key: cord-0721442-ditn3y9u
authors: Davidson, Josiah Levi; Wang, Jiangshan; Maruthamuthu, Murali Kannan; Dextre, Andres; Pascual-Garrigos, Ana; Mohan, Suraj; Sravan Putikam, Sai Venkata; Ibrahim Osman, Fujr Osman; McChesney, Darby; Seville, Jordan; Verma, Mohit S.
title: A paper-based colorimetric molecular test for SARS-CoV-2 in saliva
date: 2021-08-14
journal: Biosens Bioelectron X
DOI: 10.1016/j.biosx.2021.100076
sha: 087055c2c65da98f1e2110fcb67ab4b58b78e1bb
doc_id: 721442
cord_uid: ditn3y9u

Herein, we describe the development of a paper-based device to detect nucleic acids of pathogens of interest in complex samples using loop-mediated isothermal amplification (LAMP) by producing a colorimetric response visible to the human eye. To demonstrate the utility of this device in emerging public health emergencies, we developed and optimized our device to detect SARS-CoV-2 in human saliva without preprocessing. The resulting device was capable of detecting the virus within 60 min and had an analytical sensitivity of 97% and a specificity of 100% with a limit of detection of 200 genomic copies/μL of patient saliva using image analysis. The device consists of a configurable number of reaction zones constructed of Grade 222 chromatography paper separated by 20 mil polystyrene spacers attached to a Melinex® backing via an ARclean® double-sided adhesive. The resulting device is easily configurable to detect multiple targets and has the potential to detect a variety of pathogens simply by changing the LAMP primer sets.

In 2020, the coronavirus disease 2019 (COVID-19) pandemic infected over 79 million individuals and claimed over 1.7 million lives worldwide ("Weekly epidemiological update -29

December 2020," 2020). Additionally, it is estimated that by 2030, the COVID-19 pandemic will have caused over $16 trillion in economic damages to the United States alone (Cutler and Summers, 2020) . Most molecular tests for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, the virus responsible for are limited to the laboratory and thus have significant lag times (>24 hours) to provide a result, preventing their widespread adoption in point-of-care settings (Yüce et al., 2021) . Despite several attempts at developing a point-of-care test for SARS-CoV-2 (Tromberg et al., 2020) , some key limitations remain: i) scalability (the demand for testing is in the order of millions per week, but manufacturing new tests at that scale is challenging), ii) the need for sample processing (many tests still require an extraction step when using saliva), and iii) readability (molecular tests often require the use of fluorescence and thus, a fluorescence reader to report the results) (Parupudi et al., 2021) .

We overcome the limitations of current testing methods by developing a point-of-care test using paper-based devices and reverse-transcription loop-mediated isothermal amplification (RT-LAMP) that reports a color change in the presence of SARS-CoV-2 within 60 minutes using diluted saliva (5 % v/v in water) as the sample. RT-LAMP is a nucleic acid amplification technique conducted at a constant temperature with diagnostic performance comparable to the current gold standard, reverse-transcriptase quantitative polymerase chain reaction (RT-qPCR), especially during the acute phase of infection (Inaba et al., 2021) . Since RT-LAMP can be conducted at a constant temperature, there is no need for expensive thermal cycling equipment.

Additionally, existing colorimetric reporters for LAMP products eliminate the need for fluorescence readers. Consequently, this test is suitable for use in point-of-care settings and is amenable to rapid development and scale-up, making it appropriate for use in public health emergencies.

RT-LAMP has been implemented on microfluidic paper-based analytical devices (µPADs) to detect Zika virus (Kaarj et al., 2018) , where image analysis was performed using a smartphone to distinguish between positive and negative responses. Here, we provide a high-contrast RT-J o u r n a l P r e -p r o o f LAMP reaction on paper that provides a color change that is visible to the naked eye. In addition, instead of using wax-printing-which would require precise alignment of printed areas and dispensing of reagents-we used polystyrene spacers for preventing cross-talk between samples.

The polystyrene spacers are amenable to roll-to-roll fabrication for scale up of production. There are no other reports yet demonstrating on-paper detection of SARS-CoV-2 using RT-LAMP with colorimetric reporters and saliva as the sample matrix. Table 1 presents a selection of currently available nucleic-acid-based COVID-19 diagnosis methods. The novelty of our assay is the on-paper colorimetric detection of SARS-CoV-2 with minimal pre-processing. Our device has a sensitivity and specificity that is comparable to other RT-LAMP based assays; however, our assay is the only one that can detect SARS-CoV-2 on paper without pre-amplification. Other assays are conducted in solution which is not as scalable during manufacturing as paper-based assays. Additionally, our assay only requires a dilution step (requiring seconds to complete), whereas other assays require treatment with protease, heatinactivation, and/or RNA extraction to detect SARS-CoV-2 (steps requiring at least 10 minutes to complete and additional equipment). We demonstrate the direct detection of SARS-CoV-2 in saliva via a distinct colorimetric response that can be read using the naked eye ( Figure 1 ). This format is amenable to roll-to-roll fabrication and is anticipated to cost ~$10/test (Table S1 ). The limit of detection (LoD) of our test is 200 copies/µL saliva. The analytical sensitivity (positive predictive value) is 76%, the specificity (negative predictive value) is 100%, and the accuracy is approximately 91% as determined using contrived samples (freshly collected saliva spiked with heat-inactivated SARS-CoV-2) when evaluated visually ( Figure 1F ). Due to subjectivity in color perception of the control pad, respondents incorrectly identified a total of 20 out of the 80 devices presented as invalid, resulting in a false invalid rate of 25%. The sensitivity increases to 97% with an accuracy of 98% when the color change is quantified using image processing ( Figure 1C ).

