Microfluidic chips: Lab-on-a-chip cell culture for metabolomics

Metabolomics, the study of small molecules involved in metabolic reactions, has proven a useful tool for providing insight into a variety of cellular processes and disease states.¹ It is capable of offering both qualitative and quantitative information about multiple intra- and extra-cellular metabolic pathways. Cellular metabolic profiles can yield important information regarding biological environments which can be used to elucidate pathophysiological mechanisms of diseases ranging from diabetes to cancer.²

Advances in metabolomics research have been hindered by the lack of suitable in vitro models for cellular metabolomics experiments. Traditional cell culture methods involve the use of cell culture flasks and well plates to grow human or animal cells which are then subjected to various stimuli relevant to the system of interest. Relatively large plates (~10-20 cm2) are often used to ensure that enough cells are present to produce metabolites in detectable quantities. The volume requirements of conventional cell culture apparatuses prohibit many long-term flow-based studies in which large quantities of reagents are required. Another disadvantage of conventional cell culture techniques is the inability to simulate the dynamic conditions that are present in vivo. Cell culture flasks provide only a static two-dimensional environment for cells and offer limited flexibility in terms of integrating scaffolding and other architecture for three-dimensional cell culture.

Scaling down cell culture

MFDs are often called labs-on-a-chip owing to the ease at which valves,³ pumps,⁴ membranes,⁵ pillars,⁶ and even detectors can be incorporated into the device design. Figure 1, for example, shows an MFD used to study releasate from PC12 cells, a neuronal mimic.⁷ In this device, microdialysis sampling was employed to sample dopamine from stimulated cells cultured in a small petri dish; the dialysate was then injected into an electroosmotic flow stream via pressure-based valves. Detection took place electrochemically following the electrophoretic separation. While this system offers many advantages including online detection and fast (~20 s) separations, the device suffered from appreciable lag due to the dead volume and low flow rates associated with microdialysis sampling, which is a significant hindrance in dynamic systems where near real-time analysis is desired.

Current research in microfluidic cell analysis has focused heavily on culturing cells directly on-chip. By integrating both the cell culture platform and analysis component within the MFD, the system response time can be dramatically improved, allowing for increased throughput and preservation of biological information during analysis. Furthermore, MFDs offer the advantage of scaling down the cell culture substrate to several orders of magnitude smaller than that of traditional techniques. Thus, the required amount of media, drugs, and other reagents are 100-fold lower;⁸ this is especially beneficial for shear stress studies and other flow-based experiments as it provides cells with a constant supply of fresh media but generates small waste volumes.

One of the first considerations to be made when designing an appropriate cell culture MFD is microchip material. Microchips made out of a variety of materials including polydimethylsiloxane (PDMS), polystyrene, and glass have been described for the fabrication of cell growth chambers. Polydimethylsiloxane (PDMS) is a popular microchip material as microchannel fabrication can be easily accomplished with standard photolithography techniques. The gas permeability of PDMS also makes it an attractive material for cell culture allowing for continuous O₂/CO₂ exchange.⁹ However, cells do not adhere well to PDMS since it is extremely hydrophobic, so many MFDs used for cell studies employ microchips in which channels are fabricated in PDMS and then sealed onto a more ‘cell friendly’ base such as glass or polystyrene.¹⁰ Corona discharge and oxygen plasma treatments have shown some success in increasing hydrophilicity and thus cell adhesion to these substrates.^{11, 12}

Biological mimics

The growing popularity of MFDs for cell culture has seen with it an effort to enhance the biological nature of microfluidic culture platforms. Reproducing in vivo cellular environments for in vitro models entails attention to appropriate extracellular matrix components, co-culture of multiple cell types, three dimensional culture involving the use of scaffolds for cell adhesion, and perfusion of media or other reagents at physiological rates.¹³ These environmental elements have been shown to affect cell metabolism and behaviour, and many of these factors are impossible or very difficult to achieve in conventional culture systems. For example, one study demonstrated an increased rate of oxygen consumption of hepatoma HepG2/C3A cells cultured in a microchip under perfused flow as compared to standard petri dish-cultured cells.¹⁴ This higher oxygenation rate affected cellular metabolism by causing a higher production of acetyl-CoA, the starting point of the tricarboxylic acid cycle.

In addition to changing cellular metabolism, MFDs have been used to study cellular phenotype and morphology. The Vickerman group created a device in which endothelial cells were cultured in a channel connected to a chamber containing a collagen-based hydrogel.¹⁵ A gradient of growth factors was produced which caused the cells to grow into the gel chamber, forming capillaries. The predictable flow streams that can be generated on-chip make it possible to expose certain regions of cells to different stimuli. As cellular phenotype is often directly related to distinct biochemical signatures, MFDs offer the unique ability to analyse the influence of environmental conditions on both morphology and metabolome.  

