Team:Uppsala/Project/Microfluidics

Microfluidics

Microfluidic chips are systems that consist of small channels where volumes on the scale of nano- or microliters flow and interact. We created an easy method for making these microfluidic chips and made chips capable of transforming cells through heat shock. Scroll down to learn more!


About Microfluidics

Microfluidics is both a science and a technology that is currently an active field of academic research and study. It consists of systems that work with small volumes of fluids in the nanoliter/microliter scale, through channels ranging from tens to hundreds of micrometers in diameter. Microfluidic devices have been readily used in chemistry and molecular biology and this provides a good base for further use in life sciences. There are two essential qualities in microfluidic devices that make them attractive to life science research. Firstly, the size of the device is small which makes them ideal platforms for point-of-care diagnostics that are portable. Secondly, their small size is also convenient since the volume of the liquids required to perform complex experiments is minimal, a property that can lower the cost of reactions. Thanks to fluidic components that approach the scale of a single cell, microfluidics in cell biology increases the throughput of biological and chemical assays. Small fluidic devices can perform a wide range of experimental designs and are also able to executed fully automated computer protocols. This reduces the human error and makes experiments highly efficient andreproducible. The advantages that this field brings to life science research are proof that microfluidics shows broad potential use in medicine as well. This summer we have created an easy, affordable and convenient fabrication method for microfluidic chips. We perfected our method and created two chips capable of carrying out transformation of E. coli.

What did we do this summer?

We created two microfluidic chips that are approximately the same size as a domino piece; one of them transforms cells through heat shock and the other one through electroporation. Our first step in fabricating a microfluidics device was to plan how it is was going to function. In this process the design of the micropattern is paramount. Generally you want the channels of the microchannels kept small and use low flow rates. These parameters together with the viscosity of the fluid combine into a dimensionless number called the Reynolds number. Microfluidic devices usually have flows with low Re numbers < 2000. When flows have low Re they are labeled laminar. For a laminar flow, properties change such as its bulk motion predictability and how species in the fluid are transported via convection/diffusion as opposed to a turbulent flow where such phenomena can be very difficult to predict. How laminar the flow in your chip will be is determined by the flow rate, the fluid's viscosity and the dimensions of the channels. When designing the channels it is important to think about the range of flow rates that can be used so that the flow inside the chip are within the laminar domain.

The plan for our two chips was to utilize small volumes to minimize the amount of reagents we would need. We also wanted to make the chip itself small to keep the amount of PDMS needed for curing down. The whole work of designing our lithographic molds started in AutoCAD (AutoCAD 2015 AutoDesk), a CAD software used for drawing 2D & 3D object with scientific accuracy. Although 3D modelling software such as Blender can be used these programs tend to be more difficult to work with as they are more focused on aesthetics instead of accurate angles and lengths.

Since our CAD drawings in the end was being 3D printed we needed to consider the resolution of the printer and how to best print our design. It is important that the mold is flat so that the cured PDMS chip is flat in order to achieve a good bonding between PDMS and whatever substrate used to seal the channels. We wanted to be able to easily pour the PDMS into the mold, simplifying handling of sticky uncured PDMS, and also easily remove the cured chip without the risk of tearing it. The design should also use as little printer resin as possible to keep costs and print time to a minimum. The 3D printer in the end also determines how small the microchannels can be made, in our case we chose the shortest out of plane height that the printer could print without large deviations in the actual printed height. With these requirements in mind we designed the following chips.

The design consists of 4 main parts. The bottom plate 33 mm L / 22 mm W, the wall structure 10 mm height / 3 mm thick, the separate detachable wall and the lithographic pattern, please see our designs for more info. The design makes it possible to reuse the box and draw new patterns depending on the application for the cured chip. The benefits of reusing the box design are to minimize future unforeseen problems concerning how well the PDMS is curing, removing of chip form the mold and the flatness of the fabricated PDMS chip, increasing reusability.

Figure 1: AutoCAD design of the heatshock chip.

We designed a pattern for a heat shock transformation chip. This pattern consisted of two larger channels that would warm the smaller channel in between by circulating hot water. The cells would be injected into the middle channel and left in the channel for 45 sec and then ejected. As with the electroporation chip the idea was that the heating chip design easily could be included with other types of structures on “lab in a chips”.

