Review—Chemical and Biological Sensors for Viral Detection

Tugba Ozer; Brian J. Geiss; Charles S. Henry

doi:10.1149/2.0232003JES

The Challenge of Viral Diseases Worldwide

Viral infections represent a grave threat to public health as well as the global economy. Viral infections mostly occur through contaminated water, food, and/or bodily fluids, spread rapidly and result in death of humans and animals worldwide.^1,2 Significant concerns has raised due to the increased viral outbreaks, since the viruses could quickly spread and cause a pandemic. Thus, rapid and accurate detection can mean the difference between life and death during viral infections. Analytes such as viral nucleic acids (DNA and RNA), viral proteins, intact viral particles, and antibodies generated by the patient immune response against the virus have been used to detect viruses with the goal of being used in clinical situations. These analytes are detected using a variety of traditional methods including polymerase chain reaction (PCR), virus culture, enzyme-linked immunosorbent assay (ELISA), western blots, and serological antibody detection methods.^3,4 However, conventional techniques are often not suitable for rapid on-site analysis as they require virus isolation and biocontainment, the ability to grow cultured cells for cytopathology-related assay, and/or sophisticated and expensive laboratory tools that are difficult to transport and use at the point of care. Traditional laboratory-based assays are also time-consuming, labor-intensive, and can in some cases relatively insensitive, and in all cases require samples to be transported to centralized diagnostic laboratories for testing.^5,6 These factors increase time-to-answer and costs while reducing the quality of patient care.

Sensors consist of chemical or biological receptors that specifically interacts with a target analyte, and a transducer that converts the recognition process into a quantitative signal.⁷ Electrochemical sensors are based on reaction with the chemical solutions and generate an electrical signal that is proportional to the analyte concentration.⁸ The electrochemical techniques can be classified into four major groups including potentiometry, amperometry, cyclic voltammetry, and impedimetry.⁹ The electrochemical methods are suitable for real time viral detection because of their high sensitivity and selectivity, low cost, simple operation, portability and fast analysis.¹⁰

Viral biosensors offer exciting alternatives to traditional diagnostic assays and have the potential to provide inexpensive, sensitive, rapid, miniaturized, and portable platforms when compared to conventional laboratory-based methods. Biosensors are analytical devices that couple biological recognition elements such as enzymes, antibodies, or nucleic acids with a transducer that can detect the interaction of the analyte, and can be applied for medical diagnosis, environmental monitoring, food, water and agricultural product processing.¹¹ The goal of this review is to explore recent developments in electrochemical biosensors for viruses and viral infections. The review is divided by analyte type and is intentionally not all inclusive given the large numbers of publications in this field.

Nucleic Acid-Based Sensors

Individual viral particles usually include either an RNA or a DNA genome.¹² A DNA biosensor is based on the immobilization of a single stranded oligonucleotide on a transducer surface to detect its complementary DNA sequence due to surface hybridization. Then, the hybrid formed on the electrode surface is transformed into an analytical signal via a transducer.¹³ Electrochemical DNA biosensors have drawn attention due to their advantages such as portability, simplicity, cost-effectiveness, fast response time, high sensitivity, high selectivity, and compatibility with miniaturized detection technologies.¹⁴ The working principle of an electrochemical DNA biosensor depends on two steps including the interaction or hybridization between DNA-DNA, DNA-RNA, and protein-ligand molecules, or conversion of the changes in the DNA structure or the assembly into an electrochemical signal.^15–17 In addition, there are indirect techniques that use electrochemical active DNA intercalators, enzymes, redox mediators and particles for the amplification of signal.^16,18–20 A variety of electrochemical DNA biosensors have been developed for detecting different viruses.^21–26

West Nile virus (WNV) is a member of the Flaviviridae family (genus Flavivirus) that is endemic in many parts of the world and transmitted by Culex mosquito vectors.²⁷ WNV commonly presents with flu like symptoms but can cause severe neurological symptoms and even death in a small subset of patients. Detection of active viral infection can help with disease management as well as helping detect WNV present in mosquito populations to reduce transmission. Wang et al., developed a label-free capacitance-based DNA capacitance sensor for the Kunjin subtype of WNV using interdigitated Au electrodes (Figure 1A).²⁸ A 24-nucleotide DNA probe based on a West Nile virus sequence was immobilized on a pre-cleaned gold coated interdigitated electrode (IDE) followed by inactivation of vacant gold sites with the use of 11-Mercapto-1-undecanol (MCU). Due to the hybridization of nucleic acid targets with ssDNA probe oligomers on microelectrodes, the capacitance changes were measured and found as 70 nF in response to as few as 20 complementary DNA targets (0.25 attogram) at a concentration of 1.5 aM. Capacitance changes had excellent linear behavior toward complementary target with concentration of 20 to 2 million target DNA molecules, which enables sample analysis in a typical clinical application environment. The biosensor advantages include low detection limit and better reproducibility compared to other non-faradaic capacitive biosensors. Moreover, specificity tests of the biosensor were performed. Unlike non-complementary target DNAs, complementary DNA showed a large capacitance change in the biosensor.

Figure 1. Refer to the following caption and surrounding text. — **Figure 1.** (A) Equivalent circuit models on interdigitated electrode (IDE) sensors. (i) Bare IDEs with multiple fingers. (ii) Equivalent circuit model for bare IDE sensors. (iii) Equivalent circuit model after probe immobilization. (iv) Equivalent circuit model after MCU incubation and before target hybridization. (v) Equivalent circuit model after target hybridization for complementary targets. Adapted with permission from Ref. 15, Copyright 2016, Elsevier B.V. (B) Different steps for the fabrication of electrochemical DNA biosensor. Adapted with permission from Ref. 24, Copyright 2018 Elsevier Inc. (C) DNA hybridization sensor based on AuNP-modified SPCE employing label-free impedance for the detection of CTV-related nucleic acid. Adapted with permission from Ref. 31, Copyright 2018, Elsevier B.V.

