SaliVISION: a rapid saliva-based COVID-19 screening and diagnostic test with high sensitivity and specificity

The Coronavirus disease 2019 (COVID-19) pandemic-caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)– has posed a global threat and presented with it a multitude of economic and public-health challenges. Establishing a reliable means of readily available, rapid diagnostic testing is of paramount importance in halting the spread of COVID-19, as governments continue to ease lockdown restrictions. The current standard for laboratory testing utilizes reverse transcription quantitative polymerase chain reaction (RT-qPCR); however, this method presents clear limitations in requiring a longer run-time as well as reduced on-site testing capability. Therefore, we investigated the feasibility of a reverse transcription looped-mediated isothermal amplification (RT-LAMP)-based model of rapid COVID-19 diagnostic testing which allows for less invasive sample collection, named SaliVISION. This novel, two-step, RT-LAMP assay utilizes a customized multiplex primer set specifically targeting SARS-CoV-2 and a visual report system that is ready to interpret within 40 min from the start of sample processing and does not require a BSL-2 level testing environment or special laboratory equipment. When compared to the SalivaDirect and Thermo Fisher Scientific TaqPath RT-qPCR testing platforms, the respective sensitivities of the SaliVISION assay are 94.29% and 98.28% while assay specificity was 100% when compared to either testing platform. Our data illustrate a robust, rapid diagnostic assay in our novel RT-LAMP test design, with potential for greater testing throughput than is currently available through laboratory testing and increased on-site testing capability.


Materials and methods
Clinical sample collection and handling. The biospecimens in this study were obtained from both retrospective and prospective collections. For retrospective collection, this study used leftover nasopharyngeal swabs and saliva samples that were previously tested by the Yale Pathology Labs at the Department of Pathology, Yale School of Medicine for the clinical diagnosis of Covid-19. For prospective collection (on-site study), saliva samples were donated by symptomatic individuals who came to a Yale Health testing site for screening. An informed consent was obtained from all participating subjects through a form of verbal consent and a waiver of a signed consent, as approved by Yale's Institutional Review Board.
Nasopharyngeal swab (NP) samples were collected and transported in 3 mL of viral transported medium (VTM), according to CDC and FDA guideline. This collection is a part of routine clinical testing offered by the Yale Pathology Labs service. For saliva collection, the specimens were obtained from either passive drool or with a Micro•SAL Saliva Collection Kit (Oasis Diagnostics, Vancouver, WA), according to the manufacturer's instruction. Saliva samples collected from volunteers at the Yale Health testing sites or the Yale New Haven Sample lysis with lysis buffer. Upon collection, approximately 500 µl of saliva was aliquoted into a 1.5 mL Eppendorf microcentrifuge tube pre-loaded with 100uL of lysis buffer containing 10 mg/mL of Proteinase K (AmericanBio, Canton, MA) to release viral RNA and 2 M Guanidine hydrochloride (Sigma Aldrich, St. Louis, MO) diluted in sterile PBS. Once collected, saliva samples were lysed for 1 min at 1500 rpm. The lysed saliva samples were subsequently heat inactivated for 5 min at 95 °C. After cooled down, the treated samples were subjected to SaliVISION RT-LAMP and Thermo Fisher Scientific TaqPath COVID-19 RT-PCR assays immediately or stores at − 20 °C until further processing.
Assessment of Saliva's pH neutralization with dilution buffer. The neutralization of saliva's pH was assessed with sodium hydroxide solution at different concentrations. Briefly, negative saliva samples presenting low, or neutral pH were neutralized in different concentrations of sodium hydroxide (750 µM, 1 mM, and 1.5 mM). 1 M NaOH stock solution was prepared by dissolving NaOH pellets in nuclease-free water. To test the detection sensitivity of SARS-CoV-2 in these dilutions, these saliva samples were spiked with 20 copies/µl of synthetic SARS-CoV-2 RNA (Twist Bioscience, San Francisco, CA), in a total of 50 copies per final reaction. The diluted saliva samples were then subjected to the RT-LAMP reaction, described below. After 30 min, the reactions were terminated and transferred to an ice-cold metal block for 30 s before the final color was read and a photograph was taken with a cell phone camera.
