A 3D Microfluidic Perfusion System Made from Glass for Multiparametric Analysis Of

A 3D microfluidic perfusion system made from glass for multiparametric analysis of stimulus-secretion coupling in pancreatic islets

Torben Schulze 1,3, Kai Mattern 2,3, Eike Früh 1, Lars Hecht 2, Ingo Rustenbeck1,3, Andreas Dietzel 2,3

1 Institute of Pharmacology and Toxicology, 2 Institute of Microtechnology, and 3 Institute of Pharmaceutical Engineering (PVZ), TechnischeUniversität Braunschweig, D-38106 Braunschweig, Germany

SUPPORTING INFORMATION:

Experimental methods: S1

Fig. S1Microchannel sections containing a well of 50 µm depth (upper graph) or of 500 µM depth (lower graph). The height was set to zero (z = 0) at the bottom of each well.

Results: S2, S3

Fig. S2Distribution of freshly isolated mouse islets in dependence on well depth and flow rate. The trapping of islets by wells of increasing depth was characterized by varying the pump rate between 120 µl/min (black symbols) and 960 µl/min (light grey symbols). Note the nearly uniform settling of all islets due to their adherence to channel and well surfaces Values for each well-depth are the sum of all 8 channels.
Fig. S3Measurement of the oxygen consumption of perifused mouse islets. 30 min after introduction of the islets and assembly of the MPS the sensor responses were sufficiently stable to allow meaningful measurements. The flow rate was 40 µl/min. The first oxygen sensor gave the oxygen level in the Krebs-Ringer perifusion medium (grey trace). The second oxygen sensor (black trace) was located downstream of the islets and reported the diminished oxygen level of the medium caused by the oxygen extraction of the islets. The second trace was subtracted from the first and the difference was multiplied by the flow rate. Finally, the resulting rate was divided by the number of islets. As compared to the dynamic range of the oxygen sensors the difference between the sensor signals is small, which requires a minimization of the oxygen leakage rate. As is visible from the effect of KCN, the difference between both sensors is virtually entirely due to the mitochondrial oxygen consumption of the pancreatic islets.

Conclusion: S4

Fig. S4 Further development of the microfluidic perfusion system. This MPS has the same dimensions as a microscopic slide. Its inlet channel branches into 32 parallel channels, each housing a well of 300 µm depth for highly parallelized measurements. The oxygen sensors are directly attached to the inlet and outlet channels into which hypodermic syringes are cemented for the easier connection with the tubing. Here, the islets are injected via the capped filling port beside the left oxygen sensor.

S5 Video

The supplemental video S5shows two cultured islets in the MPS. Video sequence (1 fps) and pumping (40 µl/min) were started in parallel. One can see the islets moving through the channel passing the first well of 50 µm depth.

S6 Video

The supplemental video S6is a continuation of video S5. Itshows the two cultured islets which are now trapped in a 100 µm well. Again the video sequence (1fps) starts in parallel with the pump(40µl/min). One can track a third islet passing over the islets in the 100 µm well and settling in the subsequent 150 µm well.

S7 Video

Simulation of medium exchange in a 300 µm deep well of the MPS at 37 °C. Three-dimensional view of the displacement of fluid phase 1 (blue) by fluid phase 2 (red) at a pump rate of 40µl/min.

S8 Video

Simulation of medium exchange in a 300 µm deep well of the MPS at 37 °C. Three-dimensional view of the displacement of fluid phase 1 (blue) by fluid phase 2 (red) at a pump rate of 120µl/min.

S9 Video

Simulation of medium exchange in a 300 µm deep well of the MPS at 37 °C. Three-dimensional view of the displacement of fluid phase 1 (blue) by fluid phase 2 (red) at a pump rate of 240µl/min.