J o u r n a l P r e -p r o o f Figure 1 : Schematics and colorimetric characterization of the paper-based device. A) Schematic illustration of the workflow for device use. The control zone indicates the no-primer control. B) Colorimetric LoD on paper using heat-inactivated severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) at the indicated concentration in 5% saliva. The negative replicates are RT-LAMP reactions using nuclease-free water in lieu of heat-inactivated SARS-CoV-2. Data was taken from Figure (Table S2) were designed using Primer Explorer v5

(http://primerexplorer.jp/lampv5e/index.html) with parameters found in Table S3 . Primer sets were designed using portions of the SARS-CoV-2 genome (NCBI accession number: NC_045512). Primer sets for RegX were designed by choosing three random 2,000 nt regions.

In-silico analyses were used to the predict sensitivity and specificity of each primer set. Optimal primer sets underwent experimental cross-reactivity studies to ensure specificity to SARS-CoV-2.

Three commercial saliva collection devices were selected to evaluate their effect on the RT-LAMP reaction in saliva. The three devices were "Saliva Sampler™" (StatSure Diagnostic Systems, Inc.), "Pure·SAL TM " (Oasis Diagnostics ® ), and "Super·SAL™" (Oasis Diagnostics ® ).

Cost of goods table with source, price, and product details Heat-inactivated SARS-CoV-2 was obtained from ATCC (ATCC VR-1986HK) or BEI (NR-52286). Pooled human saliva (991-05-P-50) and saliva from individual donors (991-05-S) was purchased from Lee Biosolutions. All oligonucleotides (desalted) were purchased from Life

Technologies. Materials used in the colorimetric RT-LAMP reaction and their prices can be found in Table S1 . 

The sensitivity and specificity were determined using 30 contrived positive samples at varying multiples of the LoD (1x, 2x, 4x, 40x, and 400x with 10, 10, 4, 3, and 3 replicates, respectively) and 30 non-template control (NTC) negative saliva samples. The colorimetric response intensity was determined using ImageJ to extract the average green channel intensity of each reaction zone. A receiver-operating characteristic (ROC) curve was generated by varying the threshold cutoff between positive and negative reactions and calculating the sensitivity and specificity at each threshold value. The sensitivity was calculated as a ratio of the number of true positives to total positives, including false positives. The specificity was calculated as a ratio of true negatives to total negatives, including false negatives.

A color bar was created by averaging RGB values of phenol red on Grade 222 chromatography paper over a range of pH values. A linear gradient was created using these average RGB values and the optimal threshold value determined from the ROC curve was annotated on the color bar ( Figure 1C ).

Fresh saliva was collected from participants who enrolled in the study in accordance with Purdue University IRB Protocol # IRB-2020-527. Participants who reported a COVID-19 diagnosis within the past 60 days were not permitted to donate. Saliva was collected from participants after receiving informed consent. Samples were assigned an ID based on the donor, date of sample collection, and method used for collection. No identifying information on participants was recorded.

Colorimetric perception surveys (Additional File 1) were collected from participants enrolled in the study in accordance with Purdue University IRB Protocol # IRB-2021-375. Participants were J o u r n a l P r e -p r o o f given a color bar with a threshold annotated ( Figure 1E ). Participants were given multiple paper device scans and were asked to classify the control pad (left reaction zone) as valid or invalid and the SARS-CoV-2 reaction (right reaction zone) as positive or negative. Observations classified as invalid were discarded from assay performance analysis and the proportion of incorrectly identified invalid assays was reported as the false invalid rate.

The workflow of our assay is illustrated in Figure 1A : collect saliva, transfer sample to paperbased device, incubate at 65 °C, and read a result; a typical result is shown in Figure 1D .

We designed at least 3 primer sets for each of the following SARS-CoV-2 genes: N gene, RdRp gene, and orf1ab (Table S2) . We screened primer sets using a fluorescent RT-LAMP kit with: i)

in-vitro transcribed SARS-CoV-2 RNA for the gene in water or ii) heat-inactivated SARS-CoV-2 in 18% saliva (final reaction concentration). After screening the primer sets ( Figure S1 - Figure S3 ), we decided that orf1ab.II was our optimal primer set due to it possessing the best LoD of 200 copies/µL of reaction (reaction volume 25 µL). In case downstream testing using orf1ab.II failed, we designed primers for random portions of the SARS-CoV-2 genome, labeled RegX. These primers were screened using a fluorometric RT-LAMP kit in 18% saliva (final reaction concentration), and RegX.III was chosen due to its fast reaction time and lack of false-positives (Table 2 and Figure S4 ).

