Biomedical Microdevices

, Volume 12, Issue 3, pp 499–503

An automatic microturbidostat for bacterial culture at constant density

Authors

  • Xianjia Luo
    • Center for Microfluidics and Nanotechnology and School of Physics, The State Key Laboratory for Artificial Microstructures and Mesoscopic PhysicsPeking University
    • Ecole Normale SuperieureCNRS-ENS-UPMC UMR 8640
  • Kangyang Shen
    • Center for Microfluidics and Nanotechnology and School of Physics, The State Key Laboratory for Artificial Microstructures and Mesoscopic PhysicsPeking University
    • Center for Microfluidics and Nanotechnology and School of Physics, The State Key Laboratory for Artificial Microstructures and Mesoscopic PhysicsPeking University
    • Center for Theoretical Biology, Academy for Advanced Interdisciplinary StudiesPeking University
  • Hang Ji
    • Center for Microfluidics and Nanotechnology and School of Physics, The State Key Laboratory for Artificial Microstructures and Mesoscopic PhysicsPeking University
  • Qi Ouyang
    • Center for Microfluidics and Nanotechnology and School of Physics, The State Key Laboratory for Artificial Microstructures and Mesoscopic PhysicsPeking University
    • Center for Theoretical Biology, Academy for Advanced Interdisciplinary StudiesPeking University
    • Center for Microfluidics and Nanotechnology and School of Physics, The State Key Laboratory for Artificial Microstructures and Mesoscopic PhysicsPeking University
    • Ecole Normale SuperieureCNRS-ENS-UPMC UMR 8640
Article

DOI: 10.1007/s10544-010-9406-5

Cite this article as:
Luo, X., Shen, K., Luo, C. et al. Biomed Microdevices (2010) 12: 499. doi:10.1007/s10544-010-9406-5

Abstract

We have developed a microturbidostat for long time bacterial culture at constant density controlled by optical detection and integrated pneumatic valves. The device was fabricated by multilayer soft lithography and in-situ formation of an agarose filter. The culture chamber of bacteria was connected in one side to a single bacterial input-output channel and in another side to a nutrient channel in which the agarose filter was formed to ensure the diffusion of nutrients and metabolites without bacterial loss. The bacterial number in the culture chamber was determined by measuring the fluorescence intensity of GFP proteins of the bacteria and the redundant bacteria could be exported automatically through the input-output channel with integrated micro-valves. In order to optimize the operation performance, we investigated the bacterial exportation efficiency with different input-output channel widths. As expected, the bacterial sorting coefficient was proportional to the input-output channel width. The results also showed that with a 20 µm channel-width, a long time culture was possible with a constant bacterial number in the chamber in the range from 400 to 700.

Keywords

MicrofluidicsMicroturbidostatBacterial growth

1 Introduction

There are several reasons to be interested in the development of new tools and technologies for improved microorganism cultivation. First, the current cultivation and bio-processing are mostly relying on shake flasks, which are not flexible for the microenvironment control and easy to keep the concentration of chemical components in constant (Akgun et al. 2007). Comparing to the batch operation, continuous-flow culture systems offer the advantages of constant environmental conditions (Calcott 1981) but both conventional chemostat (Novick and Szilard 1950) and turbidostat (Martin and Hempfling 1976) have the drawback of complex step-up and hard handling. Second, the consummation for large volume systems is remarkably high, which is often not necessary for experimentations such as cultivation parameter determination. Third, it is difficult for the commonly used systems to perform parallel operation. Finally, many microorganisms found in nature are still unculturable using current techniques or conditions (Maharbiz et al. 2004) and it is therefore important to develop new systems with an improved parameter control, reduced consumption, and increased parallelism and automatism.

Lab chip technology offers a unique solution to these problems. Indeed, many groups have successfully demonstrated the relevance of fabricated microsystems in microorganism cultivation. Zanzotto et al. (2004) have developed a membrane-aerated microbireactor with integrated sensors for optical density (OD), dissolved oxygen (DO), and pH measurements. Puskeiler et al. (2005a, b), Weuster-Botz et al. (2005) and Zhang et al. (2005) have reported the integration of magnetic mixers in their microbioreactors systems. Whereas Balagadde et al. (2005) have demonstrated the operation of a microchemostat with integrated peristaltic pumps and micro valves for parallel and long-term culture of microorganisms, Zhang et al. (2005) have presented a microchemostat which is capable of OD, DO and pH real-time measurements for continuous cultivation of microorganisms. In addition, Groisman et al. (2005) have applied cell trapping technique in similar microchemostat devices.

