In patients with autoimmune diseases, the levels of specific microRNAs(miRNAs) in the blood are different. The concentrations of these miRNAs can increase or decrease during the onset of the disease. These changes in concentration can be used to diagnose patients having an autoimmune disease.¹ We planned to develop a minimally invasive, cell-free paper-based diagnostic tool that can detect these disease-specific miRNAs. For this, we divided our project into three sub-categories as mentioned in the description. The categories are (i) specific amplification and threshold detection of miRNAs, (ii) a modular tool to detect miRNAs(our detection system), and (iii) the design of the test platform and signal output. Below you can find the experimental strategy that we used in each of the three sections to make our diagnostic tool turn from an idea to a reality.

Amplification and threshold detection system

Amplification

We understood from our Human practices that similar miRNAs can change concentrations in multiple diseases. That’s why, we first used a machine learning algorithm to identify the miRNA combination that can distinguish relapsing-remitting MS patients from healthy people.

The concentrations of miRNAs in the blood range from fM to pM.² To enable our toehold switches to detect the selected miRNAs, we aim to incorporate a sequence-specific amplification step. We decided to use Nucleic Acid Sequence-Based Amplification (NASBA) as our amplification technique. NASBA produces a large amount of RNA, which consists of the selected miRNA sequence, flanked by universal sequences on both sides. This product is obtained through a cycle of 2 reactions: reverse transcription and in vitro transcription. We wanted to prove that the two NASBA reactions work to amplify our miRNAs.

The reverse transcription was performed first. During reverse transcription, the Reverse Transcriptase (RT) enzyme uses an RNA template to produce a complementary DNA (cDNA) strand. Subsequently, the RNase H domain of the RT degrades the miRNA in the RNA/DNA duplex (Figure 1a). We designed a reverse stem-loop primer based on Mader et al., 2012, that has a six-nucleotide overlap with our miRNA ‘has-miR-484’; which is found to be upregulated during relapsing-remitting MS.³ The annealing region of this primer serves as the starting point for the synthesis of the cDNA strand by RT. To confirm that the reverse transcription was successful, we performed a PCR with the reverse transcription product as a template. A PCR product can be obtained only if cDNA was produced during the reverse transcription reaction (Figure 1b). In addition, we performed an assay to determine the limit of detection (LOD) for reverse transcription. Since this is the first essential step in NASBA, it is valuable to find the lowest miRNA concentration that allows the reaction to work. For this assay, we performed reverse transcription in triplets with miRNA concentrations ranging from 70 nM to 7 fM, while the stem-loop primer concentration was held constant. The reaction products were used as a template for a PCR. We loaded the samples on the gel for analysis of the limit of detection for reverse transcription.

Secondly, we tested the in vitro transcription. During in vitro transcription, a T7 RNA polymerase transcribes a DNA template, starting from the T7 promoter sequence. In the one-pot NASBA reaction, this promoter sequence is incorporated in front of the miRNA sequence by the reverse transcriptase enzyme when adding a primer with the T7 promoter to the mixture (Figure 2). Since we focused on the sub-reactions, we prepared the DNA template separately. We performed the PCR using the cDNA strand as a template, with a forward primer including the T7 promoter sequence, and a reverse primer complementary to the cDNA strand (Figure 1b). During this PCR, the T7 promoter sequence was added to the cDNA strand upstream of the miRNA sequence. Then, we performed the in vitro transcription by adding the obtained DNA template to the in vitro transcription reaction mixture with the T7 RNA polymerase. After overnight incubation, we carried out denaturing gel electrophores and concentration measurements to confirm if the in vitro transcription was successful. Finally, we studied the effect of DNase I treatment and RNA purification on RNA stability and yield, by loading a part of the product on gel after each step. These downstream processes may be necessary for the use of the amplified miRNA sequence in our test.

Threshold detection system

Once the specific sequence is amplified, the concentration of miRNA is enough to be detected by the detection system. However, the miRNAs are always present in the blood, and the only difference between MS patients and healthy individuals is the change in concentration (Figure 3a). To detect upregulated levels of miRNAs during a disease, we must detect them above a certain threshold. Toehold switches, our detection systems, resulting in an output signal that scales linearly with the input signal (Figure 3b). Thus, we need a tuneable system that gives an output only above a certain threshold concentration, before we can use it in our detection circuit.

