Team:XMU-China/Project

Microfluidics (Opera Houses)

A Microfluidic array to cells is like an opera house to a glee that provides a cell culture environment and observation stage. To accompany the observation of microfluidic we used the fluorescence microscope, just like mass media to photograph the oscillation. We call them M & M. What a lovely couple.

Why M&M?

If we want to capture gene expression dynamics and distributions reported by fluorescence, we need to create a chemostat that provide a near-constant environment for the growth of bacteria. The near-constant environment mainly includes two aspects: continuous fresh media to ensure the optimum growth state and constant small populations of bacteria to ensure single-cell resolution in observation. Steady growth state is needed because the oscillation curve is expected to be a near simple harmonic one. Besides, bacteria tend to form biofilm which will lead to the overlapping of cells and create a false impression that the fluorescence increases, and that's why we have to prevent this to maintain a constant bacteria population with single-cell resolution.

What we mentioned above might have covered all aspects we need to consider for communication in a single colony, but if we still want to observe the coupling between colonies we have to create an array that contains many colonies and allows H₂O₂ to pass through freely.

Among many different techniques, microfluidic array seems to be our best choice with following advantages:
1) Precisely control of flowing rate and cell populations when equipped with syringe pumps;
2) Cells behavior including fluorescence strength can be captured by high resolution microscopy;
3) Trap numbers and dimensions can be adapted to different experimental need;
4) Productive: fewer resources and less time are required when compared with macroscopic cultures.

Design of Microfluidic Array

1.Pretests in "Swimming Pool"

Before our experiments on the microfluidic array we described before, a series of pretests were done on a chip whose structure looks like many swimming pool on the ground (Fig. 3-1).

Fig. 3-1 A "Swimming pool" chip in black field; those white squares (100μm×100μm×100μm) are chambers for bacteria

From the sketch (Fig. 3-2) we can see that the chamber is like a swimming pool where bacteria swim happily in the fresh LB that running from up above, and newly generated cells would be washed away.

Fig. 3-2 Diagram for swimming pool model's working mechanism

A "swimming pool" chamber model would have a lager AHL concentration when compared with a "passage" chamber because the former one has a minor dilution. According to a published research that the oscillation period will be shorter as the AHL concentration increases. So we assume that it will be more time-consuming to use this "swimming pool" model in pretests.

However, this chamber is about 100 μm deep and could grow E. coli far more than monolayer. As time goes by, the oscillation in a single colony signaling by AHL could never go back to the original baseline, and the oscillation curves have a tendency like the following graph (Fig. 3-3).

Fig. 3-3 Simulation curve for our oscillation in "Swimming Pool"

2.Advanced "Passage"

According to what we have discussed above, the microfluidic device should be able to provide flowing fresh media and maintain a constant bacteria population with single-cell resolution. Because "swimming pool" has a cell stacking problem, we designed a new passage-like device as shown below (Video 1).

Video 1

Channels and Chambers

This device is composed by two parts: the flowing channels and near 3,000 square trapping chambers, and their layout is like there are many passages with symmetry rooms distributed on both sides, so we call this array "passage" model. Firstly, cells are loaded from the cell port. When the loading stops, most cells would remain in channels, but if the loading cell density is dense enough, many cells would be pushed into chambers and form a monolayer of Escherichia coli growing in it. This is the time for running LB to get in. Fresh LB is loaded from the media port with Tween 20* and appropriate antibiotics. Flowing LB does not only permit a constant nutrition and exponentially growing cells for more than two days (that's the longest time we have tried), but also washes away newly generated cells to sustain a certain cell density, which is crucial for the determination of fluorescence strength.

Size

It has been stated before in the Mechanism part that most quorum sensing systems have a critical cell density requirement, so the trap size is designed neither too big nor too small to capture enough cells and allows for nutrients to permeate to every corner of the chamber.

You may also notice that the height of the chamber is extremely small, which is really demanding for the researcher's skills, but it's important for the observation. The oscillation could only be seen when the height of the chamber is in a particular range as our modeling predicted for many complicated reasons. For more details, please refer to the Modeling part.