SaliVISION multiplex primer design. SARS-CoV-2 genomes spanning clades G, GH, GR, GV, L, O, S, V from the global science initiative and coronavirus genomic database, GISAID, were randomly collected and aligned with Mega7. In the predominantly expressed N gene, two non-overlapping conservative regions, namely N.1 and N.2, were selected as the testing targets. N.1 primers were designed with NEB LAMP Primer Design Tool, and we used a primer set located in N.2 region from a previous study to do a multiplex availability test with the N.1 primers by using Thermo Fisher Scientific Multiple Primer Analyzer 40 . We used a primer set for RNaseP gene from a previous study as the internal control. A 4 nucleotide PloyA insert was introduced as a linker into all forward and backward inner primers, in order to cooperate with the UDG treatment reaction to decrease the risk of contamination from previous reverse transcripts. All multiplex primer sequences were listed in Table 1. BLASTN Somewhat Similar Alignment method was used to align the primers against the SARS-CoV-2 sequences from GISAID, to make sure there were no multiple mismatches for most of the genomes.
SaliVISION assay with the colorimetric reverse transcription loop-mediated isothermal amplification (RT-LAMP). Inactivated saliva samples were subjected to one-step reverse transcription loopmediated isothermal amplification (RT-LAMP), using our customized multiplex primer set that specifically targets two distinct regions on the N gene of SAR-CoV-2 viral genome. The inactivated and pH neutralized saliva samples (pH 7) were processed for RT-LAMP reactions using WarmStart Colorimetric LAMP 2X Master Mix with UDG (New England Biolabs, Ipswich, MA), according to the manufacturer's protocol. Briefly, the assays were assembled in total reaction volumes of 40 µl, including 2.5 µl of original saliva sample, 20 µl of Warmstart  TCC CCT ACT GCT GCC TGG AGG AAA ACA GTC AAG CCT CTT CTC G  1.6   BIP-N1  TCT CCT GCT AGA ATG GCT GGC AAA AAT CTG TCA AGC AGC AGC AAAG 1.6   F3-N1  GCC AAA AGG CTT CTA CGC A  0.2   B3-N1  TTG CTC TCA AGC TGG TTC AA  0.2   LF-N1  GCG ACT ACG TGA TGA GGA A  0.4   LB-N1  GGC GGT GAT GCT GCT CTT  0.4   FIP-N2  TGC GGC CAA TGT TTG TAA TCA GAA AAC CAA GGA AAT TTT GGG  SalivaDirect RT-qPCR. For SARS-CoV-2 detection in saliva samples with the FDA EUA approved Sali-vaDirect assay, saliva specimens were aliquoted into 96-well plates, in the amount of 50 µl, and subsequently treated with 2.5 µl of 50 mg/mL proteinase K (AmericanBio, Canton, MA). Sample plates were shaken for 1 min, wherein they were then incubated at 95 °C for 5 min for proteinase K inactivation. After that, the RT-qPCR was performed using TaqPath 1-Step RT-qPCR Master Mix (Thermo Fisher Scientific, Waltham, MA) with Saliva-Direct primer and probe set in a total volume of 20 µl per reaction, of which 5 µl was for the saliva sample. The samples were run using a Bio-Rad CFX96 Touch qPCR cycler. SalivaDirect's RT-PCR was used for viral nucleic acid detection with two probes: FAM probes for the presence of N1-gene amplicons, and Cy5 probes for the RNase P housekeeping gene. The threshold for positive samples was set at a Ct value of ≤ 40, in line with the SalivaDirect platform; however, exceptions were made based upon other result characteristics. Statistical analysis. All data were analyzed and graphed with GraphPad Prism 9 software (San Diego, CA). Specificity of the SaliVISION test was calculated as a percentage of the negative samples detected by RT-LAMP that were also negative in either SalivaDirect or TaqPath RT-PCR test. Sensitivity of a given Ct interval was calculated as the percentage of the positive samples detected by RT-LAMP that were also positive in either SalivaDirect or TaqPath RT-PCR test. In both cases, 95% confidence intervals were calculated by interpretating the proportion of counts as binomial rates and then computed using the modified Wald method using GraphPad Prism 9 software. All methods and experiments in this study were performed in accordance with the guidelines and regulations from Yale Environmental Health and Safety.

