Introduction

Our pathogen detection approach relies on Cas13a digesting RNA. A change in fluorescence is the most convenient way to take trace of such enzyme kinetics. This is impractical for in-field applications because commercial fluorescence detectors are expensive and inconveniently large. A straight forward approach for making our pathogen detection fit for in-field applications is a cheap and handy fluorescence detector. Other iGEM teams already constructed fluorescence detectors. We therefore rated detectors of other teams in a cost vs sensitivity diagram. Nevertheless non of them had a high enough sensitivity or the ability to measured fluorescence quantitatively. We therefore constructed a detector matching our requirements. Our detector is paper-based and can detect fluorescein concentration down to 200 nM. The detector is able measure to automatically measure fluorescence in units of equivalent fluorescein concentrations. The detector is very compact and costs less than 15 EUR. We were able to measure a time trace of Cas13a digesting RNaseAlert with our detector. For comparison we also measured a positive control containing RNase A and a negative control containing only RNaseAlert. The data are displayed in image [img:k3]. The time traces show an enzymatic reaction taking place on filter paper. This proves that our detector is sensitive enough and meets our requirements. However the detector is not limited to the detection of our specific application. It can be used for the detection of any fluorescence signal in biological or chemical systems. We think that our detector can benefit other iGEM teams and also research groups that want to make a fluorescence based detection fit for in-field applications.

Time trace of Cas13a digesting RNaseAlert. Positive control containing RNaseA and RNaseAlert. Negative control containing only RNaseAlert. Shaded areas represent uncertainties for data points with the corresponding color

Principle design

Light from a blue LED is filtered by the blue filter foil. The light from the LED excites fluorophores on a filter paper. The excitation light is blocked by an orange filter foil. The fluorescence light from the excited fluoroscopes passes through the orange filter foil and illuminates and light dependent resistor (LDR). The LDR changes its resistance corresponding to the degree of illumination with fluorescence light. An Arduino nano measures the resistance via a voltage divider and calculates the fluorophore concentration.

Components

Micro controller

We used an Arduino nano for automatized data collection. This micro controller has analogue pins that can measure voltages from 0 to 5 V. The output is an integer from 0 to 1023 corresponding to the voltage at the analogue pin. The micro controller is connected via an USB port with a computer where the data can be processed further.

Light dependent resistor (LDR)

For the detection of fluorescence light we used an light depending resistor (LDR). An LDR is a resistor which decreases its resistance $R_{LDR}$ with increasing light intensity $I$ . The dependence of the resistance $R_{LDR}$ on the light intensity $I$ is $\label{eq:k1} R_{LDR} \propto I^{-\gamma},$ where $\gamma$ is a parameter depending on the type of resistor being used and can even differ for LDRs with the same type designation.

Equation [eq:k1] is motivated from the equation $\label{eq:k2} \gamma = \frac{\mathrm{lg}\left(\frac{R_{10}}{R_{100}}\right)}{\mathrm{lg}\left(\frac{100~\mathrm{Lx}}{10~\mathrm{Lx}}\right)},$ which is given in the data sheet of the LDR. The dominator is the decadic logarithm of the fraction of two light intensities of 100 Lx and 10 Lx. $R_{10}$ and $R_{100}$ are the corresponding resistances at these light intensities. The used resistor with the type designation GL5516 NT00183 has a parameter $\gamma$ of 0.8.

The response of a LDR depends on the wavelength $\lambda$ of the incoming light. The data sheet provides information on the relative response normalized to the maximal response. The relative response is maximum for a wavelength of 540 nm and is therefore appropriate for detection of green fluorophores.

Circuit for resistance measurements

A voltage divider is the simplest way to measure resistance. Image shows the circuit of a voltage divider. Applying Kirchhoff’s laws we get $\label{eq:k20} \frac{R_{LDR}}{R_{ref}}=\frac{U_{LDR}}{U_{ref}}$ and $\label{eq:k21} U_0= U_{LDR}+U_{ref}.$ $R_{LDR}$ and $U_{LDR}$ is the resistance and voltage drop at the LDR. $R_{ref}$ and $U_{ref}$ is the resistance and voltage of a reference Resistor. $U_{0}$ is the supply voltage which we choose to be $5\mathrm{V}$ . This gives $\label{eq:k22} R_{LDR}=R_{ref}\cdot\frac{1}{\frac{U_0}{U_{LDR}}-1},$ an equation to calculate the resistance $R_{LDR}$ of the LDR from the voltage drop $U_{LDR}$ at the LDR. $U_{LDR}$ can be measured with the micro controller.

