On‐site genetic analysis for species identification using lab‐on‐a‐chip

Abstract This paper presents a microfluidic device capable of performing genetic analysis on dung samples to identify White Rhinoceros (Ceratotherium simum). The development of a microfluidic device, which can be used in the field, offers a portable and cost‐effective solution for DNA analysis and species identification to aid conservation efforts. Optimization of the DNA extraction processes produced equivalent yields compared to conventional kit‐based methods within just 5 minutes. The use of a color‐changing loop‐mediated isothermal amplification reaction for simultaneous detection of the cytochrome B sequence of C. simum enabled positive results to be obtained within as little as 30 minutes. Field testing was performed at Knowsley Safari to demonstrate real‐world applicability of the microfluidic device for testing of biological samples.


| INTRODUC TI ON
Microfluidics describes the use of systems which enable the manipulation of small amounts of fluids in channels in the micron range (Whitesides, 2006). Originally conceived from microanalytical methods, such as high-pressure liquid chromatography (HPLC), the field of microfluidics has rapidly expanded and led to significant benefits in biological research. The ability to conduct work on a miniaturized scale offers a number of advantages over conventional laboratory-based methods including small reagent and sample requirements, low cost, faster reaction times, and a smaller footprint, all of which are beneficial in producing portable systems (Whitesides, 2006).
Integrated systems capable of "sample in-answer out" genetic analysis of biological samples have previously been demonstrated, whereby biological samples are added to a microfluidic device and undergo integrated processing, usually consisting of nucleic acid extraction, amplification, separation, and detection (Liu & Mathies, 2009;Park et al., 2014). The vast wealth of literature is also reviewed in a number of more tailored articles on point-of-care testing for detection of infectious diseases (Zhang et al., 2017), bacterial pathogens (Lui et al., 2009), forensic analysis (Bruijns et al., 2016), and cancer diagnostics (Newport et al., 2006). Such systems would also be advantageous in conservation and species management settings, for example, where field-based testing could overcome problems with securing export permits for biological samples, or offering the potential for in situ genetic testing in locations where conventional laboratories may not be available.
Despite this, the use of microfluidic devices for genetic analysis in conservation has so far proven limited to single downstream analysis steps, lacking essential prior nucleic acid extraction and amplification, which are still performed using conventional means.
DNA sequencing using MinION nanopore sequencers (Oxford Nanopore Technologies) has been demonstrated in the Ecuadorian Chocó rainforest for barcoding of reptile specimens (Pomerantz et al., 2018). Microfluidic single nucleotide polymorphism (SNP) arrays have been applied to analysis of scat samples from wolverine (Gulo gulo) (Ekblom et al., 2018), gray wolves, European wildcats, and brown bears (Von Thaden et al., 2017). The use of microfluidic capillary electrophoresis chips has been shown in identifying illegal wildlife trade through restriction fragment length polymorphism (RFLP) analysis, for example, the identification of Malayan box turtle (Cuora amboinensis) in traditional Chinese medicines (Asing et al., 2016) and Macaque monkey (Macaca fascicularis sp.) in processed foods (Rashid et al., 2015). A full work flow from extraction to identification has yet to have been demonstrated in samples obtained from natural populations. The development of such a system would be extremely valuable to wildlife forensics, conservation, ecology, and taxonomy.
Biological sample collection from animals, particularly those in the wild, is most easily performed through noninvasive methods.
Dung samples in particular are extremely valuable as animals regularly defecate, the samples are relatively easy to locate, and the process of collection requires little expertise or expense (Taberlet et al., 1999). However, the use of dung samples for genetic analysis presents challenges due to the heterogeneous nature of the samples, the presence of inhibitors, low target analyte concentrations, and the potential for degraded DNA (Fernando et al., 2003). To date, the analysis of human stool samples on microfluidic devices has mainly focussed on watery samples taken from patients with diarrhea or highly diluted samples (5%-10% fecal samples) to identify infectious agents (Bunyakul et al., 2015;Fronczek et al., 2014;Phaneuf et al., 2018;Ye et al., 2018). More recently, Zhao and colleagues have used acoustic streaming to liquefy human stool samples on a microfluidic device, followed by filtering out of large debris using a micropillar array (Zhao et al., 2019). However, dung samples from herbivores present more of a challenge due to the presence of large amounts of fibrous plant material, and the presence of inhibitors, such as polysaccharides. The use of solid-phase extraction techniques is beneficial in these scenarios as they facilitate physical separation of the nucleic acids (Fernando et al., 2003), with the added advantage of enabling preconcentration of the nucleic acids which is beneficial when dealing with low target analyte concentrations (Kashkary et al., 2012). We have previously shown how using immiscible filtration assisted by surface tension (IFAST) can be successfully utilized as an example of microfluidic solid-phase DNA extraction for the detection and analysis of Helicobacter pylori in human stool samples (Mosley et al., 2016); however, these techniques have never been applied to nonhuman samples.
In many conservation settings, whether it be zoological institutions or in natural habitats, resources can be limited and therefore the advantages afforded by microfluidic devices can be beneficial.
Here, we demonstrate a microfluidic device which is capable of processing animal dung samples, from DNA extraction through to loop-mediated isothermal amplification (LAMP), to produce a "yes/ no" result for species identification. Proof of concept was demonstrated through identification of Ceratotherium simum. C. simum is a near threatened species that are susceptible in the wild due to poaching for their horns, with 1,054 reported killed in 2016 in South Africa alone (WorldWildlifeFund, 2017). The microfluidic device also incorporates controls to ensure validity of the results. Optimization of the DNA extraction and LAMP steps is demonstrated, alongside field testing of the integrated system.

