A flow-through holed PDMS membrane as a reusable microarray spotter for biomedical assays

Abstract

We propose the exploitation of a holed-designed poly(dimethyl)siloxane (PDMS) membrane as an innovative microarray spotter. The membrane is fabricated by a simple technological approach and can be reused several times. A good level of reproducibility is found upon spotting fluorescent proteins at different concentrations over large areas.

---

The possibility to dispense small amounts of liquid, down to the nanoliter scale, has allowed the discretization of functionalized surfaces to perform biological assays.¹ It is therefore possible to obtain a microarray of spots so as to reduce the number of substrates required for the assay implementation while increasing the reproducibility of measurements and result statistics.² These microarrays can be applied in several biological fields such as proteomics, genomics and lipidomics. In addition, the development of robust and effective technologies for microarray printing is strongly required in order to allow large-scale production.³

Generally, the most common technology for surface micro-spotting consists of a complex robotic system armed with needles or cusps that dispense droplets according to predefined layouts.⁴ The principle of operation can be either passive, in case the drops are loaded on the cusp tips due to the surface tension of the liquid, or active, when piezoelectric actuation or micropistons are integrated within the needles (as in inkjet printing systems).⁵ These approaches have gained popularity to become commercial products. Unfortunately, almost all of them are affected by some technological limitations. In passive devices, the quality of the spotting exhibits a strong dependence on the surface properties of liquids and substrates.⁶ In active systems, the clogging of the nozzles may occur.⁷ In the past, several alternatives have been developed in order to overcome the above-mentioned limitations.^8–13

Since the first report of Delamarche and coworkers,¹⁴ microfluidics has been largely investigated to locally guide protein solutions onto substrates.^15–21 Also in this case, microchannels can be passive (exploting self-filling by capillarity)^17,18 or active (when a pumping system was used).¹⁶ Several configurations have been proposed starting from parallel channels¹⁷ and chessboard-like patterns (by 90° re-alignment of parallel channels)^16,19 to three-dimensional microfluidic networks^20,21 in order to obtain a complex and discontinuous pattern. Among them, an interesting approach lies in the fabrication of continuous-flow microspotters. This method, first proposed by Sjölander and Urbaniczky²² and subsequently deeply investigated by Natarajan and co-workers,^23,24 allows sensibly improving the individual spot quality and increasing the sensitivity to low protein concentrations. Nevertheless, this approach also suffers from the complexity of the fabrication process (several lithographic, alignment and bonding steps are implemented) and from the control system required to manage the fluidic channels.

We propose an alternative spotting method based on the vacuum filling of vertical microchannels that are regularly distributed in a PDMS membrane to obtain a confined biological solution in contact with the target surface. Our idea introduces a remarkable novelty, since the production of the spots is performed exploiting a slight vacuum. PDMS is the peerless silicone most employed in microfluidics, lab-on-a-chip, micro total analysis systems, MEMS, and optical and electro-optical devices.^25–28 The elastomeric membrane can be fabricated by a simple technological approach and reused several times. The fabrication process and the working mechanism are schematically represented in Fig. 1.

Fig. 1 Schematic illustration showing the technological process and working mechanism: mould fabrication (a), casting and crosslinking of PDMS (b), and replica removal and attachment to the target substrate (c). After dispensing a drop of solution on the holed membrane and applying a low vacuum the membrane can be detached from the patterned substrate (d). In (e), a 3D scheme and relative photographs of the working principle are shown.

Commercial PMMA foil (5 mm thick) is used to fabricate the mould. The substrates are micro-machined with a numerically controlled milling cutter (head diameter 300 μm) in order to obtain a uniformly distributed array of pillars (Fig. 1a). The thickness of the dig is fixed to 200 μm, which is the minimum value that allows easy handling of the PDMS membrane. The choice of a pillar diameter equal to 100 μm is a good solution, allowing both dense packing of spots and enough mechanical strength for PDMS casting and replica removal without incurring damage. The pillar inter-distance is imposed by the cutter head diameter but all of the processes can be easily substituted by a single-step SU8 lithographic approach on silicon, allowing the increase in the density of the spots.¹¹

PDMS samples are prepared by mixing the polymer base and the curing agent with different weight ratios (5 : 1, 10 : 1). The mixture is poured into the moulds and thermally cross-linked (Fig. 1b). Further details about the fabrication process are collected in the ESI.† The 5 : 1 composition is chosen because its viscosity is low enough to allow the easy filling of the dig among the pillars. The tacking properties of the PDMS membrane are affected by the reduction of the mixing ratio between prepolymer and curing agent²⁵ since the elastic modulus increases.²⁶ However, the thickness of the sample also affects the elastic modulus, providing a balance of the compositional effect and allowing sufficient adhesion to the substrate.