In-silico inclusivity studies and sequence identity studies indicated over 99% conservation of orf1ab.II and orf7ab.I in available SARS-CoV-2 sequences and did not indicate cross-reactivity with other microorganisms found in human saliva or the human respiratory tract (Table S3 and   Table S4 ). We experimentally verified that the orf1ab.II and orf7ab.I primer sets were not crossreactive using genomic extracts of the analyzed microorganisms (Table S7 and Table S8 ).

We evaluated several metal-ion-chelating, DNA intercalating, and pH-sensitive colorimetric reporters for their ability to produce a robust and discernable colorimetric response during the RT-LAMP reaction using primer sets our lab had previously developed to target Histophilus somni (Mohan et al., 2021) . Some of the investigated metal-ion-chelating and DNA intercalating indicators were able to produce a colorimetric response in solution but were unable to produce a colorimetric response on paper ( Figure S5 - Figure S11 ).

We investigated several pH-sensitive colorimetric indicators for use in our assay (Figure S12 - Figure S16 ). Of the indicators screened, phenol red had the most apparent and contrasting colorimetric response ( Figure S16 ).

Optimization of colorimetric assay J o u r n a l P r e -p r o o f

We deemed an initial pH of 8.5 in solution to be optimal with respect to reported color consistency across replicates and contrast of color between positive and negative reactions ( Figure S16 ). On paper, an initial pH of 8.0 is optimal ( Figure S17 ).

We found that diluting saliva to 25% before adding to the RT-LAMP reaction (which further dilutes to a final concentration of 5% saliva) was sufficient to reduce the buffering capacity of saliva as well as the concentration of interferents (e.g., RNases) and produce a discernable colorimetric result in saliva within 60 minutes (data not shown).

Inclusion of both carrier DNA and guanidine hydrochloride as seen in the literature (Hardinge and Murray, 2019; Kiddle et al., 2012; Zhang et al., 2020) provided a colorimetric response and

LoD that was comparable in both water and 5% saliva ( Figure S18 and Figure S19 ). These components could not be added to paper, however, because they caused a significant color change upon drying the reagents (data not shown). The addition of RNase inhibitors appeared to worsen the LoD ( Figure S20 ).

We evaluated several commercially available saliva processing devices to eliminate particles in collected saliva ( Figure S21 ). We selected a sponge-based collection device (Oasis Diagnostics ® Pure·SAL TM ) due to its ability to standardize starting color and to provide the most consistent and best LoD of the observed devices.

Design of paper-based devices Inclusion of ammonium sulfate caused a color from red to yellow upon drying of RT-LAMP reagents when no template was present. We prevented this color change by increasing the phenol red concentration and replacing ammonium sulfate with betaine ( Figure S24 ). Furthermore, the addition of trehalose and bovine serum albumin (BSA) increased the reaction speed and improved LoD ( Figure S25 ).

After loading the sample onto the paper-based device, we placed the device into a 1" x 1" resealable plastic bag to prevent contamination during the RT-LAMP reaction. The plastic bag containing the paper device was then placed into an incubator at 65 °C for 60 minutes. The bag was removed, scanned with a flatbed scanner ( Figure 1D) , and compared against a color chart ( Figure 1E ) created by averaging RGB values from the phenol red response on grade 222 pads in buffers at known pH values from 6 -9 ( Figure S26 ). The threshold value in the color chart corresponds to the threshold determined from ROC analysis ( Figure 2B ).

After determining an LoD of 200 copies/µL in saliva ( Figure 1B) , we created contrived samples of 1x, 2x, 4x, 40x, and 400x LoD. We used 30 aliquots of freshly collected saliva as negative samples ( Figure S27 ). We quantified the results using image processing (Figure 2A and Figure   S28 ). We found the difference between the green channel intensity of the negative and positive colorimetric reaction pads to be significant (p < 0.001) using a two-tailed student's t-test. The J o u r n a l P r e -p r o o f specificity using image analysis was 100 %, the sensitivity was 97%, and the accuracy was 98%

( Figure 2C ).

We asked four participants to classify 10 positive and 10 negative reactions presented in Figure   S27 as valid or invalid (according to the left control zone) and positive or negative for SARS-CoV-2 (according to the right reaction zone) using a color bar ( Figure 1C ).

Observations deemed invalid by the participant were discarded. Of the 40 true positive observations (all valid using image analysis), participants incorrectly classified 19 as invalid. 

This contrasts with the 40 true negative observations (36 of which were valid using image analysis) where only 1 observation was incorrectly classified as invalid. Thus, the calculated specificity and sensitivity of our device while accounting for colorimetric interpretation were 100% and 76%, respectively, with an accuracy of 91% ( Figure 1F ) and a false invalid rate of 25%.

Overview of device design and procedure

We designed each step of the assay ( Figure 1A ) to reduce complexity and mitigate user errors in point-of-care settings. The sample collection step uses a sponge-based collection device which allows for self-collection by the patient, removes particulates from the collected saliva, and minimizes variance in results from patient to patient. The transfer step requires placing 25 µL of diluted saliva onto each of the two reaction zones on the device. For the incubation step, the paper device with the sample loaded is sealed in a resealable plastic bag and placed in an incubator set at 65 °C for 60 minutes. Finally, for the read step, the user compares the color of the control and reaction zones to the color bar in Figure 1C to determine if the results are valid and if the pathogen of interest is present.