Ebola (Family Filoviridae) infection can lead to fatal hemorrhagic fever, but death can be prevented with various therapies if the infection is detected in time.^29–32 According to the World Health Organization, there are over 25,000 cases and 10,000 fatalities caused by Ebola virus in Guinea, Liberia and Sierra Leone in 2016.^33–35 Due to the resource-poor regions of Africa where Ebola outbreaks generally occur, developing inexpensive point-of-care detection devices is highly desirable. Screen-printed electrodes (SPEs) are often used for fabrication of biosensors since they possess advantages such as disposability, mechanically robustness, inexpensiveness, stability and reproducibility for mass production.³⁶ For instance, a novel electrochemical-based nucleic acid sensor was fabricated for Ebola virus RNA detection (Figure 1B).²⁵ In the first step, the gold surface of screen-printed electrode was functionalized with a thiolated DNA capture probe sequence via S-Au bonding. Then, biotinylated target strand DNA was immobilized to the electrode surface for hybridization of thiolated DNA single strand capture. Subsequently, the streptavidin-alkaline phosphatase enzyme was exposed to the biotinylated hybrid via biotin-streptavidin conjugation bond and, covered with 4-aminophenyl phosphate solution. Electrochemical impedance spectroscopy (EIS) was used to monitor DNA immobilization and differential pulse voltammetry (DPV) method was used to detect the enzymatically-produced 4-aminophenol. The electrical signal of the fabricated biosensor exhibits a linear relationship between 10 nM-75 nM Ebola virus complementary target strand with 4.7 nM limit of detection under optimum conditions. However, this method has not been applied for detecting Ebola virus RNA in real samples yet and would require amplification for diagnostic use.

Similarly, a novel label-free electrochemical method for rapid detection of hepatitis A virus (HAV) was developed by Manzano et al.²⁶ HAV is a picornavirus (Family Picornaviridae) that possesses a small positive strand RNA genome. HAV infection, which commonly occurs through a fecal oral route with patients having consumed contaminated food, causes a range of symptoms including nausea, diarrhea, and liver disfunction. The early symptoms of HAV can often be mistaken for Influenza infection, so positive diagnosis can help avoid inappropriate treatment with anti-influenza drugs. A screen-printed gold electrode was incubated with thiolated-ssDNA probe. Cyclic voltammetry (CV) was performed using the indicator, tripropylamine (TPA), to observe CV peak current that occurred because of hybridization of complementary synthetic target DNA sequences on the electrode surface. The fabricated sensor demonstrated a LOD of 0.65 pM and 6.94 fg/μL for the complementary ssDNA and cDNA sequences characteristic for HAV target, respectively. The target HAV cDNA was also quantified by nested real-time PCR and limit of detection was found as 6.4 fg/μL. The DNA electrochemical biosensor is advantageous relative to traditional and molecular methods such as PCR-based methods due to its cost and fast detection of HAV.^37,38

Hepatitis B virus (HBV), a circular double-stranded DNA virus, infects millions of people and causes diseases such as liver cirrhosis and hepatocellular carcinoma.³⁹ An estimated 800 million people are infected with HBV globally, and almost one million people die every year due to HBV induced liver disease.^40,41 Huang et al. developed an electrochemical biosensor for detection of target DNA derived from the HBV genome.⁴² Rolling circle amplification (RCA) and molecular beacon (MB) mediated circular strand displacement (CSD), which are isothermal DNA amplification methods, were used for electrochemical detection of HBV DNA sequences. The gold surface electrode was modified with MB. RCA was employed to amplify the DPV current signal response. Since methylene blue specifically binds to the guanine bases of DNA molecules, it was used as the electrochemical redox probe. The biosensor demonstrated a linear DNA concentration range from 10 aM to 700 aM with a 2.6 aM detection limit under optimal experimental conditions. Using the RCA and MB mediated CSD approaches, the biosensor sensitivity was improved relative to traditional methods.^43,44

Various nanomaterials have been utilized in nucleic acid-based biosensors due to properties such as large surface area, high conductivity, and strong affinity toward bioreceptor probes with reactive groups such as thiols leading to high sensitivity and low limits of detection. These nanomaterials could be gold nanoparticles, carbon nanomaterials, magnetic nanoparticles or magnetic beads (MBs), silica nanoparticles, quantum dots (QDs) and hybrid nanostructures.⁴⁵ Khater et al. developed a rapid, label-free impedimetric biosensor for citrus tristeza virus (CTV) detection using gold nanoparticles (Figure 1C).⁴⁶ CTV (Family Closteroviridae) is a filamentous positive strand RNA virus that infects citrus trees (sour orange) and is transmitted through insect transmission by aphids. CTV causes stunting, stem pitting, low yields, and poor fruit quality from infected trees and is of major concern to the citrus industry.⁴⁷ Traditional methods for CTV detection use dot-blot hybridization, which is slow and labor intensive in terms of analysis time and quantitative analysis. Kather et al. used gold nanoparticles (AuNPs) deposited on the surface of a screen-printed carbon electrode (SPCE). Then, poly (AT)-thiolated ssDNA probes were covalently attached onto the AuNPs-modified electrode surface in the presence of mercaptohexanol (MCH). Here, poly (AT) and MCH were used to enhance sensor selectivity by preventing any non-specific binding of nucleobases. The hybridization process was investigated by EIS in Fe(CN₆)⁴⁻/Fe(CN₆)³⁻ red-ox system. The sensor demonstrated logarithmic behavior in the concentration range between 0.1–10 μM of CTV-related DNA with LOD of 100 nM and a 65 min assay time. The biosensor was applied in leaf extracts from healthy citrus plants spiked with various target DNA concentrations and detected the target DNA concentration as low as 500 nM. While real samples would likely never have this high of DNA concentration, it shows method viability in a complex matrix.

Chikungunya virus (CHIKV) is a small positive strand RNA virus that is transmitted by Aedes species mosquitoes. CHIKV exploded out of the Indian subcontinent in 2013, and rapidly spread through Africa, Europe, and South and Central America over the next two years.⁴⁸ CHIKV infection can cause debilitating arthralgia disease that, while rarely fatal, can incapacitate infected individuals for weeks to months.⁴⁹ Singhal and coworkers proposed a novel paper-based DNA biosensor utilizing Fe₃O₄@Au nanocubes for detection of CHIKV.²⁴ Some metal oxide nanomaterials such as zinc oxide (ZnO), iron oxide (Fe₃O₄), titanium dioxide (TiO₂), copper oxide (CuO) have been employed with AuNPs to synthesize Au-metal oxide composites for attaining better electrochemical performance.^50–53 To fabricate the paper-based device, screen- and wax-printing methods were used. AuNPs and magnetic nanoparticles (Fe₃O₄) were mixed to synthesize Fe₃O₄@Au nanocubes, which has ability to improve sensitivity of the sensor due to their higher surface area.⁵⁴ The working electrode was modified with the prepared Fe₃O₄@Au nanocubes which is responsible for electron transfer between probe DNA. After that, CHIKV probe DNA was immobilized onto this modified electrode. CV and DPV were used to analyze the hybridization. The biosensor showed LOD of 0.1 nM in a linear range between 0.1 nM–100 μM. The CHIK probe DNA/ Fe₃O₄@Au/ePAD platform was also used for detection of CHIKV target RNA in serum sample. The advantages of the biosensor include being simple to fabricate, reduced costs, and disposability of the used tests due to the use of paper.

