Overview
To explore the use of SNP-LAMP detection of F2RL3, we divided our project into 3 specific aims. First, we adapted a published method for genomic extraction via simple cellulose dipstick1. Second, we developed and optimized options for amplification of extracted F2RL3 compatible with our extraction methodology 2, 3. Finally, we adapted a published design methodology for competitive fluorescent probes to our chosen SNP 2, 3.
Dipstick Extraction of Prokaryotic Genome
We began work on our first aim by making cellulose dipsticks and testing their compatibility with downstream amplification methods. As a model, we used PCR amplification of gapA from cultured E. coli. While there are differences between mammalian and prokaryotic DNA, we felt this simple model would allow us to address any unforeseen challenges before moving to a more complex system.
Our dipsticks were made by cutting a Whatman paper filter into standardized thin rectangles and dipping in candle wax to make a handle (Figure 1a). Then, we lysed bacterial cells with a simple, SDS-based chemical lysis buffer1 and extracted genomic material by dipping the exposed end of our stick into the lysate. We found the wax crumbled a bit upon vigorous dipping and switched to a simple square of cellulose and fine-point tweezers. Besides this, however, the experiment was successful, producing a clean band of amplified gapA (Figure 1b). Total time for lysis and dipstick extraction was around 6 minutes using prepared materials.
LAMP Amplification of F2RL3
Our team hypothesized that the most difficult part of our project would be fluorescent probe optimization. To expedite this process, we ordered two fragments of F2RL3 from Twist Bioscience, one for the A allele and one for the G rs773902 allele. Then, we tested our amplification reaction on these fragments to produce material for probes testing. The rest of this results page will consider only samples which contain just one of the two alleles. We recognize that many patients in the clinic will be heterozygous, and that they will likely show an intermediate test result between the fluorescence levels from homozygous patients.
We chose LAMP as our primary method of amplification because of its speed and isothermal simplicity. However, LAMP is also known to be a finicky reaction. To develop an optimized amplification, we ordered 5 primer sets designed with standard LAMP design software 4, 5. We tested each primer set on our ordered DNA, and saw significant products from 4 of the 5 sets (Figure 2a). We chose a single primer set from those that worked and performed further confirmatory testing, including rigorous negative controls (Figure 2b) and qPCR measurements of dsDNA amplification over time using SYBR dye (Figure 2c). Note that the smear of DNA in each lane is caused by the stochastic extension of repeating LAMP amplicons, and is the expected distribution of sizes for a LAMP product.
With a seemingly well-suited reaction in vitro, we proceeded to combine amplification with dipstick extraction of F2RL3 from human cell line HEK293T. After harvesting and counting cells, we performed chemical lysis with a Tween20/TritonX-100 based buffer1 and dipstick extraction of genomic DNA into the LAMP reaction mix. We demonstrated amplification of our gene from the cells, but noted an inability to produce a clean no-template negative control over several tests (Figure 3). This was reminiscent of our earlier negative control issue (Figure 1c), but we hypothesized that it was again the product of contaminating F2RL3 gene fragments in our workspace. This is especially a concern with LAMP, as its hypersensitive, continuous amplification mechanism makes it prone to producing a disproportionately large amount of product from as little as 10 copies of contaminating template6.
Initial Testing of Fluorescent Probes on Ordered DNA
With amplified F2RL3 from our ordered fragments containing either the G or A variant of rs773902, we began testing our fluorescent probes. Our chosen system relies on a competition for binding the SNP region of F2RL3 between a DNA sequence attached to a fluorophore (F) and a similar sequence with no fluorophore, sink 1 (S1) (Figure 4). When F is bound to the SNP region, it fluoresces. However, if it is not bound to the DNA, it will bind a complementary sequence attached to a quencher (Q), decreasing its fluorescence. Likewise, when S1 is not bound to the SNP region, it has a complement sink sequence (S2) with which to bind. To produce differential fluorescence between alleles, we used an online calculator to create a set of probes which produce an increased level of fluorescence in the presence of the Ala variant vs Thr variant2.
