Fabrication of a microfluidic platform for investigating dynamic biochemical processes in living samples by FTIR microspectroscopy

Microelectronic Engineering 87 (2010) 806–809

Contents lists available at ScienceDirect

Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee

Fabrication of a microfluidic platform for investigating dynamic biochemical processes in living samples by FTIR microspectroscopy Giovanni Birarda a,b,c,*, Gianluca Grenci b, Luca Businaro b, Benedetta Marmiroli d, Sabrina Pacor e, Lisa Vaccari a a

Elettra Synchrotron Light Laboratory, S.S. 14 Km 163.5, 34149 Basovizza, Trieste, Italy TASC National Laboratory – CNR, S.S. 14 Km 163.5, 34149 Trieste, Italy Physics Department, Trieste University, Piazzale Europa 1, 34127 Trieste, Italy d Institute for Biophysics and Nanosystem Research, Austrian Academy of Science, Schmiedlstrasse 6, Graz, Austria e Life Science Department, Trieste University, Via Licio Giorgieri 7-9, 34127 Trieste, Italy b c

a r t i c l e

i n f o

Article history: Received 21 September 2009 Accepted 16 November 2009 Available online 22 December 2009 Keywords: FTIR microspectroscopy Living-cells Microfabrication Microfluidic

a b s t r a c t Here we present the optimization of fabrication steps for realizing an infrared–visible microfluidic chip to study single-living cell behaviour in physiological environment by synchrotron radiation FTIR microspectroscopy. We optimized subtractive and additive lithographic processes on CaF2 substrate, employing XARP 3100/10 photoresist both as etching-mask and for the device fabrication. Using prototype microfabricated liquid cells 9 and 5 lm thick, we measured the response of small groups of THP1 monocytic cells to mechanical compression and chemical stimulation with fMLP using conventional IR globar source, aiming to evaluate biochemical rearrangements of leukocytes during the capillary circulation or recruitment processes. Stimulated monocytes have spectral features recognizable, differentiating them from unstimulated, especially affecting the spectral region 1280–1000 cm 1, characteristic of nucleic acids and carbohydrates, and specific band ratios, such as proteins on lipids and methylene on methyl. Spectra variations have been correlated with biochemical events such as transcription, synthesis of new-proteins and variations in membrane fluidity. Ó 2009 Elsevier B.V. All rights reserved.

1. Introduction The establishment of reliable living single-cell analysis techniques is a crucial point for understanding the molecular bases of cell behaviour in response to external stimuli, either chemical or mechanical [1]. Microfluidic devices offer a unique platform for this purpose since they allow the integration of a variety of operation such as single-cell selection, positioning or lysis as well as separation and detection of cellular analytes [2]. Moreover, microfluidic devices are able to confine cells in compartments near their intrinsic volume, thus minimizing dilution effects and increasing detection sensitivity. From a detection standpoint, microchips have to be fabricated in materials transparent with respect to the probing technique. Visible and fluorescence microscopy were among the first investigation methods to be integrated in microfluidic platforms and they still are the most employed, taking advantage from the standardized fabrication protocols of suitable substrates such as glass, quartz or plastics.

* Corresponding author. Address: Elettra Synchrotron Light Laboratory, S.S. 14 Km 163.5, 34149 Basovizza, Trieste, Italy. Tel.: +39 040 3758759. E-mail address: [email protected] (G. Birarda). 0167-9317/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.mee.2009.11.081

We aim to expand cellular microanalysis in physiological environment to FTIR microspectroscopy (l-FTIR), a label free, not damaging, powerful and versatile tool for biological sample investigation. This technique has not the selectivity of a single molecule analytical tool; nevertheless, by the analysis of specific absorption bands [3], it offers a fast snapshot of cellular response to specific stimuli in terms of variation of protein structure, membrane composition and order, nucleic acid conformation as well as differences in relative concentrations of major cellular macromolecules [4,5]. Despite its great potentialities, fabrication protocols for both infrared and visible (IR–VIS) transparent materials are not well-established and technical problems related to the strong water absorption in the Mid-IR regime has be taken in consideration. Herein we present the optimization of fabrication steps for producing an IR–VIS transparent microfluidic chip, suitable for exploring biochemical changes in living-cells upon different stimulations. In particular we are interested in studying the cellular behaviour of individual leukocytes, in order to understand their internal rearrangements induced both by capillary circulation and/or chemoactivated extravasation through narrow endothelium interstices. Preliminary results on mechanical compression and chemical stimulation with fMLP (N-formyl-Met-Leu-Phe) of THP1 leukocytes in a