On-Chip sample preparation

Analysis of the intracellular metabolome often necessitates permeabilisation of the cell membrane, either reversibly or irreversibly, to extract metabolites and transfer them to the detection system. Conventionally, cell lysis has been accomplished by a myriad of approaches including physical and mechanical deformation (compression, freeze-thawing, heating, ultrasonication) and chemical methods (organic solvents, detergents, osmotic shock).¹⁶ While many of these techniques are incompatible with MFDs—for instance, many organic solvents have been shown to dissolve or swell PDMS—promising alternatives have been developed for on-chip cell lysis. Khanna et al. have shown that ultrasonic vibration induces lysis of melanoma cells introduced into a microchamber with nanocrystalline diamond microspikes; lysing efficiency was found to be increased 400% compared with non-textured surfaces.¹⁷ Another effective lysing strategy for MFDs was demonstrated by the Allbritton group, who used a pulsed laser for the rapid rupture of mouse leukemia cells.¹⁸ This method offered the advantage of confining the lysing process to a very small (5µm) zone, enabling single cell analysis.

Following lysing and metabolite extraction, many cellular metabolomics investigations require a purification step prior to metabolite detection in order to eliminate cellular debris from the sample. Solid phase extraction (SPE) involves the separation of analytes from sample matrix components according to their physical and chemical properties via retention on a solid support. Analytes can then be eluted from the SPE sorbent in a concentrated form. SPE in MFDs requires miniaturisation of packing materials and stationary phases to accommodate the small sample volumes associated with microfluidic cell culture. While microchannels packed with SPE particles have been reported, flowing mobile phases through such channels necessitates high backpressure and often leads to clogging. To this end, many researchers have focused on developing functionalised monoliths for microfluidic SPE, which are single structures with high surface area which enable analyte extraction at lower backpressures. The Woolley group has demonstrated a photopolymerised butyl methacrylate monolith for the retention of amino acids and proteins in MFDs.¹⁹ These miniaturised SPE sorbents can be easily coupled with other on-chip processes such as analyte labelling and electrophoretic separation, enhancing integration and automation for cellular micrometabolomics.      

Analysis strategies

Analytical methodologies for identifying and quantifying cellular metabolites are highly dependent on the properties of the species of interest and the desired application. For example, detection of metabolites in live cells in MFDs is often achieved with fluorescence microscopy, which offers chemical specificity while maintaining viability. Figure 2 is a fluorescence micrograph of aortic endothelial cells immobilised on glass in a microfluidic channel; the cells have been treated with a probe which crosses the cellular membrane and is converted into a fluorescent product by intracellular enzymes. The optically transparent nature of many MFDs make fluorescence microscopy an ideal choice for live/dead cell assays. Despite their sensitivity, many fluorescence-based detection schemes must be done indirectly—direct attachment of fluorescent labels to metabolites often compromises biochemical functionality.²⁰

Perhaps the most straightforward optical detection mode for metabolomics in MFDs is absorbance detection. This technique is highly suitable for proteins, which absorb light very strongly in the 280nm range, mainly due to the presence of tryptophan and tyrosine residues. Nucleotides and other nucleotide-conjugated metabolites may also be analysed using absorbance detection at 254nm without the need for prior derivatisation. However, because absorbance is directly proportional to the path length of the sample region, absorbance-based measurements in MFDs suffer from poor detection limits due to small microchannel dimensions (< 1 mm). This drawback may be circumvented by fabricating optical waveguides within the microchannels, which is especially useful for low cost and rapid cellular assays when used in conjunction with portable spectrophotometers.²¹ Nevertheless, the utility of absorbance detection schemes still remains limited by the fact that a vast majority of the metabolome is not natively absorbent. Furthermore, since some intracellular metabolite concentrations can reach picomolar levels, detection systems capable of quantifying analytes present in extremely low abundance are necessary for many applications.²²

Another popular analysis technique for microfluidic cellular metabolomics is electrochemical detection, particularly amperometry. Fabrication of electrodes in microfluidic devices frequently involves sputtercoating thin layers of indium-tin oxide, gold, platinum, or other metals on a microfluidic substrate and then patterning electrodes by wet chemical etching, although some groups have demonstrated MFDs with integrated electrode wires.²³ Although these extensive fabrication and alignment requirements have precluded the use of electrochemical detection for cellular micrometabolomics in many laboratories, amperometry remains a highly sensitive technique for monitoring rapidly-changing electrochemically active metabolites, particularly neurotransmitters such as nitric oxide. Cha et al., for example, developed a microelectrode coated with a porous polymer membrane for the selective and highly sensitive (1nM detection limit) determination of nitric oxide released from monocyte macrophages stimulated with endotoxin.²⁴

Nuclear magnetic resonance spectroscopy (NMR) is ideal for performing rapid metabolomic analyses as it does not require extensive sample preparation or a prior chromatographic separation. NMR has been used to determine label-free metabolite profiling and glycolytic flux in hepatic cells cultured in an MFD; over 40 metabolites were detected in the culture medium in this study.¹⁴ However, like absorbance detection, NMR suffers from relatively poor sensitivity (millimolar to high micromolar) and a dynamic range of only 10³ at best.²⁵ These disadvantages mean that less abundant species may not be detected, leading to an incomplete picture of the metabolic system of interest.  