| Microfluidic device fabrication
The microfluidic devices used in this experiment were 7-chamber polymer-based devices comprising of 5 interconnected chambers for sample analysis (Figure 1a, chambers A-E). The large sample chamber into molds designed in Solidworks and produced from polymethyl methacrylate (PMMA) using an M7 CNC milling machine (Datron, UK) ( Figure 1b) (Ngamsom et al., 2019). The mixture was degassed by placing in a vacuum chamber for 10 minutes before being cured at 60°C for 1 hr. Once set, the PDMS microfluidic devices were removed from the mold and plasma bonded to a plain glass microscope slide (Fisher Scientific, UK), using a Corona SB device (Blackhole Laboratories, France), to facilitate sealing of the chambers.

| Microfluidic
The chambers of the microfluidic device were filled with the following reagents, in sequential order, for optimization of the DNA

| Primer design
Species-specific primers were designed against the full C. simum cytochrome B sequence (GenBank accession number JF718874.1). A nucleotide BLAST was carried out to compare this sequence with the two full C. simum mtDNA sequences available on GenBank (accession numbers Y07726.1 and NC_001808.1), and this showed a 100% match in both cases; therefore, this sequence was taken to be representative of the species based on the available data.
PCR primers were designed using Primer3Plus (https://prime r3plus.com/cgi-bin/dev/prime r3plus.cgi), and generated an expected amplicon size of 110 bp, and were also used in the qPCR experiments for quantification. We performed in silico analysis on the PCR primers using Primer-BLAST against all mammals and this showed 1 potentially unintended match against Phocoena sinus ATP synthase subunit-a. There were 2 mismatches on the forward primer and 4 on the reverse. This is a species of porpoise which we believe is unlikely to ever be cosampled. We also used the UCSC in silico PCR tool which showed only products for C. simum.
LAMP primers were designed using PrimerExplorer V4 (https://prime rexpl orer.jp/e/) ( Table 1). LAMP specificity testing in silico was evaluated using FastPCR which enables linked searching to be carried out on more than a single primer pair (Kalendar et al., 2017). No unintended products were calculated for any other rhinocerotidae evaluated (Ceratotherium simum cottoni,

| Polymerase Chain Reaction (PCR)
PCR was used as a means to confirm the species and quality of DNA obtained from samples, prior to use in the LAMP assay. In addition to the primers for C. simum described above, species-specific primers were also purchased for D. bicornis (Brown & Houlden, 1999) and E.
ferus caballus (Tanabe et al., 2007). PCR was performed using the fol-  (Nagamine et al., 2002). Once the extracted DNA was added, the samples were heated to 65°C on a Prime thermal cycler (Techne, UK) and the reaction allowed to proceed for up to 60 minutes.