We need to find an equation to choose an optimal resistor $R_{ref}$ . We want an maximal change of $U_{LDR}$ for a certain detection range of $R_{LDR}$ . Therefore equation [eq:k22] is solved for $U_{LDR}$ giving $\label{eq:k30} U_{LDR} = \frac{U_0}{1+\frac{R_{ref}}{R_{LDR}}}.$ The change $\Delta U_{LDR}$ of $U_{LDR}$ between a maximum value $R_{max}$ and a minimal value $R_{min}$ of $R_{LDR}$ is $\label{eq:k31} \Delta U_{LdR}=U_0 \cdot \left(\frac{1}{1+\frac{R_{ref}}{R_{max}}}-\frac{1}{1+\frac{R_{ref}}{R_{min}}} \right),$ which has a maximum for $\label{eq:k23} R_{ref}= \sqrt{R_{max}\cdot R_{min}}.$ We expect a $R_{max}$ of approximately $2~\mathrm{M\Omega}$ and a $R_{min}$ of approximately $1~\mathrm{M\Omega}$ . By using equation [eq:k23] as a guideline we choose $R_{ref}$ to be $1.5\mathrm{M\Omega}$

Light emitting diode (LED)

To detect green fluorophores we choose a blue LED with peak emission at 470 nm. For optimum performance we choose the brightest LED we could find. The LED used has a luminous intensity of 12 cd, a maximum current of 20 mA and a forward voltage of 3.2 V

NPN transistor

To provide a stable illumination, it is crucial to supply the LED with a constant voltage. We therefore control it via a NPN transistor with type designation BC635. Its base-emitter-on voltage is 2 V.

Control circuit for the LED

The digital output pin is connected via a voltage divider to the base of the transistor. The voltage divider consist of the resistor $R_1$ with a resistance of $\mathrm{1~k\Omega}$ and the resistor $R_2$ with a resistance of $\mathrm{9.1~k\Omega}$ . When the output pin is set to 5 V a voltage of 4.5 V is present at the base of the transistor. This is above the base-emitter-on voltage and the LED is turned on. The resistor $R_3$ with resistance $39\mathrm{~\Omega}$ was chosen empirically to limit the LEDs working current and to power it at maximum brightness.

Light filter foils

For filtering the emission and excitation light we used filter foils from LEE filters. The producer provides transmission spectra for every filter foil. For the excitation filter we choose the filter colour “TOKYO BLUE” and for the emission filter we choose the filter colour “RUST”. We took care that the transmission spectra of the filters have no overlap. We measured the transmissions spectra of the foils our self. The data are shown in image [img:k2]. We also measured the transmission spectra of two blue filter foils combined with one orange filter foil. For this combination there is nearly no transmission up to a wavelength of 700 nm. This combination is sufficient for blocking the excitation light of the blue LED from reaching the LDR.

emission spectra of light filter foils

Filter paper

We choose glass fiber filter paper from Whatman Laboratory Products with type designation “934-AH” to detect fluorescence on. Because normal cellulose filter paper is auto fluorescent and causes a high background signal.

3D printed parts

We intended to have the LED,the fluorescence sample and the LDR in direct proximity. This ensures a maximum use excitation light from the LED and emission light from the fluorophores. We therefore choose a sandwich-like design for our sample holder that can be placed into a slot were a detection system snaps in keeps the sample in position.