The resulting PDMS samples are then manually removed from the mould and subsequently attached to the desired target surface (Fig. 1c), thus causing the formation of reversible sealing. Different substrates are tested, such as a polished silicon wafer, a glass slice, metallic thin films, polymeric substrates, so as to assess the wide exploitability of the proposed spotting system.

A 3D scheme of the micropore filling mechanism is reported in Fig. 1e. In order to produce the microspots a microliter drop of solution is dispensed onto the holed surface and a slight vacuum is applied (for example, placing the sample into a vacuum oven in order to control the temperature and pressure – see Fig. S1 in the ESI†). As the pressure decreases, the air trapped in the vertical microchannels begins to leak in bubbles within the drop of solution released on the surface. When the pressure falls below about 50 mbar, the pressure difference is sufficiently high to completely extract the air trapped inside the microchannels. At this point the surface tension of the liquid is no longer sufficient to retain the trapped air and the bubble explodes. Upon this operation, microchannels are entirely filled with the solution. The optical microscope image (top) and digital photographs (side view) collected in Fig. 1e clearly illustrate the air bubble evolution with decreasing pressure.

In the ESI† Fig. S2, a photograph of the PDMS membrane is presented, showing the possibility of a large area patterning with microarray spots.

Fig. 2a)–e) collect exemplary optical images of methylene blue (MB) aqueous solution spotted on different surfaces (silicon (a), Pt thin film on silicon (b), PMMA (c), PET (d), PDMS (e)). The spots obtained show a well-defined morphology and no evidence of lateral spreading due to the PDMS vertical channel confinement. All of these tests are done using the same PDMS membrane ultra-sonicated in acetone and ethanol after every spotting. In order to evaluate the dependence on the solvent nature, a commercial solution of iodine in acetonitrile (Iodolyte AN50, Solaronix) is used in place of water to fabricate the same pattern onto the glass surface (Fig. 2f).

Fig. 2 Optical microscope images of MB aqueous solution spotted on silicon (a), Pt thin film on silicon (b), PMMA (c), PET (d), and PDMS (e), while in (f) an acetonitrile dye solution is used on the glass surface (the scale arrows correspond to 200 μm). In (g) a digital photograph of a microscope glass slice fully patterned with microspots is shown.

Different membrane areas are fabricated in order to evidence the possibility to scale up the spottable area. In Fig. 2g) an image of a 4 cm² glass slice fully covered by microspots is presented. A zoomed microscope image of the same surface is shown in the inset.

No undesired phenomena, such as comet tailing or coffee ring (typically affecting the drop dispensing method), are evidenced due to the confinement of the solution within the vertical microchannel and to the good adhesion of the PDMS membranes to the substrates investigated.

The proposed PDMS membrane allows the eventual alignment to any substrate features by exploiting the transparency of PDMS. In this case, a common microscope or a mask aligner is required. Fig. 3 shows an example of PDMS membrane alignment on circular nanogratings recently proposed to improve the fluorescence extraction on one-dimensional photonic crystals (1D PCs).^29,30

Fig. 3 Optical microscope images collecting examples of PDMS membrane alignment on nanoscale features on 1D PCs using different layouts.

In order to check the applicability of this technical innovation for biological assays, a proof-of-principle biological binding experiment is performed on functionalized glass substrates. Details of the functionalization and activation of the surfaces are reported in the ESI† and in a previous study.³¹

These pre-activated samples are incubated with 10 μl of Protein A–Alexa 546 solution, spotted using the PDMS membrane, for 20 minutes at room temperature (only 2 minutes are required to obtain the microchannel filling by vacuum). Working standard solutions at six different concentrations (50, 10, 5, 1, 0.5, and 0.1 μg mL⁻¹) in phosphate-buffered saline (Dulbecco's PBS) are prepared by serial dilution. After the incubation, during which proteins diffuse and bind to the target surface, substrates are deeply washed with PBS–Tween 20 (0.05%), rinsed with deionised water three times and dried in a nitrogen stream.

The samples are characterized with a fluorescence microscope using a customized setup (see the ESI† Fig. S3) described in detail in the ESI.† Briefly, fluorescence is excited by means of a collimated Nd:YAG laser beam impinging on an area larger than the spot size and collected with a CMOS RGB camera through NA = 0.07 optics.

The resulting images (Fig. 4a) are analysed by dedicated software in order to quantify the intensity of the signals. Sets of different spots are collected for each concentration. For each image, the red and green channels are merged in order to have a single intensity matrix and two areas of interest are identified, the first one included in the spot and the second one outside the spot to evaluate the background level. The intensity is evaluated over these two areas by extracting the average values and the standard deviations. The limit of detection has been set to three times the standard deviation of the background noise (i.e. equal to 0.0581). All of the values are normalized to the maximum intensity level. To a first approximation the fluorescence intensity depends on the percentage of area covered by the protein. A simple and common way to model such a situation is the Langmuir adsorption model.³²

Fig. 4 RGB images of protein spots at different concentrations are shown (a). Calibration curve of fluorescence intensity versus protein concentration (b). The diamonds with relative error bars are the experimental data, while the blue continuous line is the fitting curve. The red dashed line represents the limit of detection. The inset shows a 2D intensity map of a protein spot. In (c) the FTIR spectra of the PDMS membrane and a zoom of the range of interest are shown.

