Longer amplicons provide better sensitivity for electrochemical sensing of viral nucleic acid in water samples using PCB electrodes

The importance of monitoring environmental samples has gained a lot of prominence since the onset of COVID-19 pandemic, and several surveillance efforts are underway using gold standard, albeit expensive qPCR-based techniques. Electrochemical DNA biosensors could offer a potential cost-effective solution suitable for monitoring of environmental water samples in lower middle income countries. In this work, we demonstrate electrochemical detection of amplicons as long as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${503}\,\hbox {bp}$$\end{document}503bp obtained from Phi6 bacteriophage (a popular surrogate for SARS-CoV-2) isolated from spiked lake water samples, using ENIG finish PCB electrodes with no surface modification. The electrochemical sensor response is thoroughly characterised for two DNA fragments of different lengths (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${117}\,\hbox {bp}$$\end{document}117bp and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${503}\,\hbox {bp}$$\end{document}503bp), and the impact of salt in PCR master mix on methylene blue (MB)-DNA interactions is studied. Our findings establish that length of the DNA fragment significantly determines electrochemical sensitivity, and the ability to detect long amplicons without gel purification of PCR products demonstrated in this work bodes well for realisation of fully-automated solutions for in situ measurement of viral load in water samples.

. Platinum electrodes have also been used for an in-situ electrode during microfluidic PCR platform designed to electrochemically detect amplicons during reactions 8 . All these studies require surface modification of the electrode, thereby implying increased production and operating costs due to specialised storage requirements for stability of these functionalised electrodes.
We recently demonstrated electrochemical sensing of SARS-CoV-2 amplicons with low-cost printed circuit board (PCB) electrodes, based on change in DPV and cyclic voltammetry (CV) peak current due to adsorption of MB-DNA complex on the unmodified electrode surface 11 . We reported that longer DNA fragment (N1-N2, 943 bp ) formed using CDC-recommended N1 forward and N2 reverse primers exhibited better linearity in sensor response as compared to the shorter fragment (N1, 72 bp ) formed using N1 forward and N1 reverse primer set. These studies were reported using DNA dilutions prepared in nuclease free water. The platform was also used for detection of SARS-CoV-2 amplicons in simulated wastewater sample (obtained by spiking total RNA sample with SARS-CoV-2 RNA). The amplification of longer fragments with such heterogeneous samples is difficult, since RNA is susceptible to shearing during isolation and downstream processing 12,13 . Thus, the demonstration of electrochemical sensing of SARS-CoV-2 amplicons in wastewater was limited to the shorter 72 bp N1 fragment 11 .
In this work, we study the feasibility of ENIG PCB based electrochemical sensing of bacteriophage Phi6 concentrated and isolated from lake water samples (illustrated in Fig. 1). Phi6 phage is comparable to SARS-CoV-2 in size (80-100 nm), and also possesses lipid membrane and spike proteins. Because of these reasons, phage Phi6 is a popular surrogate for SARS-CoV-2 and other enveloped pathogenic RNA viruses 14,15 . RNA isolated from the phage particles was used as template for cDNA synthesis, that was then used as template for PCR to obtain two DNA fragments of lengths 117 and 503 base pairs. Given the challenges of amplifying 943 bp N1-N2 fragment in our previous work, we have targeted fragments of intermediate length ( 117 bp and 503 bp ) in this study, based on available primers. The electrochemical sensor response was systematically studied for a wide range of concentration ( 10 pg/µl to 20 ng/µl ) for both fragments in presence of MB, and the impact of salt on sensor response was characterised and cross-validated with spectrophotometry measurements. The key contributions of this work are as follows:  Target amplification from environmental water sample spiked with bacteriophage Phi6. Water from a lake on IIT Bombay campus (Powai lake, Powai, Mumbai) was used for spiking with phage particles. The lake water was filtered through a 5 µm membrane to remove suspended particles prior to spiking with Phi6 phage. 1 ml of 10 6 PFU/ml of Phi6 phage preparation was added to 100 ml of filtered lake water and gently homogenised for 10 min at 4 • C . A small aliquot of the sample was kept aside for measuring virus load by plaque assay. We tested two different methods for concentrating the spiked Phi6 virus particles: (1) aluminium hydroxide adsorption-precipitation method 19 , that has been validated for concentration of several enveloped RNA viruses from environmental samples, and (2) a polyethylene glycol (PEG) based virus concentration method adapted from Flood et al. 20 . Since recovery efficiency of the PEG-based method was found to be better than the aluminium hydroxide method, PEG based method was used for concentrating the Phi6 particles from the lake water sample.
