranded DNA heavily depend on the presence of salts 42 , DNA-ions interactions lead to conformational (a) Fluorescent Ag-DNA are synthesized upon reduction of silver ions by sodium borohydride (NaBH 4 ) in the presence of a specific DNA template. The designed Ag-DNA structures, utilized here, show different functionalities as a fluorescent marker or as an optical sensor, as well as exhibit tunable cytotoxicity or biocompatibility. DNA sequences color code are displayed for (b) Ag-28b, (c) Ag-19b and (d) Ag-HP by Cytosine = yellow, Guanine = green, Thymine = blue, Adenine = red. changes of the hairpin loop resulting in optical emission shifts inside living cells, yielding an optical sensor. The amount of cytosine bases in the loop, and the length and sequence of the double stranded stem were optimized for sensitivity. The Poly-C-loop wrapping around the silver cluster through strong Cytosine-Ag(I)ions interactions prevents the silver surface from interacting with cellular molecules, yielding a biocompatible construct.
Fluorescence-tunable properties of Ag-DNA optical sensors. By acquiring the excitation and emission spectra, the fluorescence properties of the Ag-DNA structures were investigated in solution and in comparison in real cell environments (Figs 2 and 3). All three types of Ag-DNA were quickly internalized into living cells and showed a homogenous distribution throughout the cytoplasm and the cell nucleus ( Figure S1).
In phosphate buffer (PB), Ag-28b exhibits a single fluorescence peak at λ ex = 605 nm/λ em = 665 nm (Fig. 2a). When exposing these Ag-DNA structures to the interior of a living cell, the fluorescence properties of Ag-28b are not influenced, as indicated by the red line in Fig. 2a,b. The second peak at λ ex = 470 nm/λ em = 550 nm arises from the autofluorescence of D. discoideum wildtype cells (Fig. 2b,c) (indicated by the green line), as observed in all intracellular fluorescence spectra.
Shortening the DNA template reveals a significant effect on the environmental sensing function of the Ag-DNA construct. Inside cells, Ag-19b show a significant fluorescence shift of Δ λ ex = 130 nm/Δ λ em = 120 nm (Fig. 2d,e) (indicated by the red lines). This relocates the dominant peak observed in PB at λ ex = 495 nm/ λ em = 560 nm to λ ex = 625 nm/λ em = 680 nm in the living cell.
A fluorescence peak in this wavelength range is typically exhibited by Ag-DNA with DNA strands longer than 19 bases, such as in Ag-28b, which stabilize larger silver clusters. Hence, these larger clusters likely require higher molecular stability to reach the fluorescent state, which can be provided by the interactions of Ag-19b with the nuclear DNA, as well as relatively dense cellular structures like membrane proteins and lipids. Accumulation of Ag-19b in these cellular regions was observed, as shown in Fig. 2f-h.
The hairpin structure (Ag-HP) was designed as a biosensor specific to intracellular ions. When in PB, Ag-HP emits at λ ex = 540 nm/λ em = 590 nm (Fig. 3a). Upon internalization by cells, its fluorescence properties show a split into two peaks. The previously observed peak is largely retained. Additionally, a second peak is observed at λ ex = 580 nm/λ em = 650 nm, thus providing a shift in emission of Δ λ em = 60 nm (Fig. 3a,b) (indicated by the red lines).

Figure 2.
Fluorescence excitation/emission spectra of Ag-DNA nanoagents of type Ag-28b and Ag-19b, before (a,d) and after (b,e) internalization in D. discoideum wildtype cells. Fluorescence spectra are measured within the excitation range of 470-670 nm and the emission range of 490-760 nm, each in 10 nm steps. Ag-28b exhibits a fluorescence peak at λ ex = 605 nm/λ em = 665 nm in PB (a), which is preserved upon internalization in living cells (b) (illustrated by the red line). Without addition of Ag-DNA variants, cell interiors exhibit a peak at λ ex = 470 nm/λ em = 550 nm, corresponding to autofluorescence (c) (illustrated by the green lines). Ag-19b shows a dominant emitter at λ ex = 495 nm/λ em = 560 nm in PB (d). Upon internalization in living cells, the peak shifts to λ ex = 625 nm/λ em = 680 nm (e) yielding a shift of Δ λ ex = 130 nm/Δ λ em = 120 nm (illustrated by the red lines). The regions of interest (ROIs) for intracellular fluorescence spectra are displayed in Figure S3. Confocal fluorescence images display the accumulation of Ag-19b at cellular membranes and inside the cell nucleus (f). The nucleus is labeled by GFP-histone (green) (g) and nuclear localization is shown by the overlay (yellow) (h) of the Ag-19b signal (red) (f) with GFP-labeled histones. The scale bar is 10 μ m.