G. Birarda et al. / Microelectronic Engineering 87 (2010) 806–809

prototype microfabricated liquid cell are reported and future developments of the research activity outlined. 2. Experimental 2.1. Fabrication of IR–VIS transparent devices In order to investigate by l-FTIR leukocytes’ response to mechanical deformation during capillary circulation or induced by chemotaxis, we designed the device shown in Fig. 1a, where the narrower channels mimic capillaries and epithelium intercellular interstitials. The device was realized on calcium fluoride by photolithography with X-ARP 3100/10 (AllResist GMBH) and wet etching. First we carved larger measurements chambers in CaF2 by wet etching, and then we defined the upper level of the device by patterning a second layer of photoresist. 2.1.1. X-ARP 3100/10 characterization X-ARP 3100/10 is an experimental positive photoresist developed by AllResist GmbH. The resist was spun at thicknesses from 4 to 9 lm varying the spin rate from 6000 to 1000 rpm. Prebaking conditions were optimized in order to achieve the better aspect ratio of the structures and set at 105 °C for 2 min. Exposure doses were changed accordingly with resist thickness from 180 to 300 mJ/cm2 (Karl-Suss MJB3). The development step was done in AR 200-26 developer for 90 s and the final rinse in DI water gave the release of the lithographed pattern (Fig. 2a). 2.1.2. Wet etching of calcium fluoride For the wet etching process of CaF2, a saturated solution of NH4Fe(SO4)2 was employed, using a X-ARP 3100/10 pattern as mask. Due to the low chemical reactivity and the poor solubility of CaF2 (0.0017 mg/100 mg in water at 20 °C, insoluble in most bases and

807

acids, soluble in ammonia salts [6]), the etching at room temperature was very slow (100 nm/h) and of unsatisfactory quality. By increasing the temperature to 30 °C, enhancing both the reaction rate and the solubility of the products, we reached an etching rate of 500 nm/h and a better pattern transfer (fig. 2b). The whole etching process did not alter at all the IR transmittance of the substrate, even if no protective layer was added on the backside (Fig. 2c). 2.1.3. Fabrication of the prototype liquid cell The prototype liquid cell we used for preliminary l-FTIR measurements with conventional source, shown in Fig. 1b, was realized by photolithography of X-ARP 3100/10 spun on calcium fluoride. The device consists of two large wells for the confinement of the cells in liquid environment and two smaller holes for air background acquisition. Devices 5 and 9 lm thick have been realized and tested.

3. Measurements 3.1. Cell model The human cell line THP1 (American Type Culture Collection, Rockville, Md.) [7], in-vitro established and displaying many monocytic characteristics, was employed. Cells were cultured in RPMI medium (RPMI 1640: 2 mM L-glutamine, 1.5 g/L sodium bicarbonate, 4.5 g/L glucose, 10 mM HEPES, 1.0 mM sodium pyruvate, 0.05 mM 2-mercaptoethanol, and 10% FBS), 100U/mL penicillin, 100 lg/mL streptomycin, in incubator 37 °C with 5% of CO2. RPMI 1640 was purchased from Biowhittaker. Synthetic fMLP (N-formyl-Met-Leu-Phe) was purchased from Sigma–Aldrich. FBS (Fetal Bovine Serum) was purchased from Hyclone. Cell vitality in PBS buffer and NaCl 0.9% physiological solution, both suitable for FTIR studies, were determined by cytofluorimetric assays, and NaCl was chosen since it has no signals in the phosphate region of the IR spectrum. 3.2. Spectra acquisition

Fig. 1. The fabricated CaF2 devices. (a) The final fluidic device. The large features, like inlet channels and reservoirs are 10 lm thick while small channels are 5 lm thick, 5–20 lm wide. In the insets: SEM images of the section variable channels realized by optical lithography and of the etched features. (b) Prototype liquid cell employed for the measurements with conventional IR source.

