Low picomolar, instrument-free visual detection of mercury and silver ions using low-cost programmable nanoprobes

A single gold nanoprobe can be programmed for low picomolar visual detection of inorganic mercury and/or silver in water, soil or urine samples.

The EPA's recommended maximum allowable level of inorganic mercury in drinking water is 2 ppb (10 nM).
To our knowledge, the most sensitive colorimetric mercury sensor reported to date has a limit of detection (LOD) of 800 pM. Here, we report an instrument-free and highly practical colorimetric methodology, which enables detection of as low as 2 ppt (10 pM) of mercury and/or silver ions with the naked eye using a gold nanoprobe. Synthesis of the nanoprobe costs less than $1.42, which is enough to perform 200 tests in a microplate; less than a penny for each test. We have demonstrated the detection of inorganic mercury from water, soil and urine samples. The assay takes about four hours and the color change is observed within minutes after the addition of the last required element of the assay. The nanoprobe is highly programmable which allows for the detection of mercury and/or silver ions separately or simultaneously by changing only a single parameter of the assay. This highly sensitive approach for the visual detection relies on the combination of the signal amplification features of the hybridization chain reaction with the plasmonic properties of the gold nanoparticles. Considering that heavy metal ion contamination of natural resources is a major challenge and routine environmental monitoring is needed, yet timeconsuming, this colorimetric approach may be instrumental for on-site heavy metal ion detection. Since the color transition can be measured in a variety of formats including using the naked eye, a simple UV-Vis spectrophotometer, or recording using mobile phone apps for future directions, our costefficient assay and method have the potential to be translated into the field.
Environmental contamination by heavy metal ions is a serious global concern. 1,2 Mercury is one of the most toxic heavy metals which accumulates biologically and causes various health problems including brain damage, kidney failure, and various motion disorders. [3][4][5] Since even small quantities of mercury can be highly toxic it is critical to develop highly sensitive methodologies for its detection. 6 Mercury can exist in various forms: organic, inorganic or elemental where each species can display a different level of toxicity. 2 While the classication of each species is important for taking proper precautions, the detection of inorganic mercury, which is considered to be the precursor of organic or elemental mercury, has captured particular attention. [7][8][9][10][11] Though not considered as toxic as mercury, inorganic silver is another highly toxic metal ion which can also result in a variety of human health problems and has been found to be extremely toxic to aquatic animals. [12][13][14][15] The amount of mercury and silver can be found above its tolerable contamination level in water and soil samples, and can be ingested through drinking water or via the food chain. Therefore, it is critical to monitor natural sources for potential heavy metal contamination. [16][17][18] Particularly, the recent lead contamination incident in the drinking water supply of Flint, Michigan (USA), followed by numerous poisoning cases and fatalities, reminded us, once again, of the signicance of heavy metal ion monitoring in our natural resources. 19,20 According to the U. S. Environmental Protection Agency (EPA) the recommended maximum level of inorganic mercury in the drinking water is 2 ppb (10 nM) 21 and the tissue-based water from sh is 3 ppm (1.5 mM). 22 The Agency for Toxic Substances and Disease Registry (ATSDR) of the U. S. Department of Health and Human Service sets the highest mercury concentration as 625 ppb for normal soil. 23 The EPA recently reported a four-year long study on different species of sh from the 76 559 lakes in 48 states of the USA, of them, 49% of the sampled population of lakes (36 422 lakes) exceeded the EPA recommended concentration. 22 These results urged us to develop practical sensors that can provide on-site accurate results in shorter time frames. Such an enterprise will enable us to take quicker actions to avoid mercury poisoning through the consumption of water or food supplies from contaminated resources. Besides the detection in the natural resources, industrial sites or agricultural products it is also critical and thus required to monitor mercury content in human bodily uids to determine the level of exposure. For instance, the New York State (NYS) sanitary code requires healthcare providers to report blood or urine mercury levels to the NYS Department of Health when mercury concentration is at or above 5 ppb in blood and 20 ppb in urine. 