Figure 2: AutoCAD design electroporation chip

In our chip design for on chip electroporation we used the T-junction that combines two incoming flows connected to the pressure resistor, zigzag pattern, that evens out the pressure in the main channel, this helps generate more evenly sized droplets. Early on during planning we decided to incorporate a droplet forming section on the chip. There are many published articles that highlights the advantages of using droplets for better control of the content in the chip (Add references). By forming two phase droplets we would be able to separate the cell suspension into tiny fractions that can be manipulated independently. (Even though our currently presented prototype does not utilize the advantages of droplets, we still wanted to try and see if we could generate evenly sized and distributed droplets as this could be used for future designs. Although our droplet generation method did not allow for single cell manipulation at least it was possible to do “few cell manipulation”). After the droplets are generated they will pass two orthogonal channels that contain electrodes on either side of the main channel. The electric field between the electrodes would electroporate the passing cells inside the droplets. Since the distance between the electrodes are very small we can use a weaker electric field and control the electroporation pulse time on the cells by varying the flow rate. This design allows for continuous electroporation as opposed to current electroporators that usually only handle one cuvette at a time. We could with this design potentially electroporate how many cells we desired by automating the process and minimizing human handling.

3D-printing

To manufacture microfluidics chips you commonly use some sort of mold with channels and features on it, you then pour PDMS on this mold and bake the chip. To create these molds there are a variety of techniques but the most common ones include spin-coating or etching using electron/laser beams. The one thing these technologies have in common is that they are A: expensive and B: specialized. This means that labs which are not already using microfluidics are unlikely to have the machines and expertise to utilize them on hand. We had neither the machines, nor the expertise so we decided on a radical and little used approach for making our molds: 3D-printing!

We used the Form2 printer, which is a resin SLA printer from Formlabs, launched in 2015 as a followup from the Form1 printer. The Form2 printer was released in late 2015 and so is one of the latest models currently available. The Form2 prints using a photopolymer liquid resin which is a resin that hardens when exposed to light of certain frequencies. During the print the Form2 printer shines a laser which hardens resin selectively. This allows for the layer-wise printing of a complex 3D model. This is in contrast to many other 3D-printers which use extrusion for printing. The Form2 printer has a maximum print size of 145 × 145 × 175 mm.

Figure 1: The Form2 printer kit is sold commercially for about 4000 euro at the time of writing, which is expensive but not prohibitively so.

The reasons that we wanted to use the Form2 printer were primarily

  • Ease of use: The Form2 printer is quite easy to use, requiring almost no technical expertise beforehand. The only thing that is required is knowledge (or willingness to learn!) 3D-modelling in the CAD format. The printer is also easy to upload print jobs to using the Preform software.
  • Commercially available: The Form2 printer is already out on the market and can be bought by anyone, and although it is pricey it is not all too expensive for labs.
  • Precise: The print precision of the Form2 printer is very good, with a minimum layer size of 0.025 millimeters (25 micrometers) and the ability to create self-supporting structures at about the scale of 100 micrometers. It also allowed for selection of thicker layers if a faster, but less accurate print was needed.
  • Availability: A local lab at our university was already in possession of the Form2 printer, so we didn’t have to travel far or take contact with external organizations

Manufacturing

The production process for the Form2 prints consist of 5 basic steps.

  1. Import the 3D-shapes you want to print as an STL file to Preform and create a print job out of these, adding support structures as necessary.
  2. Make sure the printer has resin and the tank and build platform is clean. Upload the print job to the printer either via USB or via Wifi
  3. Let the printer run
  4. Pry off the finished prints from the build platform and wash the prints in isopropanol, 10 minutes in “dirty” isopropanol and then 10 minutes in clean isopropanol
  5. Cure the prints in UV-light for 4-8 hours or in daylight for several days.

In the end the Form2 printer worked quite well, although it did require a fair amount of manual maintenance, such as cleaning the resin tank and straining resin, and the prints were heavily dependent of the resin quality at the time of the print, which could decrease quite rapidly.

During the summer we printed 20-40 different moulds, during which we went through 1.5 bottle of resin. Most of the resin was not used, but we switched out the resin once the resin quality decreased too much, so that the resin could be strained to weed out smaller oligomers which form during the printing process. After straining the unused resin would have been reused.

