Rapid Detection of Zika Virus in Urine Samples and Infected Mosquitos by Reverse Transcription-Loop-Mediated Isothermal Amplification

Infection with Zika virus (ZIKV) is of growing concern since infection is associated with the development of congenital neurological disease. Quantitative reverse transcription PCR (qRT-PCR) has been the standard for ZIKV detection; however, Reverse Transcription Loop-Mediated Isothermal Amplification (RT-LAMP) may allow for faster and cheaper testing. Studies have suggested that ZIKV detection in urine is more sensitive and has a longer window of detection compared to serum and saliva. The objective of this study was to develop a urine diagnostic test that could be completed in under 30 minutes. Urine samples spiked with ZIKV or dengue virus were tested using RT-LAMP as well as by conventional quantitative qRT-PCR. These techniques were then validated using crude lysates made from ZIKV infected mosquitoes in addition to urine and serum samples from ZIKV infected patients. RT-LAMP specifically detected ZIKV in urine and serum for ZIKV infected patients and crude mosquito lysates. This test was performed in under 30 minutes and did not require RNA extraction from urine nor mosquitos. This approach could be used for monitoring of exposed individuals, especially pregnant women, couples wanting to conceive, or individuals with suspicious symptoms as well as surveillance of mosquito populations.


RT-LAMP is specific and sensitive for Zika Virus.
To establish the optimal conditions for RT-LAMP using a ZIKV PCR control, several primer sets, ranges of temperatures (57-65 °C), and incubation times (30-60 mins) were tested. The best amplification results were obtained at 61 °C for 30 mins as indicated by a banding pattern after electrophoresis on a gel (Fig. 1A). Positive reactions containing SYBR Green I could be observed by naked eye by a color change from orange to yellow, under fluorescent light in response to UV excitation, or by laddering pattern of bands after electrophoresis on a gel (Fig. 1A). The RT-LAMP reaction required all 6 primers to work under the optimized conditions; removing the forward and backward inner primers or the loop primers did not result in a positive result (Fig. 1B). In order to determine the lower detection limit of the RT-LAMP reaction for ZIKV, a dilution series ranging from 1 × 10 3 to 1 × 10 8 PFU ZIKV was amplified (Fig. 2). The limit of detection was approximately equivalent to 1 genome.
We then determined the specificity of the RT-LAMP assay by testing whole cell lysates and supernatant from Aedes albopictus C6/36 cells infected with ZIKV or DENV, a closely related flavivirus. Only whole cell lysates and supernatants from ZIKV infected cells, but not DENV, had positive RT-LAMP reactions indicating specificity of the reaction for ZIKV (Fig. 3).

Zika Virus and 18 S rRNA detection in urine.
To determine if RT-LAMP could detect ZIKV in human urine samples, human urine samples were spiked with ZIKV. To demonstrate specificity, some urine samples containing DENV were also tested. Urine samples were directly used for RT-LAMP without performing RNA isolation. RT-LAMP for ZIKV was positive for ZIKV containing urine samples for all the ZIKV strains tested to varying degrees but not for DENV positive samples, demonstrating that RT-LAMP could work in reactions containing urine and was specific (Fig. 4A). The sequence of the RT-LAMP primers were also compared to aligned sequences of related arboviruses including Chikungunya (CHIKV), DENV, Japanese encephalitis virus (JEV), West Nile virus (WNV), and Yellow Fever Virus (YFV) (Supplemental Table S1). The percent nucleotide mismatch varies from 29.6% to 52.5%, which makes it highly unlikely for there to be false positive detection with these other viruses. In order to have a quality control for RT-LAMP reactions, we designed a RT-LAMP reaction specific to human 18 S ribosomal ribonucleic acid (18 S rRNA). 18 s rRNA has good integrity in human urine and is often used for normalization of RNA [9][10][11] . Both ZIKV and DENV urine samples, but not the negative no template control (NTC), were positive for 18 S rRNA RT-LAMP reaction (Fig. 4B).