Material

Polydimethylsiloxane (PDMS), a transparent polymer material, is chosen to build the devise because of its excellent breathability, which is necessary for the permeation of H₂O₂. Once H₂O₂ diffused outside cells, it can travel from colony to colony and synchronize them.

Our device is not only environmental friendly, but also a reusable one. After the flushing of a lysis buffer, most cells can be washed away, even those attached to walls. Then the device will be washed for three times by sterilized water before reuse.

Fig. 3-4 Microfluidic array- "Passage Model"

Image Capturing & Data analysis

The green fluorescence of bacteria in chambers is captured every 5 minutes by an inverted microscope (Eclipse Ti, Nikon) with a standard CCD camera manually and usually last more than 7 hrs. (This is really a big challenge for our team and very physically demanding since this is the only available fluorescence we could find. However, we managed to finish this data collecting process anyway. Good jobs, guys！)

Data are transferred into oscillation curves by ImageJ and software designed by our teammate for digital image batch processing. (More detail for our self-designed software, please refer to protocol.)

Fig. 3-4 Microfluidic array- "Passage Model"

Data are transferred into oscillation curves by ImageJ and software designed by our teammate for digital image batch processing. (More detail for our self-designed software, please refer to Software.

Next, we are going to Improvement (For a better Glee) to see how our little friends do on stage.

Mathematical model

We modeled the oscillate gene circuit (without ndh gene) based on the paper "A synchronized quorum of genetic clocks"^[12]. In that, the author characterized the oscillator by using the four delay differential equations (DDEs), which are the basic of our modeling. DDE can perfectly simulate our circuit due to its powerful function about solving the dynamic change in biological processes.

In the first two equations, item is a Hill function equaling to , which describes the delayed reactions about the complex processes in the cells (transcription, translation, maturation, etc.). This Hill function takes the history of the system into account, i.e. the concentration of AHL at the time it binds to LuxR to form the activation complex.
This model does not include an equation for LuxR assuming that it is constitutively produced at a constant level. All parameters are normalized for the convenience of analysis.

Modeling in MATLAB

We used MATLAB to help us solve the DDEs and draw the graphs. In addition, we developed a GUI user interface after finishing the main DDE code. The user interface makes people easier to modify and adjust the parameters without recoding the .m files and any MATLAB language knowledge. We called this software "Gene OS" which is short for Gene oscillate simulator.

After travelling a long way of circuit construction, you may wonder what we are going to do with those sparkling bacteria. Of course much thought was given, too, to make our oscillation circuit both showy and substantial as a safety guard in daily life. Here we are going to show you the potential of those shinny soldiers as biosensor, cancer cell killer and gene sinusoidal signal generator.

Biosensor

To date there appears a great number of biosensors that are used for detecting certain factors, such as small organic molecules or metal ions. And we've noticed that each year many iGEM teams relate their projects to biosensors. Biological detection has strength in its low detection limit, high sensitivity and bioaffinity when compared with chemical detection methods. We can use the same gene sensing circuit to detect various metal ions or other determinant via changing different promoters which responds to a specific target molecule.

Compared with traditional single-cell biosensor, our oscillatory system biosensor has three advantages.

1) Synchronized colonies will be more sensitive to concentration changes of the target molecule and the changes in fluorescence strength are easier to observe than single cell biosensors.

2) Single cell biosensor usually works, or "switches on" when the concentration of the specific molecules which can be responded by the promoter reaches to the lowest value. Sometimes this kind of detection is limited when we want to narrow the range of detection. Because the oscillation will happen only when concentration of certain molecule is between two value, the maximum and the minimum, it can dwindle the detection range of the determinant.

3) Oscillatory system is more robust and stable when used as a biosensor. These advantages will promote oscillatory gene circuits a better use in biosensing.

Jeff Hasty from UCSD used the same kind of oscillate circuit to engineer two kinds of arsenic-sensing macroscopic biosensors. These circuits are rewired by a native arsenite-responsive promoter, which is repressed by ArsR in the absence of arsenite. Due to the different mechanism, the sensor outputs are not the same. Figure 1 shows two constructed sensing modules and Figure 2 compares the outputs of two sensors^[1].