One method to construct such a tuneable threshold is by utilizing the properties of Toehold Mediated Strand Displacement (TMSD). TMSD is based on the concept that two annealed RNAs can be replaced by another molecule that has a higher affinity. For TMSD, three RNA molecules are needed: the molecule of interest A (which is amplified in the previous step), and two partially complementary strands B(antisense strand) and C(Input for detection system). Initially, B and C are bound. Besides this, there is an excess of B present equivalent to the threshold limit. Once the amplified product A is added to the reaction, it starts binding to free B. As there is no displacement of C, there is no input for the detection system, hence no output. Once all the free B’s are bound by A, i.e. concentration of A is above the threshold, A starts displacing C because A has a stronger affinity to B. Displaced C will trigger the detection system and output will be observed (Figure 4). In this way, a threshold can be added, based on the amount of B and C RNA present, that results in a sharp transition.

In the chosen system, the output C was visualised by using an inducible spinach aptamer (D) produced by Wang et al. (2023).⁴ Spinach is an RNA structure that can stabilise the side group of a fluorophore molecule (DFHBI), which allows it to fluoresce. The inducible spinach structure (Figure 5) contains a misfolded spinach aptamer, that upon binding a trigger (in this case C), can refold into a fluorescing structure that can be quantified in a fluorescence microplate reader. By measuring the fluorescence in a plate reader, the output of C can be detected. Since the structures of D and C are set due to the refolding requirements of D, A and B had to be adapted to comply with these requirements.

NUPACK was used to model affinities of desired pairs and prevent undesired binding between unrelated RNA sequences. Based on a template form of D (an inducible spinach aptamer in this context) and the associated trigger C from a paper by Wang et al. (2023),⁴ the template parts A and B were designed. Based on sequence C, an antisense structure B was designed with a greater affinity to BC than CD and can generate a sharp threshold with binding affinities that follow AB>BC>CD. C was elongated to improve the hybridisation between AB and consequent TMSD. Strand B was designed complementary to C, with an additional toehold and elongated region to hybridise with A (Figure 6).

After we designed and confirmed the sequences in NUPACK, we converted them into DNA and combined them with a T7 promotor. This allowed us to order the sequences as DNA oligos and transcribe them using in vitro transcription (IVT) back into RNA. The T7 promotor allows RNA polymerase to bind and amplify the RNA sequence at 37°C. This was done in a thermocycler to ensure optimal temperature and minimal evaporation. RNA from IVT was directly loaded in a 96-black clear bottom plate for fluorescence detection. No purification was performed as this caused irredeemable misfolding of the fluorescent aptamer. Spinach-DFHBI-1t fluorescence was measured at 470/505 nm at 25°C, 30 minutes after adding the RNA fragments to allow the mixture to reach equilibrium.

Alternatively, RNA was transcribed in a plate reader at 37°C to directly measure the fluorescence and thereby the concentration of RNA in the system. DNA with IVT buffer and enzymes was directly loaded in a 96 black clear bottom plate for fluorescence detection. To obtain a fluorescent signal, C and D had to be co-transcribed to result in the correct folding of the fluorescent aptamer. Spinach-DFHBI-1t fluorescence was measured at 470/505 nm at 37°C in 30-minute intervals, to reduce photobleaching.

Toehold switches: miRNA Detection System

The detection of miRNA is done by toehold switches and logic gates. In our final test, the output from the threshold system (C) will act as the trigger for the detection system.

We aimed to build and test these toehold switches in vitro. We did this by cloning the toeholds and gates into a plasmid and testing expression in vitro. Firstly, we designed the plasmid containing toeholds and gates in silico. We wanted to make a single standardized modular plasmid. Doing so would help replace the toehold switches with other toehold switches detecting different miRNAs specific for other diseases. We made some cloning strategies to make this possible. The plasmid was divided into two parts:
1. The backbone with a reporter gene (lacZ)
2. The detection system (toeholds and AND gate) (Figure 7)

We considered the backbone with lacZ gene as the constant part and the toehold switches as a variable part which can be replaced with another fragment if required. To make the plasmid modular, the single-stranded DNA overhangs (Sticky ends) of the backbone should remain constant. lacZ is a part of the plasmid backbone and is placed immediately next to the toehold switches. So, the first four nucleotides of lacZ were decided as the modular single-stranded DNA overhangs on the 3’ end of the AND gate. The 5’ end of the AND gate consists of the T7 promoter. As the T7 promoter is required for in vitro testing, it will be a fixed part of the backbone (Figure 8). So, the 4nt overhang sequences for 5’ and 3’ ends at the end of T7 and the AND gate were selected as modular single-stranded DNA overhangs (Marked blue T7 and red for lacZ in (Figure 7)).