Scientific RepoRts | 6:37897 | DOI: 10.1038/srep37897 To confirm the role of ions in the observed red-shifted fluorescence peak, we exposed Ag-HP to various salt types and concentrations in PB. By the addition of 40 mM Mg(OAc) 2 in combination with 30 mM NaCl (Fig. 3c), the red-shifted intensity distribution near λ ex = 580 nm/λ em = 650 nm, as observed inside living cells in Fig. 3b, was reproduced (Fig. 3b,c) (indicated by the green line). Exposing Ag-HPs only to monovalent (Na + ) or only to divalent (Mg 2+ ) cations did not reproduce the spectrum in Fig. 3b. Cations in particular reduce the flexibility of double stranded DNA, and can induce a conformational change, depending on the type of cations bound 43 . In high salt conditions, even chloride anions have been suggested to localize on parts of the DNA surface, as well as in the grooves 44 . We observed that the addition of chloride anions was necessary for reproducing the shift of the peak in PB to higher wavelengths, that was measured in living cells, in Mg(OAc) 2 -NaCl. Noble metal nanoparticles and clusters are known to exhibit sensitivity to their environment, like the dielectric constant of a medium, as is reflected by changes in their plasmonic resonance 39 . However, for small particles these effects typically lead to relatively small shifts in resonance 40 , which has been demonstrated recently for purified, individual Ag-DNA species, leading to an absorption shift of up to 8 nm 36 . Here, we utilize the fact that synthesis of Ag-DNA yields a heterogeneous distribution of different Ag cluster sizes and DNA conformations 45 . After synthesis, the formed silver nanoclusters can occur in a fluorescent or a dark state 45 . Hence, we attribute the observed optical shifts to rearrangements in switching these optically different emitters 'on' or 'off ' , resulting from conformational changes of the DNA template. This way we facilitate much larger shifts of up to Δ λ em = 120 nm, which allow the varying responses of Ag-19b and Ag-HP to be easily separated and identified by conventional microscopy methods. In particular, DNA hairpin structures have been shown to typically stabilize a broader range of active fluorescent emitters than the single stranded variants 46 , making them very suitable for biosensing applications.
Tunable cytotoxicity of Ag-DNA. In addition to this biosensing functionality, we utilize the toxicity effect of endogenous silver clusters in Ag-DNA for live-cell applications. We show the tunability of cytotoxicity in Ag-DNA, which is controlled independently of the optical sensor ability of the Ag-DNA constructs.
In the cytotoxicity experiment, 100% cell viability corresponds to control experiments with cells not exposed to Ag-DNA or AgNO 3 agents under the same conditions. Figure 4 displays the concentration-dependent toxicity of Ag-28b, Ag-19b, and Ag-HP on D. discoideum wildtype cells, with exposures ranging between 75 nM and 36 μ M. Ag-HP exhibit biocompatibility, with cell viabilities within 80-100% at all concentrations. When incubated with Ag-19b and Ag-28b, cell survival was increasingly impaired with rising Ag-DNA concentrations. For Ag-28b a slight decrease in cell viability was observed, with over ~60% viable cells at 36 μ M. Ag-19b had the strongest negative effect on cell survival, still yielding ~30% of living cells at the highest concentration. As a reference measurement for cytotoxicity we used AgNO 3 , which readily releases free Ag(I)ions in solution. This resulted in the highest cytotoxic effect for the same concentration of silver, as compared to the DNA-stabilized silver clusters.