Figure 1.Two different constructed sensing modules.

Figure 2. Two different outputs of corresponding biosensors.

Due to the accuracy and specific condition in the field of bio-medicine, the oscillatory biosensor needs to meet the requirement such as synchronization. For example, all cells expected to twinkle at the same time only when cytotoxin existed. In addition, the period or amplitude of the oscillation needs to indicate the concentration of the cytotoxin. Our circuit has the potential to accomplish this kind of job. By further exploration in the relationship between oscillatory parameters such as period and amplitude with other environmental factors such as concentration of antibiotic or cytotoxin, we may create a oscillatory biosensor which can sense different kinds of biological toxin.

Cancer cell killer

Nowadays cancer becomes one of the greatest enemy of our human being. Our main methods used to diagnosis of cancer are based on tissue and organ level, such as MRI, CT, PET, etc. These are mainly physical diagnostic methods.

With the development of science and technology, we create various methods into molecule level concentrated on molecular recognition and signal transduction of tumor marker, such as AFP(alpha fetoprotein), CEA (carcino-embryonic antigen), PSA(Prostate Specific Antigen) and so on. Level of these special molecules in human bodies has become the main basis of diagnosis of cancer. These are mainly chemical diagnostic methods.

So how about using synthetic biology to do something for diagnosis of cancer? Our project can be a potential biological method to assist diagnosis of cancer.

ROS is reactive oxygen species in cells, which has been proved to be associated with several neurological disorders including Alzheimer's disease (AD) and Parkinson's diseases^[2][3][4]. Concentration of ROS indicates the level of cell metabolism. Normally the stronger cell' metabolism is, the more ROS it will produce. In the recent decade scientist found that the occurrence of cancer is closely related to the level of ROS. But it is not a simple positive correlation as we expect^[5].

Figure 4. ROS Level

From this picture we can get to know it clearly. As we know, we use oxide stress to describe the level of ROS in our cell. There is a threshold of oxide stress which is the critical value between normal cell and cancer cell. As a normal cell, its oxide stress is lower than the critical value while the ROS production value is over the threshold. On the one hand, when the oxide stress in our internal environment increases, normal cells may mutate to cancer cells if the ROS value is higher than the threshold value. On the other hand, cancer cells may die because of the accumulation of ROS. So the key point is, when ROS value higher than the maximum, we can kill the cancer cell while ROS value lower than the threshold we may push normal cells to change into cancer cells. So the level of ROS is in a specific range which we can measure using chemical methods.

Think about a proper range of ROS's concentration. Is it familiar to you as we mentioned a similar problem when talking about advantages of oscillatory system as biosensors? H₂O₂ is a kind of ROS. The oscillation will happen only when the concentration of is H₂O₂ between the maximum and the minimum. By regulating our circuits to let it respond in the proper concentration range which matches the ROS range of cancer cells, we can use the oscillatory system as a biological cancer detector.

In our experiments, we found that we can regulate flow rate and some parameters of microfluidic chip such as the depth of the well, the width of the channel and the concentration of H₂O₂ to change the period and amplitude of the oscillation. So by changing parameters of microfluidic chip, we can get a proper concentration range of H₂O₂ if the oscillation works. From then on, we can extract some cancer tissue to detect the ROS concentration in the microfluidic chips or get some suspected tissue on chips and diagnose whether it contains cancer cell.

2 Periodical Killer

2.1 Drug delivery

Our circuit also has a great potential application in periodic drug delivery. It is amazing when we realize that the oscillate circuit can control the drug to release in a specific period and the concentration of the drug is related to the amplitude. It is well-known that nowadays drug-release is a very hot research field. Many efforts are focusing on how to modify or delivery drug using materials such as synthetic polymers, liposome or nano shell which can respond to differences between different intracellular environments. Our oscillatory gene circuits provide a more biofriendly way for drug release. Our future plan is to make some deep exploration in this field.

2.2 Protein Synthesis

What else, a resonance may happen among the oscillators which results in a peak synthesis of protein under some conditions. This phenomenon can be used to increase the yield of some medical proteins such as antitoxic serum and antibody^[9][10].