Human papilloma viruses (Family Papillomaviridae) are small double-stranded DNA viruses that are common in the human population due to sexual transmission. Certain HPV types are strongly associated with cancer (HPV-16, HPV-18, HPV-31, and HPV-45), and detection of specific HPV types can help physicians provide appropriate patient care. Au nanotubes have drawn attention due to their outstanding properties like being physicochemically stable and capable of sensitively detecting complex analytes.⁵⁵ The 3-D nanostructured-based templates are suitable for DNA detection since they offer nanofluidic channel structures. Shariati et al. reported a human papilloma virus DNA biosensor with the lack of labelling.²² A gold nanotube decorated nanoporous polycarbonate (AuNTs-PC) template was created by electrodeposition and used as the working electrode. Then, thiolated 25-mer ssDNA probe was immobilized on the surface of the AuNTs-PC electrode. EIS was used to observe the DNA hybridization using external electric field bias, which helps probes and targets orientate flat on the surface. Due to the presence of AuNTs and the electric field, this DNA-based sensor displayed a wide linear detection range of 0.01 pM-1 μM with a very low LOD of 1 fM. The selectivity of the fabricated sensor was confirmed by mismatch, non-complementary and complementary DNA oligonucleotides.

Influenza virus is a common human pathogen that causes a high level of morbidity and mortality.⁵⁶ The World Health Organization (WHO) reported that influenza leads to 250,000-500,000 deaths annually all around the world through respiratory infections, and young children and the elderly are particularly at risk.⁵⁷ Approximately 50% of the population is affected by pandemics caused by Influenza A, whereas Influenza B and C lead to mild outbreaks.^58,59 It becomes challenging to control influenza A due to the propagation through aerosol particles, its short incubation time, and yearly mutation.^60,61 The Influenza A/H1N1 strain has resulted in more than 10,000 deaths and 275,000 hospitalizations while costing billions of dollars for health care in the United States alone.^57,62 Moreover, there is concern that pathogenic H5N1 has the potential to be used as a biological weapon, and developing sensitive and robust point-of-care diagnostics is critical for rapid response.^63,64 Tran et al. fabricated carbon nanotube based-field effect transistors (CNTFETs)-based DNA sensor for detecting Influenza virus RNA.⁶⁵ Synthetic DNA matching the RNA sequence was used as a model. Carbon nanotubes (CNTs) have been employed for fabrication of electrodes because of their large surface to volume ratio and excellent electronic characteristics as well. The surface of carbon nanotubes was modified with strong nitric acid.⁶⁶ After the pretreatment, probe DNA strands were deposited to the CNT network followed by immersing in a solution of the target DNA for the DNA hybridization. The sensor exhibited high reproducibility and fast response time of less than a minute. They reported a selectivity and LOD of 1 pM to 10 nM and 1 pM, respectively. The biosensor has a promising potential for detection of Influenza type A DNA. Recently, nanoparticle (NP)-carbon nanomaterials (CNMs) hybrid structures have become popular owing to their physical and mechanical properties, and excellent electrical conductivity. CNTs are the most widely used nanomaterials to generate Au based-nanocomposites hybrid structure since they can be easily modified and have distinguished properties. Au-CNT nanocomposites can be produced either by attaching Au nanostructures directly to CNTs or by forming some binding between Au nanostructures and CNTs via Au–S bond or π-π stacking.^67,68

Over 31 kinds of foodborne pathogens caused 600 million foodborne illnesses and 420,000 deaths globally with norovirus responsible for approximately 37% of these illnesses.^69,70 Therefore, there is great interest in rapid and sensitive detection of noroviruses. Lee et al. developed an electrical resistance biosensor based on multi functionalized-CNTs for detection of influenza A and norovirus target DNA.⁷¹ First, gold/magnetic iron-oxide nanoparticle (MNP)-decorated CNTs (Au/MNP-CNT) were synthesized and deposited on the surface of a commercial planar Pt-interdigitated electrode using an external magnetic field. Then, this electrode was conjugated with thiol (-SH)-modified probe DNAs for influenza virus and norovirus. Linear sweep voltammetry was used for target DNA detection. While single and fully mismatched DNA sequences were used to examine the selectivity of the sensor toward influenza A virus, Zika virus RNA and influenza A RNA were used to investigate the selectivity for norovirus against RNA viruses. The Au/MNP-CNT-based DNA sensing platform demonstrated high selectivity against target RNAs. Electrical resistance of the sensing channel displayed a linear behavior with the concentration of target DNA in the range from 1 pM to 10 nM with LOD of 8.4 pM and 8.8 pM for influenza virus and norovirus, respectively. While these levels are low, they are still well above what would be required for detection without amplification.

Challenges with current nucleic acid biosensors

The development of new nucleic acid biosensors that can accurately and sensitively detect specific sequences is quite encouraging, and many of the detection modalities provide potential ways to sense DNA and RNA. However, there several challenges remain for nucleic acid biosensors. First, the effective use of nucleic acid biosensors requires that the target nucleic acid be accessible to the biosensor surface. Virus nucleic acids in patient samples (serum, saliva, urine, etc.) are generally protected from their environments inside viral particles. For the biosensor to detect the nucleic acid, the virus particles need to be disrupted in a way that releases the nucleic acid from the viral particle. Disruption of virus particles can be performed via heating or chemical treatment, which adds additional sample preparation procedures into the testing paradigm needed to detect target nucleic acids. Heat disruption is the gentlest way of disruption particles, although solvent conditions during heating need to be well controlled to avoid base-mediated hydrolysis of RNA genomes. Chemical disruption using detergents to dissolve membranes (sodium dodecyl sulfate, Triton X 100, etc.) and chaotropic agents to unfold proteins (urea, guanidinium hydrochloride, etc.) can be effective, but can affect subsequent steps in analyte detection and must be removed. Protein removal is particularly important in negative sense RNA viruses, such as Influenza and Ebola viruses, as the genomes present in viral particles are tightly bound to nucleoproteins that must be removed before sequence-specific probes can recognize viral RNAs. Nucleic acid purification is essential for detecting viral genomic nucleic acids in real patient samples, but increases the time and effort required to sense nucleic acid analytes.