Experiments were carried out at the infrared beamline SISSI (Synchrotron Infrared Source for Spectroscopic and Imaging) at the Elettra Synchrotron Laboratory, Trieste, Italy [8] according to the following procedure. After being removed from the incubator, the cells were counted in a Bürker chamber and a vitality test with trypan blue was performed. The growth medium was removed by centrifugation (Eppendorf Microcentrifuge) at 1600 rpm for 5 min, and then substituted with NaCl 0.9% physiological solution; the procedure was repeated twice, ensuring the complete buffer exchange. The cells were then re-suspended in NaCl 0.9% physiological solution supplemented with glucose 5 lM. For the experiments, 1 lL of cell’s suspension was dropped in the device and the chemical stimulation was induced by adding 1 lL of fMLP 1 lM. FTIR transmission spectra were acquired using a Bruker Hyperion 3000 Vis–IR microscope mounting a mid-band HgCdTe detector, coupled with Bruker Vertex 70 interferometer. Both interferometer and microscope, sealed with an in-house designed box, were purged with nitrogen in order to reduce spectral contributions for environmental water vapour and carbon dioxide. Repeated spectra on THP1 cell groups were collected with globar source setting knife-edge apertures at 40  40 lm, using 15X Schwarzschild condenser and objective, co-adding 512 scans with a spectral resolution of 4 cm 1, starting to collect data 30 min after the cells were dropped in the device. The delay time is imposed by the liquid cell assembling, the sample screening and the IR

808

G. Birarda et al. / Microelectronic Engineering 87 (2010) 806–809

focusing. The Buffer spectrum, acquired near the measured cell, and the air background, were collected using the same parameters. 3.3. Data analysis Raw spectra were corrected for water vapour and CO2 absorption; buffer subtraction was performed using OPUS 6.5 (OPUS NT 6.5, Bruker Optics GmbH, Ettlingen, Germany), setting the buffer on cell ratio in the range between 0.97–1:1. The scaling factor was chosen in order to obtain the best fit of the combination band of H2O centred at 2100 cm 1 [9]. 4. Results and discussion The deformability of leukocytes has long been studied with biological and biochemical approaches. However, several aspects on the mechanobiology of these cells remain to be investigated as well as other cellular events involving transformations and sustenance of mechanical forces. It is know that circulating leukocytes experience many cycles of compression-release during their whole life, still retaining their full functionality. Moreover, their ability to shrink for passing through narrow capillaries and interstices between endothelial cells is the fulcrum of human immune response. In order to monitor the biochemical signatures associated to the deformation of a single white blood cell, the fabrication of IR-VIS transparent devices is needed for cell visualization, tracking and analysis. Calcium fluoride is one of the few materials that satisfy both conditions, having a water solubility low enough for performing measurements in liquid (Kps = 3.4  10 11). Among the other IR suitable materials, ZnSe is even less soluble than CaF2 (Kps = 3.6  10 26) but less transparent to visible light and toxic for cellular samples (EU classification: T) while BaF2 (Kps = 1.7  10 6) is too soluble to perform experiments in physiological environment. Also silicon would be a good choice, because of the

Fig. 2. (a) Optimized lithographic results on X-ARP 3100/10. (b) Result of the etching of CaF2: after 16 h, 8 lm thick structure was obtained. (c) Transmittance spectra of the same CaF2 window before (left panel) and after (right panel) completion the fabrication process, showing no detectable differences in transparency.