The Backbone

Initially, we did not have the backbone with the required constructs like T7 promoter, T7 terminator, and lacZ gene. So, we amplified them separately and later added them to the backbone along with the toehold switches. As the T7 promoter was small, we amplified it along with the backbone. The T7 terminator and the lacZ gene were amplified separately. We did the amplification using PCR. All primers were flanked by BsaI restriction sites so that the amplified fragments could directly be used during the Golden Gate reaction.

The toehold

We constructed the toeholds using overlap extension PCR (OE-PCR) to test them in vitro. The whole construct was produced by designing four primers, with partial overlap. To perform the OE-PCR, these four primers are used at the same time using a Q5 PCR program, with a Tm of 65°C, followed by DNA extraction of the product. The same PCR program is then used again on the product, together with the first and last primer, to amplify the whole construct. At the same time, the backbone is amplified. Both the construct and the backbone are flanked by BsaI sites. The region where the enzyme cuts overlaps with the backbone. Thus, the construct could directly be used in a golden gate reaction with a linearized SEVA backbone (Figure 9).

We designed the toeholds specific to hsa-miR-484. This miRNA was shown to be upregulated in the blood of multiple sclerosis patients. Three toeholds were designed to detect this miRNA namely Toehold A, Toehold B, and Toehold C. Toehold A and B were designed using our toehold designer, and Toehold C was designed using only NUPACK. In the lab, we used a similar method of construction for all the toeholds as mentioned above.

The AND gates

During the onset of a disease, several disease-specific miRNAs vary in their concentration. To increase the accuracy of our test, detecting multiple disease-specific miRNAs is advantageous. Therefore, we want to use toehold switches performing logic gate functions such as an AND gate. These AND gates can detect multiple miRNAs and express output only in the presence of all the miRNAs. For our test, we constructed a two-input AND gate that detects two different trigger RNAs.

A two-input AND gate is a single toehold switch that gets activated by two trigger RNAs. The input for the respective toehold switch is a combination of two input RNA molecules that are partially complimentary to each other. These partially complementary inputs hybridize to form an input trigger RNA that can anneal to the trigger binding region of the toehold switch. Once the trigger RNA is bound to the trigger binding region, the loop in the toehold switch opens up and the RBS is available for the ribosome to bind, allowing gene expression. The absence of any one trigger RNA will not lead to the unfolding of the loop structure, thus causing no gene translation (Figure 10).

Instead of directly ordering the AND gate, we constructed it using Q5 high-fidelity PCR. The AND gate was divided into two parts. Both the parts were amplified with Q5 PCR separately. Primers were constructed in such a way that the 3’ end of the first part and the 5’ end of the second part have the same single-stranded overhangs. Thus, after the BsaI cut during a Golden gate reaction, these single-stranded DNA overhangs can anneal making one complete AND gate (Figure 11).

Flow of experiments

Two parts of the AND gate and the backbone along with the reporter gene were amplified using Q5 high-fidelity PCR. DNA was extracted using Gel electrophoresis and purified using Zymoclean™ Gel DNA Recovery Kit. All amplified fragments were ligated using the Golden Gate method. We directly ordered trigger RNAs for testing the toeholds and AND gate in vitro (Figure 12).

Cell free in-vitro test

Our diagnostic test relies on cell-free transcription and translation. To test the functionality of our detection system, we used the commercial PURExpress In Vitro Protein Synthesis Kit (NEB E6800) to perform in vitro experiments. The kit contains two solutions and control DHFR Plasmid. This is a reconstituted form of gene expression that works by isolating the necessary components, such as ribosomes and amino acids, purifying them, and assembling them into a reaction mixture.⁶ It consists of two solutions, which can be supplemented with other reagents like substrate or RNase inhibitor. PURExpress is based on the T7 RNA Polymerase, which has a high specificity for the T7 bacteriophage promoter. To test this system, a plasmid was assembled containing the T7 promoter, the lacZ gene and the T7 terminator. lacZ was selected as the output gene (Figure 13). This gene encodes the enzyme \beta-galactosidase, which catalyses the hydrolysis of the yellow Chlorophenol Red-b-D-Galactopyranoside (CPRG) into Chlorophenol red. If the gene is expressed, the colour changes from yellow to purple, which can be measured at 570 nm.