We report in this work the design and fabrication of an automatic microturbidostat which can be used to control the biomass concentration by feedback system and thus to provide the microorganisms with constant physiochemical environment. Our device consisted of integrated culture chamber, nutrient channel, bacterial input-output channel and pneumatic valves. An agarose filter integrated between the nutrient channel and the culture chamber allowed the diffusion of nutrients and metabolites without bacterial losing (Beebe et al. 2000; Wu et al. 2006; Cheng et al. 2007; Kim and Beebe 2007). To evaluate the bacterial exportation efficiency, five devices have been fabricated with different input-output channel-widths. Experimental data were then analyzed and compared with a simple bacterial exporting model. For turbidostat culture, the bacterial number was calculated by image-processing algorithm and then changed with a feedback control system.

2 Materials and methods

2.1 Device design and fabrication

The microturbidostat used in this work was fabricated by multilayer soft-lithography (Unger et al. 2000). It included two layers (Fig. 1(a)): the culture layer (lower layer) and the valve layer (upper layer). The mold of air-liquid channel was fabricated by photolithography with photo-resist AZ9260, whereas the mold for the remaining parts of the lower layer (reservoirs and connecting channels) as well as the pneumatic valve channels were obtained by photolithography with photoresist SU-8 2007. After photolithography, the lower-layer resist pattern was baked at 120°C for 5 min, which made a round shape of the AZ9260 part. A commercial kit of polydimethylsiloxane (PDMS) was used for soft lithography (GE KV615). For the lower layer pattern fabrication, a PDMS mixture at ratio of 20:1 was spun on the mold at 2,000 rpm and then cured in 80°C for 23 min. For the upper layer pattern fabrication, a PDMS mixture at ratio of 10:1 was cast onto the mold with a thickness of 5 mm and then cured in 80°C for 1 h. Afterward, the PDMS of the upper layer pattern was peeled off and connection holes were punched before placing it on the top of the PDMS film of the lower layer pattern. After curing at 80°C for 3 h, the two layer PDMS assembly was peeled off. Finally, the connection holes for the lower layer pattern were added and the whole system was irreversibly bonded to a glass slide after O2 plasma treatment of both PDMS and glass surfaces. As results, we obtained a multilayer PDMS device with pneumatic valves, a square culture unit of 200 µm side-lengths which is linked to a nutrient channel of 300 μm width and an air-liquid channel of 100 µm width (Fig. 1(b)). The length of input-output channel was 200 μm.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-010-9406-5/MediaObjects/10544_2010_9406_Fig1_HTML.gif
Fig. 1

Schematic diagram of a microturbidostat. (a) Microturbidostat consists of a lower layer for bacterial loading, culture and exporting (1–4, 9–11) and an upper layer for pneumatic valve channels (5–8). (b) Culture chamber connected on its left to a nutrient channel through a diffusion channel with agarose filter and on its right to an air-liquid channel through a bacterial input-output channel. (c) Three dimensional view of the culture chamber with the two connection channels

An agarose filter was then introduced in the short channel which linked the wide nutrient channel and the bacterial culture chamber (Fig. 1(c)). Figure 2 shows a schematic process for the agarose filter integration. First, the fabricated device was left at room temperature for 2 days in order to make the PDMS surface hydrophobic. Aqua solution with 2% agarose was then prepared. After heating the solution and the PDMS device in oven at 80°C for 10 min, the solution was injected into the nutrient channel (Fig. 2(a)). Since the connection channel between the nutrient channel and the culture channel was designed to have a 50 µm wide opening, a 20 µm entrance and then a 30 µm-wide segment, the agarose solution can be stopped in the middle of channel (Fig. 2(b)) because of the fluidic resistance (Lee et al. 2006). Afterward, air was carefully injected into the nutrient channel to remove the agarose solution except the part block the in connection channel (Fig. 2(c)). After the device was brought to the room temperature for 10 min, the remaining agarose was fixed. Then, filtered deionized water was injected into the nutrient channel to make the agarose wet.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-010-9406-5/MediaObjects/10544_2010_9406_Fig2_HTML.gif
Fig. 2