We fit the experimental data with a Langmuir equation having the following expression:

where c is the protein concentration, I is the fluorescence intensity and α is a fit parameter (whose value is 0.23). The resulting curve, together with experimental data and corresponding error bars, is reported in Fig. 4b. It can be seen from this figure that the intensity of the fluorescence images increases with increasing protein concentration from 0.1 μg mL⁻¹ up to 50 μg mL⁻¹. The resulting calibration curve is linear below 5 μg mL⁻¹. When the protein concentration increases to a certain level, the fluorescence intensity starts to level off, indicating that the binding affinity reaches saturation.

Compared to other protein microarray experiments our results reveal that the lowest concentration detectable by fluorescence analysis is lower than the minimum one (0.05 mg ml⁻¹) observed by Ho et al. using a micro-contact printing approach.³³ Indeed, a protein detection ability down to a concentration of 0.1 μg ml⁻¹ is demonstrated and this result is in accordance with the minimum protein concentration (0.12 μg ml⁻¹) detected by Natarajan et al. using a continuous-flow microspotter.²³

With the purpose of assessing the uniformity of the spotted proteins, fluorescence images of the spots have been analysed by dedicated software in order to obtain the 2D profile of the fluorescence intensity spatial distribution. An exemplary image (inset of Fig. 4b) shows a uniform intensity inside the spot area with a rapid fluorescence drop at the boundary of the spot.

In order to check the reusability of the PDMS membrane, the sample cleaning method (described in the ESI†) can be addressed by means of FTIR spectroscopy. Results are presented in Fig. 4c. The spectrum of a freshly prepared PDMS displays its characteristic absorption peaks assigned elsewhere.^25,26 New absorption bands appear in the spectra of PDMS after Protein A–Alexa 546 incubation, indicated by arrows, in particular, the OH stretching broad band at 3600–3150 cm⁻¹ characteristic of carboxylic groups, C=O stretching mode at 1713 cm⁻¹ and the typical amide I at 1645 cm⁻¹, amide II at 1563 cm⁻¹ and amide III at 1348 cm⁻¹.³⁴ After the cleaning procedure the protein characteristic bands have completely disappeared, thus demonstrating the possibility to reuse the PDMS membrane for multiple processes.

In addition, the proposed approach could also be exploited for multiple protein microarray printing and cascade reactions by simply coupling the membrane or the spotted surface with an ad hoc designed microchannel array as depicted in Fig. S4 and S5 in the ESI.†

Conclusions

Herein we have proposed for the first time the exploitation of a holed-designed PDMS membrane as an innovative passive microarray spotter. The elastomeric membrane is fabricated taking advantage of a very simple technological approach, and only a slight vacuum is required to activate the procedure. Moreover, the PDMS can be easily cleaned and reused several times. To demonstrate the effectiveness of this method, aqueous and organic solvent-based dye solutions are spotted onto several substrates and the alignment ability on the surface feature is demonstrated. The proposed approach is practically almost independent of the substrate of immobilization since PDMS presents a good self-adhesion to almost every kind of substrate usually employed in biological assays. In order to verify the employability of this technical innovation for biological assays, fluorescent proteins at different concentrations are patterned onto the functionalized substrate, and the calibration tests by fluorescence measurements confirm the high sensitivity (detection ability down to concentration of 0.1 μg ml⁻¹) and the good degree of reproducibility of this microarray spotting approach.

Acknowledgements

The authors would like to thank Dr. Alessandro Chiadò for the help in surface functionalization and Dr. Emiliano Descrovi for fruitful discussions.