In detail, two blue filter foils are stacked and glued with tape in front of two excitation windows of the upper halve of the sandwich. One orange filter foil is glued in front of the two detection windows on the lower half of the sandwich. The two detection and emission windows enable us to measure a blank sample and an actual sample with the exact same set-up. Care needs to be taken that no tape is covering the detection or excitation windows because tape is usually auto fluorescent and causes a high background signal. A piece of filter paper is placed between the two halves of the sandwich. The upper and lower part of this sandwich and pressed together with magnets and hold the filter paper in position. The magnets ensure an user-friendly exchange of filter papers. The sandwich can now be inserted into the slot of the detection device. The LED and the LDR are mounted onto beams at opposite sites of the detection device. The sandwich snaps in when the LED and the LDR are at the right position under the excitation and over detection window. Four magnets apply an additional force to the beams and press the LED and the LDR close together. The design ensures that the distance between filter paper, LED and LDR is only limited by the thickness of the filter foils.

List of materials and cost calculation

List of materials
used item	price of components per detector in EUR
10 Magnets	4,8
8 Resistors	0,48
2 NPN Transistors	0,12
Green LED	0,1
Blue LED	0,1
Geekcreit ATmega328P Arduino Kompatibel Nano	2,51
LDR	0,15
USB cable	1,5
Hole raster board	0,42
ca 100 g PLA	3
M3 screws	0,01
M3 nut	0,01
Filter foils	0,1
Total cost	13,3

[tab:k20]

We keept an eye on using only low cost and easy available items for the construction of our detector.

Calibration

Derivation of a calibration function

We want to find an equation that relates fluorophore concentration $c$ and the corresponding resistance $R$ of the LDR.

The light intensity I at the LDR during a fluorescence measurement is a sum of signal intensity $I_s$ and background intensity $I_b$ : $\label{eq:lk1} I = I_s + I_b$

We assumed that $I_b$ is the intensity for a water sample. Importantly, wet samples give different background signal than dry ones, suspectedly due to different light scattering in the filter paper. With equation [eq:k1] this gives an equation for the resistance $R_b$ of a water sample and for the resistance $R_{LDR}$ for a fluorescence sample: $\begin{aligned} \label{eq:k11} R_{b} \propto & I_{b}^{-\gamma}\\ \label{eq:k12} R_{LDR} \propto & \left( I_{b}+ I_{s} \right)^{-\gamma}\end{aligned}$ Equation [eq:k12] is normalizes by equation [eq:k11]: $\label{eq:k13} \frac{R_{LDR}}{R_{b}}=\left(\frac{I_{b}+ I_{s}}{ I_{b}}\right)^{-\gamma}=\left(1+\frac{I_{s}}{I_{b}}\right)^{-\gamma}$

We assumed that the intensity $I_{s}$ of fluorescence light depends linearly on the concentration $c$ of fluorophores and the light intensity $I_{0}$ produced by the LED: $\label{eq:k15} I_{s}\propto I_{0}\cdot c$ And we assumed that the background intensity $I_{b}$ depends also linearly on $I_{0}$ . $\label{eq:k16} I_{b} \propto I_{0}$

The fraction $\frac{I_{s}}{I_{b}}$ does not depend on the intensity $I_{0}$ of the LED and depends linearly on the concentration $c$ . $~\label{eq:k17} \frac{I_{s}}{I_{b}} = k\cdot c$

$k$ is a constant that depends on the transmission spectra of the filter foils, the spectra of the fluorophore, the spectra of the LED and light scattering effects of the filter paper. $k$ can be assumed to be constant for one specific measurement set up. Equation [eq:k17] inserted into equation [eq:k13] gives the final equation relating fluorophore concentration $c$ and the resistance $R_{LDR}$ of the LDR. $\label{eq:k18} \frac{R_{LDR}}{R_{b}}=\left(1+\ k\cdot c\right)^{-\gamma}$

Determination of the calibration parameter k

We want to find a value for $k$ from equation [eq:k18]. To determine that $k$ does not depend on the intensity $I_0$ we made two measurement series. We changed the resistance $R_3$ in series to the LED to change the light Intensity $I_0$ . We used a $39~\mathrm{\Omega}$ resistor and a $60~\mathrm{\Omega}$ resistor.

