Proof of Concept

Moving through the proof of concept process

DIAS utilizes the upregulated levels of miRNAs in liquid biopsies of patients with lung cancer as a detection method to identify patients with lung neoplasms in the early stages of the disease. The DIAS detection platform is based on the functionality of the LbuCas13a enzyme in combination with the optimal crRNA sequence which hybridizes with the target miRNA biomarker. In addition, the fluorophore quencher labeled reporter RNA should be properly designed to allow for efficient trans cleavage by the enzyme. Therefore, all system's elements should function in a coordinated way as individual components of the overall system which should display a proven efficacy in detecting the target miRNA even if presents in very low concentration in the test sample. Through the integration of the microfluidic chip, the system detection analytical sensitivity can be significantly enhanced, improving the limit of detection (LOD) of the method.

As described in more detail in the Results section, all system components have been properly designed, produced and purified in a long "synthetic biology" journey starting from complicated cloning procedures and ending in the final LbuCas13a and IVT crRNA production and purification. In a few words, the following milestones regarding our system's basic components were successfully completed:

• The LbuCas13a CDS cloning in pSB1C3 backbone retaining the RFC [10] compatibility.
• The efficient production and purification of LbuCas13a protein in E.coli in a relatively high protein yield so as to be incorporated into the DIAS detection platform.
• The efficient cloning of 6 different crRNA sequences in pSB1C3 backbone retaining the RFC [10] compatibility. We even proposed for the first time an easy and efficacious methodology to change the crRNA sequence in case the desired target miRNA changes, just by designing a different primer and following the guidelines provided by the DIAS crPrep Sample preparation kit.
• All the 6 different crRNA were in vitro transcribed and further purified to be integrated into the DIAS detection platform.

Towards implementation of the DIAS detection platform

Bulk fluorescence assay

Since the basic experimental milestones of our project were successfully reached, we tested our system's efficacy in detecting the fluorescence intensity of known added quantities of synthetic miRNA-17-3p corresponding to the target of the assay. To accomplish this, a detection standard curve was constructed plotting the miRNA concentration in known tested samples (standards) against fluorescence intensity (Shan et al., 2019). The calibration curve is a bioanalytical method that represents a linear relationship between the concentration of an analyte (independent variable-miRNA added) and response (dependent variable-fluorescence intensity). This relationship is used to predict the unknown concentration of the analyte in a complex matrix such as the blood or serum (Azadeh et al., 2017). The standard curve-based quantification method is efficiently applied by other analytical quantification assays and allows us to identify the miRNA concentration of an unknown sample just by acquiring its fluorescence intensity through the DIAS detection assay.

Standard curve construction

For the standard curve construction, we followed the procedure described in the DIAS detection assay subsection of the Design page. As for all bulk fluorescence experiments, 50μl of Cas13a mix was prepared with 20nM LbuCas13a, 20nM crRNA, 0.5μΜ FQ5U Reporter, different concentrations of target miRNA (0-negative control, 1pM, 5pM, 10pM, 50pM, 100pM) and 1X Reaction buffer (10mM Tris-HCl, 50mM NaCl, 10 mM MgCl2, pH 7.9). The fluorescence detection was performed at 37 °C for 60 min using a fluorescence plate reader with fluorescence collected every two minutes. To construct the final standard curve for the detection system we plotted the fluorescence intensity of each miRNA tested concentration versus time analyzing the data from the plate reader. For each time point, a standard curve was constructed along with the corresponding linear trendline and coefficient of determination (R2). The curve showing the best linearity and R2 was chosen as the final standard curve for the DIAS detection platform. The experimental workflow followed to construct the standard curve is illustrated in Figure 1.

But is this enough as a proof of concept of our system's functionality? As clearly depicted in Figure 2 the coefficient of determination R2, which is a measure of how well a linear regression line fits the data, approaches 1 indicating a perfect fit. This linear calibration curve demonstrates the DIAS assay performance in a validated analytical range (1-100pM) and verifies the remarkable quality of the CRISPR-based bioanalytical method. The best standard curves with R2 approaching the 0.995 and 0.999 values correspond to the time point of 6 and 8 minutes after the initiation of the fluorescence detection. Therefore, we concluded that the fluorescence measurements should be made in the time interval from 5 to 10 minutes after the addition of the reagents to the microplate, confirming the short required test execution time. Furthermore, our system is capable of quantifying miRNA concentrations as low as 1pM which indicates our system's exceptional sensitivity and allows for further optimization to achieve an even higher sensitivity at lower target miRNA concentrations on the fento scale.