Schematic diagram of the fabrication process of agarose filter in the microfluidic device. (a) An agarose solution (2% w/w) is injected into the nutrient channel at 80°C. (b) The solution is introduced into the broad part of the connection channel. (c) The solution is removed from the nutrient channel with air keeping a portion of the agarose solution in the connection channel. After cooling down to the room temperature, a solidified agarose filter is formed

2.2 Experimental setup

The experimental set-up is shown in Fig. 3. The microturbidostat was placed on an inverted microscopy (Olympus IX71) and connected to a syringe pump (TS2-60), a home made pneumatic valve controller and PC. Images were taken with a digital camera (Panasonic CCD). For the turbidostat culture, the bacterial number in the culture chamber was deduced by analyzing the images with a Matlab program. If the bacterial number was larger than 600, a signal was yielded and sent to the valve controller to begin a flow of bacterial exportation. After 90 s, the valve was closed to stop the bacterial flow. The system repeated automatically the imaging analyses and flowing control each 25 min to maintain a bacterial culture in constant density.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-010-9406-5/MediaObjects/10544_2010_9406_Fig3_HTML.gif
Fig. 3

Schematic diagram of experimental setup. The fluorescence signals of bacteria, cultured in the microfluidic chamber, are recorded by a CCD camera and then analyzed using a custom build Matlab program. The switch of the bacterial input-output channel is controlled through micro-valves according to the analytic results

2.3 Strain and culture

E.coli MG1655 (wt), kindly provided by Dr. Francois Taddei of University Rene Descartes, was transformed with plasmid I7001 expressing ampicillin resistance and GFP proteins. Luria-Bertani (LB) medium used consisted of 5 g/L yeast extract, 10 g/L NaCl and 10 g/L tryptone. Single colony of E.coli MG1655 (I7001, Amp) on LB plate (containing 2%(wt/vol) agar and 250 µg/mL ampicillin) were cultured overnight in 2 mL LB medium with 250 µg/L ampicillin on a shaker at 180 rpm in 37°C. Then, 10 µL of culture medium was transferred to a test tube containing 1 ml of fresh, which was also incubated on a shaker at 180 rpm in 37°C until the optical density (OD600 nm) reached 0.6. The culture medium was injected into the microfluidic device at normal temperature, which had been sterilized by injecting 75% ethanol for 2 h and rinsed with filtered deionized water to remove the ethanol.

2.4 Cell loading

The PDMS device was first degassed in a vacuum desiccator for 5 min. Then, filtered deionized water was injected into both of the nutrient channel and the air-liquid channel. Since the degassed PDMS absorbs the gas, water could be introduced into the bacterial culture chamber (Luo et al. 2008). Afterward, the microfluidic channels were sterilized by 75% ethanol for 2 h and rinsed with filtered deionized water. The sterilized LB medium was finally injected into the nutrient channel. When the bacteria were injected into the air-liquid channel, they swam into the culture chamber through the bacterial input-output channel, due to bacterial chemotaxis.

3 Results and discussion

In our designed devices, the bacterial input-output channel width is one of the most important parameters for the determination of the operation time sequences. In order to optimize the performance of the operation, devices having the channel-width 20 µm, 35 µm, 65 µm, 80 µm and 100 µm had been tested. In the experiment with 80 µm and 100 µm wide channels, the agarose filters were easily broken by the pressure introduced from the air-liquid channel. Figure 4 shows the bacterial exportation efficiency of the device with 20 µm, 35 µm and 65 µm channel-widths. As expected, a large input-output channel-width allowed a more efficient bacterial exporting. We also noted that the initial bacterial number in the chamber with 20 µm input-output channel width is significantly lower than that with wider channels. This is probably due to that the bacteria are more difficult to swim into the chamber through a narrow channel. In order to decrease the influence of bacterial exportation on their growth environment, it is also important to keep the channel width as small as possible.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-010-9406-5/MediaObjects/10544_2010_9406_Fig4_HTML.gif
Fig. 4