We prepared a dilution series of fluorescein by 10-fold dilution from $100 \mathrm{nM}$ to $1\mathrm{mM}$ . For each measurement we pipetted 30 $\mathrm{\mu l}$ of sample on a fresh filter paper, placed it in the detector and turned on the LED. We measured the resistance $R_{LDR}$ directly with a multimeter. For the set-up with the $39~\mathrm{\Omega}$ resistor we additionally measured a $200 \mathrm{nM}$ sample because this is the expected final working condition of the detector. After each measurement the detector was cleaned gently with ethanol. The first and last measurement of each series was conducted with plain water to determine $R_b$ and to confirm the absence of contaminations. The measured resistances can be seen in table [tab:k1]. We normalized every resistance of one measurement series with the resistance $R_b$ for plain water. A plot of the normalized resistances is shown in image [gra:k1]. We fitted the data with equation [eq:k18] to determine a value for k . For the fit we used an value of 0.8 for $\gamma$ . The fitted value for k with absolute uncertainty is $k= (3.758 \pm 0.133)\frac{1}{\mathrm{\mu M}}$ and the relative uncertainty $\sigma_k$ is therefore $0.035$

$normalized resistances \frac{R_{LDR}}{R_b} vs fluorescein concentration c and corresponding fit function$

normalized resistances $\frac{R_{LDR}}{R_b}$ vs fluorescein concentration c and corresponding fit function

Calibration measurement
Concentration in $\mathrm{\mu M}$	Plain water	0.1	0.2	1	10	100	1000
$R_{LDR}$ in $\mathrm{k\Omega}$ for bright LED	1200	960	748	340	58.0	10.2	2.4
$R_{LDR}$ in $\mathrm{k\Omega}$ for dimmed LED	2050	1620		529	80.6	12.9	3.1

[tab:k1]

Read out

Data collecting and processing with the micro controller

We want to verify if a reaction with a fluorescence product is taking place on a filter paper. Therefore we need to measure the resistance $R_{LDR}$ of the LDR over a certain time certain period. We choose to measure a data point every 5 min over a time period of 1 h to take trace of the reaction kinesics. Before such a measurement series a blank measurement with plain water is performed to determine the resistance $R_b$ .

As a first step the exact value of the supply voltage $U_0$ needs to be measured. The supply voltage is connected via an additional voltage divider with an analogue pin. The voltage divider consists of a $100~\mathrm{\Omega}$ resistor and a $910\mathrm{k\Omega}$ resistor. The analogue pin measures still a correct value for the supply voltage $U_0$ because the first resistor is negligible small compared to the second resistor. We did not connect the analogue pin directly with the power supply to prevent the micro controller from damage in case of a short circuit or a peak voltage caused by an other component of the overall device. The micro controller measures the supply voltage 50 times with a delay time of 50 ms between measurements. It calculates the average of $U_0$ and the empirical standard deviation $\sigma_{U_0}$ in percent, which is used as measurement uncertainty for further calculations.

To initiate a resistance measurement of the LDR the LED needs to be turned on by setting a digital output pin to 5 V. An additional green LED on the device is turned on as well to indicate that a measurement is taking place. After a waiting time of 30 s the actual measurement starts. This waiting time was determined empirically and is required because of the slow response of the LDR. $U_{LDR}$ is measured as $U_0$ , 50 times with a delay time of 50ms between measurements. The average of $U_{LDR}$ and the relative empirical standard deviation $\sigma_{U_{LDR}}$ are calculated. Equation [eq:k22] is used to calculate $R_{LDR}$ from the average of $U_{LDR}$ . We derived an equation for the propagation of the relative systematic and the relative statistic uncertainty of $U_0$ and $U_{LDR}$ . For the relative statistic uncertainty $\sigma_{stat}$ of $R_{LDR}$ we get $\sigma_{stat}=\frac{R_{LDR}}{R_{ref}}\cdot \frac{U_0}{U_{LDR}}\cdot \sqrt{\sigma_{U_0}^2+\sigma_{U_{LDR}}^2}.$ We used an value of 1 digit for the absolute systematic uncertainty for a voltage measurement. The relative systematic uncertainty is $\frac{1}{U}$ for a measured voltage U. For the relative systematic uncertainty $\sigma_{sys}$ of $R_{LDR}$ we get therefore $\sigma_{sys}=\frac{R_{LDR}}{R_{ref}}\cdot \frac{U_0}{U_{LDR}}\cdot \left(\frac{1}{U_{LDR}}+\frac{1}{U_{0}}\right).$ The equation for the total uncertainty $\sigma_{R_{LDR}}$ is then $\sigma_{R_{LDR}}=\sigma_{sys}+\sigma_{stat} = \frac{R_{LDR}}{R_{ref}}\cdot \frac{U_0}{U_{LDR}}\cdot \left(\sqrt{\sigma_{U_0}^2+\sigma_{U_{LDR}}^2}+\frac{1}{U_{LDR}}+\frac{1}{U_{0}}\right).$