The limitations of the 3D printer and its resin forced us to revise the AutoCAD/design files several times to better achieve our goals. Our early designs were flat, thin and quite large chips with design patterns embossed directly on the plates. To save resin we printed these directly, without including supporting structures. This led to a batch of moulds that warped and twisted alarmingly during the UV-curing process, which in turn meant that any PDMS chip that we made based on these had crooked surfaces and would not properly bond to glass slides using a plasma-bonding process. Furthermore we discovered that the chips became unnecessarily large and that we had no use for quite so much space on our chips. Our second round of designs were smaller, almost square and used a grid of supporting structures underneath to counteract the warping. These worked somewhat better but we still got some warped moulds and the supporting grids sometimes made the moulds hard to use during the chip baking and pouring process. In the final designs that we settled on we discarded the grids, thickened the bottom and added thick walls for support and to use during the chip-baking process, allowing us to save on our PDMS and dispense with using petri dishes as containers.

Fabrication process

Mixing of PDMS

First and foremost, all workspaces used for mixing of PDMS was covered in aluminium foil to protect them from untreated PDMS. The preparation of PDMS was carried out using unpowdered gloves due to the high risk of contamination. The PDMS for the chips was the SYLGARD® 184 Elastomer KIT. The base and the curing agent comes separately and mixing is required. The base and the curing agent were mixed in a falcon tube in the ratio of 10:1. The volume of PDMS for one chip is approximately 2.4 ml, thus about 2,2ml of the base and 0.22 ml of curing agent. The ratio 1:10 is standard but if a stiffer chip is preferred the amount of curing agent can be increased, for example a 10:2 ratio.

The PDMS was mixed thoroughly with a solid rod until the solution was full of bubbles. The solid rod should not be made of glass since they can easily break. For balance in the centrifuge, a second falcon tube was filled with water to the same weight as the tube with PDMS. Both tubes were centrifuged for 30 seconds at 3000g force.

The complete mould was prepared by taping the detachable wall to the rest of the mould. Roughly 2.4 ml of the mixture was poured into the mould until it covered the whole bottom and halfway up the walls. Any excess mixture was stored in a -20°C freezer for a couple of days and used for other chips.

Degassing

The PDMS was degassed to get rid of the bubbles. Depending on the amount of bubbles this was done using a vacuum chamber, a fridge or both ways simultaneously. Construction advice for the vacuum chamber can be found in the manual. When using the vacuum chamber the mould with PDMS was put in the chamber and connected to a vacuum pump. The mould or the whole vacuum chamber was at times put in the fridge overnight. Any excess topmost bubbles were occasionally burst with nitrogen gas.

Baking

The baking of the PDMS was done in several ways. During degassing the oven was preheated and a petri dish was prepared to stabilize the mould in the oven. If the chip was baked at 100°C tin foil was used instead of a petri dish made of plastic, since it melts at 100°C. Depending of the amount of curing agent that had been used the PDMS cured at different times and temperatures. The initial designs were baked in the oven for 2 hours at 100°C. The later chips were baked for three hours at 80°C. If the PDMS was not cured after three hours the mould was left longer in the oven, for instance overnight at a lower temperature.

Taking out the chip & making holes

Once the chip was baked the tape was removed and the chip was cut out from the mould with a scalpel. A hole maker in form of a needle with a blunt end was attached to a syringe, and the syringe was thereafter filled with air. The needle was compatible with the tubing connectors for the chip. The holes were created by sticking the sharpened, blunt needle into the PDMS all the way to the other side. Air was blown out of the syringe with force in order to take out the little piece of gel inside of the created hole. The needle was then taken out from the PDMS and the procedure was repeated for all six inlet and outlet holes in the design.

Assembly & cleaning

After baking the PDMS was cleaned. The side of the chip that contains the channels was covered with scotch tape to get rid of dust. The tape was taken away straight away and the chip was washed in isopropanol.

Two 4x6 centimeter glass slides were made using a saw. The glass slides were prepared by making holes compatible with the holes on the PDMS with a drill. 6 additional holes were also made along the side of the slides. Furthermore, the glass slides were washed in isopropanol and dried with nitrogen together with the PDMS. The two slides were put together with m3 screws in the additional holes and adding the PDMS in the middle. Lastly, the tubing connectors were inserted to the inlet and outlet holes on the chip.

Results

Heat Shock Chip

During the summer we made over 30 chips and numerous successful heat shock transformations using only 8.4 ng of DNA and 5.9 μL of competent cells.

