Improvement in Detection Limit for Lateral Flow Assay of Biomacromolecules by Test-Zone Pre-enrichment

Lateral flow assay (LFA) is one of the most prevalent commercially available techniques for point-of-care tests due to its simplicity, celerity, low cost and robust operation. However, conventional colorimetric LFAs have inferior limits of detection (LODs) compared to sophisticated laboratory-based assays. Here, we report a simple strategy of test-zone pre-enrichment to improve the LOD of LFA by loading samples before the conjugate pad assembly. The developed method enables visual LODs of miR-210 mimic and human chorionic gonadotropin protein, to be improved by 10–100 fold compared with a conventional LFA setup without introducing any additional instrument and reagent except for phosphate running buffer, while no obvious difference occurred for Aflatoxin B1 (AFB1). It takes about 6–8 min to enrich every 50 μL of sample diluted with phosphate running buffer, therefore we can get visual results within 20 min. We identified a parameter by modeling the entire process, the concentration of probe-analyte conjugate at test zone when signaling unit being loaded, to be important for the improvement of visual limit of detection. In addition, the test-zone pre-enrichment did not impair the selectivity when miR-210 mimic was adopted as target. Integrated with other optimization, amplification and modification of LFAs, the developed test-zone pre-enrichment method can be applied to further improve LOD of LFAs.

To make a conjugate pad with DP-AuNPs, 250 μL of DP-AuNPs stock solution was sprayed onto a conjugate pad (0.65 cm  30 cm) using a 3D spray point platform, dried in a vacuum at room temperature, stored in a dryer at 4 o C and cut into 6.5 mm  4 mm.

Characterization of DP-AuNPs
The effective combination of AuNPs and DP molecules is one of the key factors for the success of LFA. The optimal ratio of DP to AuNPs for the preparation of DP-AuNPs was studied in presence of 0.2 M NaCl, with an ionic strength level high enough to precipitate AuNPs that were not fully protected (Fig. S1). Upon addition of NaCl, solutions with molar ratios lower than 10 pmol DP to 9.1 nmol Au showed color change from wine red to blue immediately, which is the typical phenomenon of AuNPs aggregation, while those with higher molar ratios tolerated the presence of 0.2 M NaCl well. The produced solution with a molar ratio S-3 of 10 pmol DP to 9.1 nmol Au turned to bluish violet after 24 h of storage/aging, indicating this molar ratio was still not high enough to prohibit the self-assembling of AuNPs into bigger aggregates induced by high concentration of salt in long-term storage, while molar ratios of 50 pmol DP to 9.1 nmol Au seemed high enough. Then the performance of as-synthesized DP-AuNPs in sandwich-like LFA was verified. From the inset of Figure S1, DP-AuNPs synthesized with higher molar ratio of DP/AuNPs showed better signal at both control and test zone with the presence of target oligonucleotide at the same concentration. So the molar ratio of 100 pmol DP to 9.1 nmol Au was chosen for further synthesis [1]. The AuNPs and DP-AuNPs were characterized by UV-vis absorption spectrometry, and transmission electron microscopy (TEM) (Fig.   S2). Both AuNPs and DP-AuNPs were about 20-30 nm in diameter according to their TEM image and their maximal absorption at 521.5 nm for AuNPs and 526.5 nm for DP-AuNPs. The UV-vis spectrum of DP-AuNPs showed slight red shift as compared with that of AuNPs, verifying the successful fabrication of DP-AuNPs conjugates.