Zika virus and Aedes aegypti Actin detection in mosquitos.
We sought to determine if our ZIKV RT-LAMP was sensitive enough to detect ZIKV in a single infected mosquito. Quantitative reverse transcription PCR (qRT-PCR) was done to confirm that ZIKV was able to infect this colony of mosquitoes (Fig. 5A). Crude mosquito lysates were used for RT-LAMP without RNA isolation. In the ZIKV specific RT-LAMP reaction, there was robust detection of ZIKV from just a single ZIKV infected mosquito, but no signal from the mock or DENV infected mosquitos (Fig. 5B). Thus, the ZIKV RT-LAMP is specific and sensitive for ZIKV in as few as one single infected mosquito. To again establish a RT-LAMP reaction that could serve as a quality control, a RT-LAMP reaction was also designed against Aedes aegypti Actin. Actin was detected in all samples containing a mosquito but not in the no template negative control (Fig. 5C). Lastly, qRT-PCR was done using the crude mosquito lysates without RNA isolation. All crude single mosquito lysates had detectable levels of housekeeping gene s17; as expected only the positive control and ZIKV infected samples had detectable ZIKV mRNA (Fig. 5D).   Table 1). Patient #Z02, a women at 8 weeks pregnant, had a positive urine sample and a weakly positive serum sample by RT-LAMP (Fig. 6A). This patient also had a positive IgM and a weakly positive IgG test (Table 1). Patient #Z03, who had no symptoms, had a positive serum and negative urine for RT-LAMP. The IgM test was reported to be positive for this patient. Six other urines from patients were also tested with 4 of these being positive by RT-LAMP without RNA isolation (Fig. 6A, Table 1). Patient #Z08 was positive by RT-LAMP and not qRT-PCR, but only after RNA isolation ( Fig. 6B,C, Table 1). Patient #Z09, whose urine sample was collected on day 15 post-infection, had milder symptoms than the other patients and no longer had detectable ZIKV in their urine as confirmed by qRT-PCR testing after RNA isolation ( Fig. 6B,C, Table 1). Detection of 18 S rRNA by qRT-PCR demonstrated that RNA could be successfully detected in this sample. 20 asymptomatic controls from areas of low risk for ZIKV transmission were also tested with no positive results (Fig. 6B, not all data shown).