Another critical issue with nucleic acid biosensors is viral genome structure. Many of the nucleic acid sensors listed above test their system using short oligonucleotide DNAs that contain viral sequences. Short RNA or DNA sequences purchased from commercial vendors are inexpensive, easy to work with, and available at high concentrations but are very different than purified viral genomes. Long single-stranded viral genomes can adopt a wide range of thermodynamically stable secondary structures (including stems, loops, and pseudoknots) and these structures can strongly interfere with sequence-specific DNA probes from recognizing the target nucleic acid.⁷² Hybridization of DNA probes with long single-stranded nucleic acids requires the target first being completely denatured to remove inhibitory secondary structure, followed by time for limited numbers of denatured target genomes to interact with surface-bound probe DNAs. Hybridization in this context is similar to hybridization performed in nuclease protection assays, which use long RNA probes (200–500 bases) with a short 95°C denaturation step followed by 12–16 hr incubations at high temperatures (55–65°C) in high ionic strength solutions to reduce secondary structure formation and achieve efficient and specific hybridization. These conditions are not required for the short oligonucleotide targets but are critical for detecting longer genomes. Incorporation of rapid and efficient hybridization conditions are essential for nucleic acid biosensors to be useful at the point of care. The use of molecular crowding chemicals (such as polyethylene glycol) or applying electrophoretic potentials to biosensors to increase local concentrations of DNA probes and target nucleic acids to increase hybridization rates may help decrease hybridization time and increase efficiency.

Double-stranded RNA and DNA genome structures (such as reoviruses and papilloma viruses, respectively) also pose similar issues. Base-paring between both strands of dsRNA and dsDNA viral genomes generally precludes biosensor-bound probe molecules from binding in a sequence-specific manner, making recognition impossible. Variants of traditional nucleic acids, such as peptide nucleic acids (PNAs) and bridged nucleic acids (BNAs), have the potential to undergo strand invasion and perform local denaturation of nucleic acids, but these tend to be relatively expensive. Additionally, while it is possible to denature double-stranded nucleic acids with heat or chemical treatment, the affinity of the long single-stranded genome strands for each other is generally much higher than the affinity for short DNA probes bound on biosensor surfaces. This leads to complementary strands of viral genomes preferentially re-hybridizing to each other rather than to the biosensor-bound DNA, reducing the sensitivity of the biosensor.

Finally, the amount of viral genomes present in clinical samples is generally fairly small and in many cases much lower than the LOD of reported biosensors. For example, Faye et al. performed a retrospective study of viremia (viral genome copies/ml plasma) in Ebola virus infected individuals and compared viral loads to disease outcomes.⁷³ This study found that patients with less than 10⁵ genome copies/ml had a low probability of death, patients with viral titers between 10⁵ and 10⁶ had moderate probability of death, and patients with titers over 10⁶ had a high probability of death. Viral titers over 10⁸ genome copies/ml were not observed in patients. 10⁸ genome copies/ml appears to be quite a large number, but actually only represents approximately 166 attomoles of viral genomes per ml of plasma. Biosensors that demonstrate nanomolar and picomolar LODs are 6 to 9 orders of magnitudes less sensitive than necessary to detect Ebola virus even in very highly infected patients as described above. For nucleic acid biosensors to be useful for detecting viral genomes without additional amplification, they need to be able to detect between 10⁹ genomes/ml (1.6 femtomoles) and 1000 molecules/ml (1.6 zeptomoles). In other words, detection limits for biosensors should be below ∼100 aM for the biosensor to be able to detect clinically relevant infections.

Each of the factors listed above can significantly degrade the ability of a nucleic acid biosensor from binding and detecting viral genomes in real-world patient samples, and often multiple of these factors make detection in real patient samples impossible. Testing biosensors with short target oligonucleotides is only the first step in diagnostic development, and using more authentic genomes for later-stage testing is critical. Additionally, appropriate sample preparation and nucleic acid handling is absolutely essential for any diagnostic assay to function correctly, so for DNA-based biosensors to continue to evolve into useful diagnostic platforms and be used in detecting viral genomes in diagnostic settings, these issues need to be addressed early in the development process.

Antibody-Based Sensors for Viral Antigen and Particle Detection

The working principle of an electrochemical immunosensor is based on the generation of an electrical signal due to the specific interaction between antibody and antigen in the presence of an electrochemical transducer. Antibodies are attached to the biosensor surface with either covalent or non-covalent bonding. Biotin-streptavidin or conductive polymers (polypyrrole) are commonly used examples for immobilizing antibodies on electrodes for these applications.⁷⁴ After the capture antibodies are immobilized on a sensor surface, a complex between the immobilized capture antibodies, the antigen, and detection antibodies are used for sandwich-type immunosensors.⁷⁵ Immobilizing antibodies on a solid surface can also improve their stability.^76,77 Different immunoassay electrochemical sensing platforms have become available owing to their high affinity and sensitivity.^78–82

Assays targeting viral antigens

Haji-Hashemi et al. presented the first label-free electrochemical immunosensor to detect CTV antigens.⁸³ First, a gold electrode was modified using 11-mercaptoundecanoic acid (MUA) and 3-mercaptopropionic acid (MPA). Then, the terminal carboxylic groups on surface of the modified electrode were activated using N-(3-dimethylaminopropyl)-N-ethyl carbodiimide hydrochloride (EDC) and N-hydroxysuccinamide (NHS). The modified gold electrode was incubated with the antibody against the coat protein (CP) of CTV. After that, a solution of BSA was dropped on the electrode to block the non-specific sites. DPV method was used to evaluate the antibody-antigen interaction on the electrode surface in the presence of [Fe(CN)₆]^3−/4−. The detection limit was found to be 0.27 nM, which is lower than ELISA and other immunoassay techniques for CTV detection.^84,85