$R_{LDR}$ , $R_b$ and their uncertainties are read by the computer and saved for further analysis in a text file.

Data analysis and final result

With equation [eq:k13] and the fitted value for k the measured resistances can be translated into fluorescein concentrations $c$ . Equation [eq:k13] solved for $c$ is

$\label{eq:k41} c= \frac{\left( \frac{R_b}{R_{LDR}} \right)^{\frac{1}{\gamma}}-1}{k}.$

The equation for the absolute uncertainty $\sigma_c$ of the fluorescein concentration $c$ is $\label{eq:k42} \sigma_c = \sqrt{\left(\frac{R_b}{R_{LDR}}\right)^{\frac{2}{\gamma}}\cdot\frac{1}{k^2}\cdot \left(\sigma_{R_b}^2+\sigma_{R_{LDR}}^2\right)+c^2\sigma_k^2}.$ We are now enabled to measure fluorescence in units of equivalent fluorescein concentrations $c$ .

As a first proof of principle we measured a kinetic curve of Cas13a digesting RNaseAlert. For a negative control we measured a kinetic curve for Cas13a without target RNA. and for a positive control we digested RNaseAlert with RNase A. We analysed the data with equation [eq:k41] and [eq:k42]. Image [img:k4] shows fluorescence in equivalent fluorescein concentrations vs time. The time traces show an enzymatic reaction taking place on filter paper. This proves that our detector is sensitive enough for our application.

Time trace of Cas13a digesting RNaseAlert. Positive control containing RNaseA and RNaseAlert. Negativ control containing only RNaseAlert. Shaded areas represent uncertainties for data points with the corresponding color

CascAID

Our project, which we named Cas13a controlled assay for infectious diseases (CascAID), features the recently identified CRISPR/Cas effector Cas13a¹⁰. Unlike other proteins in the familiy, Cas13a has the unique ability to bind and cleave specific RNA targets rather than DNA ones. Moreover, after cleaving its target, Cas13a is able to unspecifically cleave RNA molecules. By using this collateral activity from Cas13a, our system is capable of detecting virtually any RNA target. This is done by changing the crRNA in the protein, that is a short RNA sequence that determines what is recognized as target.

Cas13a binds specific target RNA depending on the crRNA sequence. After activation, Cas13a cleaves RNA indiscriminately.

We wanted to start our project by showing that Cas13a's collateral activity could be used to detect the presence of specific RNA. For this, we used the RNAse alert system, as done in a recent publication¹¹, to detect RNA digestion. In this assay, the presence of RNAse-like activity is detected by an increase in green fluorescence. Our experiments yielded a convincing proof-of-principle which we went on to model. Moreover, CascAID can be used to detect a wide spectrum of pathogens, as our experiments with gram-positive and viral targets suggested. As we wanted to make CascAID available for everyone, we focused on building an inexpensive fluorescence detector to measure the presence of the target. Our detector “Lightbringer” was designed to be able to detect the fluorescence produced by the fluorescein in the Rnase alert system¹², but we theorize that changing the filters allows detection of other fluorophores. In addition, we experimented with freeze-drying on paper to make CascAID durable and easily portable.

Cas13a can be used to detect specific RNA sequences

Picture of the Thermocycler

For RNA extraction from the samples we tested three methods: extraction with silica beads, extraction with silica membrane and heat lysis. We custom-built an affordable thermocycler for signal amplification by RT-PCR to improve the detection limit. We explored recombinase polymerase amplification (RPA), an isothermal amplification procedure, to use over more conventional PCR methods as its simplicity makes it the more attractive option.

