Fully Dried Two-Dimensional Paper Network for Enzymatically Enhanced Detection of Nucleic Acid Amplicons

Two-dimensional paper networks (2DPNs) have enabled the use of paper-based platforms to perform multistep immunoassays for detection of pathogenic diseases at the point-of-care. To date, however, detection has required the user to provide multiple signal enhancement solutions and been limited to protein targets. We solve these challenges by using mathematical equations to guide the device design of a novel 2DPN, which leverages multiple fluidic inputs to apply fully dried solutions of hydrogen peroxide, diaminobenzidine, and horseradish peroxidase signal enhancement reagents to enhance the limit-of-detection of numerous nucleic acid products. Upon rehydration in our unique 2DPN design, the dried signal enhancement solution reduces the limit-of-detection (LOD) of the device to 5 × 1011 nucleic acid copies/mL without increasing false positive detection. Our easy-to-use device retains activity after 28 days of dry storage and produces reliable signal enhancement 40 min after sample application. The fully integrated device demonstrated versatility in its ability to detect double-stranded and single-stranded DNA samples, as well as peptide nucleic acids.

To test the sensitivity of our 2DPN, we designed model DNA probes that would mimic sample behavior in our device. Samples need a FITC tag in order to bind to the test line, and they require a biotin tag in order to bind the SA-HRP-AuNP and produce a visual signal. ssDNA probes were purchased from IDT.
To create a dsDNA sample, we hybridized a probe tagged with FITC, indicated by the IDT sequence label FAM, to a complementary probe labeled with biotin. For an ssDNA sample, we designed a single probe tagged with both FITC and biotin.  TARA is a novel chemical transfer reaction that allows template-dependent molecular amplification directly from a biological sample with minimal sample preparation, and the detection of colorimetric signals with the eye on lateral flow paper strips 1 . TARA can rapidly amplify nucleic acid targets directly from blood, urine, saliva or nasopharyngeal swabs without nucleic acid purification without the use of PCR. (Fig.  S2). TARA offers significantly better sensitivity than simple hybridization-based detection because of its ability to repeatedly label one of the probe RNA strands (G probe, Fig. S2) templated by the same target RNA.
As shown in Fig. S2A, the reporter group (biotin, shown as a blue circle) is transferred from the A probe to the G probe through a thiol exchange reaction depicted in Fig. S2B 1 . This reaction is dependent on the hybridization of the probes to adjacent sequences in the target DNA or the RNA. Importantly, the thiols in the PNA probes will not conjugate with the AuNP in the assay because they are protected with glycans.
Biotin-TCCGCCAA-FITC A note on NA product size: Within reason, we do not expect products below 200 base pairs in length to produce extremely different signal intensities, as the signal produced in our test is dependent primarily on the strong and highly reliable interactions of streptavidin-biotin binding, FITC-anti-FITC binding, and DAB/HRP. Additionally, LAMP amplicons, which can reach lengths of up to 500 base pairs, (but typically remain <300 for detection reasons 2 ) have been extensively studied in other LFIAs. Additional data not included in this publication showed no significant effect (no difference in final signal) even when we used samples containing large products produced by LAMP (~200 base pairs) on our 2DPN.
We would, however, expect differences in signal intensity for very large samples (> 300 base pairs), which would be difficult to pass through the nitrocellulose membrane. We do not expect samples of this size to be produced via TARA or CRISPR. Therefore, we are not concerned with effects that might be observed with these samples larger than 300 base pairs.

One-Dimensional Flow Characterization Results
Flow parameters for conjugated AuNP and DAB in nitrocellulose channels of varying widths were tested in one-dimensional flow studies (Table S1). Further optimization studies were subsequently performed to decide the optimum reagent concentrations and the locations of the fluidic inputs (Table S3). V = 2 cos( ) ( + ℎ) 6 (Eq. S1) Clean delivery is defined as the sequential delivery of reagents with minimal parallel flow or mixing of discrete fluid samples 3 . The sequential delivery of fluids can be optimized by controlling the distance between the fluid sources. To determine the optimal distance between source pads, 5 mm x 100 mm FF80HP nitrocellulose strips were cut with Silhouette Studio scrapbook device. 50 µL of blocking solution (5% sucrose, 2% BSA, 0.25% PVP, and 0.05% Tween in PBS) was applied to strips and allowed to dry. 5mm x 8mm source pads were cut from Millipore glass fiber and placed at the start of the strip and varying distances from the start of the strip (10 mm, 15 mm, 20 mm). Cellulose wick was placed at the end of the nitrocellulose strip. 25 µL of SA-HRP-AuNP solution was deposited on upper source pad and 25 µL of PBS wash (with food coloring for visualization) was placed on lower source pad simultaneously.

S9
Images of 5E11 copies/mL ssDNA on 2DPN Replicate Results Figure S5. Replicates of ssDNA LOD tests at a concentration of 5E11 copies/mL. Detection region of 2DPN imaged at 10 minutes after sample binding (AuNP signal) and at 60 minutes after signal enhancement (DAB enhanced signal). Figure S6. Replicates of TARA sample (1:10 dilution in PBST) on dried 2DPN. Imaged at 10 minutes after sample binding (AuNP signal) and at 60 minutes after signal enhancement (DAB enhanced signal).

Images of TARA 2DPN Replicate Results
Signal enhancement solutions have been used in one-dimensional and two-dimensional LFIAs to improve device sensitivity. The improvement in the signal intensity produced by these enhancement solutions is typical quantified as a percent increase or as a fold difference. In order to compare the performance our device to other literature, we calculated the percent difference of the enhanced signal intensity relative to the AuNPs signal intensity, shown below. From this data the average percent difference for all sample concentrations above the LOD (5E11 copies/mL) was 116%. Four-parameter logistic (4PL) models are typically used to model signal intensity as a function of sample concentration in ligand-binding assays, such as ELISAs 4 . We selected this model to evaluate our LOD produced by the DAB Enhanced Signal ( Figure S3).
where y is signal intensity, x is log of sample concentration, a is the response at 0 concentration, d is the response at infinite concentration, c is the concentration that gives a signal intensity halfway between a and d, and b is a slope parameter typically near 1 5 . The 4PL model was fit to the LOD data in GraphPad Prism v8 with the slope parameter (b) constrained to 1. The signal output of the 2DPN showed test line intensity changes as a function of sample concentration. The fit of the ELISA (R 2 =0.9771) with our data, indicates that our paper-based dried reagent ELISA assay can generate results like that of a traditional ELISA ( Figure S3). This demonstrates the potential of this paper diagnostic to quantitatively determine sample loads, which is crucial for chronic disease monitoring applications. Figure S7. The DAB Enhances signal on our fully dried 2DPN produces a predictable signal increase in relation to sample concentration (ssDNA) increases.

Cost Estimation
Providing easy-to-use diagnostics at an affordable price is crucial for successful translation of the device into clinical use. In academic research, low-cost design specifications are typically assessed by a bill of materials. For example, Richards-Kortum's paper reporting on the development of a paper-based HIV-1 drug resistance detection device included a bill of materials that summed $12.69 6 . However, the academic prototype cost estimation often varies widely from actual commercial prices. While a pregnancy test could be made in a lab from $0.22, commercially available pregnancy tests in the US start at a price of $3.50 7 . Although such a cost analysis does not take into account production costs and factory overhead, the bill of materials does provide some insight on whether or not the devices would be within a feasible cost range in low income areas. We therefore still found it useful to assess our raw materials cost, although we acknowledge that commercial costs of this device after manufacturing would be higher. We estimate the total material cost of our 2DPN to be approximately $1.82.