Discussion
Given the rapid emergence of ZIKV and the severe birth defects that can result in children of infected mothers, improved diagnostic options that are fast, reliable, easy, and affordable are required. This is critical since ZIKV infection in adults can be asymptomatic or result in non-specific, flu-like symptoms that could also be the result of other common diseases endemic to the area, including DENV. Moreover, couples trying to conceive may not know they are infected, and as ZIKV RNA can remain detectable in semen for up to 6 months after onset of symptoms 3,12 , there is risk of unknowingly passing on ZIKV infection in utero. Conventional qRT-PCR, while specific and sensitive, must be done by trained personnel on specialized equipment at a qualified laboratory. This study demonstrated that RT-LAMP allows rapid detection of ZIKV in clinical urine and serum samples as well as mosquito samples. Furthermore, we have developed two robust RT-LAMP assays that can serve as a sample quality control for the RT-LAMP reactions by amplifying housekeeping genes-18 S rRNA in human samples and Actin in Aedes aegypti mosquito samples.
Currently, clinical testing for ZIKV in the United States is done by central testing laboratories, resulting in a 2-4 week turn-around to patients for results. This study sought to improve upon this by developing a potential point-of-care test. Point-of-care testing has several advantages for emerging infectious diseases like ZIKV and must be easy to use and analyze, low cost, fast, and require little if any laboratory infrastructure without compromising on sensitivity or specificity. RT-LAMP, as a nucleic acid based test, meets these requirements and therefore has large value for screening and testing for ZIKV and other infectious diseases in potentially exposed populations, both human and endogenous vectors including Aedes aegypti. Only a few kits are commercially available for   mosquito testing for DENV, and those are antigen based which may lack the specificity a nucleic acid test could provide. There is currently no commercially available test for ZIKV surveillance in mosquitos.
In some of the experiments in this paper, urine preservative (Norgen Biotech) was added to urine samples immediately after collection. This was done as some urine samples were collected where refrigeration was not immediately possible; therefore, preservative was required to stabilize any RNA in the samples. The presence of urine preservative in samples did not have any noticeable effect on RT-LAMP reactions compared to samples with no urine preservative. Furthermore, we found that adding urine at more than 20% of the total volume of the RT-LAMP reaction was inhibitory. An unexpected benefit was that we did not have to isolate RNA from the samples; unprocessed samples were directly used in the RT-LAMP assays. From the clinical samples tested, 7 of 8 symptomatic patients were positive by RT-LAMP and this was verified in some patients by a positive IgG or IgM tests. In cases where both urine and serum samples were tested, the urine samples gave stronger RT-LAMP signals than the serum samples. One patient, who was asymptomatic, was positive by RT-LAMP in the serum sample only. Samples of one patient first required RNA isolation for ZIKV to be detectable by RT-LAMP. However, qRT-PCR was still not able to detect ZIKV from this isolated RNA sample, suggesting that RT-LAMP may be more sensitive. One patient did not test positive for RT-LAMP, however ZIKV was also not detectable in their urine by qRT-PCR even after RNA isolation and purification, suggesting that ZIKV was no longer detectable in the urine. Although urine has a longer detection window for ZIKV than serum or saliva, it does decrease with time and this urine sample was collected 15 days after symptom onset 3 . Lastly, all 20 asymptomatic control patients, taken from geographical regions with low ZIKV infection risk, were negative for the ZIKV specific RT-LAMP reaction.
Primers were designed for a highly conserved sequence of the nonstructural protein 5 of ZIKV that was found in 16 strains and were verified in 5 different ZIKV strains and in three sample types (human urine and serum, and mosquitos). We tested a range of temperatures from 57 °C to 65 °C; all the temperatures gave comparable results thereby this particular RT-LAMP reaction had a large temperature range. We also tested a range of times from 5-60 minutes. The optimal time for detection of RT-LAMP products by color change alone was 30 minutes, however we could detect by UV light excitation or banding patterns on gels in a total incubation time of as little as 8 minutes. The addition of DMSO in our reaction completely inhibited the RT-LAMP reaction. Lastly, the use of Bst polymerase 2.0 WarmStart polymerase was used as this has been reported to obtain amplification signals quicker and have increased stability at room temperature compared to wild-type Bst DNA polymerase 13,14 .
Other groups have also developed new technologies for detection of ZIKV 15-21 , however many technologies have only been demonstrated in spiked samples, require RNA isolation and purification, have not been demonstrated to work in mosquitos, or have not been demonstrated to work in urine samples which may contain inhibitors for the proposed technology. None are yet commercially available. In contrast, we demonstrated that our diagnostic test works in both spiked and endogenously infected samples, in both human and mosquito samples, and works in both urine and serum samples. The limit of detection was approximately equivalent to 1 genome. Typical viremia of a symptomatic ZIKV infected patient is 103-106 PFU/mL 22 . Importantly, this method was able to detect ZIKV in samples without first doing a RNA isolation and purification, which significantly decreases the time and cost of this assay.
Although RT-LAMP reactions are highly specific, there are several limitations including that it is not a quantitative test. There can be a higher rate of false positives, however we did not experience this in any of our no template negative control reactions, known ZIKV negative control samples, DENV positive controls, or mock mosquito samples. The primers selected also have poor nucleotide alignment with other flaviviruses, making it highly improbable to detect these other viruses. We also took the extra precautions of having a lateral work flow for all experiments including analysis and we included Antarctic Thermolabile UDG (Uracil-DNA Glycosylase; NEB) in all reactions. UDG removes the uracil-base from DNA, thereby preventing possible carry-over contamination from previous reactions. Lastly, this study was not powered to determine sensitivity in a clinical population. All experiments were performed in accordance with relevant guidelines and regulations. Midstream urine samples were collected by the patient and had preservative (Norgen Biotek) added immediately after collection then shipped to Beaumont and stored at room temperature. Some urine samples did not contain the preservative; after collection they were immediately centrifuged for 10 minutes at 700 g and the supernatant frozen at −80 °C until analysis. Mosquito experiments. Aedes aegypti were maintained on raisins at 27 °C and 80% humidity according to standard rearing procedures. Mosquitos were propagated by feeding defibrinated sheep blood using a Hemotek artificial membrane feeder 23   Analysis of RT-LAMP. RT-LAMP reactions were divided in half. One half of the RT-LAMP reaction was electrophoresed along with Invitrogen Low DNA Mass Ladder on a 1 × Nancy-520, 2% agarose gel in 1 × TAE buffer (40 mM Tris, 20 mM acetic acid, 1 mM EDTA) at 90 V for 90-120 minutes. Gels were imaged under UV light using the BIO-RAD ChemiDoc XRS + Imaging System. Lanes containing a laddered banding pattern were qualified as a positive amplification. In addition, a 1:10 SYBR green I (Life Technologies) dilution was made in TAE buffer, then 2.0 μL of the SYBR dilution was added to the remaining half of the reaction. The subsequent visual change of color (orange to yellow) was then also used to identify positive amplifications. The SYBR green I PCR tubes were also imaged under UV light in the BIO-RAD ChemiDoc XRS + Imaging System as the reaction creates a fluorescent output. Data availability. All relevant data are within the paper.