Results
Assessment of pH neutralization in original saliva samples. The normal range human saliva's pH is 6.2-7.4, with an average at 6.7. Additionally, approximately 10% of human saliva is acidic 41 . This naturally low pH in primary samples pose a significant problem in the development of a pH-dependent assay, such as the Sali-VISION test. To solve this problem, we tested the ability of sodium hydroxide solution (NaOH) in neutralizing the acidity of human saliva. Saliva samples with low pH were diluted with different concentrations of NaOH solution. While the low pH saliva samples diluted in either H 2 O or 750 µM NaOH caused a color change from pink to orange immediately after being added to the WarmStart reaction mix, the color only showed a subtle change when these samples were diluted in 1 mM or 1.5 mM NaOH, indicating that higher concentration of NaOH could neutralize the acidity of saliva and thus maintain the original pH of the WarmStart master mix prior the LAMP amplification (Fig. 1A). Strong base NaOH solution can neutralize the low pH in acidic saliva and enhance the specificity of the test. On the other hand, it may also result in increased alkalinity in saliva with neutral pH, and thus, reduce test sensitivity to these samples. To test this possibility, we performed a SaliVISION assay with spiked SARS-CoV-2 RNA in acidic and neutral saliva diluted with either 1 mM or 1.5 mM NaOH solution. At the NaOH concentration of 1 mM, all acidic and neutral saliva samples showed a strong positive result after 30 min of incubation at 65 °C. The sensitivity of this assay, otherwise, was significantly diminished in neutral saliva samples diluted with 1.5 mM NaOH, when compared to the results of low pH saliva (Fig. 1B). Our data showed that the Dilution Buffer with 1 mM concentration of NaOH is effective to neutralize the natural acidity of saliva sample and stabilize the original pH of the reaction after sample input, without interfering with the sensitivity of the assay. In this test, the assay is designed to tolerate saliva samples with pH ≥ 4.9 while still ensuring stability and reliability for on-site performance.

Analytical performance of SaliVISION screening and diagnostic test. The newly developed
saliva-based RT-LAMP test's limit of detection (LoD) was initially assessed with synthetic SARS-CoV-2 RNA (Twist Bioscience, San Francisco, CA) at different concentrations at 10 copies/µl, 5 copies/µl, 2.5 copies/µl, 1.25 copies/µl, and 0.625 copies/µl diluted in dilution buffer. With 10 µl of sample input per 40 µl of total reaction mix, 2.5 copies/µl, or 25 copies total, was determined as the lowest detectable concentration of genomic SAR-CoV-2 RNA, at which 100% of replicates were detected in 10 replicates (Fig. 2 A and B). An additional 10 samples containing only dilution buffer, served as negative controls and to rule out contamination (data not shown). www.nature.com/scientificreports/ The LoD of SaliVISION was subsequently verified on 20 negative saliva samples contrived with the same concentrations of synthetic SARS-CoV-2 RNA (Twist Bioscience, San Francisco, CA). These results were in 100% concordance with the results of the aforementioned analytical sensitivity assay, with 2.5 copies/μl being the lowest viral RNA concentration at which 100% of replicates were detected, therefore confirming the LoD of the SaliVISION assay ( Fig. 2

C and D).
Clinical sensitivity and specificity of SaliVISION compared to SalivaDirect testing platform. While analytic performance provided a quantitative information on the minimal copy number of the        (Fig. 5). This inconsistency in detection may account for the difference in testing materials. In fact, several studies have shown that SARS-CoV-2 can be detected in the saliva of asymptomatic persons, while it was not detected in the corresponding NP swab samples 42,43 . Of interest, a recent finding confirmed SARS-CoV-2 infection in the salivary glands and mucosae and thus, the oral cavity is an important site for SARS-CoV-2 infection 19 .