To facilitate this process, we combined raw amplification products with a mixture of each of the probes at 1 µM, according to the procedure described in Hyman et al. 2 After mixing, the temperature was increased to 95 °C and then allowed to slowly cool back to room temperature. We expected to observe differential fluorescence between samples with different F2RL3 alleles. However, as seen below, we did not observe any meaningful differences between samples over numerous trials (Figure 5).
This led us to suspect an issue with the conditions of the probe reaction. To address this, we tried to optimize our probe conditions by testing various combinations of probe and DNA concentrations. This included testing the following hypotheses:
- Our LAMP reaction was producing SNP sites in excess of our fluorophore, such that 100% of F was bound to template DNA in the presence of both variants (Figure 6a). We were able to titrate DNA between maximum possible concentration (no DNA dilution, results of which were not significantly different than the 1:25 dilution pictured here) and a 1:200 dilution of our LAMP product, which did not produce a curve with a significant difference from conditions with no amplicon added. None of the tested conditions produced significant differential fluorescence.
- Concentrations of S1, S2, & Q were not optimal, and did not produce differentially fluorescent equilibria under the present conditions. We tested various increased and decreased concentrations of each sink component, together and individually. No combination convincingly produced improved results (Figure 6b, c, d). While thermodynamic modeling may be able to predict correct concentrations for each component, we did not have a working model and relied on a simple trial and error approach of conditions relatively close to those successfully used by Hyman et al. 2
Discovery of Incorrect Amplicon and Reoptimization of LAMP
Faced with an unexplained failure to show differential fluorescence with our probes, and persistent amplification in no-template LAMP controls, both with dipstick samples from cultured cells and in various amplification reactions on ordered F2RL3 fragments, we began to question the identity of our product (Figure 2c, 3). We were concerned that we may have observed amplification of an incorrect amplicon, especially in light of the contamination concerns explained above.
As a preliminary experiment, we performed a diagnostic digest with a restriction enzyme targeted to break up the repeating structure of LAMP products into similar-sized fragments of around ~250 bp. We failed to observe digestion. Having evidence of an incorrect LAMP product, we returned to optimization of our LAMP conditions. We again tested all 5 primer sets, adding a diagnostic digest to our evaluation criteria of each condition and found that there was one primer set that produced a digestible result. This was a different primer set than we had been using to that point (Figure 7a).
We proceeded to again evaluate the efficacy of our probes on the products of each primer set, including the digestible product, but again failed to observe convincing differential fluorescence in any condition (Figure 7b). We recognize that our probe reaction optimization experiments should be repeated on this improved product, but did not have time in the short iGEM season to conduct such testing.
Asymmetric PCR As Alternative Amplification Strategy
While our initial literature review revealed LAMP as a promising candidate for F2RL3 amplification due to its speed and sensitivity, we were concerned that our probe experiments may have been affected by the primary or secondary structure of the LAMP amplicon. To address this, we adapted a method from another research paper that used a similar competitive fluorescence method to detect an SNP in a gene amplified by asymmetric PCR3. Asymmetric PCR is standard PCR amplification with an excess of one primer, producing a large number of single stranded products complementary to our probe sequence. Although asymmetric PCR requires a thermocycler and is not as user-friendly as LAMP, it addresses several of LAMP's drawbacks. In addition to a more simple amplicon structure, the quantized nature of PCR amplification means a small amount of amplified contamination present at the beginning of the reaction was less likely to compose a disproportionately large amount of the product.
We designed and tested PCR primers for an asymmetric PCR reaction, producing a correctly sized, diagnostically digestible product (Figure 8a). However, testing this product with our probes again failed to produce a convincingly differential result between F2RL3 alleles (Figure 8b). As above, we recognize that our probe optimization experiments should be repeated on this improved product, but did not have time in the short iGEM season to conduct such testing.
Development of Improved Probe Design Algorithms
When we began our project, we realized that the single probe and sink combination, even if produced by the online calculator from Hyman et al. may not be optimal or effective for our test. To address this, we developed our own algorithm to design and evaluate alternative probe and sink options. We have produced a command-line tool to implement our algorithm, which is explained in detail on our software page.