The PEG method used is as follows: PEG 8000 and NaCl were added to the Phi6-spiked lake water sample to obtain a sample with 8 % PEG 8000 and 0.2 M NaCl . The samples were incubated in a shaker at 4 • C for 4 h , followed by centrifugation at 4700 g for 45 min . The supernatant was discarded and the pellet was resuspended in 1 ml of the same supernatant. All spiking and virus concentration experiments were conducted in triplicate. After concentration, a small aliquot was kept aside for measuring the recovery efficiency by plaque assay method. RNA was isolated as described earlier and eluted in 40 µl of elution buffer provided with the kit. Since the RNA concentration would differ from sample to sample in the triplicate, 2 µl of RNA was used for cDNA synthesis for all three samples, irrespective of their concentrations. cDNA synthesis was carried out as described earlier. 1 µl cDNA was used as a template for 20 µl PCRs for 35 cycles, to amplify 117 bp and 503 bp fragments. These samples are denoted '1:1' i.e. no dilution. A no-template control (NTC) was set up as a negative control, while positive control (PC) was set up using the cDNA synthesized from RNA isolated from purified phages as template. Quantitative PCR (qPCR) was performed using Brilliant III Ultra-Fast SYBR Green QPCR Master Mix (Agilent Technologies) in a Stratagene Mx3000P RT-PCR instrument. Reactions were set up as mentioned earlier, in triplicates. Cycle threshold (Ct) values were recorded for all samples. Additionally, diluted samples were prepared using 1 µl of 1:100 dilution of cDNA in filtered lake water as template for 20 µl PCRs for 35 cycles. These samples are denoted '1:100' .
Electrochemical biosensor. The PCB electrodes were manufactured using commercially available, lowcost electroless nickel immersion gold (ENIG) process without additional gold electroplating. The ENIG PCB electrode specifications are detailed in our previous work 11 . Conventional cleaning recipes for electrodes, such as using piranha solution or cyclic voltammetry with sulphuric acid are not recommended for ENIG PCB electrodes, as they may cause stripping of the thin gold layer (thickness ≈ 100 nm ) and expose the underlying copper layer that is susceptible to corrosion [21][22][23][24][25] . The electrodes were therefore cleaned using lint-free wipes dampened with IPA. The samples to be measured are incubated with 50 µM MB at 4 • C for 1 h to facilitate intercalation. www.nature.com/scientificreports/ In our previous work, we observed that the sensitivity and linearity of the sensor improved by increasing the MB concentration 11 . Based on the optimisation reported in our earlier work, we used 50 µM MB concentration for intercalation with DNA in this study. Electrochemical detection of double stranded DNA (ds-DNA) can be achieved using anionic or cationic intercalators. Although anionic intercalators detect DNA with better selectivity they require overnight incubation, resulting in longer assay time. Cationic intercalators such as MB, on the other hand, require lesser incubation time of approximately 1 h for electrochemical detection of ds-DNA 6 . Each measurement involved dispensing 5 µl of the sample to be tested on the electrode, followed by cleaning with IPA dampened wipes before performing the next measurement with another sample. Each sample was tested on 5 different electrodes, unless otherwise mentioned. DPV and CV measurements were performed using PalmSens Sensit Smart potentiostat, and PSTrace software was used for potentiostat configuration and data acquisition, including peak current calculation. The following settings were used for DPV and CV measurements:  Fig. S1 in supplementary information. Figure 2 shows the results (peak current) obtained for DPV and CV measurements with gel purified PCR products. DPV measurements show higher sensitivity (change in current with increase in DNA concentration) as compared to CV measurements, since the background capacitive current in CV measurement hides the Faradaic current 26 . The data for each box in the box plots contains measurements from 5 electrodes. The same set of electrodes were used for all measurements, to avoid measurement errors due to electrode-to-electrode variations. We observed increasing trend in the peak current for DPV and CV measurements for lower concentrations of DNA, with the increase more pronounced for the longer ( 503 bp ) DNA fragment as compared to the 117 bp fragment. This is consistent with the trend expected for adsorption on the electrode, reported in our previous work 11 . Adsorption of the MB-DNA complex facilitates charge transfer at the electrode, thus contributing to rise in peak current. Other studies have shown the influence of oligonucleotide size and sequence on MB-DNA intercalation [27][28][29][30] . The Guanine-Cytosine (GC) content of both the amplicons ( 117 bp and 503 bp ) is approximately 50 %, sug- Presence of salt in sample reduces electrochemical sensitivity at lower DNA concentrations. Salt present in PCR master mix can interfere with the electrostatic interactions between MB and DNA, and therefore the DPV and CV measurements on 5 electrodes were performed separately by adding 2 mM MgCl 2 to the gel purified products with 50 µM MB, to study the impact of salt on the MB-DNA interaction. As seen in Fig. 3, we observe that the decreasing trend in peak current for higher DNA