long-standing experience in its micro fabrication, but it is not transparent to visible light [10]. Unlike to silicon, CaF2 is difficult to handle, due to its high sensitivity to thermal shocks [11] and low reactivity towards common reagents for wet processes. However, we demonstrated that it can be patterned preserving its optical proprieties unaltered (see Fig. 2c) by a fine tuning of fabrication protocols. To reach our goal, we used X-ARP 3100/10 (AllResist GMBH). This resist showed the better adhesion properties on the substrate in comparison with SU-8 and PMMA, without the need of any adhesion promoter, which could have undesirable IR features interfering with the signal of the cells. X-ARP 3100/10 photoresist exhibited also appreciable resistance to wet etching of CaF2, making it suitable for each fabrication step. l-FTIR is an absorption spectroscopy with diffraction limited spatial resolution. For the Mid-IR region extending from wavenumbers 4000 to 800 cm 1, the one of major interest for biological studies, the spatial resolution is of some microns (2.5–12 lm), sufficient for single-cell analysis and even investigation of sub-cellular events. However, this theoretical limit is achievable maintaining an appreciable signal to noise ratio only by exploiting the brightness advantage of IR synchrotron radiation sources. The optimization of the single-cell measurement set up suitable for SR experiments at SISSI beamline at Elettra is still ongoing. However, in order to test the experiment feasibility with conventional IR sources, we designed and realized a simplified liquid cell for studying small groups of THP1 monocytes. THP1 cells, having an average size of 8–10 lm, have been measured in slightly deformed (9 lm thick devices) and deformed conditions (5 lm thick), activated or not by fMLP chemoattractant (see Fig. 1b).

Fig. 3. (a) Infrared subtracted spectrum of a group of cells collected with globar source in a 5 lm thick microfabricated liquid cell, highlighting the main cellular components’ bands: amide, I 1700–1600 cm 1; amide, II 1590–1480 cm 1; lipids, 3000–2800 cm 1: methylene groups (CH2), 2946–2884 cm 1, methyl (CH3) 2995– 2946 cm 1; phospholipids, 1700–1740 cm 1; carbohydrates, 1480–900 cm 1; phosphate asymmetric stretching, 1275–1196 cm 1; phosphate symmetric stretching, 1130–1000 cm 1. (b) Comparison between THP1 mean spectra in phosphate region in deformed (red) and un-deformed (orange) states. Spectrum line thickness is proportional to standard deviation for the analyzed groups. (c) THP1 phosphate region in deformed (blue) and un-deformed (cyan) conditions after fMLP stimulation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

G. Birarda et al. / Microelectronic Engineering 87 (2010) 806–809

Several repeated spectra of the same cell group (3–4 cells) were acquired to observe the spectral variations associated to purely mechanical stimulation in the 5 lm liquid cell and compared with the control (9 lm liquid cell). The deformation affects the behaviour of the cells mainly leading them to an activated state in which DNA transcription and protein synthesis start (see Fig. 3b). As a matter of fact, by analyzing the average mean spectra of the cells, the major variations could be observed in the spectral region attributed to nucleic acids (see Fig. 3). The phosphate asymmetric stretching band shifted to lower wavenumbers, from 1236 to 1234 cm 1, and the asymmetric stretching from 1084 to 1087 cm 1. The appearance of a shoulder at 1120 cm 1, attributed to RNA C–O symmetric stretching [12], demonstrated an increased content in RNA molecules caused by deformation, in accordance to the activation of transcriptional phenomena. This hypothesis was also supported by the appearance of a shoulder at 1122 cm 1, a contribution tentatively assigned to Z-DNA form that is strongly enhanced when transcription processes are ongoing. fMLP was used for simulating the first response to inflammation and infection. As a matter of fact, fMLP mimics the activity of bacterially derived peptides with formylated N-terminal methionine groups, which are strong chemoattractants and induce, among other effects, adherence, degranulation and production of tissuedestructive oxygen-derived free radicals in phagocytic cells. fMLP binds to G-protein coupled receptor and would lead to two different pathways of cellular response [13]: the activation of nuclear factor-jB (NF-jB), resulting in the transcription of genes encoding cytokines and growth factors, as well as the modification of membrane fluidity and cytoskeleton, allowing extravasation from the blood stream to reach the inflammation site. fMLP effects after 30 min (see section 3.2) were detected in the region of nucleic acids, according to the activation of NF-jB, and they were comparable for un-deformed and deformed states, suggesting that the chemical stimulation was the one dominating the cell behaviour (Fig. 3c). The appearance of a shoulder at 1220 cm 1, attributed to ribose ring vibration of RNA, was seen as in purely deformed conditions. Moreover, the shoulder at 1209–1211 cm 1 (assigned to PO2 stretching of DNA in Z-form) and the increasing of bands’ contributions at 1060 and 1057 cm 1 (tentatively assigned to furanose C–O backbone stretching) [14], are all signals related to RNA transcription and Z-conformation of DNA, commonly believed to provide torsional strain relief while DNA transcription occurs. Moreover IR signals at 1066–1068 cm 1 could be assigned to phosphorylation of proteins, one of the main pathways for signal transduction in cells. Differences in membrane composition were also detected for chemically stimulated cells: a variation in –CH2/–CH3 ratio could be appreciated indicating a change in membrane fluidity (data not shown) [15]. The effect of the chemoattractant on the membrane composition could also be