Our platform comprises three main components: primer sets imposing specificity to the assay, a paper device containing two reaction zones (one control and one reaction), and a heating source used to heat the paper device to reaction temperatures. The primer sets determine what pathogen the assay targets. Therefore, our platform can be reconfigured to target a different pathogen by redesigning the primer sets while keeping all other aspects of our device and formulation the J o u r n a l P r e -p r o o f same. Additionally, the paper-based device can be configured to accommodate numerous reaction zones allowing for multiple targets to be detected simultaneously.

We designed primer sets targeting the N gene (Corman et al., 2020) , RNA-dependent RNA Polymerase (nsp12) gene , and portions of the orf1ab genes (Huang et al., 2020) .

Our assay utilized a no-primer control to ensure that the reaction zones do not change color when heating. Other assays utilize a Human RNaseP mRNA control which increases diagnostic power Yaren et al., 2021) . We screened primer sets in water using in-vitro transcribed RNA of the gene to assess performance and ability to dimerize. We further screened primer sets in saliva (18% reaction concentration) to assess performance in complex samples and off-target interactions. Since only one primer setwith a relatively poor LoDpassed both stages of screening, primer sets targeting random regions of the SARS-CoV-2 genome were designed and screened in 18% saliva. We selected primer set RegX.III (renamed to orf7ab.I) for further assay development alongside orf1ab.II. For further discussion on primer screening, see Supplemental Information.

Although fluorescent reporters work well for quantitative RT-LAMP in a real-time thermocycler, they require an additional ultraviolet (UV) light source to be read by the naked eye which adds technical complexity to our assay. Thus, we developed a colorimetric RT-LAMP assay using indicators previously reported in the literature for both LAMP and PCR (Tanner et al., 2015) . We evaluated three classes of indicators: i) metal ion indicators (Rodriguez-Manzano et al., 2016) , ii)

DNA intercalating indicators (Roy et al., 2017) , and iii) pH indicators (Tanner et al., 2015) .

Whereas some of the screened metal-ion and DNA-intercalating indicators produced a colorimetric response in solution, none were able to produce a colorimetric response on paper ( Figure S5 - Figure S11 ).

Therefore, we screened pH dyes used as indicators in literature ( Figure S12 -Figure S16 ). Of the pH indicators we examined, cresol red and phenol red ( Figure S16 ) produced the most vibrant colorimetric contrast between positive and negative LAMP reactions. We used phenol red as our indicator since it was used frequently as a reporter in literature (Ahn et al., 2019; Tanner et al., 2015; Thi et al., 2020) .

The initial pH also has a drastic effect on the 60-minute end-point values. For initial pH values of 8.1, all negative reactions are false positives. For pH values of 8.3, some negative replicates at the 60-minute time point also appear to be changing color ( Figure S16 ). Consequently, an initial pH value of around 8.5 is necessary to achieve a consistent, distinguishable colorimetric response, despite the contrast between positive and negative reactions being lower at this pH.

Previous studies have found that the dependence of the final colorimetric response on initial pH is likely due to the diversity of compositions and buffering capacities of complex samples (Uribe-Alvarez et al., 2021) . We speculate a similar explanation is responsible for the observed dependence of the final colorimetric response on initial pH.

To standardize initial pH and to remove particulates from collected saliva, we evaluated three commercially available saliva collection devices to pre-process saliva. We chose Pure·SAL TM due to consistent initial starting pH, the best LoD, and the highest contrast between positive and negative reactions at the end of the reaction using our optimized LAMP formulation ( Figure   J o u r n a l P r e -p r o o f S21). We found reports of Pure·SAL TM used in lateral flow strips and other investigations (Dalirirad et al., 2020; Sa et al., 2020) , but not in isothermal amplification assays. Literature investigations of Pure·SAL TM suggest that it removes greater than 70% of mucinous material from saliva, which would act as an inhibitor to our amplification assay (Khurshid et al., 2017) .

When using our optimized LAMP formulation, processing saliva with Pure·SAL TM provides a four-fold better LoD when compared to unprocessed saliva accompanied by a more robust colorimetric response. ( Figure S23 ).

We determined that diluting the saliva to 25% with nuclease-free water and further diluting to a final concentration of 5% saliva upon addition to the RT-LAMP reaction was needed to obtain results within 60 minutes. Dilution reduces the buffering capacity of saliva and decreases the concentration of inhibitory components, both of which would delay colorimetric reporting. We found dilution preferable to more commonly used pre-treatment steps found in a variety of studies, such as pre-treatment with proteases (Janíková et al., 2021; Wei et al., 2021b) , Chelex® 100 (Yaren et al., 2021) or RNA extraction steps to inactivate inhibitory components of saliva (Garneret et al., 2021; Yamazaki et al., 2021) , as it is less complex on the end-user.