concentrations ( > 2 ng/µl for 503 bp and > 10 ng/µl for 117 bp ), in both DPV and CV measurements, is not significantly impacted by addition of salt (see Fig. S2 in supplementary information for representative voltammograms). However, at lower DNA concentration, the sensitivity is greatly reduced by addition of salt, resulting in no significant change in current with DNA concentration. Other researchers have previously reported similar negative impact of salt on MB-DNA interactions and intercalation 33,34 . The Mg 2+ cation binds with the negative phosphate backbone of DNA thereby hindering the electrostatic interactions between MB and DNA 35 . At higher DNA concentrations, steric inhibition of redox active MB causes reduction in peak current, and therefore electrostatic interactions do not significantly influence the sensor response. The key takeaway is that such biosensors are more suitable for detection of higher DNA concentrations (few ng/µl or higher) for use-cases such as fully-automated processing of environmental water samples, where gel purification of the PCR products may not be feasible.
Longer DNA results in more pronounced change in optical absorption for higher concentrations. To further validate the results discussed above, we performed optical measurements using a UV/Vis spectrophotometer (Thermo Scientific Multiskan GO), with 50 µl of sample used for each measurement. The absorption signature decreases with increase in DNA concentration, as seen in the trend for area under the absorption curve for wavelength range 600 nm to 700 nm , shown in Fig. 4 (absorption spectra shown in Fig. S3 in supplementary information). There is no noticeable difference in the absorption for samples containing DNA compared to sample containing only MB, for DNA concentration lesser than 1 ng/µl (for both 503 bp and 117 bp length fragments), indicating the absence of steric inhibition of redox active MB. At higher DNA concentrations, we observed gradual decrease in the absorption signal, and noticed that the extent of reduction in absorption is lesser in presence of salt. These results are attributed to the molecular interactions with base stacks in the DNA hybrid, and steric inhibition. Our results are in concurrence with reports on spectroscopic studies of MB-DNA www.nature.com/scientificreports/ intercalation in literature, that have associated the hypochromism with decrease in the energy level of π-π * electron transition due to intercalation 36-38 . Longer DNA fragment from Phi6 bacteriophage isolated from lake water can be detected with high sensitivity using unmodified PCB electrode. We evaluated the utility of our sensor with water sample from Powai Lake spiked with Phi6 bacteriophage. The concentration of the RNA isolated from the phage spiked water samples ranged from 15.8-19.4 µg/ml , while the RNA isolated from purified phage suspension was estimated to be 1945 µg/ml , giving a recovery efficiency of approximately 1 %. RNA was reverse transcribed to cDNA and used as template for both PCR and qPCR. The products sizes were confirmed by agarose gel electrophoresis (Fig. 5)    www.nature.com/scientificreports/ ous environmental samples is challenging given the shortcomings due to low efficiency of virus concentration, and RNA degradation. However, with our virus concentration and PCR amplification protocols, we were able to successfully amplify the 503 bp fragment for electrochemical sensing. Figure 6 show the electrochemical sensor results obtained for 503 bp fragment amplicons, both with undiluted cDNA as template (1:1) and hundred-fold diluted cDNA as template (1:100) for PCR, compared with NTC and PC (see Fig. S4 in supplementary information for representative voltammograms). Each box in the box plots in Fig. 6 comprises of measurements of triplicate samples on 5 electrodes. The same electrodes were used for measuring all samples, to avoid errors due to electrode-to-electrode variation. DPV measurements show better resolution for distinguishing test and PC samples from NTC, as compared to CV measurements, due to hiding of the Faradaic current by background capacitive current in the latter, as discussed before. For the longer amplicon we observed that the negative control (NTC) resulted in higher CV and DPV peak current relative to the positive control, whereas the positive and undiluted test samples revealed similar peak heights for DPV peak current. The mean and median value of the measurements for each undiluted (1:1) test sample and PC are clearly resolved from the sensor output for NTC samples, while the resolution for samples with 1:100 dilution are not as highly pronounced. For hundred-fold dilution of the cDNA, we did not observe any band during gel electrophoresis (lane not shown in Fig. 5) and the corresponding DPV and CV peak currents were similar to NTC, as expected. The results for 117 bp fragment are presented in supplementary information. The negative control prompts an electrochemical response from the PCB sensor due to free MB adsorbing on the electrode, and the interaction of MB with single stranded primer oligonucleotides. Thus, each time a sample is to be tested, one must run a negative control and compare the peak current for test sample to that obtained for the negative control in order to achieve a differential (relative) measurement 39,40 to classify the test sample as positive or negative.