Validation of miRNA detection in complex matrices

Based on the standard curve constructed with known added miRNA concentrations (standards), we continued with additional experiments aiming to detect the miR-17-3P levels in more complex matrices. The following proof of concept experiments were conducted to assess the detection system's performance:

• Quantification of known added miR-17-3P concentrations in different samples based on the standard curve.
• Quantification of known added miRNA-17-3P concentrations in different samples to which a known concentration of other non-selective miRNAs (miR-1246 & miR-143) has previously been added.
• Relative quantification of miRNA-17-3P in 2 different cell lines, including human fibroblast normal cells (MRC-5) and human lung adenocarcinoma cells (A549).

Quantification of miR-17-3P concentrations in different samples based on the standard curve

Based on the DIAS platform standard curve (Figure 2) we tested the system's detection accuracy using different reaction samples which contain a specified known concentration of the miR-17-3P. After the fluorescence measurement, the obtained fluorescence intensity values corresponding to each miRNA concentration are processed by subtracting the background fluorescence corresponding to the negative control value (no miRNA added). The resulting values were entered on the equation \( y = 498.8 x + 117.0 \) of the standard curve (Figure 2, at 8 min) in place of the y variable. The resulting x value corresponds to the predicted miRNA concentration. Based on the analysis results we constructed a plot (Figure 3) where the y-axis shows the actual added miRNA concentrations while the x-axis shows the predicted values of the corresponding miRNA concentrations based on the standard curve. Four different miR-17-3P concentrations (2pM, 20pM, 40pM & 200pM) were investigated and finally plotted on the plot. As described previously, for all bulk fluorescence experiments, 50μl of Cas13a mix was prepared with 20nM LbuCas13a, 20nM crRNA, 0.5μΜ FQ5U Reporter and the different concentrations of target miRNA (standards) in 1X Reaction buffer (10mM Tris-HCl, 50mM NaCl, 10 mM MgCl2, pH 7.9). All reactions were conducted in triplicate.

Analyzing the graph and observing the values listed in Table 1, we can conclude that the detection system can accurately quantify the miRNA concentrations in a range of 2pM to 20pM. However, in higher miRNA concentrations (40pM & 200pM) the quantification error between the real and predicted values significantly increases deviating from the diagonal line which indicates complete agreement.

Quantification of known added miR-17-3P concentrations in "non-selective miRNA-enriched complex environments"

Going one step further, we examined the system's detection specificity by adding a specified same concentration (8pM) of miR-1246 and miR-143 in all the different reaction samples each containing a different concentration of the target miR-17-3P (2pM, 50pM, 40pM, 200pM) as described previously. The aim of this study is to investigate the LbuCas13a enzyme's ability to efficiently bind with the crRNA and subsequently with the target miRNA in an "artificial environment" that contains a pool of non-specific miRNAs. If the detection accuracy is retained in such a "miRNA-enriched environment", then we can proceed with more complex biological matrices such as the total RNA extracted from different cell lines. As extensively described in the previous section, the obtained processed fluorescence intensity was entered on the equation \( y = 498.8 x + 117.0 \) of the standard curve in place of the y variable to predict the miRNA concentrations. Based on the analysis results we constructed a plot (Figure 4) where the y-axis shows the actual added miR-17-3P concentrations while the x-axis shows the predicted values of the corresponding miRNA concentrations based on the DIAS platform standard curve. As described previously, for all bulk fluorescence experiments, 50μl of Cas13a mix was prepared with 20nM LbuCas13a, 20nM crRNA, 0.5μΜ FQ5U Reporter, 8pM of miR-1246, 8pM of miR-143 and the different concentrations of target miRNA in 1X Reaction buffer (10mM Tris-HCl, 50mM NaCl, 10 mM MgCl2, pH 7.9). All reactions were conducted in triplicate.

The quantification results in “non-selective miRNA-enriched reaction environments” (Table 2) agree with the quantification results provided in the previous section. Specifically, the detection system's performance is notably good in a wide range of miRNA concentrations (2-40pM), with small differences observed between the predicted and the corresponding actual values. This demonstrates that the LbuCas13a/crRNA complex can efficiently recognize the target miRNA triggering the enzyme's trans cleavage ability independently of the non-selective miRNAs presented in its "reaction environment". However, as observed in the previous experiments, in higher miRNA concentrations (>40pM) the quantification accuracy decreases.

Relative quantification of miRNA-17-3P in total RNA extracts isolated from 2 different cell lines.

To validate the detection method's performance in more complicated biological matrices which more closely resemble the clinical samples, we isolated total RNA extracts from human fibroblast normal cells and human lung adenocarcinoma cells. The aim of this experiment is to demonstrate the LbuCas13a/crRNA system's feasibility in detecting the miR-17-3P in a pool of total RNAs as well as investigate the target miRNA expression levels in lung cancer compared to lung normal cells. In addition, as a positive control, we added a specified concentration of the target miR-17-3P (36pM) in A549 cell-derived total RNA extract to demonstrate the system's performance.