In another example of CTV detection, a microfluidic electrochemical device (μFED) was proposed by Freitas et al.⁸⁶ The μFED included an array of eight screen-printed electrodes fabricated. Poly(diallyldimethylammonium chloride solution) (PDDA) was applied to electrode surface to create a positively charged film. Next, the electrodes were modified with AuNPs resulting in formation of a PDDA/AuNP bilayer. Monoclonal antibodies specific for the capsid protein of CTV (Ab₁) were covalently immobilized on AuNPs by EDS/NHS coupling procedure. To block the non-activated carboxyl groups, BSA was used. After that, horseradish peroxidase enzyme (HRP) and polyclonal capture antibody (Ab₂) were conjugated to carboxyl-functionalized magnetic beads (MBs). The prepared Ab₂/MB/HRP was used to capture CP-CTV biomarker from sample. The μFED array modified with Ab₁ was incubated with the CP/Ab₂/MB/HRP, forming a sandwich-type immunoassay. The electrochemical detection of the magnetically captured biomarker was achieved by simultaneously applying a potential of −0.2 V to the 8-immunosensors via amperometry. The detection mechanism is based on the enzymatic activity of HRP using H₂O₂ and hydroquinone (HQ), and shown as following:

An LOD of 0.3 fg mL⁻¹ for the detection of CTV was obtained. The immunosensor was successfully used to detect Citrus tristeza virus with the use of small volumes of healthy and infected plant samples.

Graphene (GR), which is a monolayer of carbon atoms formed in a honeycomb lattice, has received attention because of its high electrical conductivity, fast electrode transfer kinetics, and strong mechanical strength.⁸⁷ Graphene-based biosensors are more advantageous than conventional carbon electrodes since they have higher sensitivities, lower LODs, faster response time and longer life time than traditional carbon surfaces.⁸⁸ Islam and coworkers demonstrated a smart graphene-based field-effect transistors (FETs) to detect human immunodeficiency virus (HIV) biomarkers (Figure 2A).⁸⁹ HIV is a globally important virus that infects patients T-cells and can result in a severe immunodeficiency syndrome. According to World Health Organization recent data, 36.9 million people have been affected by HIV worldwide.⁹⁰ According to recent studies, it is possible to decrease the transmission up to 96% by early antiretroviral treatment so rapid and accurate point-of-care (POC) detection of HIV infection status is critical for protecting patients.⁹¹ For FET fabrication, graphene was exfoliated on SiO₂/Si substrate and then electron beam lithography and thermal evaporation techniques were applied to pattern the electrodes. Graphene was reacted with EDC and NHS to activate its carboxylic groups on the surface. Anti-p24 antibody for HIV was covalently conjugated by dropping on the activated graphene FET. While the specific antibody bound the p24 antigens, the graphene layer converted the chemical signal from antibody-antigen interactions into a measurable electrical signal. The sensor exhibited a linear response to p24 protein from 1 fg/mL to 1 μg/mL with an LOD of 100 fg/mL. The graphene-based FET immunosensor exhibited better response time and sensitivity compared to traditional methods but does require more expensive fabrication methods.

Figure 2. Refer to the following caption and surrounding text. — **Figure 2.** (A) Fabrication of the graphene immunosensor. Adapted with permission from Ref. 71, Copyright 2018, Elsevier B.V. (B) Reaction scheme for Au microwire electrodes. Adapted with permission from Ref. 79, Copyright 2018, American Chemical Society. (C) The three-electrode immunosensor chip with three detection zones. Adapted with permission from Ref. 83, Copyright 2015, Elsevier B.V.

Graphene oxide (GO) provides enhanced electron transport, mechanical, and chemical characteristics for biosensors. GO contains large number of oxygen with functional groups so that it can be assembled with biomolecules on the sensor.^92,93 A screen-printed graphene oxide textile biosensor for point-of-exposure detection of Influenza A virus was developed by Kinnamon and coworkers.⁹⁴ After the electrode array was screen printed using conductive silver ink on textile, the silver electrodes were patterned with a layer of GO. 1-Pyrenebutyric acid-N-hydrosuccinimide ester (PANHS) as crosslinker was attached on surface of the sensor via π–π stacking. Subsequently, influenza protein-specific antibodies were incubated on the textile-biosensor surface leading to an amide bond with the PANHS. EIS was used to detect Influenza A in a biofluid analog buffer and the sensor displayed a linear range from 10 ng/mL to 10 μg/mL with a limit of detection of 10 ng/mL. Application to real samples was not demonstrated.

Lim et al. developed a sensitive, label-free electrochemical biosensor to detect the dengue fever biomarker, NS1.⁹⁵ M13 phage display technique was used to identify high affinity peptides for Dengue virus type 2 NS1 (DENV2 NS1), which is a protein biomarker for specific detection of dengue fever. While phage peptides are not the same as antibodies, they work in many of the same ways from a binding perspective. Gold electrodes were coated with the peptides and their affinity tested using CV and EIS. Due to its high content of basic residues (two His and Arg), the R3#10 phage peptide with amino acid sequence EHDRMHAYYLTR was chosen to modify the electrode surface to bind DENV2 NS1 protein. The resulting LOD was 1.5 μg/mL, which is below the normal level of NS1 in bloodstream from dengue patients (50 μg/mL).⁹⁶ However, the biosensor was not validated with real samples.

Assays targeting viral particles

Direct detection of intact virus particles can provide information to clinicians about the phase of the infection. Traditionally, viral particle concentrations (or "titers") have been determined by culturing patient samples on permissive mammalian or insect cell lines and looking for cell death. This approach is slow and is not amenable to POC applications. The Henry and Geiss groups developed a microfluidic paper-based analytical device with integrated microwire Au electrodes for trace detection of West Nile virus particles (Figure 2B).⁹⁷ The Au microwire electrodes were treated with lipoic acid for attachment of the carboxyl terminated alkane dithiol. Next, the Au microwires were modified with amine-PEG2-biotin, EDC, and NHS to obtain carbodiimide cross-linking for capturing antibodies. For characterization of the device, biotin-modified microwires were utilized to bind to streptavidin modified microparticles to optimize performance. West Nile virus particles were then detected with the antibody-modified electrodes and the LOD was found as 10.2 particles in 50 μL of cell culture media. The assay sensitivity was verified in the presence of 1.0 × 10⁷ particles mL⁻¹ Sindbis virus particles. The detection platform has better performance characteristics when compared to current methods.

