Team:EPF Lausanne/Notebook/Microfluidics


Making 768 MITOMI chips



1       Material and methods

MITOMI Device Fabrication by Multilayer Soft Lithography

  1. PDMS resin: Heat curable silicone elastomer (Dow Corning Sylgard 184).
  2. Trimethylchlorosilane (TMCS) (Sigma-Aldrich).
  3. Mixing and degassing of PDMS: Thinky Mixer ARE-250 equipped with adaptor for 100 ml disposable PP beakers (C3 Prozess- und Analysentechnik GmbH).
  4. Degassing of PDMS control layer: Vacuum desiccator (Fisher Scientific AG).
  5. Spin coating of PDMS flow layer: Programmable spin coater SCS P6700 (Specialty Coating Systems Inc.).
  6. Stereomicroscope, SMZ1500 (Nikon AG).
  7. Manual hole punching machine and pin vises, 21 gauge (0.04’’ OD) (Technical Innovations, Inc.)



  1. The control layer mold is placed in a glass Petri dish lined with aluminum foil to facilitate easy removal. Care must be taken that the aluminum foil lining does not contain any holes.
  2. To generate a hydrophobic surface, both flow and control mold are exposed to vapor deposits of TMCS for 10 min by placing them into a sealable plastic container with 1 ml TMCS filled into a plastic cap.
  3. For the control layer, 30 g (60 g) of a 5:1 Sylgard mixture (25 g Part A:5 g Part B) (50 g Part A:10 g Part B) is prepared, mixed for 1 min at 2,000 rpm (~400 × g) and degassed for 2 min at 2,200 rpm (~440 × g) in a centrifugal mixer.
  4. The mixture is poured onto the control layer mold and degassed in a vacuum desiccator for 30-50 min.
  5. For the flow layer, 20 g (10.5 g) of a 20:1 Sylgard mixture (20 g Part A:1 g Part B) (10 g Part A:0.5 g Part B) is prepared, mixed for 1 min at 2,000 rpm (~400 × g) and degassed for 2 min at 2,200 rpm (~440 × g) in a centrifugal mixer.
  6. The mixture is spin coated onto the flow layer with a 15 s ramp and a 35 s spin at 2,200 rpm. 
  7. After removing the control layer mold from the vacuum chamber any residual surface bubbles are destroyed by blowing on top of the PDMS layer. Any visible particles on top of the control channel grid are carefully removed using a toothpick.
  8. Both layers are cured in an oven for 30 min at 80°C.
  9. Following polymerization, both molds are taken from the oven and allowed to cool for 5 min.
  10. The control layer is then diced with a scalpel and holes (1–8 and B, S, C, O in Fig. 1a) are punched at the control input side using a hole puncher or a 21 gauge luer stub.
  11. The channel side of the control layer is thoroughly cleaned with Scotch Magic Tape.
  12. The cleaned control layer is then aligned to the flow layer on the stereomicroscope.
  13. The device is bonded for 90 min at 80°C in an oven.
  14. Bonded devices are removed from the oven and allowed to cool for 5 min.
  15. Following the outline of the control layer each individual device is cut with a scalpel and peeled off the flow layer. Holes are punched for the sample inlet and outlet (S1–S7 and O in Fig. 1a) using a hole puncher.
  16. Holes are punched for the sample inlet and outlet (S1–S7 and O in Fig. 2a) using a hole puncher.
  17. The flow channel side is cleaned thoroughly with tape before aligning the device to a spotted glass slide.
  18. The flow mold is cleaned of any residual polymerized PDMS either by peeling off the thin layer of PDMS using a pair of tweezers or by an additional PDMS layer. For the latter, 11 g of a 10:1 Sylgard mixture (10 g Part A:1 g Part B) is mixed for 1 min at 2,000 rpm (~400 × g), degassed for 2 min at 2,200 rpm (~440 × g), poured on the flow mold cured in the oven for 30 min at 80°C, and peeled off after cooling down to room temperature. The control mold is cleaned with a nitrogen air gun of any PDMS debris.

Glass slide preparation:

 Cleaning procedure:

  1. All glassware is prepared by rinsing with Milli-Q water.
  2. 750 ml Milli-Q water and 150 ml ammonium hydroxide are heated to 80°C in a staining bath.
  3. 150 ml hydrogen peroxide is carefully poured to the ammonium solution.
  4. Glass slides are added into the staining bath and incubated for 30 min.
  5. After removal from the staining bath, the glass slides are allowed to cool for 5 min.
  6. Glass slides are then rinsed with Milli-Q water in the staining bath.
  7. Clean glass slides are dried with nitrogen and stored in a dust free box.