2.3 Prevention pathogen

In the applications of controlling pathogens, quorum sensing system plays an important role in regulating growth and reputation of microorganism. So we could restrain quorum sensing system to control microorganism to make our life more healthy and nice. The most important part in the metabolism of pathogens controlled by quorum sensing system is the synthesis of toxin and the formation of membrane system. If we can restrain these two kinds of effect may we decrease some potential diseases resources and improve the effects of antitoxins. What's more, it is a particularly attractive point to prevent pathogens through restraining their quorum sensing system instead of killing them directly as it offers the prospect of combating effectively infections resistant to all existing drugs. As we all know, just treat diseases via using antibotics or antitoxin can help these pathogens improve their resistance to various drugs, which is very dangerous for us human being. To date, although we've found that different fungus form different signal molecules, the quorum sensing system almost takes part in cell morphological transformation and information communication. We think our oscillatory gene circuits provide a very useful and meaningful model to promote more and more research in this field^[11].

Gene sinusoidal signal generator

With the development of science and technology, computer had been here and there with larger RAM, faster CPU and smaller volume. All these changes owe to VLSI （Very Large Scale Integrated Circuites）, which consists of different electronic component.

Just image what if we can control cell as a biocomputer. Cellular computer is a device consists of different kinds genetically engineered bacteria, which can do the calculation, such as add and subtract like a normal computer. To finish such a device, scientists need various gene circuits to complete different operations. These circuits are just like electronic component working in an electronic computer.

At the beginning of this century, biocomputer is still a crazy dream far more than a possible blueprint. But now scientists have had the technology to make this "impossible mission" comes true. In recent months, scientists from Stanford constructed a new "bio-crystal valve" by DNA and RNA, which can do the logical operations in the living cells^[6].

Two months later, scientists from MIT used less than 3 gene devices to do the digital operation in division, logarithmic and square root. Thus, constructing gene circuits as electronic devices have become possible. From their research, their cellular calculator which uses analog computation has more advantages in computing efficiency in part count, speed and energy consumption compared with digital computation^[7].

Figure 1: Different kinds of log-domain analog computation.

Due to the fact that our circuits can generate a steady oscillation, it is possible that using our circuits as a sinusoidal signal generator.

Signal generators, such as function generator (including sinusoidal signal generator), are electronic devices that generate repeating or non-repeating electronic signals. As an important part of basic circuit studies, they are generally used in designing, testing, troubleshooting, and repairing electronic or electroacoustic devices.

What's more, Li Guang-bi from TUST reported that they have constructed two kinds of electronic circuits which well simulated the gene oscillator. This indicates that our circuit can develop to a new kind of gene sinusoidal signal generator^[8].

Figure 2.This picture shows a kind of electronic circuit generating the oscillation wave.

Figure 3.This graph shows the generated oscillation wave.

Future Version Trilogy

We blueprint our next step developing oscillatory gene circuits in following steps:

Step One

Complete our project by cleaning every obstacle on our way to oscillation.
Confirm that every part is functioning as expected;
Amplify the effect of different copy numbers;
Complete the comparison of different strains' effect on oscillation;
Complete the comparison of different EGFP & sfGFP effect on oscillation;
Confirm most suitable microfluidic parameters for oscillation;
Optimize our model

Step Two

We will make our oscillation a controllable one by integrating the catalase gene into the circuit.

Catalase is a common enzyme found in nearly all living organisms exposed to oxygen. It catalyzes the decomposition of hydrogen peroxide to water and oxygen. It is a very important enzyme in protecting the cell from oxidative damage by reactive oxygen species (ROS). Since our lab is studying Shewanella oneidensis, and its catalase protein has showed outstanding performance in catalyzing H₂O₂. So we thought we could integrate the catalase gene into our circuit and under an artificial inducible promoter, by which we can decide when to start this catalase gene and stop the oscillation.

Step Three

Realize everything we imagined in APPLICATION!

Make our circuit function as biosensor, cancer cell killer and gene sinusoidal signal generator.

Table 1. The Bio-bricks used in the study