Avian leukosis virus (ALV) is an avian retrovirus that can infect and cause cancers in poultry, and 1.5% excess mortality in chicken flocks has been attributed to ALV-J subtype in commercial broiler-breeder flocks.⁹⁸ There is no therapeutic for ALV, so detection of the virus in new animals introduced to flocks is the only way of controlling infection in naïve flocks. Ning and coworkers developed a simple, sensitive electrochemical immunosensor to detect ALV virus particles.⁷⁸ A ferrocene-functionalized gold nanoparticle (Fc-AuNP) nanocomposite was prepared. Then, β-cyclodextrin (β-CD) was covalently bound to secondary antibodies (Ab₂) and incubated with Fc-AuNP resulting in Fc-AuNP-β-CD-Ab₂ bioconjugates. Next, perylene-3,4,9,10-tetracarboxylic acid (PTCA) was combined with graphene via π–π stacking interactions to overcome the limitations of graphene itself for primary antibody (Ab₁) attachment. While GR-PTCA was utilized as the sensing platform to capture Ab₁, Fc-AuNP- β-CD -Ab₂ was used the label. DPV was performed to obtain the amperometric responses of the immunosensor. The Au NPs improved the sensitivity of immunosensor due to their ability to bind multiple Fc molecules. The immunosensor exhibited a linear current response against ALV-J in the range from 10^2.0 to 10^4.0 TCID₅₀/mL with a detection limit of 10^1.93 TCID₅₀/mL.

Bhardwaj et al. developed a novel a vertical flow assay (VFA)-based paper immunosensor to detect Influenza H1N1 viruses electrochemically and colorimetrically.⁹⁹ A double pore size (DP) sample pad, a conjugate pad, and a nitrocellulose (NC) membrane strip were attached together for fabrication of the VFA sensor. Wax printing was used to pattern the NC membrane strip. A gold paper electrode, carbon paste electrode and Ag/AgCl electrode were used as working, counter and reference electrodes, respectively. The DP sample pad included both large pores and small pores that allowed rapid detection (∼6 min) via concentrated antigen-antibody complexes on the conjugate pad. In addition, small particles like viruses can pass through the sample pad, which gave the ability for virus detection in air samples. An LOD of 3.3 plaque forming units (PFU)/mL and 4.7 PFU/mL (saliva) was obtained using EIS. The integrated device is useful for sensitive, on-site detection of H1N1 viruses in real samples while decreasing false results using simultaneous quantitative and qualitative analysis.

Han et al. proposed a nano-flow electrochemical immunosensor chip with three different sensors for simultaneous detection of H1N1, H5N1, and H7N9 influenza viruses (Figure 2C).¹⁰⁰ First, SU-8 was used to coat a bare Si wafer for fabrication of master mold. Then, PDMS was poured onto the master mold and cured to create a microfluidic chamber. Then, ZnO nanorods (NRs) were grown in the three sensor regions of the PDMS channel via hydrothermal method and lift-off process. After the ZnO NRs growth, O₂ plasma treatment was applied to adhere ZnO NRs to the PDMS sensor. An immunosensor chip with three detection zones was achieved by photolithography. Subsequently, capture antibodies of H1N1, H5N1, and H7N9 were immobilized onto the ZnO NRs grown on three sensor regions of the PDMS surface. Finally, H1N1, H5N1, and H7N9 antigens were incubated on sensors followed by injection of detection antibodies conjugated with horseradish peroxidase (HRP). 3,3,5,5-tetramethylbenzidine (TMB) substrate was oxidized by HRP enzyme resulting in signal production. H1N1, H5N1, and H7N9 influenza viruses were simultaneously detected in a mixture of three virus antigens using three-gold electrodes via amperometry. The limit of detection of each virus using this technique was 1 pg/ml. Real samples were not tested using the system.

Hushegyi et al. developed a glycan-based impedimetric biosensor for detection of influenza virus.¹⁰¹ A gold electrode was modified with self-assembled monolayer consisting of oligoethylene glycol (OEG)-COOH and OEG mixture of with a ratio of 1:5. Next, EDC-NHS was used to activate the COOH to covalently link an amine-terminated glycan to the electrode surface. Lectin Maackia amurensis agglutinin (MAA), which is a glycan binding protein, was analyzed in the presence of two nonspecific probes namely Datura stramonium lectin (DSL) and human serum albumin (HSA). The glycan-based biosensor was able to sensitively detect MAA using EIS in the concentration range of 8 aM-0.8 nM with an LOD of 5 aM. Influenza H3N2 and H7N7 viral particles were used to evaluate the selectivity of biosensor as well. The biosensor had an LOD of 13 viral particles in 1 μl for H3N2 viruses, it was capable of selectively detecting H3N2 viruses with a sensitivity ratio of 30 over influenza viruses H7N7.

Baek et al. fabricated a highly sensitive and selective peptide based-biosensor for detection of human norovirus.¹⁰² Gold screen-printed electrode (Au-SPE) was functionalized via Au-S covalent bond by separately applying eight novel synthetic noroviral peptides. Among these peptides, Au electrode modified with the NoroBP peptide showed the best selectivity toward norovirus using EIS technique with a detection limit of 1.7 copies/mL, which is 3-fold lower than the previously reported methods. Finally, electrochemical behavior of gold electrode coated with 0.3 mg/mL NoroBP peptide was successfully evaluated due to the increased impedance signal with increasing concentration of norovirus extracted from oyster and showed LOD of 2.47 copies/mL.

Challenges with antibody-based biosensors

There are numerous challenges associated with antibody-based biosensor that can largely be divided between reagent and target. The first challenge is generation of consistent high affinity antibodies that selectively target the analyte of interest. Antibodies are notoriously variable from lot to lot, meaning sensors can work with one batch but not the next. Monoclonal antibodies generally have better reproducibility but bind only a single epitope giving rise to potential problems with number of binding sites. Another antibody related problem is immobilization on the surface of the electrode. If care is not given to the method, a significant fraction (up to 75%) can be either denatured or immobilized in a manner that makes antigen binding impossible. Careful consideration of immobilization techniques and blocking for non-specific adsorption is required. Finally, antibodies can lack stability relative to DNA/RNA reagents and often require cold storage to maintain shelf life. Increased temperatures, ionic concentrations, and reducing agents can inactivate antibodies on sensor surfaces and dramatically reduce the efficacy of the sensor. Therefore, appropriate buffer composition and temperature control are critical considerations in the use of antibody-based sensors.