The experimental procedures followed for the detection assay are briefly described below.

The two cell lines were cultured in high glucose Dulbecco's modified Eagle Medium (DMEM), enriched with 10% FBS and 1% PS, and incubated under the following circumstances: 37°C, 5% CO2. After the cell density reached 80%, the cells were detached, and centrifuged for collection. 5x105 cells were collected from each cell line and washed with PBS 1%. Afterwards, RNA was isolated from each sample according to the RNeasy micro kit (Qiagen Inc., Chatsworth, CA, USA) which enables the recovery of the total RNA content including the miRNAs. The extracted RNA was assessed for purity and integrity via nanodrop 2000.

For the fluorescence experiments, 50μl of Cas13a mix was prepared with 20nM LbuCas13a, 20nM crRNA, 0.5μΜ FQ5U Reporter and 83.3ng of total isolated RNA extract from each cell line respectively in 1X Reaction buffer (10mM Tris-HCl, 50mM NaCl, 10 mM MgCl2, pH 7.9). All reactions were conducted in triplicate.

The results of the detection assay revealed that the miR-17-3P expression levels in A549 lung cancer cells are slightly elevated compared to MRC-5 normal cells (Figure 6).

In addition, the positive control sample containing the same quantity of total RNA extract with the addition of a predefined concentration of miR-17-3P (spike-in) revealed an approximately 10 fold elevation in miR-17-3P detection levels compared to MRC-5 and A549 cells (Figure 5). Therefore, we can conclude that the detection platform's performance is retained in this biological complex matrix (MRC-5 & A549 cells) which contains the target miRNA molecules in a pool of several RNA molecules including mRNAs, rRNAs, tRNAs, miRNAs, etc. The next of our experimental pipeline involves the verification of miR-17-3P elevated levels in A549 compared to MRC-5 with qPCR.

Microreactors fluorescence assay

Since the fluorescence detection assay and the microfluidic chip construction were successfully implemented, we attempted to transfer our molecular system from the bulk solution to droplet-based microreactors. The aim is to compartmentalize the molecular components of the assay into picoliter-scale microreactors to enhance the sensitivity of the detection method without nucleic acid amplification (NAA) steps. The concept is to imitate the natural confinement effect which means that the local concentration of a single molecule is elevated inversely with decreasing analytical volume (Tian et al., 2020). Our aim was to generate water in oil (w/o) emulsion via the droplet microfluidic chip where the water phase would contain all the reagents (Cas13a protein, crRNA, miRNA, reporter RNA) and the oil phase would consist of an oil for emulsion generation. The architecture of the droplet microfluidic chip was designed based on the principles of the flow focusing structures, as described in more detail in the engineering success section. The droplet microfluidic chip enables the droplet formation with a high monodisperse index.

Since we had previously configured the detailed detection assay method based on the standardized SD, RS, CC and MM mixtures, the whole procedure was carried out automated in a premeditated way. The detailed protocols for the production of the mixtures mentioned above are described in the measurement section. Firstly, we had to dilute the miR-17-3P in water to achieve a concentration of 5nM. Then, the RST mixture and CC mixtures were prepared in a scaled-up solution (x20) to succeed 1ml final reaction volume, as shown in Tables 3-4 below. The volume of the final reaction was adopted in 1ml to be sufficient for the flow control by the syringe and the pump.

The last step was the addition of the CC mixtures into RST mixture generating the MM master mixture performing the water phase of the emulsion. As for the oil phase we chose mineral oil which is an ideal choice for the emulsion formation. The surfactant that we chose for the sufficient emulsification of the samples and the enhanced stability of the emulsion was Tween 20 at a final concentration of 1% in the water phase. The syringes which were filled with the aqua and oil phase were connected with two syringe pumps, as shown in Figure 7. On our first trial, the flow rates of the two phases were controlled in Qw = 24μl/min and Qo = 120μl/min, representing a flow rate ratio FRR (quotient of the dispersed phase flow rate over the continuous phase flow rate) of 1:5.

Afterwards, the emulsion derived from the microfluidic chip was loaded onto a Neubauer improved counting chamber to visualize the shape of the droplets as well as to estimate the number of the fluorescent droplets utilizing a fluorescent microscope. As shown in Figure 8, following these steps we received an emulsion with fluorescent droplets showing high polydisperse index with a size range from 1μm to 200μm.

Since the chosen flow rates generated an emulsion with high polydisperse, we tested a panel of flow rates to clarify the most efficient, maintaining the FRR=1:5. We observed that on reduced flow rate we could formulate an emulsion showing appropriate disperse. So, we ended up on the following flow rates Qw = 8μl/min and Qo = 40μl/min , creating an emulsion with an estimated size distribution from 10μm to 100μm, as shown in Figure 9.