Simulation and modeling
The model assumptions include: (1) diffusion limits the delivery of DP-AuNPs to the test site, (2) reaction ultimately limits the capture of DP-AuNPs to the test site in the LFA systems, and (3) the reaction of DP-AuNPs capture is kinetically limited second order reversible interactions.
As the reaction of DP-AuNPs capture was reported to be the rate-limiting step to improve the LFA sensitivity, we focused on the reaction rate comparison between the direct sampling and the test-zone pre-enrichment method.
In molecular beacon sandwich-like format we used, only when the analyte bound with the ring part of the molecular beacon (MB, as capture probe) at test zone probe and disassembled the self-binding at the stem of MB, the detecting probe on AuNPs could bind with the 3' end of MB and showed a visible signal at test zone. Thus there are two successive essential reactions for the capture of AuNPs on the test zone, 1)

2)
In reaction 1, two species including analyte (A) and test zone probe (T) are involved, and the product the conjugate of test zone probe and analyte (TA) is one reactor of reaction 2. The formation of TAP (conjugate of TA and P (DP-AuNPs) ultimately limits the capture of DP-AuNPs to the test zone and is the decisive factor of final color intensity. [T] 0 is constant and is assumed to be 100 M. The forward reaction rate constants (k on ) for the reaction 1 are predicted to be 1.28-2.70×10 6 M -1 s -1 with salinity of 0.01-1 via a web-based software tool available at http://nablab.rice.edu/nabtools/kine constructed by Zhang et al. in 2018 [6]. There must be some deviation from the reality because the hybridization experimentally characterized in their work were all performed in 5×PBS buffer solution but the reactions takes place on NC membrane here, and all their target and probe sequences were 36 nt long but ours are only 24 nt long.
[TA] versus reaction time curves are simulated by assuming hybridization rate constants to be 110 6 M -1 s -1 to 10 3 M -1 s -1 as almost all the k on in ref 6 (k Hyb was used in the literature) were within that range ( Fig. 6a to 6d).
As the former one of a two-step continuous reaction, the more fully balanced reaction 1 will benefit the subsequent reaction a lot. Therefore, when the reaction rate constant is fixed and the concentration of the target is constant, appropriate extension of the reaction time of the analyte on the test strip is conducive to the more sufficient reaction 1, so that more analyte molecules can be captured in the test zone. Usually, the LFA takes 10-30 minutes to complete the test. If the combined reaction can reach a balance within this period, the best detection effect may be achieved. As for different [A] 0 (100 M, 10 M, 1 M, 100 nM, 10 nM, 1 nM and 0.1 nM) at certain assumed T and k on , the higher [A] 0 is, the faster TA forms, and vice versa ( Fig. 6a). Given k on is 110 6 M -1 s -1 , it takes less than 0.1 s for the reaction of TA formation to reach balance Thus, when the target concentration is high, the reaction 1 can quickly reach balance, and the subsequent signaling unit's capture becomes the speed limit step. However, when the concentration of the target is low, the reaction 1 can hardly reach balance within one test duration, then the reaction 1 turns to be the speed limit step. If the sample solution is pre-loaded for enrichment for 10-20 minutes, the analyte with the concentration at 10 nM or above will be able to fully react with the capture probe and reach the reaction balance. The enrichment of analyte and the increase of TA products in the test zone undoubtedly provide a great promotion to the follow-up reaction 2 in the test zone.
The time difference of the binding reaction with different reaction rate constants to reach equilibrium on the test strip will also affect the result of the LFA. While [A] 0 is fixed as 100 M, it takes only 0.1 s for the reaction of TA formation to reach balance if k on is 110 6 M -1 s -1 , however up to 100 s is needed if k on is only 110 3 M -1 s -1 (Fig. 6d). Therefore, in order to realize the rapid detection of test strip, it is necessary to select antibodies or adapters with a binding rate constant high enough to facilitate the rapid TA and TAP formation in a limited time.

