Viral diseases of plants cause significant economic losses and are a threat to the world’s food security. Plant virus detection and identification is key to the management of plant diseases for growers in their fields, but also for quarantine agencies, nurseries, and for diagnosis laboratories. Plant viruses are unculturable, which poses a major complication, thus they need to be detected directly from the infected specimens without culturing in growth media. Plant viruses are particularly difficult to manage since infected plants cannot be cured (except through lengthy tissue culture) and their early diagnosis is key to controlling their dissemination. Furthermore, plant viruses are typically low in concentration in plant tissues at the beginning of infection, thus, to identify viral infections promptly and effectively it is required to boost the viral titer of plant samples for the effective use of molecular biology techniques. Classical laboratory techniques such as enzyme-linked immunosorbent assay (ELISA), the polymerase chain reaction (PCR) and real time polymerase chain reaction (RT-Q PCR) require specific antibodies against the target viruses or the knowledge of their nucleotide sequence, and, thus, are only suitable to detect known viruses. Recently, the use of unbiased technologies such for whole genome sequencing such as Next Generation Sequencing (NGS) or High-Throughput Sequencing (HTS) constitutes the best approach to detect novel and unknown viruses both in plants (Malapi-wight et al., 2021, Mollov et al., 2018, Malapi-wight et al., 2018, Roossinck et al., 2015, Ho et al., 2014) and insects (Nouri et al., 2015).

Although many quarantine programs around the world are validating the use of HTS, there is still the need to validate the optimal sequencing depth for the identification of all pathogens present in the samples, including those viruses that may occur at low titer, this task tis particularly challenging, since it must be validated for each crop. This bottleneck arises from the fact that there are multiple sizes of host genomes, and the viral titer can change dramatically among plant species and their associated pathogens.

The optimization and harmonization of HTS throughout quarantine programs in the US and the rest of the world has been the focus of reviews (Waage and mumford, 2008, Waage, 2001) and initiatives, such as the FAO 2016 Procedures for Global Harmonization of Plant Quarantine; or the North America Plant Protection Organization, (NAPPO), that, as testimony to its commitment to quarantine harmonization, “each year recognizes representatives from its three member countries—the United States, Canada, and Mexico—for their efforts to promote and implement regionally harmonized approaches to managing plant pest threats”.

A major problem for the implementation of unbiased virus identification methods such as HTS is the implicit requirement of sequencing and analyzing all the genetic material comprised in the sample. A widespread problem when applying HTS to detect/identify viruses is the high occurrence of host material reads and, in less measure, reads from other pathogens including bacteria and other viruses (Roossinck et al., 2015). There are traditional virus purification techniques that are used to increase the number of viral reads while simultaneously diminishing the host reads. Among the most widely used we find ultracentrifugation, immunocapture and membrane filtration.

The main disadvantage of ultracentrifugation is that each virus/host combination requires a specific protocol; for instance, the protocol used to separate ZYMV (Lisa et al., 1981) could not be used for TSWV (Joubert et al., 1974), thus, this procedure that could not be readily applied to unknown viruses.

Another virus separation technique is ultrafiltration using membranes, which requires large area filters (cm2) due to the low porosity shown by the membranes. In membrane ultrafiltration, the virus are accumulate on the upstream side of the system and is limited by the low porosity of the membranes which is around 10%, also requiring large sample volume (100 ml) to be able to separate enough virus material [(Jeon et al., 2014)], Finally, immunocapture is highly efficient, but very specific as it requires the knowledge of the molecular structure of the target virus in order to synthesize and use the antibody for capture [(Lien et al., 2007)].

To address several of the shortcomings described above, we have developed a carbon nanotube microfluidic device (CNTMD) capable of retaining virus from a plant extract, decreasing host nucleic acids, and ensuring that no low titer virus would be missed during HTS analyses. Regardless of the molecular technique used to identify viruses, having a tool to select and enrich viruses vs. host material would undeniably increase the success of plant virus diagnosis. In addition to improving virus processing and increasing identification efficiency, CNTMDs simplify the entire plant virus identification process. In the long term, we envision that CNTMDs would be shared with quarantine programs in the US.