The LoD of the unoptimized colorimetric assay in 5% saliva that has been processed using Pure·SAL TM is 1000 copies/reaction (reaction volume 25 µL), which corresponds to 800 copies/µL of patient saliva after accounting for dilution ( Figure S21 ). This LoD is several times worse than other LoDs found in the literature for other RT-LAMP based reactions on whole saliva (generally ~50 copies/reaction, but some reported values are as poor as 500 copies/reaction) and is several orders of magnitude higher than RT-PCR assays or other assays utilizing RNA extraction (on the order of 1 copy/reaction) . However, J o u r n a l P r e -p r o o f almost all of these studies are accompanied by pretreatment protocols and/or RNA extraction steps to achieve the reported LoD (Chomczynski et al., 2021; Ganguli et al., 2020; Lalli et al., 2021; Reynés et al., 2021; Yang et al., 2021) . To improve this LoD, we investigated the use of RNase inhibitors (Lalli et al., 2021) , Guanidine HCl , and carrier DNA which had previously been reported to increase the sensitivity of RT-LAMP and RT-PCR (Hardinge and Murray, 2019; Kiddle et al., 2012) . The addition of RNase inhibitors seems to worsen the LoD in 5% saliva ( Figure S20 ). This result directly contradicts literature reports where RT-LAMP assays in saliva utilize RNase inhibitors improve the LoD (Janíková et al., 2021; Wei et al., 2021b) ; this discrepancy may be due to the type of RNase inhibitor used. Both Guanidine

HCl and carrier DNA improved the LoD ( Figure S18 and Figure S19 ) and was added to our RT-LAMP reaction formulation for colorimetric solution reactions (Lalli et al., 2021; Wei et al., 2021b) . We did not include these components in our paper RT-LAMP formulation, however, as they resulted in a color change when drying on paper (due to unexplained reasons). Finally, uracil-DNA glycosylase (UDG) and deoxyuridine triphosphate (dUTP; Figure S22 ) were included to reduce carryover contamination (Hsieh et al., 2014; Tang et al., 2016; Wei et al., 2021b; Yaren et al., 2017; Zeng et al., 2020) . When including Guanidine HCl, carrier DNA, and UDG, the LoD of our optimized RT-LAMP colorimetric assay in 5% processed saliva in solution improves to 250 copies/reaction ( Figure S23 ).

Following optimization of the colorimetric RT-LAMP in solution, we transferred our assay onto paper. Paper is widely used in pH indicators and urine strips primarily due to the inexpensive cost, low technical complexity, and ease of production using roll-to-roll manufacturing (Yetisen et al., 2013) . For our devices, we evaluated the use of several types of papers and selected J o u r n a l P r e -p r o o f chromatography paper ( Figure S7 - Figure S9 ). Chromatography paper is typically used in paperbased biosensors due to its improved wicking ability compared to other papers (Gutiérrez-Capitán et al., 2020; Singh et al., 2018; Verma et al., 2018) . Most paper-based devices in literature use Grade 1 chromatography paper (Busa et al., 2016) ; however, due to the large area (5 mm x 20 mm) of the Grade 1 chromatography paper needed, we noticed that the distribution of solution across the paper was uneven. Thus, we selected a different type of chromatography paper, Grade 222 (0.83 mm), which is approximately 4.6 times thicker than Grade 1 (0.18 mm).

Due to its increased thickness, the same volume of liquid can be loaded onto a 70%smaller reaction area by reducing the size of our device (5 mm x 6 mm), allowing for even distribution.

To reduce the complexity of our device, we dried the components of the RT-LAMP reaction (without the template) on the paper. Drying is one of the hallmark advantages of using paperbased devices as it allows for stable distribution of the device and easy operation without compromising diagnostic performance (Ratajczak and Stobiecka, 2020) ; the user simply adds the sample to rehydrate the reagents. Upon drying our reagents on paper, however, we noticed that the papers changed color from red to yellow over time without any template being present, indicating a decrease in the pH of the paper. Through a series of leave-one-out experiments, we were able to determine that ammonium sulfate was responsible for this change ( Figure S24 ). We speculate this color change could result from the oxidation of cellulose caused by heating and the oxidizing nature of ammonium sulfate or the acidification of reagents by degassing of ammonia from the RT-LAMP mixture. To prevent the color change in the absence of amplification, we replaced ammonium sulfate with betaine and increased the concentration of phenol red which acts as an antioxidant ( Figure S25 ). Literature reports vary concerning the effectiveness of betaine in LAMP reactions (Foo et al., 2020; Zhou et al., 2014) ; however, evidence exists J o u r n a l P r e -p r o o f indicating that betaine reduces oxidative damage, and we, therefore, included it in our formulation (Willingham et al., 2020) . Trehalose has been reported to act as a protein stabilizer by resisting dehydration when drying (García-Bernalt Diego et al., 2019; Hayashida et al., 2015) .

Furthermore, bovine serum albumin (BSA) has been reported to act as a protein stabilizer when freeze-drying (Shimizu et al., 2017) . As both trehalose and BSA have been shown to improve the performance of isothermal amplification methods (Mok et al., 2016) , we added them to our formulation on paper ( Figure S25 ).