Cross-reactivity analysis of SaliVISION assay.
Due to the employment of large primer-sets in multiplex PCR assays, the success of a given protocol is largely contingent upon primer design. One such issue is cross-reactivity with other pathogens, given the potential overlap for any one of more than several pairs of primers. Therefore, to ensure our primer design provided adequate specificity for SARS-CoV-2, we performed in silico cross-reactivity analyses of primer sequences. This was done by using BLASTN Somewhat Similar Alignment method to align the 12 primers against the 24 respiratory disease pathogens genomes ( Table 2). The aver- www.nature.com/scientificreports/ age nucleotide identity rate between 314 bps of primer sequences and SARS-CoV-2 is 100%. SARS-CoV has an average nucleotide identity rate of 74.7% with the primer sequences, while in each primer set, at least one primer has an identity rate lower than 25% with SARS-CoV. All other pathogens have no significant overall homologous sequences with the primers, except MERS-CoV, which has a 66% identity rate with one single primer, and Chlamydia pneumonia CWL-029, which has an 84% identity rate with another single primer. Since each primer set needs 6 primers to coordinate for a successful amplification, it is unlikely for these partial homologous primers to cause false positives. Following in silico analysis of the SaliVISION multiplex primer set, we conducted wet testing of our primer set to definitively rule out the possibility of cross-reactivity with our assay. The specificity of SaliVISION multiplex primer set against common respiratory and viral pathogens was tested with purified, intact viral particles, cultured RNA, or bacterial cells commercially available from ZeptoMetrix Corporation (Table 3). 50 μl of each stock was spiked into a pooled negative saliva samples and tested with the SaliVISION assay. No positive signal was detected in any samples, indicating that this assay is specific for detecting SARS-CoV-2, with no cross-reactivity to common respiratory or other viral pathogens, which would generate potential false-positive results (Table 4).
In Silico analysis for the inclusivity of SaliVISION multiplex primer set. Given the rapid evolution of pervasive, and in some cases, increasingly infectious SARS-CoV-2 variants, the robustness of Covid-19 testing methods is of paramount importance. To account for the potential presence of different SARS-CoV-2 strains, we performed in silico analysis for the inclusivity of our multiplex primer sets. In our assay design, there are two primer sets targeting 2 non-overlapping N gene regions of the SARS-CoV-2 sequence, set N1 and set N2, with each primer set comprised of 6 primers. A total of 200 complete high coverage SARS-CoV-2 sequences spanning clades G, GH, GR, GV, L, O, S, V from GISAID were randomly collected, with specimen collection time from  www.nature.com/scientificreports/ December 2019 to November 2020. BLASTN Somewhat Similar Alignment method was used to align the primers against the SARS-CoV-2 sequences. BLASTN results showed that the N1 primer set has a primer carrying 1 mismatch for 3% of the strains and another primer carrying 1 mismatch for 2% of the strains. In comparison, the N2 primer set has a primer carrying 1 mismatch for 2% of the strains and another primer carrying 1 mismatch in 1% of the strains. None of the mismatches are at the 3' end of the primers, and no multiple mismatches were found simultaneously in the same primer set. The multiple mismatches causing false positivity are evaluated with a low possibility, therein revealing the SaliVISION primer sets to be of robust points of detection for the presence of SARS-CoV-2 viral RNA (Table 5).