Future Directions
While we did not produce convincing differential fluorescence in the short iGEM experimental season, we learned a great deal about our methodology that is helpful for our own future work and other iGEM teams that may pursue similar projects.
Towards the simple, point-of-care extraction of patient genomic information, our future work will include optimization of extraction conditions from human cells by testing the compatibility and specificity of our improved amplification reactions with a dipstick extraction model. Given the resources, we would hope to conduct this testing on more realistic patient samples, such as human blood or saliva.
Towards rapid, point-of-care amplification of F2RL3, we would like to improve our methodology of confirming product identity through the use of a Sanger sequencing experiment on our PCR product. LAMP amplicon sequencing is not as straightforward due the complexity of its primary structure; however, we think properly analyzed next-gen sequencing could produce reliable results. Additionally, we have considered the idea that the total fluorescence signal of a sample may be dependent on the amount of DNA in solution. This means a valid test must standardize the amount of DNA produced in an amplification reaction. Our plan to address this is to ensure reagents, such as primers or dNTPs, for each reaction are limiting, such that each amplification produces the maximum amount of product. The sigmoidal curve of our SYBR experiments suggests that this is feasible, although we recognize more rigorous testing must be conducted with a finalized amplification reaction.
Additionally, to convert our methodology into a usable point-of-care test, further work must be done to show that the amplification procedure can be performed with simple equipment. This would include testing a LAMP reaction in a simple, temperature controlled heat block, and developing strategies to robustly minimize contamination of samples with non-patient-derived F2RL3. We hypothesize that this could be done by including a restriction enzyme like the one we used for diagnostic digestion in a commercial LAMP master mix. This enzyme could destroy contaminating DNA, and can be denatured at temperatures which do not affect LAMP polymerase, such that a restriction enzyme could keep the mix contamination free up until deactivation immediately before addition of a patient sample.
Towards developing reliable differential fluorescent probes and sinks, we will perform concentration and buffer optimization with our confirmed amplification product. Additionally, we recognize that our system relies on a manageable number of well-characterized chemical interactions, making it a good candidate for mathematical modeling. We realized this late in the season, and while our team is currently developing this capability through DNA interaction chemistry and thermodynamic modeling, we do not yet have a working model.
Once probe conditions are optimized, development of a usable point-of-care test further work must include demonstration of efficacy via simple heat block melting and annealing. Further, since fluorescence is dependent on temperature, a temperature range must be selected to be the end-point at which fluorescence is measured.
Finally, extensive trials must test the statistical properties of this test to determine whether controls for each F2RL3 allele must be included in each test. Further, heterozygous individuals will likely show intermediate fluorescence, which may be labeled with less confidence than a homozygous sample. Testing must be conducted to demonstrate test efficacy in this scenario.
References
- Mason, M.G., Botella, J.R. (2020). Rapid (30-second), equipment-free purification of nucleic acids using easy-to-make dipsticks. Nat Protoc 15, 3663-3677. https://doi.org/10.1038/s41596-020-0392-7
- Hyman, L. B., Christopher, C. R., & Romero, P. A. (2022). Competitive SNP-lamp probes for rapid and robust single-nucleotide polymorphism detection. Cell Reports Methods, 2(7), 100242. https://doi.org/10.1016/j.crmeth.2022.100242
- Wang, J., Zhang, D. (2015) Simulation-guided DNA probe design for consistently ultraspecific hybridization. Nature Chem 7, 545-553. https://doi.org/10.1038/nchem.2266
- NEB LAMP Primer Design v1.4.1, New England BioLabs. https://lamp.neb.com/#!/
- Primer Explorer V5, Elkien Chemical Co, https://primerexplorer.jp/e/
- Loop-Mediated Isothermal Amplification (2023), New England BioLabs https://www.neb.com/en-us/applications/dna-amplification-pcr-and-qpcr/isothermal-amplification/loop-mediated-isothermal-amplification-lamp