For our device, we elected to use two reaction zones; one targeting SARS-CoV-2 and one providing a no-primer control to determine the stability of our reagents on paper (which should not react with any sample). A schematic of our final device is shown in Figure 1C and results from our device with and without heat-inactivated SARS-CoV-2 spiked into saliva at 5% reaction concentration are shown in Figure 1D . As can be seen in Figure 1B , the LoD of our assay on paper (250 copies/reaction) is comparable to the LoD observed in solution for our optimized colorimetric RT-LAMP formulation ( Figure S23 )

For the construction of our device, we used a Melinex® backing to provide structural support.

Using a double-sided adhesive, we were able to attach two reaction pads to the backing without altering their pH. The number of reaction zones can be arbitrarily increased to allow for multiplexed detection without altering the design of our device. We found similar designs in literature where a solid substrate provided structural support to the reaction sites and utilized an adhesive in some cases (Choi et al., 2016; Reboud et al., 2019; Verma et al., 2018) . Finally, we added 20 mil polystyrene spacers between reaction pads to provide a physical barrier inhibiting leakage from one reaction zone to an adjacent reaction zone, thus eliminating crosstalk during both reagent and sample addition. Typically for paper-based devices, the reaction zones are J o u r n a l P r e -p r o o f separated by hydrophobic barriers resulting from wax printing preventing sample from crossing (Altundemir et al., 2017; Carrilho et al., 2009; Martinez et al., 2007) ; however, to enable roll-toroll manufacturing, we wanted to eliminate the usage of wax and thus settled on the incorporation of spacers (Kaarj et al., 2018) .

To evaluate the analytical sensitivity and specificity of our assay on paper, we utilized contrived samples with heat-inactivated SARS-CoV-2 at multiples of the LoD for orf7ab.I to run our RT-LAMP assay on paper. We used a total of 30 positive samples and 30 corresponding negative samples which is the minimum number required for emergency use authorization (EUA) in the United States (Ravi et al., 2020) . To determine the optimal colorimetric threshold to differentiate positive from negative reactions, we constructed an ROC curve by calculating the sensitivity and specificity at varying green channel intensity threshold values (Figure 2 ). At the optimum threshold, our assay has the following analytical metrics: sensitivity of 97%, specificity of 100%, and accuracy of 98% (Figure 2A and Figure 2B ). We found the difference between the positive and negative groups to be significantly different (p < 0.001). The paper strips were cut by hand and small differences in the size of the paper can cause differences in the colorimetric response.

Therefore, we speculate large-scale fabrication and quality control will further improve the consistency within the two groups (positive and negative). This sensitivity is comparable to assays using RNA extracts (sensitivity ~95%), and better than that reported for crude samples where significant decreases in sensitivity (to ~80%) are common (Subsoontorn et al., 2020) .

Due to restrictions imposed by the Purdue University Institutional Biosafety Committee, we were unable to test samples obtained from COVID-19 positive and negative patients in a timely J o u r n a l P r e -p r o o f manner, and thus, we were unable to establish diagnostic sensitivity, specificity, and accuracy for our assay. We aim to perform this characterization in future studies.

To observe the effect of color perception on the performance of our device, we surveyed four participants and asked them to interpret the results of our device. We provided each participant with a color bar and scans of our device after 60 minutes (Additional File 1) and asked them to classify the result as valid or invalid (using the control pad) and positive or negative based on the threshold marked on the color bar. The sensitivity and accuracy of our device decreased to 76% and 91%, respectively when we introduced user interpretation to our analysis, which is lower than other assays using crude samples ( Figure 1F ). This low sensitivity stems from respondents identifying many positive reactions as invalid based on the control pad, leading to a false invalid rate of 25%, which may be attributed to contamination of the control pad with amplicons during the reaction. Additionally, user interpretation of the pads where regions of both yellow and red exist could introduce ambiguity, resulting in an increased false-positive rate or false invalid rate.

Recent findings suggest this ambiguity is due to a third, intermediate color cluster (along with positive/negative clusters), that is not adequately addressed in colorimetric assays. de Oliveira Coelho et al., 2021) . Since invalid results were discarded from further analysis, this elevated false invalid rate may artificially influence the specificity and accuracy metrics of our device.

We developed a platform capable of on-paper detection of SARS-CoV-2 from saliva using colorimetric reporters that produce responses visible to the naked eye. Our platform has the following eight advantages: i) it uses saliva, ii) it requires minimal operator training, iii) it can be J o u r n a l P r e -p r o o f fabricated using roll-to-roll methods to achieve millions of tests, iv) it performs similar to a RT-qPCR assay in terms of analytical sensitivity and specificity, v) it provides a colorimetric response visible to the naked eye, vi) it is amenable to point-of-care use, vii) it provides results in less than 60 minutes, and viii) it is estimated to cost ~$10/test.

Four limitations of the current approach are: i) LAMP-based assays are prone to false-positives due to the large amount of DNA product produced and carryover contamination in the form of aerosols and thus, the reaction needs to be sealed well externally to avoid cross-contamination, ii) an external heater is still needed for conducting the assay (although we envision that an integrated battery-operated heater could be developed for each test strip), iii) although redyellow color transitions are vivid, variations in perception of color could affect interpretation of results and their validity (this could be overcome using a cellphone or camera to capture images and quantify color), and iv) sample needs to be added directly to each reaction zone (development and incorporation of a spreading layer would enable the user to add sample to a single site and the spreading layer would distribute the sample evenly to all reaction zones).