To exclude other sources for cytotoxicity beside Ag-DNA, the following steps were taken. We removed potentially harmful byproducts or educts possibly causing toxicity by filtration of the synthesized Ag-DNA. Also we confirmed experimentally that a toxic effect caused by oligonucleotides is unlikely. Furthermore, for the cytotoxicity experiments, incubation steps and subsequent cell viability assays were performed in the dark, to exclude possible light-induced toxicity, like ROS generation. These points allow us to identify the silver in the Ag-DNA constructs as the only significant source of cytotoxicity, which is supported by the biocompatibility of Ag-HP.
We conclude from these results that the DNA templates can protect cells from silver and the selected DNA-sequence determines the toxic effect, ranging from non-toxic for Ag-HP to highly toxic for Ag-19b. This mechanism of protection results from distinct interactions of silver with DNA bases and the steric conditions of the Ag-DNA, varying the degree of silver exposure to cellular structures. Due to high binding energies of silver clusters to cytosine and guanine bases, which are several times stronger than Watson-Crick base pairs 46 , we can assume that the silver clusters mainly remain contained in the Ag-DNA structures. Hence, the observed cytotoxicity effect of single stranded Ag-DNA nanomaterials most likely results from direct contact of cellular structures and molecules with exposed areas of the silver clusters as well as possibly released individual Ag(I)ions.
In Ag-HP, the silver cluster is surrounded by a Poly-C loop. As guanine and cytosine bases in particular show a high affinity to Ag(I)ions 45,47,48 , multiple Ag(I)ions-DNA base connections are formed which lead to a strongly shielded silver cluster, allowing for a decrease in Ag-DNA cytotoxicity. Both Ag-28b and Ag-19b constructs are single stranded DNA templates, which are expected to wrap around the silver cluster upon reduction with NaBH 4 , due to Ag(I)ions-base bonding. The sequences in Ag-28b and Ag-19b lack the high density of cytosines of the Poly-C loop in Ag-HP, hence a larger area of the silver nanocluster is exposed in Ag-28b and Ag-19b, which can interact with surrounding molecules and impact cellular processes. Since Ag-28b provides a higher amount of cytosine and guanine bases, and additionally, its longer strand offers the possibility of a more stable Ag-DNA conformation, it yields a lower cytotoxicity than its shorter counterpart Ag-19b. In these three variants, we show the potential for cytotoxicity tuning by adapting the cytosine and guanine amount and location in the implemented DNA template.

Conclusion
In conclusion, we show that the unique properties of Ag-DNA allow for nanomaterials with multiple functionalities, targeted at optical sensing in combination with tunable cytotoxicity, by simple modifications of Ag-DNA designs. Ag-28b exhibits stable fluorescence properties and is thus suitable for bright and persistent labeling of cells, as well as localization experiments. Its shorter counterpart, Ag-19b, is environmentally sensitive with an optical emission shift of Δ λ em = 120 nm and exhibits higher cytotoxicity, making it interesting as an anti-bacterial or anti-cancer agent. The biocompatible, non-toxic Ag-HP optically responds to intracellular salt concentrations, visible as a large fluorescence shift of about Δ λ em = 60 nm, which allows a convenient detection of signal changes for both, Ag-HP and Ag-19b, by commonly used fluorescence microscope systems.