24 These standards, along with many other studies, stress the importance of and need for easy-to-use, fast and cost-efficient mercury sensors for environmental and biological screenings. 25 At present, there are various methods available for the detection of heavy metal ions: atomic absorption spectroscopy, 26 inductively coupled plasma mass spectrometry (ICP-MS), 27 cold vapor atomic uorescence spectroscopy (CV-AFS) 28 and atomic uorescence spectroscopy. 29 Though these wellestablished methods are very sensitive and provide accurate results, they require large sample volumes, time-consuming sample processing prior to the detection and trained personnel for the operation of sophisticated instrumentation. Moreover, the instruments needed for these detection approaches are costly and not portable, and therefore are not suitable for onsite quick screening. There have been numerous attempts to develop methodologies for mercury detection using uorescence, Raman or absorbance spectroscopies as an alternative to the existing methodologies. [30][31][32][33] Among those the most attractive ones for on-site applications are the ones with colorimetric read-out signals, however most of these approaches suffer from low sensitivity or selectivity. 32,[34][35][36] Here we combined the programmable features of DNA nanotechnology with the plasmonic properties of metallic nanoparticles for the development of one of the most sensitive inorganic mercury and/or silver sensors using specic pyrimidine interactions in DNA duplexes as described below. [37][38][39][40][41][42] Studies have shown that thymine-thymine (T-T) and cytosine-cytosine (C-C) base pairs are very selective for capturing Hg 2+ and Ag + and form T-Hg 2+ -T and C-Ag + -C bridges in DNA duplexes, respectively. [43][44][45] These pyrimidine base pairings with Hg 2+ and Ag + , which are stronger than typical Watson-Crick base-pairings, are central for our reprogrammable and selective detection scenario described herein. In this study, the pyrimidine pairing is followed by a hybridization chain reaction (HCR) for the amplication of the colorimetric signal readout.
The hybridization chain reaction (HCR) is an enzyme-free DNA polymerization process where two metastable species of DNA hairpins (H1 & H2) coexist in solution without binding to each other (Fig. 1a, step 1). 42,46,47 However, in the presence of a specic short single stranded (ss) DNA (initiator ¼ I) molecule the hairpins are activated and self-assemble to form a long double stranded (ds) DNA polymer. Briey, the HCR process is as follows: the initiator (I) binds and opens the rst hairpin (H1, Fig. 1a, step 2) which immediately binds and opens the second hairpin (H2, Fig. 1a, step 3). The opening of H2 forms a dsDNA with a sticky end and induces the activation of another H1 initiating the HCR, (Fig. 1a, step 4). As a result, two hairpin pairs assemble into a long DNA polymer, which was unable to happen in the absence of the initiator strand. Because of its extraordinary signal amplication properties, this enzyme-free DNA polymerization step is essential for achieving the outstanding sensitivity observed in our approach. [40][41][42]

Results and discussion
The studies were performed using two different initiators (I Hg and I Ag ), each one used for programming the sensor template for a different metal ion detection; I Hg for mercury or I Ag for silver ions. Prior to the detection of the metal ions, the HCR was validated using gel electrophoresis using various amounts of initiator (I Hg ) with respect to the H1 & H2 hairpin pair (0.1, 0. . Aer conrming the affinity of HCR in the presence of the designed initiator strands (I Hg and I Ag ) and both hairpins (H1 & H2), we used these components for the detection studies illustrated below. Fig. 2a illustrates the schematics of the programmable and visual detection of the Hg 2+ and Ag + by combining the ampli-cation feature of the HCR with the surface plasmon properties of the gold nanoparticles. First, the 20 nt-long thiolated capture probe (Cp Hg-Ag ) was immobilized on a gold nanoparticle forming the AuNP-Cp Hg-Ag nanoprobe. The rst half of the Cp Hg-Ag , from the 5 0 end, was designed for the detection of Hg 2+ whereas the second half was used for the detection of Ag + (Fig. 2a). Separately, two different short single stranded initiators (I