Epoxysilane deposition:

  1. Before epoxysilane deposition, all glassware is rinsed with acetone and dried at 80°C.
  2. Cleaned glass slides are incubated for 20 min in 891 ml toluene with 9 ml 3-GPS.
  3. After rinsing with fresh toluene to remove unbound 3-GPS, the glass slides are dried with nitrogen.
  4. Glass slides are baked at 120°C for 30 min.
  5. Following sonication in toluene for 15 min, glass slides are rinsed with fresh IPA.
  6. Coated glass slides are dried with nitrogen and stored in a dust-free box under oxygen free conditions until usage.
  7. In case of systematic PDMS chip delamination: Prior to DNA spotting, glass slides are rinsed with toluene and dried with nitrogen.


Adapted from Rockel, S., Geertz, M., & Maerkl, S. J. (2012). MITOMI: A Microfluidic Platform for In Vitro Characterization of Transcription Factor–DNA Interaction. In Gene Regulatory Networks (pp. 97-114). Humana Press.

sfGFP induced by arabinose (pBAD promoter) and constitutive GFP expression in MITOMI chip



1      Material and methods


-768 MITOMI chip

-Lysis buffer (30mM NaOH + 12% w/v SDS)

-LB with chloramphenicol + arabinose (0.2% w/v)

-Microfluidic material



-GenePix software for data analysis


The aim of the experience was to test the expression of sfGFP (BBa_I746908) induced by arabinose (pBAD promoter) and constitutive GFP (BBa_K584001). Each plasmid was transformed individualy in E.Coli bacteria. After an overnight culture of these bacteria in LB medium containing chloramphenicol and arabinose, they were flown in the chip and subsequently scanned.

2      Results

2.1     Data

The picture shows the chambers of the chip containing the bacteria. Fluorescence is only visible in the chambers because lysis buffer was flown in the channels, thus removing all signal comming from the bacteria outisde the closed chambers. First line contains the sfGFP, second line the const. GFP and this is repeated over the whole chip by pair of lines.

2.2     Interpretation

Great signal of both superfolded GFP and constitutive GFP was obtained. In theory superfolded GFP should be more intense than conventional GFP. A slight difference can be seen on the scan, however as the cell density could not be checked precisely this could be due to different concentrations of bacteria in the chambers.


CheY-rLucN CheZ-rLucC bioluminescence assay



1       Material and methods


-768 MITOMI chip

-Lysis buffer

-LB with chloramphenicol + Arabinose(0.2%)

-Microfluidic material



-Glucose solution (20ul): 1mM diluted in water

-Coelenterazine (clz) solution (20ul): 500uM diluted in PBS ph 7

-Genepix and Excel softwares for data analysis

-pYNZC and pRluc plasmids from Prof. Waldor's Lab (link)


Aim of the Experiment
Addition of coelenterazine in the medium was expected to increase the luminescence of the cells that were in arabinose. 
Cells that did not have arabinose were expected to not emit any luminescence and cells with full luciferase were also loaded on the chip as a positive control



Cells were loaded into the MITOMI chip and grown overnight at room temperature

We used cells containing following plasmid:

  • pYNZC (from Waldor's Lab) grown overnight in LB + arabinose
  • pYNZC (from Waldor's Lab) grown overnight in LB (without arabinose)
  • pRluc (from Waldor's Lab) as positive control = full luciferase

Medium composition: 15ml LB solution with chloramphenicol and 4*670ul = 2680 ul arabinose 100mM (following Waldor's Lab paper)

Compartments were as follows:


1st loading: with ara (M2 + M4 closed), 5 min

2nd loading: without ara (M2 + M3 closed), 5 min

-Wash with lysis buffer 5 min

-Wash with medium 5 min

3rd loading: positive control (full luciferase) (M1 + M4 closed)

The only difference between cells with arabinose and without arabinose is that they were cultured overnight with or without arabinose (before chip). Once the cells were loaded into the chip, they were all supplied with arabinose.


The chip was scanned before starting the experiment in the morning, (Fig 1)

A scan of lysis buffer and medium flowing in the chip was also made to check if these solutions gave a fluorescent signal. (Fig 2 and 3).
Medium was also used as a control. (Fig 4)

Clz was added and incubated for 10 minutes with closed sandwiches and 2 scans were taken at interval of 5 minutes (Fig 5 and 6).
A third one was also taken with the lamp turned off, this scan did not produce a decent picture.

Once the scans post clz addition were taken, we flowed glucose in the chip. The expected response to glucose is a fading of luminescence.