Colorimetric read-outs

To couple CascAID with an easy read-out method we explored three colorimetric read-outs:

AeBlue: The RNA strand in a specially designed RNA/DNA dimer is cut by Cas13a's collateral activity. After digestion, the interaction between the two strands is too weak to hold the dimer and it decays. We can then use the DNA-strand as template to translate the chromoprotein aeBlue.

Diagram of aeBlue

Intein-Extein: By binding TEV-protease with a RNA-linker we can use Cas13a's collateral activity to regulate the protease's diffusion and use it to cleave a TEV tag separating the intein regions of a modified chromophore. After the first cleavage, the intein segment excises itself¹³, bringing together the halves of the chromophore. Only then is the chromophore functional and produces the colorimetric read-out.

Diagram of Intein-Extein

Gold nanoparticles: Gold nanoparticles coated with short DNA sequences are held closely together by a complementary linker RNA, which makes the solution intense blue¹⁴. Activated Cas13a cuts the linker RNA, causing the nanoparticles to diffuse away from each other. This increase in distance causes a color change to intense red.

Gold nanoparticles

Software

To help facilitate the design of crRNA, the sequences that give CascAID its specificity, we developed a software tool that checks crRNA for unwanted secondary structures. This gives valuable insight on whether the sequence is suited to use with Cas13a or whether some modifications are needed. Together with Team Delft's software tool which designs the corresponding crRNA based on the target, we collaborated to develop a powerful tool that suggests crRNA sequences and checks their usability

References

Cohen, Limor, and David R. Walt. "Single-Molecule Arrays for Protein and Nucleic Acid Analysis." Annual Review of Analytical Chemistry 0 (2017).
Nakano, Michihiko, et al. "Single-molecule PCR using water-in-oil emulsion." Journal of biotechnology 102.2 (2003): 117-124.
Taniguchi, Yuichi, et al. "Quantifying E. coli proteome and transcriptome with single-molecule sensitivity in single cells." science 329.5991 (2010): 533-538.
Rissin, David M., et al. "Single-molecule enzyme-linked immunosorbent assay detects serum proteins at subfemtomolar concentrations." Nature biotechnology 28.6 (2010): 595-599.
Pardee, Keith, et al. "Rapid, low-cost detection of Zika virus using programmable biomolecular components." Cell 165.5 (2016): 1255-1266.
Slomovic, Shimyn, Keith Pardee, and James J. Collins. "Synthetic biology devices for in vitro and in vivo diagnostics." Proceedings of the National Academy of Sciences 112.47 (2015): 14429-14435.
Tang, Ruihua, et al. "A fully disposable and integrated paper-based device for nucleic acid extraction, amplification and detection." Lab on a Chip 17.7 (2017): 1270-1279.
Vashist, Sandeep Kumar, et al. "Emerging technologies for next-generation point-of-care testing." Trends in biotechnology 33.11 (2015): 692-705.
Gubala, Vladimir, et al. "Point of care diagnostics: status and future." Analytical chemistry 84.2 (2011): 487-515.
Abudayyeh, Omar O., et al. "C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector." Science 353.6299 (2016): aaf5573.
Gootenberg, Jonathan S., et al. "Nucleic acid detection with CRISPR-Cas13a/C2c2." Science (2017): eaam9321.
https://www.idtdna.com/pages/docs/technical-reports/in_vitro_nuclease_detectionD325FDB69855.pdf (retrieved: 13.10.17)
Anraku, Yasuhiro, Ryuta Mizutani, and Yoshinori Satow. "Protein splicing: its discovery and structural insight into novel chemical mechanisms." IUBMB life 57.8 (2005): 563-574.
Link, Stephan, and Mostafa A. El-Sayed. "Size and temperature dependence of the plasmon absorption of colloidal gold nanoparticles." The Journal of Physical Chemistry B 103.21 (1999): 4212-4217.

Team:Munich/Hardware