Due to the simplicity and scalability of this test, we envision that it could be used in a wide variety of settings, potentially including in-home diagnostics. Our platform can be readily reconfigured to target different pathogens simply by screening primer sets in solution and multiplexing is enabled by adding additional reaction sites to the device. The reconfigurable nature of our platform makes it an ideal tool for deployment and detection of emerging pathogens in future public health emergencies.

J o u r n a l P r e -p r o o f

Rapid and simple colorimetric detection of multiple influenza viruses infecting humans using a reverse transcriptional loop-mediated isothermal amplification (RT-LAMP) diagnostic platform

A review on wax printed microfluidic paper-based devices for international health

Colorimetric RT-LAMP SARS-CoV-2 diagnostic sensitivity relies on color interpretation and viral load

Advances in Microfluidic Paper-Based Analytical Devices for Food and Water Analysis

Understanding Wax Printing: A Simple Micropatterning Process for Paper-Based Microfluidics

An integrated lateral flow assay for effective DNA amplification and detection at the point of care

Rapid processing of SARS-CoV-2 containing specimens for direct RT-PCR

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

The COVID-19 Pandemic and the $16 Trillion Virus

Aptamer-Based Lateral Flow Biosensor for Rapid Detection of Salivary Cortisol

Essential properties and pitfalls of colorimetric Reverse Transcription Loop-mediated Isothermal Amplification as a pointof-care test for SARS-CoV-2 diagnosis

Loopamplification (LAMP) reaction as viable PCR substitute for diagnostic applications: a comparative analysis study of LAMP, conventional PCR, nested PCR (nPCR) and real-time PCR (qPCR) based on Entamoeba histolytica DNA derived from faecal sample

Rapid isothermal amplification and portable detection system for SARS-CoV-2

Progress in loop-mediated isothermal amplification assay for detection of Schistosoma mansoni DNA: towards a ready-to-use test

Performing point-of-care molecular testing for SARS-CoV-2 with RNA extraction and isothermal amplification

Electrochemical Paper-Based Biosensor Devices for Rapid Detection of

Reduced False Positives and Improved Reporting of Loop-Mediated Isothermal Amplification using Quenched Fluorescent Primers

Direct Blood Dry LAMP: A Rapid, Stable, and Easy Diagnostic Tool for Human African Trypanosomiasis

Simultaneous elimination of carryover contamination and detection of DNA with uracil-DNA-glycosylasesupplemented loop-mediated isothermal amplification (UDG-LAMP)

RT-LAMP for rapid diagnosis of coronavirus SARS-CoV-2

Diagnostic accuracy of LAMP versus PCR over the course of SARS-CoV-2 infection

Loop-mediated isothermal amplification for the detection of SARS-CoV-2 in saliva

Detection of SARS-CoV-2 with SHERLOCK One-Pot Testing

Simpler, Faster, and Sensitive Zika Virus Assay Using Smartphone Detection of Loop-mediated Isothermal Amplification on Paper Microfluidic Chips

Human salivary protein extraction from RNAPro·SAL TM , Pure·SAL TM , and passive drooling method

GMO detection using a bioluminescent real time reporter (BART) of loop mediated isothermal amplification (LAMP) suitable for field use

Rapid and Extraction-Free Detection of SARS-CoV-2 from Saliva by Colorimetric Reverse-Transcription Loop-Mediated Isothermal Amplification

A Novel Reverse Transcription Loop-Mediated Isothermal Amplification Method for Rapid Detection of SARS-CoV-2

EasyCOV : LAMP based rapid detection of SARS-CoV-2 in saliva

Patterned Paper as a Platform for Inexpensive, Low-Volume, Portable Bioassays

Loop-Mediated Isothermal Amplification for the Detection of Pasteurella multocida, Mannheimia haemolytica, and Histophilus somni in Bovine Nasal Samples

Comprehensive evaluation of molecular enhancers of the isothermal exponential amplification reaction

Rapid molecular diagnostics of COVID-19 by RT-LAMP in a centrifugal polystyrene-toner based microdevice with end-point visual detection

Evidence-based point-of-care technology development during the COVID-19 pandemic

High-performance modified cellulose paper-based biosensors for medical diagnostics and early cancer screening: A concise review

Diagnostics for SARS-CoV-2 detection: A comprehensive review of the FDA-EUA COVID-19 testing landscape

Paper-based microfluidics for DNA diagnostics of malaria in low resource underserved rural communities

Rapid visual detection of SARS-CoV-2 by colorimetric loop-mediated isothermal amplification

Rheonix COVID-19 TM MDx Assay: Instructions for Use

Reading Out Single-Molecule Digital RNA and DNA Isothermal Amplification in Nanoliter Volumes with Unmodified Camera Phones

Colorimetric Nucleic Acid Detection on Paper Microchip Using Loop Mediated Isothermal Amplification and Crystal Violet Dye

Effect of Sodium Bicarbonate Mouth Wash on Salivary pH and Interleukin-1β Levels among Smokers