The plasmid encoding \beta-galactosidase was assembled using HiFi Assembly (NEB, E2621), with the DHFR control plasmid from the PURExpress kit as the backbone. We designed primers to create homology arms for both the backbone and the full LacZ gene. The plasmid was assembled and transformed into E. coli DH10\beta and miniprepped.

The PURExpress reactions were carried out according to the manufacturer’s instructions, but we scaled down the volume from 25 \muL to 2 \muL. By working in smaller volumes, fewer materials are required per test. This means that the production is more cost-efficient. To prevent the breakdown of RNA, an RNase inhibitor was supplemented (Promega, 2515). Additionally, the substrate CPRG was added to the reaction mixture. The reaction was tested at the recommended 37°C and at room temperature. This same method was used to test toehold switches and logic gates.

Paper-based testing

Our test platform is paper-based. A paper-based method has several advantages, such as low production costs, easy storage, and stability of biochemical components.⁷ Therefore, we wanted to integrate cell-free expression with a paper matrix. This method was described previously by Pardee et al. (2014), which we also used for our test.⁸ First, the cellulose filter paper is treated with 5% BSA, to prevent any unwanted side reactions due to pH-dependent charges. After drying, the filter paper is cut into 2 mm discs using a biopsy punch, and these disks are placed in a 384-well plate. Next, cell-free reactions are assembled by combining the two PURExpress solutions, containing the cell-free reaction components, with the substrate, the DNA template, and the RNase inhibitor. These reactions are applied to the discs. Then, the plate is dipped in liquid nitrogen to flash-freeze the discs, after which it is freeze-dried overnight. After 24h, the discs can be rehydrated with nuclease-free water with or without trigger miRNA. The plate is then incubated at 37°C for two hours. Colourimetric output can be seen by the eye; a negative result will remain yellow, and a positive result will be purple (Figure 14).

By assembling the tests in 384-well plates, they can easily be stored and distributed at room temperature. Furthermore, no biotic components are present, contributing to the biosafety of the product.

In vitro tests

To evaluate the performance of the MS-specific toehold switches and the AND gate, we performed in vitro tests over time. As explained previously, the toehold switches and the AND gate were put under the control of the T7 promoter and terminator. Synthetic input RNA was added to imitate the final testing environment. The lacZ gene (Figure 15) was used as the output protein for the toehold switch and the AND gate (Figure 11)(Figure 15).

Cell-free reactions were set up in a 5 \muL volume. The toehold switch and AND gate plasmids were supplied at a final concentration of   30 ng. For all samples, 1 \muL of trigger RNA was added at 5 \muM. For the AND gate, 1 \muL was added to both triggers. The reactions were assembled in PCR tubes and incubated at 37°C. At eight time points, a 0.5 \muL sample was taken from the reactions and diluted 151x in a 96-well plate. The absorbance was measured at 570 nm with a plate reader.

DNAzymes

As an alternative to the conventional enzymatic reporter signal, we investigated the use of DNAzymes, short ssDNA oligonucleotides that form catalytically active structures. Especially the G-quadruplex (G4) / hemin DNAzyme has been gaining increasing recognition in biosensing/nucleic acid sensing platforms. This quadruplex contains four stretches of three guanines, a G4 sequence, and forms a tertiary structure which is able to interact with hemin as a cofactor (Figure 16). It acts as a horseradish peroxidase-mimic, catalysing H2O2-mediated oxidations of various substrates. For colourimetric measurements, 2,2-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) can be used to turn the colourless ABTS^2- to the green-coloured ABTS^- radical.

A common approach for designing such a DNAzyme-biosensor is by splitting the G4 sequence, usually in a 1:3 fashion, and extending these with probe sequences that hybridise with their target adjacently, which thereupon reconstitutes the DNAzyme (Figure 17).

For the detection of the activated toehold switch, we designed our probes by splitting the EAD2 sequence (BBa-K3343000) and attaching 15 – 20 nucleotides complementary to the sequences that are normally sequestered in the hairpin / stem-loop, namely the RBS and start codon (Figure 18).

With this approach, we could theoretically at least leave out the protein translation aspects of the cell-free test which is the most component-heavy and energy-dependent part of a PURE/cell-free transcription and translation (TXTL) system. The toehold switch protocol is to be followed suit, but without the translation components of the cell-free test. The in vitro transcription (IVT) can be conducted with the switch DNA template, trigger RNA, and probes present.