First glucose was flown for 5 min with M3 closed, meaning that only “without ara” and “medium” (negative control rows) compartments were supplied with glucose.
Incubation of 5 min with closed sandwiches followed this addition. (Fig 7)

Next, to bring glucose in the +control compartment, glucose was flowed for 5 min with M1 and M4 closed.
The addition of glucose should not decrease the signal for the full luciferase. After 5 min incubation, a scan was taken. (Fig 8).

A last scan was taken after 5 more minutes of incubation, without any new addition of glucose. (Fig 9)


A final scan was taken in the morning (Fig 10). This should check if the luminescence is completely gone. The chip was left without any medium supply overnight.

Data analysis started on 24.07 and finished on 25.07 using Genepix and Excel

2       Results


Figure1 : scan after overnight culture, before experiment start (595 and 530 nm)


Figure2 : Scan after overnight culture, lysis buffer flow before experiment start (595nm)


Figure3 : scan after overnight culture, medium flow before experiment start (595nm and 530nm)


Figure4 : Scan aftern overnight culture, medium control (595 and 530 nm)


Figure5 :  t= 10min after coelenterazine addition ( 600, 595, 530nm)


Figure6 :  t= 15min after coelenterazine addition ( 600, 595, 530nm)


Figure7 : t=5min after glucose addition for "without ara" and "medium" (685, 595 and 530nm) 


 Figure8 : Glucose added to +control rows, t=15min after first addition (685, 595 and 530nm)


 Figure 9 : Last scan at t=20min after first flucose addition, t = 10min after second addition (685, 595 and 530nm)


 Figure10 : Final scan, chip left with no supply overnight (685, 595 and 530nm)


3    Interpretation

Figure11 : Genepix analysis results, luminescence intensity vs time for cells supplied with coelenterazine. The dashed line represents the addition of glucose. The F595 value of every chamber was subtracted with the F595 value of it’s background noise ( we scanned next to the chamber ) and then we also subtracted the F595 value emitted by the medium.

Even though the scanning method wasn't optimised for luminescence, we managed to reproduce part of what was done in the reference paper (link). The positive control's signal (full renilla luciferase) is also decreasing upon addition of glucose even if in theory it should not be affected. This observation was also made in Prof. Waldor's results.


Induction of RFP by CpxR sensor



1   Material and methods


1.1 Material

-768 MITOMI chip

-Lysis buffer (30mM NaOH + 12% w/v SDS)

-LB with chloramphenicol

-Microfluidic material



-GenePix software for data analysis


1.2 Methods

The aim of this experience was to test the expression of RFP (BBa_K339007) induced by the CpxR responsive promoter. The plasmid was transformed in E.Coli bacteria. After an overnight culture of these bacteria in LB medium containing chloramphenicol they were flown in the chip and subsequently scanned after stressing the bacteria.

CpxR is synthetised upon membrane stress (see our Cpx pathway description for more informations). The idea here was to stress bacteria by applying pressure on them in order to make the bacteria produce their Cpxr protein and thus inducing RFP synthesis. In the biobrick we used (BBa_K339007), there is a CpxR protein sensor upstream of the RFP coding sequence. This sensor should trigger the RFP synthesis once a CpxR protein binds the sensor.

Even though The MITOMI chip isn't oprimised for applying mechanical pressure on bacteria, this experience was performed as follows.

Cells were flown into the chip afer an overnight culture in LB + chloramphenicol. Then the buttons of the chip were activated, thus applying pressure on the upper part of the chambers. As most of the bacteria sit in the lower part of the chambers, pressur affected only a small amount of them.

Scans were taken at regular time intervals.

2   Results


2.1 Data

 Zoom in four chambers of the chip. Even though we had quite a lot of background noise, we can distinguish some brighter spots corresponding to bacteria potentialy expressing RFP.



 This scan was made the next morning of the eperiment. We only see signal on the lower chambers as lysis buffer was flown on the upper channel (closed chamber valve)

2.1 Interpretation

A couple of weeks after these RFP experiments, we sequenced the plasmid that was used to double check the part. We found out that the most important part of it, that is the CpxR sensor senquence, was missing. Thus this experiment is unconclusive and we could not give an explanation about the signal we obtained.

This image shows the deletion in the CpxR sensor coding sequence.

cpxR linked with GFP on the N terminal, induced by arabinose


The aim of this experience was to induce CpxR expression by arabinose. The Cpxr is fused to a GFP reporter (BBa_K1486002) allowing us to determine if the protein is really expressed. The experiment was previously done on wet bench and this one repeats it on a microfluidic chip.