Notes and references

1. G. MacBeath and S. L. Schreiber, Science, 2000, 289, 1760.
2. M. F. Templin, D. Stoll, M. Schrenk, P. C. Traub, C. F. Vöhringer and T. O. Joos, Drug Discovery Today, 2002, 7, 815.
3. T. Kodadek, Chem. Biol., 2001, 8, 105.
4. T. Martinsky, in A Beginner's Guide to Microarrays, Springer, US, 2003, pp. 93–122.
5. D. Rose, Microarray Biochip Technol., 2000, 19, 1.
6. M. K. McQuain, K. Seale, J. Peek, S. Levy and F. R. Haselton, Anal. Biochem., 2003, 320, 281.
7. T. Martinsky, PharmaGenomics, 2004, 2, 42.
8. X. Zhou, L. Lau, W. W. L. Lam, S. W. N. Au and B. Zheng, Anal. Chem., 2007, 79, 4924.
9. D. A. Chang-Yen, D. G. Myszka and B. K. Gale, J. Microelectromech. Syst., 2006, 15, 1145.
10. K. Campbell, T. McGrath, S. Sjölander, T. Hanson, M. Tidare, Ö. Jansson, A. Moberg, M. Mooney, C. Elliott and L. Buijs, Biosens. Bioelectron., 2011, 26, 3029.
11. J. Wang, Y. Zhou, H. Qiu, H. Huang, C. Sun, J. Xi and Y. Huang, Lab Chip, 2009, 9, 1831.
12. M. A. Witek, M. L. Hupert, D. S.-W. Park, K. Fears, M. C. Murphy and S. A. Soper, Anal. Chem., 2008, 80, 3483.
13. V. Romanov, S. N. Davidoff, A. R. Miles, D. W. Grainger, B. K. Gale and B. D. Brooks, Analyst, 2014, 139, 1303.
14. E. Delamarche, A. Bernard, H. Schmid, B. Michel and H. Biebuyck, Science, 1997, 276, 779.
15. E. Delamarche, A. Bernard, H. Schmid, A. Bietsch, B. Michel and H. Biebuyck, J. Am. Chem. Soc., 1998, 120, 500.
16. C. A. Rowe, L. M. Tender, M. J. Feldstein, J. P. Golden, S. B. Scruggs, B. D. MacCraith, J. J. Cras and F. S. Ligler, Anal. Chem., 1999, 71, 3846.
17. Y. D. Kim, C. B. Park and D. S. Clark, Biotechnol. Bioeng., 2001, 73, 331.
18. A. Papra, A. Bernard, D. Juncker, N. B. Larsen, B. Michel and E. Delamarche, Langmuir, 2001, 17, 4090.
19. X. He, D. S. Dandy and C. S. Henry, Sens. Actuators, B, 2008, 129, 811.
20. D. T. Chiu, N. L. Jeon, S. Huang, R. S. Kane, C. J. Wargo, I. S. Choi, D. E. Ingber and G. M. Whitesides, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 2408.
21. Z. H. Wang, Y. H. Meng, P. Q. Ying, C. Qi and G. Jin, Electrophoresis, 2006, 27, 4078.
22. S. Sjölander and C. Urbaniczky, Anal. Chem., 1991, 63, 2338.
23. S. Natarajan, P. S. Katsamba, A. Miles, J. Eckman, G. A. Papalia, R. L. Rich, B. K. Gale and D. G. Myszka, Anal. Biochem., 2008, 373, 141.
24. S. Natarajan, A. Hatch, D. G. Myszka and B. K. Gale, Anal. Chem., 2008, 80, 8561.
25. A. Lamberti, M. Quaglio, A. Sacco, M. Cocuzza and C. F. Pirri, Appl. Surf. Sci., 2012, 258, 9427.
26. A. Lamberti, M. Di Donato, A. Chiappone, F. Giorgis and G. Canavese, Smart Mater. Struct., 2014, 23, 105001.
27. L. Pasquardini, C. Potrich, M. Quaglio, A. Lamberti, S. Guastella, L. Lunelli, M. Cocuzza, L. Vanzetti, C. F. Pirri and C. Pederzolli, Lab Chip, 2011, 11, 4029.
28. E. Descrovi, F. Frascella, M. Ballarini, V. Moi, A. Lamberti, F. Michelotti, F. Giorgis and C. F. Pirri, Appl. Phys. Lett., 2012, 101, 131105.
29. A. Angelini, E. Barakat, P. Munzert, L. Boarino, N. De Leo, E. Enrico, F. Giorgis, H. P. Herzig, C. F. Pirri and E. Descrovi, Sci. Rep., 2014, 4.
30. A. Angelini, E. Enrico, N. De Leo, P. Munzert, L. Boarino, F. Michelotti, F. Gioprgis and E. Descrovi, New J. Phys., 2013, 15, 073002.
31. C. Ricciardi, R. Castagna, I. Ferrante, F. Frascella, S. L. Marasso, A. Ricci, G. Canavese, A. Lorè, A. Prelle, M. L. Gullino and D. Spadaro, Biosens. Bioelectron., 2013, 40, 233.
32. R. I. Masel, Principles of adsorption and reaction on solid surfaces, John Wiley & Sons, 1996, vol. 3.
33. C. E. Ho, C. C. Chieng, M. H. Chen and F. G. Tseng, J. Microelectromech. Syst., 2008, 17, 309.
34. P. Rivolo, S. M. Severino, S. Ricciardi, F. Frascella and F. Geobaldo, J. Colloid Interface Sci., 2014, 416, 73.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4lc01027b

---