Discussion
While public health efforts to curtail the spread of COVID-19 have continuously yielded promising results, continued diagnostic testing constitutes a foremost requirement in the prevention, and early detection of viral infection. Here, we report a new, non-invasive RT-LAMP-based assay for the rapid detection of SARS-CoV-2 in saliva, encompassing novel components for enhanced test accuracy. Following comparative analyses of the Sali-VISION assay, in conjunction with both SalivaDirect and Thermo Fisher Scientific TaqPath FDA EUA-approved diagnostic testing platforms, the SaliVISION assay offers 94.29% and 98.28% accuracy, respectively. Moreover, we report 100% specificity for SARS-CoV-2, with no presentable cross-reactivity with other pathogens. Additionally, careful primer design for the targeted N-gene of the SARS-CoV-2 genome, with the inclusion of novel spacer-loop elements, have yielded greater amplification fidelity, and have contributed to the robustness of our test, in spite of a myriad of pervasive SARS variants. Furthermore, the versatility and scalability for which this assay presents allows for increased point-of-care and high-throughput testing capabilities, with a robustness that allows for consistent detection of SARS-CoV-2 44 .
Thus far, molecular diagnostics that have been designed for the diagnosis of COVID-19 have presented an array of challenges which have been further compounded by a myriad of different factors. The newly developed SaliVISION test is not an exception. The primary challenge in this pH-based assay is the original pH of the saliva samples that can significantly impact to the stability of the reporter system and thus affects the specificity of the  www.nature.com/scientificreports/ result. The normal range human saliva's pH is 6.2-7.4, with an average at 6.7. Additionally, approximately 10% of human saliva is acidic 41 . The pH in these samples could reduce the pH of the WarmStart master mix and cause a color change in the final reaction, despite the absence or presence of the SARS-CoV-2 DNA, leading to a false positive result. To solve this problem, we diluted the inactivated saliva samples with a defined concentration of NaOH to neutralize the (1) natural acidity of saliva sample, and (2) stabilize the original pH of the reaction after sample input, without interfering with the sensitivity of the assay. In addition, a precise dilution factor was optimized to provide an adequate sample input to the final reaction, preventing inhibitory effects due to overwhelming concentrations, thus enhancing both efficiency and efficacy of the assay. In this test, the assay is designed to tolerate saliva sample with pH ≥ 4.9 while still ensuring adequate stability and reliability for on-site performance. In the interest of preventing cross-contamination, we have incorporated Uracil-DNA Glycosylase (UDG) into our assay as a means of preventing amplicon contamination from antecedent reactions 45,46 . Moreover, each reaction strip is completely closed prior to amplification and remains so as visualization of results occurs simultaneously with amplification. Lastly, because the success of RT-LAMP reactions is largely contingent upon successful primer design, several amendments to our primer sets, over the course of test development, were made necessary to maximize test specificity. As it stands, the SaliVISION assay provides 100% specificity for SARS-CoV-2, utilizing a two-primer set for distal locations on the N-gene. The specificity for which this test presents is also, in part, attributed to a novel spacer-loop element which is purposed for increased amplification fidelity for viral SARS-CoV-2 RNA. Given the projected prevalence of asymptomatic persons carrying COVID-19, designated as asymptomatic spreaders, as well as the increase of new variants with more contagious characteristics, high testing sensitivity in molecular diagnostics for COVID-19 is of paramount importance. Moreover, as some studies have suggested high false-negative rates among RT-PCR diagnostics for COVID-19 [47][48][49][50] , the importance of high-sensitivity testing is further underscored as the continuation of false-negative diagnoses perpetuate the spread of COVID-19. RT-LAMP has since served as a popular alternative to traditional laboratory diagnostics due to its relative ease-of-use and overall versatility. However, current RT-LAMP tests typically provide a 75%-91% clinical sensitivity 28,51,52 and offer a limit of detection-in congruence with the reported sensitivity-corresponding to a range of Ct values from ~ 30-33.5 [53][54][55] . We report that the SaliVISION assay has an overall clinical sensitivity of 99.46% for saliva specimens with a Ct value of ≤ 36, as confirmed on both SalivaDirect and TaqPath testing platforms. Moreover, when specifically compared to the TaqPath NP swab testing platform using 98 paired patient samples, collected