CNTMDs use vertically aligned carbon nanotubes arrays as virus capture and filtration elements. Carbon nanotubes are ultra-thin (1-100 nm) hollow fibers made of carbon; the wall of a carbon nanotube consists of one or multiple sheets made of hexagonal lattice of carbon atoms rolled onto itself (Ebbesen and Ajayan, 1992). It has been three decades since CNTs (carbon nanotubes) were discovered, nowadays there is a wide variety of synthesis methods and applications in biology, among those are tissue scaffolds, cell substrates, drug delivery (Heister et al., 2013) and virus filtration (Brady-Estévez et al., 2010) (Shannahan et al., 2013). CNTMD technology is based on aligned carbon nanotubes, such alignment is a result of the simultaneous and collective growth of millions of individual CNTs on a flat substrate (Terrones et al., 1997). Fig. 1a depicts a representative low magnification SEM (Scanning Electron Microscope) micrograph of a vertically aligned CNT array that we use for trapping viruses in our microfluidic devices. CNTs originate from catalyst nanoparticles (iron) pre-deposited on a silicon substrate, the quantity and size of such catalyst nanoparticles defines the morphology of the individual CNTs and the density of the array, conveniently, the porosity of our CNT arrays is remarkably high as it easily reaches 90%. In these CNT arrays the average distance between individual nanotubes can be regulated from 20 to 500 nm; that spacing range matches the size of most known viruses. Tuning the spacing between nanotubes makes it possible fabricate devices for more efficiently capturing viruses of specific size.

Nowadays, most bio-separation techniques based on low-porosity membrane filtration are intended to eliminate the virus particles from the sample, in these methods the virus is discarded along with the membrane and further analysis is not possible (Mostafavi et al., 2009, Kim et al., 2016). A few years ago, we implemented a microfluidic device using high porosity CNT arrays and demonstrated enrichment of avian viruses. In that study the average intertube distance was around 80 nm intended to match the dimensions of avian viruses (Yeh et al., 2016). The high porosity of the nanotube array prevents clogging and allows the processing of larger sample volumes and consequently the capture of more viruses. Fig. 1b shows a photograph of one completed CNTMD which is compact and portable.

Since this technology is based on size exclusion, it can be used to identify new virus variants because it does not require antibodies or pre-knowledge of their genome sequence. CNT arrays have been successful in detecting and identifying known and unknown animal viruses in birds. However, this is the first time it has been evaluated for detecting plant viruses. Plant tissues contain a much higher amount of complex carbohydrates, polyphenolics, and tannins that make them a difficult substrate for extraction of nucleic acids (Rezadoost et al., 2016), compared with animal and cell tissues, in consequence the protocols for virus detection from vegetal samples often need to be modified from their use in animal virus counterpart.

In this work, we present the protocols for capturing plant viruses using optimized CNTMDs and the analysis and identification by RT-Q PCR and NGS to show the feasibility of using unbiased diagnostic techniques. The viruses that were used in the study are tomato spotted wilt orthotospovirus (TSWV), which has spherical shape and diameter from 80 to 120 nm (Scholthof et al., 2011) and the long flexuous zucchini yellow mosaic potyvirus (ZYMV) which is an elongated rodlike virus ~700 nm in length and ~12 nm in diameter (Wong et al., 1994).

Fig. 2 shows a schematic depiction of the virus retention in the CNT arrays. Vertically aligned CNT arrays are the core element of this virus enrichment system. CNT arrays are fabricated by selectively patterning films of iron catalyst on silicon substrates (see Fig. 1a). The CNTs grow by a chemical process known as chemical vapor deposition (CVD) in which a liquid solvent, which acts as carbon source, is supplied into a to quartz tube at 850 °C where the silicon substrates are located. In the CNT growth process, the amount of catalyst pre-deposited on the silicon surface strongly affects the density and length of the resulting aligned carbon nanotube array. In general, thinner iron catalyst (few nm) results in densely packed small-diameter nanotubes while thicker iron catalyst films (>10 nm) yield larger-diameter nanotubes broadly distributed in space. A detailed description of the CNT growth and related processes can be found in (Yeh et al., 2016). In a CNT forest the tubes are randomly distributed, Fig. 3b depicts the distribution of the spacing between tubes. It can be noticed that for CNT arrays formed by small-diameter nanotubes the dispersion if the spacing distance is smaller while for large-diameter CNT arrays the distribution is broader. The data obtained in this report was generated from three different average inter tube distances: 105 nm, 155 nm, and 293 nm.