Bacterial growth was seen on nearly all of the chip transformed plates. In other words, the transformation worked on our chip and it greatly reduced the amount of reagents needed since colony growth was observed on plates were only 6mL of cell/DNA suspension was heat shocked. The number of colonies for each trial was added by counting all of the five plates for each trial as one replicate. That left three replicates for each chip. Calculations of colony forming units per microliter DNA were made and the mean for each chip is shown in Figure 1. We found that the transformation efficiency was much higher in the conventional heat shock than on our chip.

The plates with bacteria transformed on the chip showed many colonies, considering the small amount of cells and DNA that were plated. A large variation in number of colonies between different trials was observed. That is, two transformations carried out on the chip in the same manner could yield a varying number of colonies.

The cleanliness of our chip was good as little to no colonies grew on the plates with only SOB run through the chip nor on the plates with only cells (negative control).

Figure 1. Mean transformation efficiencies of the different chips (1-3) and conventional transformations (0). Transformation efficiencies calculated by: # colonies on plate/g of DNA plated. Error bars show standard deviation, n=3.



Figure 2. Plates after transformation with chip. Two different trials show the variation in efficiency. Variation could be due to human error.
Raw data

Raw data

Here the raw data of every transformation is presented alongside the comparison transformations we did.

Table 1.
Trial. 34567891011
Chip model123
No. of colonies9641096505911107455857711029
Transformation efficiency (cfu/ng DNA)34.939.618.23338.81.993.0727.937.2


Table 2.
DNA concentration(ng/µl) 4747470.56
No. of Colonies>10001707264
Transformation efficiency (cfu/ng DNA)42.5304129114

According to the results presented in Table 1 and Table 2 the heat shock chip did not reach higher efficiency than conventional methods, at least with the comparison metrics used. There was also very high variation in the transformation efficiency among the experiments done on on the chip(which were done using similar/identical reagents). The efficiency varied from 1,99 in experiment number 8 up to 39,6 in experiment number 4, which is a 20-fold variation in efficiency. On a per plate basis the highest efficiency we had was 123 000 and the lowest was 0 (8 out of 50 plates had 0 colonies), with the average transformation efficiency being 36.

During the experiments we could identify several potential causes which could cause this large variation of results. The setup of the chip involved running heated water from a beaker through the chip and this means that there can be a chance that there is higher or lower thermal leakage than expected, leading to the transformation efficiency being affected. Also we were not sure whether heat shocking at 42 degrees for 35-40 seconds is the optimal way to do it. Since the heat is conducted much better and faster through the chip from the heating channels and because the cells are in a much smaller volume of media the cells are heated more rapidly maybe a lower/higher temperature should be used a shorter/longer time. Furthermore we noticed that there was a very large variance due to the human factor(i.e. who was actually conducting the experiment) and that experience in the procedure played a large role in how successful the transformation was. This was likely due to the primitive and direct way cells were fed into and out of the chip (using a syringe to push in/out cells and eye-measuring to see whether cells have entered the chip makes for inaccurate timing and a lot of back and forth).


Results for the temperature comparisons

For each transformation at a set temperature the amount of formed colonies can be viewed in figure 3. Figure 3 shows the data points for the triplicates at each temperature having a large spread of the amount of formed colonies. Only the transformations at temp. 55 ◦C showed formed colonies for each of the three replicates.

Figure 3. Number of colonies formed after transformation at each temperature. The crosses represent each replicate, the circle is a contamination check, and the rhombs are the average amount of colonies at each temperature.



The highest average value for formed colonies was at 55 ◦C for the heating fluid, giving 30 colonies whilst heat shocking at 65 ◦C and 75 ◦C yielded 21 and 14.7 colonies. The average (avg) number of colonies at each temperature drops slightly when going from 55 ◦C, 65 ◦C to 75 ◦C since the latter two contain replicates with zero number of colonies.

The avg value for the number of colonies at each temperature was used to determine the average transformation efficiency, represented by the cfu/µg value and is presented in Table 3.

Table 3. Average transformation efficiencies at different temperatures.
Temperature (°C)Transformation efficiency (cfu/ng DNA)
554.11
652.88
752.01


Our designs

Since we are competing for the Design Award within iGEM, we have a more detailed description of how our designs were produced and the final rationalization and measurements for our designs in the Design page.