on site, the SaliVISION assay displayed an overall specificity and sensitivity of 97.62% and 85.71%, respectively; however, among samples with a Ct value of ≤ 36, assay sensitivity was 100%. Although this is not ideal-as the standard convention for ruling out the possibility of COVID-19 infection is confirmed when a Ct value of viral RNA is > 40,-the relative increase in sensitivity for which the SaliVISION assay provides, coupled with its ease-of-use, make it a reliable alternative to traditional laboratory diagnostics, which may be less cost and time efficient.While saliva has been established as a less reliable source of bodily excretion for COVID-19 molecular diagnostics, as opposed to nasopharyngeal excretions, it has become a promising alternative due to less invasive sampling procedures. Further, it has been reported that while saliva typically carries a lighter viral load, the SARS-CoV-2 virus is capable of infecting, and replicating, within cells lining the oral mucosa 19,20 . In addition to infected cells of the lining of oral mucosa, it has also been observed that even in asymptomatic individuals, an acellular fraction of SARS-CoV-2 from infected glands is capable of making de novo virus 19 , therein contributing to infectiousness as well as providing a source of live virus for saliva-based COVID-19 testing. What's more, the viral longevity in saliva has been reported to be, on average, 18-20 days 10,56,57 , which allows for saliva-based testing to be a valid diagnostic modality for the entire duration of an infected individual's contagious window. Therefore, saliva has become an appealing source for molecular diagnostic testing; coupled with the associated non-invasive sampling procedures, as well as array of saliva-based rapid testing that currently exists, the use of saliva in COVID-19 testing is rapidly becoming an attractive alternative to other forms of testing.
In the interest of expanding the scope of COVID-19 rapid saliva-based testing, it is imperative to establish a feasible methodology that encompasses stringent safety measures to prevent further spread of the virus. Currently, the body of rapid molecular diagnostics for COVID-19 is primarily comprised of antibody testing; however, such testing is markedly inadequate as the production COVID-19 specific antibodies may take up to several weeks, and thus will not inform an individual of active infection. Other rapid testing modalities include nucleic acid-based testing and antigen testing; the latter has yet to prove reliable in individuals present with low viral loads, while the former, albeit reliable, may still take up to a couple days for results, and is considerably more expensive. Given the foregoing, testing efforts, on all scales, would greatly benefit from an expanded repertoire of readily available rapid testing. Our SaliVISION assay provides enhanced specificity and sensitivity when compared to other modes of rapid diagnostic testing, while proving more time efficient, providing test results within 45 min. In congruence with these benefits, RT-LAMP provides a cheaper alternative testing method due to its minimalistic approach with regards to equipment and reagents. Comparatively, both SalivaDirect and TaqPath testing platforms require a minimum of several hours to process samples, and often take up to 24-48 h for the results to made available. Additionally, the necessity for specialized equipment and instrumentation (e.g. biological safety cabinet, qPCR machine, qPCR consumables/reagents, etc.) can easily triple sample processing costs when compared to the SaliVISION assay. Moreover, we utilize an intuitive, convenient self-sample collection process that allows for increased throughput, while mitigating health-exposure risks (Fig. 6). Further, self-collection accommodation with pre-loaded lysis buffer inside the collection tube allows for specimens to be lysed and inactivated in a closed tube after sample collection, thus circumventing the need for expensive biological safety cabinets and prolonged safety procedures. Therefore, expansion of the RT-LAMP testing platform for the rapid diagnosis of COVID-19 is a promising avenue by which large-scale testing efforts can achieved, more efficiently.
As ongoing efforts to curb the spread of the COVID-19 global pandemic-such as mass vaccination-continue, the demand for readily accessible diagnostic testing remains to be of unequivocal importance, until the efficacy (3) A fixed volume of inactivated saliva is processed through a serial dilution step in a microtube test strip preloaded with Dilution Buffer (#0), negative control (#1), SARS-CoV-2 test with multiplex primers targeting the viral N gene (#2), and internal control with human RNP gene (#3). The RT-PCR reaction will be processed at 65 °C for 30 min; (4) The result is read and interpreted with the cooled reaction mixes after a 30-min incubation with a REPORT CARD.