These results will allow us to extend these Ag-DNA constructs to more complex DNA-origami structures, and the investigations from single cells to living organisms. Therefore, Ag-DNA represent a new nanomaterial with high potential for a multi-functional nanoagent, combining bio-sensing and effector functionality. Possible targeted DNA-sequence specific applications would open a large field of patient-specific treatment including cancer or rare and degenerative diseases, as well as specific antimicrobial targeting.  Ag-DNA preparation. For synthesis of (a) Ag-28b and Ag19b, and (b) Ag-HP solutions, 50 nmol of the respective DNA molecules were mixed with AgNO 3 in a ratio of (a) 9.6:1 silver ions:DNA strands, or (b) 7:1 silver ions:DNA strands in 1 ml 20 mM NH 4 OAc. The solutions were incubated for 30 min at 4 °C and then reduced with NaBH 4 at 4 °C over night. The reduction ratios were (a) 0.5:1 and (b) 0.3:1 NaBH 4 :Ag + . To exclude formed silver clumps and unbound Ag(I) ions as well as other salts, Ag-DNA constructs were filtered using a 3 kDa centrifugal filter and subsequently a 50 kDa centrifugal filter, simultaneously exchanging the buffer with phosphate buffer (PB; 3.57 mM KH 2 PO 4 , 3.46 mM Na 2 HPO 4 , pH 6.0). The concentration of the DNA was calculated based on the absorption at 260 nm, measured with a Cary 50 UV-Vis spectrophotometer (Varian Medical Systems, Netherlands). For the reproduction of the observed peaks, various concentrations of Mg(OAc) 2 and NaCl were added to a 15 μ M solution of Ag-HP in PB and the fluorescence properties measured. The excitation and emission properties of the Ag-DNA in PB were determined by Cary Eclipse fluorimeter (Varian Medical Systems, Netherlands).

Methods
Cell culture. Dictyostelium discoideum HG1729 histone-GFP in AX2 and Dictyostelium discoideum Wildtype in AX2 cells were cultured in HL5-C medium adjusted to pH 6.7 at 21 °C; medium for HG1729 histone-GFP was supplemented with 10 μ g/ml Geneticin. For the experiments the cells were harvested below a confluency of 40% and washed three times with PB by centrifugation at 400 g for 5 min. Subsequently, cells dispersed in PB were seeded onto an observation chamber (composed of a cover glass and a custom-made Teflon ® frame) and rested for 30 min, allowing the cells to settle down.
Microscopy. Imaging of cells incubated with Ag-DNA was performed by confocal microscopy set-up equipped with a Yokogawa 10,000 rpm spinning disc unit (Andor Technology Ltd., UK), Eclipse TiE Nikon inverted microscope (Nikon Corporation, Japan) and an IXON Ultra EMCCD camera (Andor Technology Ltd., UK). Ag-28b and Ag-19b were added to D. discoideum HG1729 histone-GFP cells to an end concentration of 4.5 μ M and 3 μ M, respectively, in an observation chamber and immediately imaged with a 100x oil immersion objective with NA 1.45. Ag-28b and Ag-19b were excited at 640 nm, histone GFP at 488 nm. As negative controls, also cells without Ag-DNA were imaged, as well as cells incubated in 36 μ M AgNO 3 and 4.5 μ M 28b-DNA, separately and mixed. Local fluorescence spectra measurements were performed with a Leica TCS SP8 STED 3X Microscope and with a 63x objective lens (Leica Microsystems, Germany), additionally equipped with HyD detectors (Leica Microsystems, Germany). The excitation wavelengths ranged from 470 to 670 nm in 10 nm steps and the emission wavelengths ranged from 480 to 770 nm, with a step size of 10 nm and a detection window of 10 nm. For this assay, solutions of Ag-28b (4.5 μ M), Ag-19b (3 μ M) and Ag-HP (15 μ M) were incubated with D. discoideum wildtype cells.
Cytotoxicity assay. D. discoideum wildtype cells were harvested and washed three times with a maltose-loaded phosphate buffer (3.57 mM KH 2 PO 4 , 3.46 mM Na 2 HPO 4 , 52.58 mM Maltose, pH 6.70) by centrifugation at 400 g for 4 min. Cells were incubated for 2 h with Ag-28b, Ag-19b and Ag-HP, with concentrations ranging between 75 nM and 36 μ M, as triplicates with 2.5 × 10 4 cells/condition in the dark. As a positive control, incubation with AgNO 3 was performed under the same conditions, as well as a negative control without agent incubation. To determine cell viability, cells were incubated with 50 μ g/ml Fluorescein diacetate for 15 min, and the average fluorescence intensity was measured using a TECAN Infinite M200 well plate reader (Tecan Trading AG, Switzerland). To avoid background fluorescence from the Ag-DNA, 50 mM NaCl was introduced to all samples prior to measurements.