Hg for Hg 2+ and I Ag for Ag + ) were designed for programming the nanoprobe for individual or simultaneous detection of the metal ions. The 10 nts at the 3 0 ends (light colored blue region in I Hg and green region in I Ag , Fig. 2a) of the initiator strands serve as the target-recognition regions, while the remaining 24 ntlong regions (orange colored, Fig. 2a) are able to trigger the HCR and are identical for both initiators. The visual detection assay begins as follows; the underlined ve T bases in the rst half of the Cp Hg-Ag (dark blue region) and in the 3 0 tail of I Hg (light blue region) are bridged in the presence of Hg 2+ forming T-Hg 2+ -T pairs whereas the ve C pairs in the second half of the C Hg-Ag (dark green region) and in the 3 0 end of I Ag (light green region) are paired with Ag + forming and C-Ag + -C bridges (Fig. 2a inset). Aer the initiators bind to the AuNP-Cp Hg-Ag , the resulting puried nanoprobe assembly initiates the HCR by opening the rst hairpin, H1 (black colored hairpin, Fig. 2a), in the presence of the H1 & H2 hairpins. The opening of H1 triggers the opening of H2 (gray colored hairpin, Fig. 2a) and starts the HCR. As a result, the surface of the nanoprobe is covered with the DNA polymers composed of alternating H1 & H2 strands illustrated as black and gray stripes on AuNP-Cp Hg-Ag (Fig. 2a * in the lower le panel). This assembly occurs only in the presence of a specic metal ion (Hg 2+ or Ag + ) with its initiator pair, increasing the hydrodynamic size of the AuNP-Cp Hg-Ag from 16.9 AE 0.7 nm to 153.6 AE 20.3 nm aer the polymerization on the nanoprobe surface.
Finally, in order to monitor the effect of the HCR-induced DNA polymerization on the nanoparticle surface, and therefore evaluate the detection of Hg 2+ or Ag + , Mg 2+ ion salt ($55 mM nitrate salt) was added to the resulting nanoparticle assembly to aggregate the gold nanoparticles. 47 While AuNPs with low and/ or short DNA coating density can aggregate under this high salt condition, DNA polymers anchored on the surface of the nanoparticles can protect the nanoparticles from aggregation by shielding this charge effect. To demonstrate the working principle of the assay the studies were performed separately for both metal ions (Fig. 2b-d, shown only for Hg 2+ in the gure). As anticipated (Fig. 2a ** in the lower right panel), an immediate salt-induced AuNP aggregation was observed in the form of a red-purple-gray color transition (Fig. 2b) and shi in the surface plasmon band at $520 nm (Fig. 2d) in the absence of the target metal ion. On the other hand, in the presence of the target metal ion the DNA polymers on the gold surface protected the nanoparticles from Mg 2+ -induced aggregation, which was recorded by no change in the spectrum of the gold nanoparticles (Fig. 2c) and a retained red color of the colloidal suspension (Fig. 2b).
In our assay the H1 & H2 hairpins are the building blocks of the DNA polymers on the nanoprobe surface, the initiator strands (I Hg or I Ag ) are necessary to initiate the HCR and target metal ions (Hg 2+ and Ag + ) are required to anchor the HCRinduced DNA polymers on the nanoparticle surface. In order to validate the necessity of each component, we have performed absence tests for each target metal ion (Hg 2+ and Ag + ), separately. As seen in Fig. 3a and b, in the absence of any of the components (initiator, H1 & H2 hairpins or target metal ion), salt-induced AuNP aggregation was observed which was recorded as a color transition and a change in the absorbance spectra. On the other hand, only the nanoparticles with essential components for the HCR retained the original spectral information and color of the assay (Fig. 3a and b). Aer characterizing our system and validating it for the detection of Hg 2+ and Ag + separately, we next demonstrated the programmability feature of our nanoprobe for each target in a 2 Â 4 array as illustrated in Fig. 4. The AuNP-Cp Hg-Ag was programmed for the detection of Hg 2+ , but not Ag + , when I Hg was used as an initiator strand (programming unit). This was observed as a retained red color with Hg 2+ and a red-purple-gray color transition with Ag + . On the other hand, using I Ag instead of I Hg programs the same AuNP-C Hg-Ag for Ag + , but not Hg 2+ , (second row in Fig. 4a and b). In order to use the same nanoprobe for the simultaneous detection of Hg 2+ and Ag + , both initiators were included in the assay. The observed retained red color with both metal ions demonstrates that the same nanoprobe can be programmed for the simultaneous detection of both ions. Finally, in order to illustrate that