With regards to the target, the most significant challenge comes from the biology itself. Viruses are notorious for both mutating to change their coat proteins and for having coat proteins that are very similar. For the former, one need only consider influenza. New flu vaccines are needed each year because a new influenza variant appears almost every year. This makes creating antibodies against the mutating species very difficult. Likewise, the similarity of coat proteins across virus serotypes is a significant challenge for selective antibody-based sensors. For example, there are four main dengue virus, and creating antibodies that can distinguish between these serotypes is non-trivial. Careful selection and treatment of antibodies used on biosensors is critical for successful detection.

Aptamer-Based Sensors

Aptamers are single-stranded oligonucleotides or peptides possessing high affinity toward antigens and they are derived from the sequential exponential enrichment (SELEX) method.¹⁰³ Compared to antibodies, aptamers have advantages including low cost, reproducibility, simple production, high specificity, stability, and low toxicity.¹⁰⁴ Aptamers are capable of binding to their target molecules via molecular shape complementarities, electrostatic or van der Waals interactions, stacking of aromatic rings, and hydrogen bonding.¹⁰⁵ Aptamers have good stability and can repetitively denature and refold.¹⁰⁶ Aptamer-based biosensors (aptasensors) has been developed for sensitive and selective virus detection.^107–110 Aptasensors with high affinity and specificity aptamers have advantageous such as sensitive, selective, real-time, fast, label-free and low-cost analysis.¹¹¹

Ghanbari et al. developed an electrochemical aptasensor for sensitive detection of hepatitis C virus (HCV) core antigen (Figure 3A).¹¹² At first, graphene quantum dots (GQD) were synthesized according to their previous work.¹¹³ A glassy carbon electrode (GCE) was modified with GQD and the aptamer was adsorbed to the surface of GQDs due to the interactions between hydroxyl and carboxyl groups at GQDs edges and amine groups of the aptamer.¹¹⁴ EIS was used to detect HCV core antigen using ferricyanide/ferrocyanide as a redox probe. The aptasensor exhibited two linear concentration ranges 10–70 pg mL⁻¹ and 70–400 pg mL⁻¹ with a detection limit of 3.3 pg mL⁻¹. The aptasensor was used for detection of HCV core antigen concentration in human serum samples. The aptasensor is more advantageous than previously reported aptasensors for detection of HCV that it is easy to fabricate, rapid, cost effective and has low detection limit.

Figure 3. Refer to the following caption and surrounding text. — **Figure 3.** (A) Electrochemical nano-aptasensor for HCV core antigen detection. Adapted with permission from Ref. 95, Copyright 2017, Elsevier Inc. (B) The scheme of recognitions between the aptamers and viruses on the gold electrode surfaces with low and high surface probe densities. Adapted with permission from Ref. 100, Copyright 2018, Elsevier B.V.

Wang et al. reported a label-free nanowell-based quartz crystal microbalance aptasensor for detecting H5N1 avian influenza virus (AIV).¹¹⁵ A nanoporous gold film was prepared using metallic corrosion technique and immobilized onto a gold electrode surface. Then, a mixture of mercaptohexadecanoic acid (MHDA) and 1,6-hexanedithiol (HDT) were applied on the gold electrode surface to form a nanowell-based electrode. Moreover, EDC and NHS were used for further treatment to immobilize NH₂-aptamers via amide bond on the electrode surface. The immobilization of aptamers was enhanced fivefold using a nanowell-based electrode compared to the electrode without the nanowells. The aptasensor detected AIV H5N1 as low as 2⁻⁴ hemagglutination units (HAUs)/50 μl and had linear range 2⁻⁴ to 2⁴ HAUs/50 μl. The aptasensor specificity was demonstrated with non-target AIV subtypes H7N2, H1N1, H5N3, and H2N2 and was verified by detection of AIV H5N1 in chicken tracheal swab samples. The detection time using the nanowell-based QCM aptasensor was reduced to 10 min from 2 hr using a label-free assay compared to other reported QCM immunosensors/aptasensors for detection of AIVs.¹¹⁶

Bai et al. developed an electrochemical impedance aptasensor to detect H1N1 virus particles (Figure 3B).¹¹⁷ The aptamer selective to H1N1 was identified and used on the surface of modified gold electrode. The detection limit was 0.9 pg/μL with the higher probe density. Since the affinity of the selected aptamer depended on probe density, the biosensor demonstrated response toward H1N3 and influenza B virus. Therefore, this work was not suitable for subtyping. However, Bhardwaj and coworkers reported more specific label-free electrochemical aptasensor for subtyping of influenza A H1N1 virus.¹¹⁸ A ssDNA aptamer, which enables to detect a wide range of influenza A H1N1 subtypes, was developed. Indium tin oxide (ITO)-coated glass strips were pretreated with NH₄OH, H₂O₂, and H₂O to generate hydroxyl groups on the ITO surface. The electrode was modified with 1% polyethylenimine followed by aptamer adsorption. DPV was performed for detection of the H1N1 virus using [Fe(CN)₆]^3−/4− as the redox probe between the potential −0.8 V and +0.8 V with a scan rate of 100 mVs⁻¹. The LOD of the aptasensor for H1N1 viruses was 3.7 PFU mL⁻¹. The selectivity of the aptasensor was confirmed by distinguishing six strains of H1N1 viruses from four different subtypes of influenza A viruses. The detection of H1N1 virus using this aptasensor is faster than the standard ELISA and nucleic acid-based techniques.

Challenges with aptamer-based biosensors

Aptasensors have the advantages of DNA/RNA based sensors in terms of chemical stability and the ability to be reproducibly synthesized making them attractive for targeted binding relative to antibodies. Aptamers have the benefit of being relatively easy to mass produce through chemical synthesis techniques and provide reasonably high specificities and binding affinities, similar to antibodies.¹¹⁹ Synthetic oligonucleotide aptamers can be easily produced with multiple different attachment chemistries (thiol, NHS, amines, carboxylates, etc.) at the 5' or 3' ends of the aptamer or on internal bases if desired, allowing very precise control of the orientation that the aptamer binds to the sensor surface. Oligonucleotide-based aptamers have the added benefit of being able to re-fold into their active structures if the sensor surface encounters harsh conditions (increased ionic strength solutions, denaturing reagents, increased temperatures).¹²⁰ This can allow sensors using aptamers as the affinity reagent to potentially be re-used, increasing the utility and decreasing the costs associated with the assays. Therefore, aptamers possess significant advantages as sensor-coupled affinity reagents compared to protein-based affinity reagents. However, several challenges do exist with the use of aptamers that should be considered. The most significant problem with aptamers, however, is the lack of known aptamers for virus detection related sensing. The number of high affinity aptamers is very limited relative to the number of antibodies and development of new aptamers takes time and specialized knowledge limiting the rate at which new reagents are reported. Another issue that should be considered with nucleic acid-based aptamers, especially RNA based aptamers, is degradation by RNAse enzymes. RNAses are relatively ubiquitous and care should be taken to protect RNA aptamers from degradation by RNAses, which would reduce the sensitivity of the sensor. This can be mitigated by addition of RNAse inhibitors (such RiboLock) with the analyte during sensing. DNA and 2'O-modified aptamers tend to be less sensitive to nuclease digestion and can be used without additional nuclease inhibitors.