Bacterial number as a function of time for three types of microturbidostats with different input-output channel widths. Scatters are experiment data and solid lines are linear fits for the experiment data less than 300 min. The widths of the bacterial input-output channels are 20 µm, 35 µm, and 65 µm respectively. R2, the goodness of fit statistics, are 0.9930, 0.9665, 0.9884, and σ, the standard deviation of residual, are 0.0035, 0.0111, 0.0191 for 20 μm, 35 μm, 65 μm wide input-output channels

To determine quantitatively the bacterial exporting speed through micro channels of different sizes, we used the following phenomenological model, a logistic equation, to fit the experiment data for the chosen time range of interest (0–300 min.):
$$ \frac{{dN}}{{dt}} = - {\hbox{v}}N $$
(1)
where N and v are the bacterial number and the exporting coefficient of the microturbidostat. Figure 4 also shows the fitting curves obtained by using the above logistic Eq. 1, resulting in a bacterial exporting coefficient of 0.00127, 0.00187 and 0.00465 for 20 µm, 35 µm and 65 µm wide input-output channels, respectively.
Knowing that for a given input-output channel width and exporting time, only a part of bacteria is capable to escape the culture chamber. The parameter v should then depend on the channel geometry and the migration capacity of the bacteria, which we suppose to be geometry independent. Then,
$$ {\hbox{v}} = hdC $$
(2)
where h and d are the height and width of the input-output channels, C is the bacterial migration capacity due to diffusion and chemotaxis effect. Figure 5 shows the parameter v as a function of channel width d, linearly fitted as predicted by Eq. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-010-9406-5/MediaObjects/10544_2010_9406_Fig5_HTML.gif
Fig. 5

Parameter v as a function of input-output channel width d. The red solid line is linear fit. R2, the goodness of fit statistics, is 0.9726

The fitting curves in Fig. 5 were obtained from data in the range of 0–300 min. For more than 300 min, the bacterial number is much smaller and there appear large deviation from the exporting model because of the following two reasons. First, the bacteria located in the culture chamber closer to the nutrient channel can not be easily exported. In fact, these bacteria undergo the attraction of nutrients from both sides. Thus, the migration capacity of bacteria toward the input-output channel is much reduced so that the reduction of the bacterial number is less rapid with time. Second, when the exporting time becomes long, the influence of cell growth on exporting effect increases, resulting in the deviation from the model. More laborious are expected for a more quantitative interpretation of the present results.

By using the device with an input-output channel width of 20 µm, we were able to control the bacterial number in the culture chamber in the range between 400 and 700. Figure 6 shows a comparison of the bacterial growth behavior with and without the turbidostat control. Without the turbidostat control the bacteria population continually increased in the culture chamber. With the turbidostat control, the images were taken every 25 min and analyzed with a Matlab program to calculate the bacterial number. When the bacterial number exceeded 600, the pneumatic valves were opened to export the bacteria for 90s. The result indicated that the microturbidostat could control the biomass concentration in constant by feedback system, and provide constant physiochemical environment for bacterial culture.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-010-9406-5/MediaObjects/10544_2010_9406_Fig6_HTML.gif
Fig. 6

Bacterial number as a function of culture time in the micro chamber with and without turbidostat control (the width of the input-output channel is 20 µm). Without the turbidostat control the bacteria population continually increases in the culture chamber (black square). With the turbidostat control the bacteria number remains quasi constant (black circle) and their population varies in the range of 400–700

4 Conclusion

We designed and realized an automatic microturbidostat for bacterial culture in constant density. The results showed that the bacterial sorting coefficient in our devices was proportional to the width of the input-output channel which connected the culture chamber and the input-output channel. Under optimal conditions, the designed system can be used for bacterial culture in a well-controlled micro-environment. We believe that such a microturbidostat has unique advantages of easy and robust sample handling with much reduced reagent consumption. Furthermore, both device fabrication technology and operation strategy proposed in this work can be extended for high throughput screening by considering either parallel or sequential system integration.

Acknowledgement

The authors would like to thank F Taddei, C.B. Lou, L.P. Xu and M. Ni for useful discussion and help. This work is partially supported by the Chinese Natural Science Foundation (NO. 10704002, 10634010, 10721403) and National 973 project (NO. 2003CB715900).

Copyright information

© Springer Science+Business Media, LLC 2010