Characteristics of proteinaceous additives in stabilizing enzymes during freeze-thawing and -drying

Paper-Based Sensors: Emerging Themes and Applications

Structural Genomics of SARS-CoV-2 Indicates Evolutionary Conserved Functional Regions of Viral Proteins

The diagnostic accuracy of isothermal nucleic acid point-of-care tests for human coronaviruses: A systematic review and metaanalysis

Advanced uracil DNA glycosylase-supplemented real-time reverse transcription loop-mediated isothermal amplification (UDG-rRT-LAMP) method for universal and specific detection of Tembusu virus

Visual detection of isothermal nucleic acid amplification using pH-sensitive dyes

A colorimetric RT-LAMP assay and LAMP-sequencing for detecting SARS-CoV-2 RNA in clinical samples

Rapid Scaling Up of Covid-19 Diagnostic Testing in the United States -The NIH RADx Initiative

Low saliva pH can yield false positives results in simple RT-LAMP-based SARS-CoV-2 diagnostic tests

Sliding-strip microfluidic device enables ELISA on paper

Direct diagnostic testing of SARS-CoV-2 without the need for prior RNA extraction

Fielddeployable, rapid diagnostic testing of saliva for SARS-CoV-2

Betaine Supplementation May Improve Heat Tolerance: Potential Mechanisms in Humans

Development of a point-of-care test to detect SARS-CoV-2 from saliva which combines a simple RNA extraction method with colorimetric reverse transcription loop-mediated isothermal amplification detection

Saliva TwoStep for rapid detection of asymptomatic SARS-CoV-2 carriers

Point of sampling detection of Zika virus within a multiplexed kit capable of detecting dengue and chikungunya

Ultra-rapid detection of SARS-CoV-2 in public workspace environments

Paper-based microfluidic point-of-care diagnostic devices

COVID-19 diagnosis -A review of current methods

Uracil-DNA-glycosylase-assisted loop-mediated isothermal amplification for detection of bacteria from urine samples with reduced contamination

Enhancing colorimetric loop-mediated isothermal amplification speed and sensitivity with guanidine chloride

Establishment and application of a loop-mediated isothermal amplification (LAMP) system for detection of cry1Ac transgenic sugarcane

contributed the following: design and screening (using fluorometric methods) of all primers, all in-silico analysis, writing -original draft (lead), writing -formatting, review, and editing (lead), development and administration of saliva collection IRB protocol (lead), methodology (equal), and project coordination (support). J.W. contributed the following: development of paper-based devices (equal), optimization of formulation for RT-LAMP on paper (equal), characterization of paper-based device (lead), development and administration of colorimetric interpretation survey IRB protocol (lead), writing -original draft (support), and methodology (equal). M.K.M. contributed the following: screening (colorimetric methods) of primers (equal), development and optimization of formulation for RT-LAMP in solution (lead), screening and selection of saliva processing and devices

A.P.G. contributed the following: screening of colorimetric indicators (equal) and writing -original draft (support). S. M. contributed the following: screening (colorimetric methods) of primers J o u r n a l P r e -p r o o f (equal), screening of colorimetric indicators (equal), development of protocol for colorimetric data collection on BMG, troubleshooting the heating and sealing of RT-LAMP solutions reactions (equal), and writing -original draft (support). S.V.S.P contributed the characterization of heating devices. F.O.I.O. contributed by troubleshooting the heating and sealing of RT-LAMP solution reactions (equal) and writing -original draft (support). D.M. contributed by screening and selection of paper materials and pH indicators (equal)

contributed the following: conceptualization, writing -original draft (support), writingformatting review, and editing (support), development of saliva collection IRB protocol (support), development of colorimetric interpretation survey IRB protocol (support), project administration

Funding Sources

We are grateful for collaborations with Raytheon BBN Technologies (Bryan Bartley, Aaron Adler, Allison Taggart, Jake Beal, Miles Rogers), PortaScience (Mike Gavin, Hanna Richman), LaDuca LLC (Frank LaDuca), and Cortex Design. We are also grateful to Jacqueline Linnes from Purdue University for her help with troubleshooting. The following reagent was deposited by the Centers for Disease Control and Prevention and obtained through BEI Resources, NIAID, NIH: SARS-Related Coronavirus 2, Isolate USA-WA1/2020, Heat-inactivated, NR-52286. Other reagents obtained from BEI are located in Table S7 and Table S8 .

The authors declare the following competing financial interests: M.S.V. has interests in Krishi J o u r n a l P r e -p r o o f  Paper-based device utilizes RT-LAMP to detect SARS-CoV-2  Extraction-free detection of SARS-CoV-2 in whole saliva with minimal processing  Primer sets target the orf7a, orf7b, and orf1ab regions of SARS-CoV-2  Configurable reaction zones enable multiplexed detection of pathogens J o u r n a l P r e -p r o o f

☐ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☒The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:M.S.V. has interests in Krishi LLC, a company interested in licensing the technology developed here. The work performed here was not funded by Krishi LLC. Employees of Raytheon BBN (sponsor) were involved in the conceptualization and analysis of the data presented here. The contents of this manuscript have been filed for provisional patent protection by Purdue University and Raytheon BBN.