the programmability of the nanoprobe depends on the initiators, studies were carried out in the absence of initiators. In the presence of any of the metal ions (Hg 2+ or Ag + ), nanoprobes aggregated (last row Fig. 4a and b) which conrms the necessity of the initiator strands for programming the nanoprobe for the detection. The results of the 2 Â 4 array tests in Fig. 4 overall suggest that the same gold nanoprobe (AuNP-Cp Hg-Ag ) offers visual and spectroscopic detection of Hg 2+ and Ag + in all four possible combinations which was achieved by only changing the compositions of the initiator strands in the assay while keeping every other parameter unchanged. The system is highly selective to the programmed settings with no false-positive or false-negative results.  The approach described here has a remarkable signal amplication feature because the intercalation of ve Hg 2+ or Ag + ions can trigger the HCR and anchor a DNA polymer; sized 1000 bps or longer (Fig. 1b); on the AuNP-Cp Hg-Ag surface that can protect the nanoprobes from a salt-induced color transition. Furthermore, since a single AuNP-Cp Hg-Ag is conjugated to hundreds of capturing probes, this protection is far more pronounced with multiple polymerization sites. The visual sensitivity of our system was tested using 10, 100 and 500 pM and 1, 2, 5 and 10 nM of Hg 2+ and Ag + , separately. The results demonstrate that the assay undergoes a color transition within minutes in the absence of the target metal ion (blank, reference well) and as low as 10 pM (0.4 fmol) of Hg 2+ or Ag + can be differentiated with the naked eye (rst row in Fig. 5a and b). Aer the color transition has stabilized, the difference becomes exceptionally obvious, about an hour aer the addition of Mg 2+ , and as low as 100 pM of Hg 2+ or Ag + can be visually detected. Additionally, the studies performed by monitoring the aggregation rate (Fig. 5c and d) and degree ( Fig. 5e and f), obtained by recording Abs 520/700 with various concentrations of Hg 2+ and Ag + , illustrate the sensitivity of the assay spectroscopically.
Later the selectivity of the assay for Hg 2+ and Ag + was determined using 1 mM of Cd 2+ , Mn 2+ , Cu 2+ , Pb 2+ , Zn 2+ , Co 2+ , Ni 2+ and 1 mM of Ca 2+ and Mg 2+ . As seen in Fig. 6a and b, any of these metal ions or a cocktail prepared by combining all of these metal ions in a single well gave a negative result as predicted.   6 Selectivity for Hg 2+ or Ag + . The selectivity of the assay for (a) Hg 2+ or (b) Ag + is evaluated by testing the nanoprobe with Cd 2+ , Mn 2+ , Cu 2+ , Pb 2+ , Zn 2+ , Co 2+ , Ni 2+ , Ca 2+ , Mg 2+ and a soup of all metal ions (last well), where the Ca 2+ and Mg 2+ concentrations were chosen as 1 mM and the remaining metal ions' concentrations were 1 mM. In the second row of the assay the studies were performed after the addition of 1 nM of (a) Hg 2+ or (b) Ag + into each metal ion mixture. Abs 520/700 was recorded for each well 1 h after the addition of Mg(NO 3 ) 2 . Experiments were performed in triplicate.
With only the addition of 1 nM of Hg 2+ (Fig. 6a) or Ag + (Fig. 6b) in each well, including the last well with the metal soup, were the nanoprobes protected from aggregating and gave a positive signal. The results overall indicate that the nanoprobe is highly selective to the programmed metal ion and can recognize the target ion selectively in a complex metal ion soup.
Finally, we tested our assay for inorganic mercury detection in real water and soil samples. The array (Fig. 7a) was prepared as follows: experimental well (E) was used to monitor Hg 2+ in the real sample, well S was used to detect Hg 2+ in the real sample spiked with 1 nM of Hg 2+ and a reaction buffer prepared from ultrapure water was used for the reference well (R). The water samples (w 1 , w 2 , w 3 and w 4 ) were obtained from tap water, pond water, river water and lake water, respectively, near regional resources. The results illustrate that the detection of inorganic mercury in real water samples can be performed using this colorimetric methodology (Fig. 7b). Since detection of mercury is important biologically, industrially and agriculturally we also demonstrated the detection of Hg 2+ in soil and urine samples which were initially spiked with 1 nM of Hg 2+ (Fig. 7c-d). As predicted the samples spiked with 1 nM inorganic mercury retained the nanoprobe's original color whereas in the absence of the target ion a clear color transition was observed. The results overall demonstrate that our assay is capable of detecting inorganic mercury contamination in different environmental or biological matrices.