ii) AuNPs diffusion/[P] profile
There are two consecutive phases of flow in the LFAs: "membrane" and "absorbent pad" phase, and the transition point from "membrane flow" to "absorbent pad flow" is the moment that the solution frontier reaches the absorbent pad. Time dependent velocity ( , U is the diffusion velocity, a is a constant related with the pore diameter of NC membrane, t is the diffusion time) could be obtained from "membrane flow", while the slower velocity from "absorbent pad flow" is constant. Exactly, with the direct sampling method, DP-AuNPs travels through NC membrane in "membrane flow" within the first minute and in "absorbent pad flow" in the following minutes (Fig. 6e). However, DP-AuNPs travels only in "absorbent pad flow" with the test-zone pre-enrichment method because the NC membrane is already thoroughly wetted after sample pre-enrichment (Fig. 6f). We calculated the diffusion of AuNPs according to the integrated density of gray of the squared part assuming the velocity of AuNPs here is the same as that at the test zone and achieved the dynamics of AuNPs in different sampling method (Fig. 6g). In the direct sampling method, the concentration of AuNPs in the front end of the solution was the highest, and it fell to less than 82% within 50 s. However, in the pre-enrichment method, the concentration of AuNPs peaked and fell more slowly, and it maintained at more than 87% for at least 210 s. In addition, the peak concentration of AuNPs  In direct sampling method, DP-AuNPs migrates along with the analyte toward the test zone and reacts there.
When they arrive at the test zone, the reaction to generate TA has just started, thus [TA] is quite low at the initial stage of LFA (Fig. 6a). Therefore, the reaction rate of TAP formation is mainly controlled by [TA] and is very slow at the initial minutes.
However, in the test-zone pre-enrichment method, when DP-AuNPs is loaded a large amount of TA has already been generated in the test zone after minutes of pre-enrichment of analyte under the same concentration of analyte as that of the direct sampling method. Therefore, the rate of TAP formation is greatly increased with the larger [TA] and the final [TAP] is also improved in the test-zone pre-enrichment method compared with that of direct sampling method.

S-7
According to the simulation, [TAP] 10 ([TAP] achieved 10 min after loading of DP-AuNPs) versus [A] are analyzed (Fig. 6h to 6k). In the direct sampling method (group 1), given k on is 10 6 or 10 5 M -1 s -1 , there is no difference in [TAP] when [A] is 1 M or 10 M, while given k on is 10 3 M -1 s -1 , the target seems undetectable even at 10 nM sampling 50 or 100 L with pre-enrichment method, which are both not consistent with the experimental results in Figure 2a. Only the statistic results with the assumption of k on as 10 4 M -1 s -1 seem to be consistent with most of the experimental results: 1) there is a linear relationship between the signal at test zone and log[A] from 10 5 nM to 10 nM in direct sampling method and from 10 3 nM to 1 nM in pre-enrichment method; 2) the signal of 50, 100 and 400 L analyte at 10 nM and 400 L analyte at 1 nM in pre-enrichment method is comparable with that of 50 L analyte at 100 nM in direct sampling method and larger than that of 50 L analyte at 10 nM in direct sampling method, which is undetectable in experiment .
Therefore, we may say that k on is possibly 10 4 M -1 s -1 for reaction 1), and with the pre-enrichment of 50, 100 and 400 L 10 nM sample solution, [TAP] 10 could be increased by 3.7-fold, 6.0-fold and 18.6-fold compared with that of direct sampling method, leading to improved LOD.
[TAP] 10 is still detectable with pre-enrichment 400 L 1 nM sample solution. Given both methods were set to read the signal intensity 10 min after loading DP-AuNPs, the test-zone pre-enrichment method was able to generate more TAP to obtain higher sensitivity due to the much higher initial [TA].
According to the modeling results, the capturing rate of DP-AuNPs was higher in the test-zone pre-enrichment method because the concentration of test zone probe-analyte conjugate was much higher when DP-AuNPs was loaded. Given both direct sampling method and test-zone pre-enrichment method were set to read the signal intensity 10 min after loading DP-AuNPs when the gray intensity of NC membrane recovered to the original value ( Fig. S9 to S11), the capturing rate of DP-AuNPs dominated the amount of DP-AuNPs captured, the decisive factor of sensitivity. Thus, higher sensitivity could be obtained due to the much higher concentration of test zone probe-analyte conjugate in the test-zone pre-enrichment method than that of direct sampling method. This modeling gives us insight into the design and optimization of pre-enrichment enhanced LFAs. S-8