Currently the most powerful cellular analysis tool in terms of ability to cover the largest fraction of the metabolome simultaneously with high sensitivity is mass spectrometry (MS). Another advantage of MS is structural confirmation, which can be accomplished by fragmenting ionised metabolites and analysing the product ions. The most common ionisation methods for MS detection used for metabolomics are electrospray ionisation (ESI) and matrix-assisted laser desorption ionisation (MALDI). MALDI is a batch technique in which metabolite samples are spotted onto a plate which is pulsed with a laser; as such, MFDs cannot be directly interfaced with MALDI, although offline coupling of the two methods is easily achieved. ESI, in contrast, is a continuous ionisation source which has the potential to be coupled directly with MFDs for cellular metabolomics. The greatest challenge associated with using online ESI-MS detection with MFD cell culture is interfacing flow from the MFD with an emitter tip to direct flow into the MS. The Lin group has demonstrated the analysis of vitamin E metabolites from a suspension of epithelial cells by combining a cell culture MFD with another MFD containing SPE beads for desalting.²⁶ The flow from these two chips was interfaced with ESI-MS via a capillary with a cannula on the end. Currently, the development of fully-integrated cell culture MFDs for metabolomics with online MS detection is still in its infancy, but advances such as fabrication of nanospray emitter tips directly on MFDs offer promise for further progress in this area.

Conclusion

Conventional approaches to cellular metabolomics have proven inadequate both in their ability to mimic the physiological system of interest and in their capacity to detect molecules of low abundance. Recent advances in bioengineering have led to the advent of microfluidic platforms for all-in-one cell culture, sample preparation, and metabolite analysis. These MFDs have the potential to be used in conjunction with metabolite flux analysis to reveal the pathophysiological mechanisms of various metabolic disorders. Once these biochemical changes are known, efforts can be undertaken to identify biomarkers and develop therapeutic targets to treat these often debilitating diseases.