Antigen-Based Immunosensors

Another strategy for virus detection is to modify electrodes with antigens to detect circulating antibodies. Wang et al. developed a sensitive method for detecting Zika virus specific antibodies using non-faradaic capacitance sensing (Figure 4A).¹²¹ Zika virus (ZIKV) is a member of the Flaviviridae family (genus Flavivirus) that has been rapidly spreading across the globe through Aedes mosquito vector and sexual transmission. ZIKV was declared a Public Health Emergency of International Concern in February 2016 by the World Health Organization (WHO) due to the expanded outbreaks throughout much of the Americas.^122,123 ZIKV has been detected in blood,¹²⁴ semen,¹²⁴ urine,¹²⁵ saliva¹²⁶ and breast milk¹²⁷ of infected patients. The biosensor consisted of a glass substrate with a polydimethylsiloxane (PDMS) layer, and a two electrode system including gold and Ag/AgCl microwires. The gold microwires were modified with MUA. The electrode surface was activated using NHS/EDC followed by functionalizing with Zika virus (ZIKV E) or Chikungunya virus (CHIKV E) envelope antigen. The change in the total capacitance was measured due to the binding of antibody to antigen. The detection limit was approximately 10 antibody molecules in a 30 μL sample and the sensor was able to distinguish antibody isotypes. Also, the platform was used for specific, sensitive detection of polyclonal anti-ZIKV antibodies available in mouse serum. The developed technique is more advantageous than other immunosensors or ELISA assays because of its superior detection limit.

Figure 4. Refer to the following caption and surrounding text. — **Figure 4.** (A) (i) Capacitive immunosensor design and (ii) working principles. Adapted with permission from Ref. 104, Copyright 2019, Elsevier B.V. (B) The fabrication steps of the impedimetric immunosensor for dengue virus. Adapted with permission from Ref. 111, Copyright 2018, Elsevier Ltd.

Palomar et al. developed an impedimetric biosensor to detect the dengue virus antibody (Figure 4B).¹²⁸ Dengue is a major health issue due to its transmission by mosquitoes, and it is estimated that there are 50 to 100 million new cases every year according to WHO.¹²⁹ In this work, multi-walled CNT were deposited on a glassy carbon electrode and then functionalized with pyrrole-NHS via electropolymerization. Dengue virus 2 NS1 glycoprotein was subsequently immobilized via amide coupling. The GCE/CNT/poly (pyrrole-NHS)/Dengue Virus NS1 sensor was used to sensitively detect the antibody of Dengue Virus 2 NS1 in PBS and bovine plasma using EIS in the presence of [Fe(CN)₆]^3−/4−. The biosensor had a linear range between 10⁻¹² and 10⁻⁵ g mL⁻¹ with a detection limit of 10⁻¹² g mL⁻¹ in PBS and a linear range between 10⁻¹¹ and 10⁻⁵ g mL⁻¹ with a detection limit of 10⁻¹² g mL⁻¹ in plasma.

Mikula et al. reported an electrochemical biosensor for sensitive detection of anti-hemagglutinin antibodies.¹³⁰ A gold electrode was modified with 4-mercapto-1-butanol (MBT) and synthesized dipyrromethene derivatives (DPM). Cu (II) was attached to the electrode surface with a surface coverage of 7.35 ± 0.4 × 10⁻¹¹ mol cm^{− 1} due to affinity of DPM toward transition metal ions. The recombinant histidine-tagged hemagglutinin (HA) from the H1N1pdm09 influenza virus (His6-H1 HA) was immobilized on the DPM-Cu (II) redox activated electrode. The change in Cu (II)/Cu (I) redox current was measured via Osteryoung square-wave voltammetry to detect anti-hemagglutinin H1 monoclonal antibodies against swine virus present in mice sera vaccinated with mixture of His6-H1 HA in monomeric and oligomeric form. It was possible to detect the antibodies in sera sample diluted 10⁹ fold and this sensitivity was a few order of magnitude better than ELISA test.

Challenges with antigen-based biosensors

Antigen-based sensors share many of the challenges associated with antibody-based sensors such as reagent stability and immobilization concerns. In addition, synthetic proteins or peptides are often used as antigens and may not have the correct structure for which the antibody was generated in the body, so careful validation of these synthetic peptides and antigens in non-biosensor assays is important. These factors can make creating a viable sensor difficult, but consideration of these factors at the start of development will significantly improve the probability of a successful assay being produced.

Conclusions

Electrochemical biosensors for early detection of viral infections have drawn attention due to their selectivity and sensitivity as well as wide dynamic range, simple fabrication, fast analysis and small sample volumes. This review describes the recent developments in fabrication and application of electrochemical biosensors for detection of viruses. These examples prove that electrochemical biosensors can offer advantages over conventional methods such as enzyme-linked immunosorbent assay, polymerase chain reaction or western blot due their characteristics such as simplicity, portability, miniaturization, disposability, fast response, real-time analysis. Also, radioactive labels, toxic dyes, and expensive instrumentation are not necessary with electrochemical biosensors. Nowadays various nanomaterials such as gold nanoparticles, carbon nanomaterials, magnetic nanoparticles or magnetic beads, quantum dots and hybrid nanostructures have been employed for fabrication of biosensors to enhance their sensitivity and stability. Also, microarrays integrated with microfluidic systems usually provide low-cost fabrication and simultaneous viral detection. Although electrochemical biosensors demonstrate promising features, some of the reported biosensors need to be validated in real samples.

Acknowledgments

Present work was supported by funding from the National Science Foundation (CHE-1710222) and the National Institutes of Health (NIH/NIAID R41 AI141047 and R01 AI132668).

ORCID

Charles S. Henry 0000-0002-8671-7728

Author affiliations

Author notes

ORCID iDs

Dates