The basis of the prototype remains the same as the prototype 3.0- the bioreporter, a GMO bacteria expressing Green Fluorescent Protein is incubated with a water sample, where the fluorescence is detected optically and can be quantified in order to measure the concentration of Arsenic in the water. The main difference with this version of the prototype is that beneath the vial where the excitation LED is placed, we add a second red LED, which does not excite the fluorescent molecule, to try and measure the quantity of light that is scattered due to the bacteria. During the measurement of the fluorescence measurement we want to be able to assess the quantity of measured light that comes from the scattering of the blue light and not from fluorescence. By implementing this second LED we do not have to add a filter to the device to separate the fluorescent light.
The device contains:
- A vial containing the water sample we want to test is positioned on a socket through which a fluorescent excitation LED (blue 488nm for eGFP) and a scattering LED (red 600 nm) passes.
- GFP absorbs this blue light (λ=475 nm) and emits green light (λ=504 nm) which is detected by a photosensor (light to frequency converter) on which the light is concentrated with the help of two lenses that avoid loss of intensity.
- A red LED is placed in-line with the photosensor to measure the transmittance, which can be converted into turbidity. The measurement of turbidity will allow us to normalize our results with respect the density of bacterias, because fluorescence intensity is a function of the density of the bacteria.
- A light sensor to measure the quantity of fluorescence.
We can easily control this system with the help of a microcontroller board such as an Arduino and simple electronics. We use a simple program to control the device that follows the following steps. The program is initiated by the push of a button, the first three cycles last for the same predetermined amount of time. The last one lasts 300 ms. The predetermined amount of time can be anywhere between 20 and 180s, the optimal excitation duration is not yet evident for this prototype.
- The blue LED is turned on, to measure fluorescence.
- The red LED beneath the sample is turned on, to measure the effect of scattering.
- No lights are turned on to measure the random noise in the device. By doing a blank experiment we can assess the quantity of noise in the device and remove this step if the noise intensity is constant and low.
- The turbidity LED is turned on for 300 ms.
To run this prototype we need some electronics: The main parts of the device are listed below:
- An Arduino board
- Blue LED, to excite the fluorescent molecule. Shop
- 2 Red LED's, one for the scattering measurement the other for the turbidity measurement. Shop
- A light sensor, light to frequency converter TSL235R. Shop
- Transistors, to power the LED's. We need to use those because the output pins of the Arduino do not provide enough current. Shop
- Terminals, wires, resistors, buttons.
Step by step building
To build the device the steps are nearly identical to those followed while building prototype 4.0, the only different step being that there are two LED's beneath the vial.
Here is a representation of the circuit that is required to make the device work:
Proof of concept for the prototype
We are going to present a few result that helped us confirm that such a simple device could indeed be used to detect low levels of fluorescence in a liquid sample.
On the detection of fluorescence
The bioreporter that we use produces eGFP when in the presence of arsenic, but in low quantity. We have to know if the type of device we are using has the sensitivity for low levels of fluorescence or if the intrinsic noise of the device does not cover the fluorescent signal. To prove this we set up an experiment using fluorescently labeled Dextran molecules.
The dextran was diluted to concentrations that produce the same intensity of fluorescence the bioreporter produces in the presence of a concentration of 50µg/l of arsenic. The samples were then measured with the prototype.
We can clearly see that during excitation with the Blue LED increases the amount of light measured with the dextran concentration, and excitation with the red LED produces constant light intensity. Therefore we conclude that we can detect low fluorescence levels with this device and that there is no scattering in a dextran solution. This proves that this type of device can be used to measure fluorescent bioreporters.
Separating fluorescence and scattering
The LED we use to excite the fluorescent molecule will also produce scattering. Therefore during the fluorescence measurement the collected light comes from fluorescence and from scattering. To know how much fluorescence is measured we have to separate the two components of the light. This is why we use a second red LED that will tell us how much scattering is measured, and afterwards we can extrapolate the fluorescence measurement from the blue LED. Below a small drawing to illustrate the problem.
An important result was found by comparing our bioreporter bacteria (wich is not fluorescent in the absence of aresenic) with a bacterial solution that naturally produces eGFP. By measuring similar dilutions we could evaluate the effect of scattering in both solutions and check if we can detect a fluorescence signal in the fluorescent sample. We found some important results
- Bacterial density makes scattering vary linearly (as was expected).
- The linear relationship between scattering and density is very noisy, and we do not know why yet.
To extract the fluorescence part of the blue light excitation we measured the ratio between the red and the blue excitation in the non fluorescent sampl, a value that we called ∂NF (for non fluorescent scattering), in function of turbidity. This value was then used in the fluorescent samples to subtract the scattering part of the blue LED excitation measurement.
Bioassay with prototype 3.1
Prototype v3.1 was tested in a bioassay context, which means with arsenic. The bacteria were incubated with concentrations of arsenic from 0 to 100 µg/l.
The experiments was conducted with 5 samples per concentration, we therefore conclude that the measurements are very variable. but nonetheless the signal increases with the concentration of arsenic so it's a good sign. Unfortunately it's nowhere near enough for something that can be used. For a device to be approved by government officials, a positive measurement has to have a value that is more than 3 standard deviations away from the background signal which is not at all the case here.
To find these fluorescent values, the experiment were always run together with a set of blank samples. This means that the samples containing 0 µg/l of arsenic were used as the background signal level as these samples would contain no fluorescence at all (there would be some residual but none that interest us to find the arsenic concentration). The measurements from the blank samples were used to find the fluorescent part of the blue excitation in the other samples.
What interests us is to find a function that tells us how much blue scattering will be produced, according to the density of the sample. As we can see from the graph, the simple function that are found from the blank samples results very noisy measurements and are not sufficient for precise values. The noise of the values could come from:
- LED noise
- Instability of the scattering effect
We have found that the variability does not come from these sources.
- Instability of the bioreporter: Each experiment done in 96-well plate had results that was very stable between triplicates.
- Noise in the system: We measured the system in different settings (no vial, empty vial, vial with water) or different configurations (smaller aperture, larger aperture, more focusing of the light). These modifications changed a little bit the intensity of the background signal but not its variability, and we did not find an important improvement that reduced the noise considerably.
Adjusting the power of the LED's
To obtain the best ration between scattering and fluorescence we know that we had to adjust the intensity of the excitation sources. Our hypothesis is that the linear relationship between scattering-intensity and fluorescence-intensity is not the same and that there is an optimal light intensity where the signal-to-noise ratio is the highest. Unfortunately it was not possible to make extensive tests to get a definite answer.