1      Material and methods


1.1 Material

-768 MITOMI chip

-Lysis buffer (30mM NaOH + 12% w/v SDS)

-LB with chloramphenicol

-LB with chloramphenicol and arabinose (5mM)

-Microfluidic material



-GenePix software for data analysis

-Plasmid with pBAD promoter followed by cpxR linked with GFP on the N-terminal (BBa_K1486002)


1.2 Method

Cells were inoculated and grown overnight in 3 ml LB medium with chloramphenicol.

The next morning the cells were loaded on chip and the upper half of the chip had a flow of LB with arabinose and chloramphenicol whereas the lower half of the chip had a flow of LB with only chloramphenicol.

The cycle was the following :

-take a scan

-close chamber valves and flow lysis for 10min

-close M2 and flow LB with arabinose and chloramphenicol for 5min

-open M2, close M1 and flow LB with only chloramphenicol for 5min

-open M1 and close the outlet and main inlet valve

-culture time of 40min

Repeat 6x for a total of 3h.

This is a schematic representation of how the bacteria were divided on the chip. M1 and M2 stands for multiplex 1 and 2 and are valves that allow to block the flow of bacteria (or medium). Thanks to these multiplexes up to 16 different types of cells/bacteria can be flown in distinct rows. In this experiment we only used a separation in two region (ara+ and ara-). 


2      Results


2.1     Data

Figure 1. Scan of the microfluidic chip at t = 0min. No signal is detected


Figure 2. Scan of the microfluidic chip at t = 300min. The upper half of the chip has medium with arabinose and the lower half doesn't. Expression is detected on the upper half



After GenePix analysis, where we calculated the intensity of each chamber and also the intensity of the area next to the chamber (to subtract as background value) we calculated an average of fluorescence expression. Figure 3 shows the results.

Figure 3. Evolution of CpxR-GFP fluorescence over time

2.2     Interpretation

A great GFP signal was detected which confirms the expression of our portein of interest Cpxr.

The increasing standard deviation for the cells with arabinose can be explained as some chambers did not have a lot of cells and so there was a low intensity thus increasing the deviation. As it can be seen in the following picture :

Figure 4. These are chambers with arabinose in the medium, you can see that there are different cell density and thus different intensity in the chambers. Inducing a high standard deviation

On-chip IFP detection


The aim was to prove that we are able to detect IFP complementation on-chip. Cells containing the split IFP fused to CpxR (BBa_K1486056) were either stressed with PBS + 80 mM KCl or left unstressed in neat PBS and imaged on-chip after 20 min.

Material and methods


-Smash-coli microfluidic chip

-Lysis buffer (30mM NaOH + 12% w/v SDS)

-PBS 1X, pH 7.4 (Gibco)

-Biliverdin hydrochloride 25 mM in DMSO (Sigma)

-Arabinose 20% w/v

-Bacteria transformed with BBa_K1486056 or BBa_K1486013 (negative control)

-Microfluidic material



-ImageJ software



An overnight culture of E. coli transformed with BBa_K1486056 (named CpxR-IFP) or BBa_K1486013 (named IFP-neg) was prepared in 5 ml LB supplemented with chloramphenicol and inoculated with cells from fresh agar plates. The culture was incubated at 37°C with shaking at 180 rpm. The following day the overnight culture was diluted 1:100 in 5ml fresh LB + chloramphenicol and grown for 4 hours at 37° C on a rotary shaker to reach an OD of ~0.6.

1.8 ml of the culture was then diluted in 1.2 ml arabinose 20% (final concentration of 8%) and grown for 2 more hours. 1 μl of biliverdin hydrochloride 25 mM in DMSO was added and thoroughly mixed until homogenization. The culture was grown for 4 more hours, then centrifuged at 2800 rpm for 10 min. The pellets were resuspended in either 320μL PBS or 320μL PBS + 80 mM KCl and aliquoted into a white 96-well plate in triplicates.

Once the difference in intensity between stressed and non-stressed cells on the plate reader was high enough, cells of each batch were taken from one well and transferred on the microfluidic chip as shown below.

Cells were scanned after 20 min using a Cy5 filter.




Cy5 scan of a chamber containing non stressed IFP-neg bacteria

Cy5 scan of a chamber containing KCl-stressed IFP-neg bacteria


Cy5 scan of a chamber containing non stressed CpxR-IFP bacteria

Cy5 scan of a chamber containing KCl-stressed CpxR-IFP bacteria

The ratio of fluorescent cells was calculated on 5 chambers for each batch using ImageJ (the total number of cells was measured on brightfield images). The mean and standard deviation of these ratios were then calculated for each batch and represented as shown below:


This experiment shows that it is possible to detect IFP complementation following KCl stress on-chip, which is the prerequisite for further on-chip studies of CpxR-split IFP cells.