809

seen in the protein versus lipids ratio, whose decreasing indicated an increase in lipid content, in line with modifications that would lead the cell to shrink up in narrow intercellular interstices (1.67 ± 0.05 vs. 1.43 ± 0.04 at 9 lm and 1.81 ± 0.02 vs. 1.61 ± 0.02 at 5 lm). 5. Conclusions The preliminary results here presented do not claim to be conclusive but for sure they prove that the synergic match between lfabrication and l-FTIR could extend the frontiers of FTIR microspectroscopy to fields of science previously unexplored with this technique, such as mechanobiology of leukocytes and living-cell measurements in general. Although being a prototype, our microfabricated liquid cell represents an improvement in comparison to commercial IR liquid measurement chambers, guaranteeing good pathlength reproducibility and design versatility. From the spectroscopic point of view, the acquired data are among the first that allow the evaluation of the biochemical response of livingcells to different stimulations [4]. This result, even if quite remarkable itself, has to be seen as a first step toward the wider application of microfabrication techniques to IR transparent materials. By jointly exploiting the capabilities of the microfluidic device we presented and the synchrotron radiation brightness, we will have the opportunity to follow in real-time a broad range of biochemical events at single-cell level, shining a light on several biological problems still poorly understood. References [1] O.P. Hamill, B. Martinac, Physiol. Rev. 81 (2001) 685. [2] J. El-Ali, P.K. Sorger, K.F. Jensen, Nature 442 (2006) 402. [3] H.H. Mantsch, D. Chapman (Eds.), Infrared Spectroscopy of Biomolecules, Wiley-Liss, Inc., New York, 1996, p. 279. [4] P. Heraud, B.R. Wood, M. Tobin, J. Beardall, D. McNaughton, FEMS Microbiol. Lett. 249 (2005) 219. [5] W.K. Surewicz, H.H. Mantsch, D. Chapman, Biochemistry 32 (2) (1993) 389. [6] International Crystals laboratories, Calcium Fluoride Optical Crystals, . [7] S. Tsuchiya, M. Yamabe, Y. Yamaguchi, Y. Kobayashi, T. Konno, K. Tada, Int. J. Cancer 26 (1980) 171. [8] S. Lupi, A. Nucara, A. Perucchi, P. Calvani, M. Ortolani, L. Quaroni, M. Kiskinova, J. Opt. Soc. Am. B 24 (2007) 959. [9] K. Rahmelow, W.H. Bner, Appl. Spectrosc. 51 (2) (1997) 160. [10] International Crystals laboratories, Silicon (Si) Optical Crystals, . [11] D.C. Harris, Infrared Phys. Technol. 39 (1998) 185. [12] M. Banyay, J. Sandbrink, R. Strömberg, A. Gräslund, Biochem. Biophys. Res. Commun. 324 (2004) 634–639. [13] M. Banyay, M. Sarkar, A. Gräslund, Biophys. Chem. 104 (2003) 477–488. [14] L.-Y. Chen, A. Ptasznik, Z.K. Pan, Biochem. Biophys. Res. Commun. 319 (2004) 629. [15] I. Dreissig, Spectrochim. Acta Part 71 (2009) 2069.