References

1. Yuan, W., Edwards, J.L., 2010. Capillary separations in metabolomics. Bioanalysis 2:953-63
2. Filla, L.A., Yuan, W., Feldman, E.L., Li, S., Edwards, J.L., 2014. Global metabolomic and isobaric tagging capillary liquid chromatography–tandem mass spectrometry approaches for uncovering pathway dysfunction in diabetic mouse aorta. Journal of Proteome Research 13:6121-34
3. Araci, I.E., Quake, S.R., 2012. Microfluidic very large scale integration (mVLSI) with integrated micromechanical valves. Lab on a Chip 12:2803-6
4. Bowen, A.L., Martin, R.S., 2010. Integration of on-chip peristaltic pumps and injection valves with microchip electrophoresis and electrochemical detection. Electrophoresis 31:2534-40
5. Neeves, K.B., Diamond, S.L., 2008. A membrane-based microfluidic device for controlling the flux of platelet agonists into flowing blood. Lab on a Chip 8:701-9
6. Green, J.V. et al., 2011. Microfluidic pillar array sandwich immunofluorescence assay for ocular diagnostics. Biomedical Microdevices 13:573-83
7. Mecker, L.C., Martin, R.S., 2008. Integration of microdialysis sampling and microchip electrophoresis with electrochemical detection. Analytical Chemistry 80:9257-64
8. Noon, J.P. et al., 1997. Impaired microvascular dilatation and capillary rarefaction in young adults with a predisposition to high blood pressure. Journal of Clinical Investigation 99:1873-9
9. Martin, R.S., Root, P.D., Spence, D.M., 2006. Microfluidic technologies as platforms for performing quantitative cellular analyses in an in vitro environment. The Analyst 131:1197-206
10. McDonald, J.C. et al., 2000. Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis 21:27-40
11. Filla, L.A., Kirkpatrick, D.C., Martin, R.S., 2011. Use of a corona discharge to selectively pattern a hydrophilic/hydrophobic interface for integrating segmented flow with microchip electrophoresis and electrochemical detection. Analytical Chemistry 83:5996-6003
12. Guruvenket, S., Rao, G.M., Komath, M., Raichur, A.M., 2004. Plasma surface modification of polystyrene and polyethylene. Applied Surface Science 236:278-84
13. Meyvantsson, I., Beebe, D.J., 2008. Cell culture models in microfluidic systems. Annual Review of Analytical Chemistry 1:423-49
14. Ouattara, D.A. et al., 2012. Metabolomics-on-a-chip and metabolic flux analysis for label-free modeling of the internal metabolism of HepG2/C3A cells. Molecular BioSystems 8:1908-20
15. Vickerman, V., Blundo, J., Chung, S., Kamm, R.D., 2008. Design, fabrication and implementation of a novel multi-parameter control microfluidic platform for three-dimensional cell culture and real-time imaging. Lab on a Chip 8:1468-77
16. Brown, R.B., Audet, J., 2008. Current techniques for single-cell lysis. Journal of the Royal Society Interface 5(Suppl 2):S131-S8
17. Khanna, P. et al., 2009. Nanocrystalline diamond microspikes increase the efficiency of ultrasonic cell lysis in a microfluidic lab-on-a-chip. Diamond and Related Materials 18:606-10
18. Lai, H-H. et al., 2008. Characterization and use of laser-based lysis for cell analysis on-chip. Journal of the Royal Society Interface 5(Suppl 2):S113-S21
19. Nge, P.N., Pagaduan, J.V., Yu, M., Woolley, A.T., 2012. Microfluidic chips with reversed-phase monoliths for solid phase extraction and on-chip labeling. Journal of Chromatography A 1261:129-35
20. Zenobi, R., 2013. Single-cell metabolomics: analytical and biological perspectives. Science. 342(6163)
21. Ruano-López, J.M. et al., 2006. A new SU-8 process to integrate buried waveguides and sealed microchannels for a lab-on-a-chip. Sensors and Actuators B: Chemical 114:542-51
22. Sun, G. et al., 2008. Matrix-assisted laser desorption/ionization-time of flight mass spectrometric analysis of cellular glycerophospholipids enabled by multiplexed solvent dependent analyte-matrix interactions. Analytical Chemistry 80:7576-85
23. Johnson, A.S. et al., 2013. Integration of multiple components in polystyrene-based microfluidic devices part 1: fabrication and characterization. The Analyst 138:129-36
24. Cha, W., Tung, Y-C., Meyerhoff, M.E., Takayama, S., 2010. Patterned electrode-based amperometric gas sensor for direct nitric oxide detection within microfluidic devices. Analytical Chemistry 82:3300-5
25. Want, E.J., Cravatt, B.F., Siuzdak, G., 2005. The expanding role of mass spectrometry in metabolite profiling and characterization. ChemBioChem 6:1941-51.
26. Gao, D., Wei, H., Guo, G-S., Lin, J-M., 2010. Microfluidic cell culture and metabolism detection with electrospray ionization quadrupole time-of-flight mass spectrometer. Analytical Chemistry 82:5679-85
27. Baudoin, R. et al., 2007. Trends in the development of microfluidic cell biochips for in vitro hepatotoxicity. Toxicology In Vitro 21:535-544
28. Liu, B-F. et al., 2005. Microfluidic chip toward cellular ATP and ATP-conjugated metabolic analysis with bioluminescence detection. Analytical Chemistry 77:573-8
29. Mazutis, L. et al., 2013. Single-cell analysis and sorting using droplet-based microfluidics. Nature Protocols 8:870-91
30. Cheng, W., Klauke, N., Smith, G., Cooper, J.M., 2010. Microfluidic cell arrays for metabolic monitoring of stimulated cardiomyocytes. Electrophoresis 31:1405-13

Biographies

Laura A. Filla, a native of St. Louis, Missouri, completed both a bachelor’s and a master’s degree in chemistry at Saint Louis University. She then accepted a position at Monsanto developing flow-based systems for the high-throughput genotyping of corn and soy seeds. She is currently pursuing a doctorate degree in analytical chemistry under the direction of James L. Edwards at Saint Louis University, where she makes biomimetic microfluidic models for diabetes research. Laura’s graduate research experience encompasses microfluidics, electrochemistry, cell culture, and LC-MS for biological applications.  

James L. Edwards received BA and MS degrees in chemistry from Saint Louis University. He earned a PhD at the University of Michigan under the guidance of Robert T. Kennedy and later performed post-doctoral research at the Juvenile Diabetes Research Foundation Center for the Study of Complications in Diabetes under the direction of Eva Feldman.  He was previously an assistant professor at the University of Maryland and has been an assistant professor in the chemistry department at Saint Louis University since 2012. His research group develops metabolomic methods focused on advancing sample processing, capillary separations and mass spectrometry for diabetes research.