Discussion.
Our findings illustrate different mechanisms that influence electrochemical sensor performance for different length of amplicons at varying DNA, concentrations validated by optical measurements using UV/ Vis spectrophotometer. Our observations highlight the insight that longer DNA fragments up to ≈ 500 bp can be detected with higher sensitivity, and the presence of salt in the sample does not compromise sensitivity for higher DNA concentrations (few ng/µl and beyond. Additionally, we studied the impact of different types of samples including gel purified amplicons with and without added salt, and spiked lake water samples, on DPV and CV measurements. We observed that DPV offers better resolution as compared to CV, since the background capacitive current also impacts CV measurement, rendering it less sensitive. Amplification of longer fragments depends on integrity of the viral genomic RNA. Several studies have shown that amplification of longer fragments is not always efficient due to degradation of RNA in the environment and Table 1. Ct values of samples as measured by qPCR 1 NTC has no template added in the reaction. 2 Positive control has cDNA isolated from purified Phi6 bacteriophage particles as template.  11,[41][42][43][44] . We observed that PEG-based virus concentration method works better than aluminium hydroxide based virus concentration method to concentrate bacteriophage Phi-6 spiked in lake water samples. Demonstration of ability to detect long DNA fragment overcomes the requirement of multiplex PCR for amplifying multiple shorter length templates, and reduces the possibility of cross-specificity. Biological samples are scarce, and therefore designing a biosensor that requires minimal sample for testing is desirable. The ENIG PCB electrodes used in this study require only 5 µl of sample for testing to cover the active area of the electrode. Moreover, the same electrode can be reused after cleaning, before dispensing the next sample. The amplified sample does not require addition of any chemicals other than methylene blue, an inexpensive and commonly available chemical. Since the cost of manufacturing of each electrode is approximately USD$0.55 (i.e. INR ₹ 40), this biosensor can be a cost effective alternative to existing detection techniques. Table 2 shows a comparison of this work with other sensors for long DNA fragments in heterogeneous samples reported in literature.
The main limitation of this method is the possibility of non-specific amplification in heterogeneous samples such as wastewater and lake water, or with low purity primers, given the reliance of the MB based electrochemical detection scheme on PCR for specificity. To improve the sensitivity of the electrochemical detection method using unmodified ENIG PCB electrodes for DNA detection with unpurified PCR products, it is necessary to better understand errors introduced by the ununsed dNTPs and primers, optimise the reaction conditions and testing protocols. Measurement of auxiliary physicochemical parameters such as pH, temperature and biological oxygen demand (BOD) of the water samples may also be necessary to improve the accuracy of measurements.

Conclusion
In summary, we have presented a low-cost electrochemical ENIG PCB sensor for detection of viruses from environmental (lake water) samples. Unlike electrodes with immobilised oligonucleotides that need low-temperature storage to retain sensitivity, or custom fabricated substrates for DNA sensing 53,54 , our technology utilises unmodified PCB electrodes, with longer shelf life and no specific storage needs, and therefore suitable for developing automated sample processing and measurement solutions for deployment in LMICs. The biosensor utilises an inexpensive DNA intercalating redox dye (MB) for rapid detection of target amplicon. Since MB non-specifically binds to single and double stranded oligonucleotides, the non-specific amplification commonly found in environmental samples would reduce the specificity of such sensing method. Thus, the specificity of this test depends on the optimisation of primers and PCR reaction conditions. Further, the CV and DPV peak currents obtained from sample under test should be interpreted relative to response obtained from negative control (NTC) for every test. The electrochemical sensor design and approach presented in this work can be integrated with an auto-sampler for developing a fully automated and low-cost solution, that can collect and analyse samples, and communicate the findings back to the lab wirelessly.