Though we only focused on using 13 nm sized gold nanoparticles for this study, the performance of the sensor could be further improved by using different sizes and shapes of gold nanoparticles. The surface coverage and the size of the HCRproduct on the nanoparticle surface can be increased with bigger nanoparticles, which could, in turn, offer greater sensitivity.
The thymine-thymine (T-T) and cytosine-cytosine (C-C) base pairs in the DNA designs are very specic to Hg 2+ and Ag + , respectively, however a small-or macro-molecule that has a binding affinity to either one of the metal ions could interfere with the sensor's performance. This could be circumvented by either nitri-cation or a UV digestion procedure prior to detection. 48,49 Conclusions In this study we developed an instrument-free, ultrasensitive, cost-efficient and colorimetric assay, which offers visual detection of as low as 10 pM of inorganic mercury and silver. To our knowledge this is the most sensitive instrument-free detection of each metal ion. We have used our assay for the detection of metal ions in complex metal ion cocktails, water, soil and urine samples. The assay can be used for the individual or simultaneous detection of each metal ion. The assay utilizes the unique pyrimidine-pyrimidine base pairing for specicity, the hybridization chain reaction for sensitivity and programmability, and the plasmonic properties of gold nanoparticles for visual detection. The components of the nanoprobe are inexpensive and easy to obtain or synthesize which makes it ideal for the development of practical sensors for on-site detection. A single test costs less than a penny per well of a 384-well microplate. Measurement of the color change in the assay can be achieved in a variety of formats including by naked eye, UV-Vis spectroscopy and in the future possibly using mobile phone apps. Considering that the heavy metal ion contamination of natural resources is a major concern, this highly practical colorimetric approach could be ideal for on-site heavy metal ion detection not only for resourcelimited areas but also for developed industrial sites.

Materials
All oligonucleotide sequences were purchased from Integrated DNA Technologies (IDT), USA with the following sequence information. used for the study. The color-transitions were recorded 45 min aer the addition of $55 mM Mg(NO 3 ) 2 .
Detection in water. To test real samples with the proposed method, water samples (w 1 : tap water in UAlbany campus, w 2 : UAlbany Indian Pond, w 3 : Hudson River and w 4 : Rensselaer Lake) from regional sources were collected. All samples were ltered through a 0.22 mm syringe lter prior to the tests. The unspiked samples were used as is and the spiked samples were prepared by the addition of 50 mL of stock Hg 2+ solution into the 450 mL water samples to a nal concentration of 10 nM of Hg 2+ . 20 mL of these samples were added into 180 mL of sensor solution to a nal concentration of 1.0 nM of Hg 2+ (for the spiked sample). The experimental phosphate buffer prepared using ultrapure water was used as a reference. The remaining HCR steps were performed as described above.
Detection in soil. Soil samples were collected from the UAlbany campus. One gram of soil was mixed with 5 mL of ultrapure water, which was followed by a two-step ltration process using standard lter paper and a 0.22 mm syringe lter, respectively. For the spiked sample 10 nM of Hg 2+ was added prior to the ltration process. 20 mL of the samples were added into 180 mL of sensor solution to a nal concentration of 1.0 nM of Hg 2+ (for the spiked sample). The experimental phosphate buffer prepared using ultrapure water was used as a reference. The remaining HCR steps were performed as described above.
Detection in urine. The unprocessed urine sample was used to illustrate the detection in biological uids. Prior to the test, a fraction of the specimen (200 mL) was spiked with 10 nM of Hg 2+ . Both spiked and unspiked samples were centrifuged at 12 000 rpm at 4 C for 20 min to eliminate any cell pellets or large biological content. The supernatant was collected from the top two thirds of the microtube container and used without further processing. 20 mL of samples were added into 180 mL of sensor solution to a nal concentration of 1.0 nM of Hg 2+ (for the spiked sample). The experimental phosphate buffer prepared using ultrapure water was used as a reference. The remaining HCR steps were performed as described above.