| Detection
All DNA amplification products were run on a 2% (w/v) agarose gel,

| Integration and field testing
Once the conditions for DNA extraction and LAMP had been optimized, the molecular biology grade water on the microfluidic device ( Figure 1; chamber E) was replaced with LAMP reagents to enable integrated "sample in-answer out" analysis to be performed. Following confirmation that the integrated procedure was successful in the laboratory, testing was carried out on-site at Knowsley Safari.

| Microfluidic DNA extraction
Optimization of the DNA extraction processes for dung samples was

| Specificity testing
In addition to the in silico specificity testing described in the methodology, experimental evaluation against closely related species identified through phylogenetic analysis (Price & Bininda-Emonds, 2009) was carried out on those samples which could be physically obtained and included D. bicornis and another Perissodactyl, E. ferus caballus.
Visual examination showed a positive reaction for C. simum, and no reaction against D. bicornis, E. ferus caballus or for the negative control ( Figure 4a). This was then confirmed using gel electrophoresis ( Figure 4b). All DNA extracts were also successfully amplified using the polymerase chain reaction with species-specific primers to ensure DNA of amplifiable quality was present.

F I G U R E 2
Schematic showing operation of the microfluidic device in the field: (i) dung sample collection with minimal off-chip preparation and chemical cell lysis; (ii) DNA binding to PMPs; (iii) washing steps; and (iv) DNA amplification using colorimetric LAMP, including positive and negative controls

| Speed of analysis
The time taken to achieve a positive amplification result using the LAMP primers, with DNA extracted from C. simum, was also investigated and showed products from as little as 30 minutes when visualized using gel electrophoresis ( Figure 5). Visual analysis of the color change (pink to yellow) was also carried out, and RGB analysis showed statistically significant differences from 45 minutes compared to control samples (p < .05).

| Field testing of the integrated microfluidic device
Once optimized, the microfluidic devices were taken to Knowsley Safari for field testing. Fresh dung samples were collected and then analysis took place on the back of a pick-up truck within the park The use of the microfluidic device offers advantages in terms of reduction in the overall speed of analysis (<5 minutes compared to approximately 1 hr) and portability for field-based applications but without a reduction in the amount of DNA recovered.

| LAMP DNA Amplification
A species-specific LAMP reaction has been developed for C. simum, which offers a number of advantages over conventional PCR-based methods of identification. These include simpler operating systems, small DNA template targets, and faster, visual detection (Becherer et al., 2020).

| Field Testing of the Integrated Microfluidic Device
Field testing of the microfluidic devices demonstrated that species identification from dung samples could be confirmed using the system within 1 hr, compared with the previously reported DNA sequencing methods carried out in the field which were able to analyze specimens within 24 hr of collection (Pomerantz et al., 2018). Dung samples are added to the device, and a color change, from pink to yellow, is produced if the target species is present. This enables operation by nonspecialist personnel who can visualize the control and sample amplification chambers to check for a reaction, overcoming technical and language barriers. The work presented here demonstrates proof of concept in using such systems within the field, although the easy adaptation of the microfluidic device using alternative LAMP primers opens up possibilities in many areas of conservation where rapid, cost-effective, portable genetic testing would be beneficial particularly in scenarios where a simple "yes/no" result is required. More broadly, it could benefit the conservation and management of threatened taxa with particular utility in wildlife forensics, for example, the identification of animals in traditional Asian medicines or species confirmation from products seized as part of the illegal wildlife trade, and population monitoring of species with overlapping range whose dung is difficult to reliably identify, for example, similarities between eld's deer (Rucervus eldii siamensis) and F I G U R E 6 (a) Photograph showing field testing of the microfluidic device at Knowsley Safari, UK, including staff participation; (b) example image of a microfluidic device that was tested in the field following DNA amplification (n = 4); and (c) gel electrophoresis image showing results from samples tested in the field and brought back to the laboratory for confirmation analysis, where L is DNA size ladder, +ve is the positive control with lambda DNA, S is the extracted DNA from the dung sample processed on the integrated microfluidic device, and -ve is the negative control with no DNA present (a) (b) (c) muntjac (Muntiacus muntjac) dung, making field collection of samples more effective.

ACK N OWLED G M ENTS
The authors would like to thank Dr. Alexander Iles from the University of Hull for manufacturing of the PMMA molds used for creation of the microfluidic devices and Joana Borges for collecting the black rhino dung samples.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest to report.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analyzed in this study. All experimental data are contained within the manuscript.